2013美赛ICM优秀论文

For office use only

T1________________ T2________________ T3________________ T4________________Team Control Number

18492

Problem Chosen

C

For office use only

F1________________

F2________________

F3________________

F4________________

Two-tier Communication Network Model of Global Health

Summary

Modeling and predicting the Earth's health condition is an intricate problem, which should embrace the complexity of Earth's interrelated systems, and take into the consideration of global impacts on local conditions and vice versa. With the assumption that the population and human civilization act as the dominating driver of environment degradation, we build a novel people-oriented two-tier communication network model (TCNM) to consider the environment degradation.

Firstly, we define the elements of the global network. The nodes are the abstraction of geographic division, whose area share analogous trait regardless of the nuance of traits within the area. Besides, the links are the geographic adjacent relation between two nodes. Here we introduce the concepts of signal channels and signals derived from communications network to represent the links and impacts transmitted between nodes respectively, so as to make it possible to consider all the impacts as a whole in the network.

Secondly, we study the node tier of TCNM. Several driving factors that could result in environment degradation are identified inspired by Forrester's World Model, and also quantified so as to enable further analysis. Then we use Bayesian Belief Network to study the impact generated by the interaction of these factors with respect to the inherent trait of nodal area. Also, a novel analogy with nodal character and communication protocol is introduced to study how nodes' condition evolves. The impacts imposed on the nodes are compared to the protocol events in communications field, while the impact exported by the node is compared to protocol actions executed when status transferring, which simplify the study of node state evolution when every iteration occurs. Moreover, we develop a nodal health measurement, which measures the local health of the network. The measurement standard takes into account three environmental factors that can quantify the health condition and then weights them using an analytical hierarchy process (AHP) approach. The nodal tipping point is also mentioned and quantified, which is 0.301 for our model.

Thirdly, the global tier of TCNM is studied. We distinguish the influence of a node’s inactive/active state on the signal transmission. Besides, the diffusion of atmosphere pollution and population movement over the network are studied and modeled respectively. This process is critical because it makes it possible for us to study the global impact on local conditions and vice versa. The global health measurement is studied with lattice network structure. We use the percolation ratio of the active nodes as the global health measurement. By using network theory, we come up with a global tipping point, that is, when percolation ratio of active nodes is under 0.682, then the tipping point has been reached.

Finally, we discuss the requirements of necessary data in order to validate or apply our model to practical use. We also study the methodology of parameter estimating. Furthermore, to better explain our model, we also discuss in detail how certain kinds of policies impact the model in a general scenario and vice versa. Sensitivity is also discussed in this section. Moreover, the critical nodes and edges are studied using network structure.

In the end, we conclude the strengths and weakness of TCNM. The model is quite robust because of the network structure of our model and would probably work well with enough necessary data, according to the intrinsic trait of machine learning approach. However, the correctness of our model remains to be verified lacking of necessary data. Also, the introduction of AHP approach brings in some subjective factors of our model.

Contents

1. Introduction (2)

1.1 Background (2)

1.2 Our Work (2)

2. Assumption (3)

3. Two-tier Communication Network Model (3)

3.1 Node (5)

3.1.1 Node definition (5)

3.1.2 Driving factors of environment degradation (5)

3.1.3 Node characterizing with communication protocol (6)

3.1.4 A Revised Bayesian Belief Network study of factors (7)

3.1.5 Nodal Health Measurement (9)

3.2 Link (10)

3.3 Topology (10)

3.3.1 Influence of active/inactive node condition (11)

3.3.2 Migration Model (11)

3.3.3 Air Pollution Dispersion Model (12)

3.3.4 Policy and emergency impact (12)

3.3.5 Node status evolve in every iteration (12)

3.4 Global Health Measurement (13)

4. Discussion and Simulation Run (14)

4.1 Dataset Requirements and Collect Suggestion (14)

4.2 Parameters Estimate (15)

4.3 Impact of policy and decision support (15)

4.4 Sensitivity and Uncertainty Analysis (17)

4.5 Critical nodes and links (17)

5. Conclusion (18)

5.1 Strength (19)

5.2 Weakness (19)

Reference (20)

1. Introduction

1.1 Background

Human beings are blamed for global environment degradation. Many biological forecasting now depends on projecting recent trends into the future assuming various environmental pressures, or on using species distribution models to predict how climatic changes may alter presently observed geographic ranges (Anthony D. B., et, al., 2012).

Study have shown that threshold effect can cause 'Critical Transitions' (Scheffer, M.et al., 2009) and thusly lead to state shifts, which lead to inevitable biotic great change. Some measurements have been addressed to anticipate a planetary state shift, including quantified land use. An estimate of 0.68 was use for the year 1700. However, researchers still are not clear about how much land would have to be directly transformed before a state shift was imminent (Anthony D. B., et, al., 2012). An empirical landscape-scale studies shows that the critical threshold may be between 50 and 90% (Jackson, S. T., et, al., 2009).

However, few models are able to take into consideration complex global factors or determine the long-range impacts of potential polices. To build better predictive models, firstly the modeler must embrace the complexity of Earth's interrelated systems and study the mutual effect of global and local systems. Secondly the factors that put press on the environment must be identified (Anthony D. B., et, al., 2012).

In this problem, as members of International Coalition of Modelers, we are required to build a people oriented dynamic global network model of some aspect of Earth's health by identifying network nodes and appropriately connecting them to track relationship and attribute effects. We are also expected to run the model to see how it predicts future Earth health and discuss the factors produced by the model, the state change or tipping point prediction. Also, the model is expected to help decision makers about how their potential policies influence the long-range global Earth health. The network structure, say, critical nodes or relationships should be examined at the same time.

However, the model is hard to be verified, lacking of real-world dataset. Two important reasons are proposed. Firstly, although the Earth has suffered "Big Five" extinctions (Barnosky, A. D.et al., 2011) before, the datasets at that time are hard to obtain, and the environment indexes are difficult to measure the Earth health today. Besides, the antediluvian Earth environment evolves without human beings and civilization, which act as the key cataclysm of today's environment degradation. Secondly, although many kinds of data our model uses can be obtained somehow, some necessary data remains absent without the support of authority or decision makers. We will discuss the data requirements later. Thus, good creative ideas and theories are very important for our model in this case. Moreover, a more clear and intuitive running process should be given compared to data-based model.

1.2 Our Work

The development of our model was inspired by two ideas, which respectively contributes to the two tiers, with respect to both the complexity of the dynamic global network and simplification of the modeling process.

●Firstly, we define the node based on geographic division; whose intrinsic trait and its impact on the

interaction of driving factors are studied using Bayesian Belief Network. The states of Bayesian Belief Network (inner nodes) represent the driving factors of ecological degradation. We defined five driving factors with their dependency derived from famous models of world dynamic. Moreover, we elaborate the environment dimension as the local health measurement for the node. Here we use an analytic hierarchy process (AHP) to weight the various aspect of environment, such as species diversity index, etc. Since each aspect of the environment factor has its own threshold, the measurement thusly also has a general threshold.

By using a Bayesian Belief Network model to reveal the impact of the area’s intrinsic trait on the interaction of factors, we could train the Bayesian Belief Networking to get the probability distribution the change of factors. Besides, considering the policy effect or emergency events which can be represented by numerical change of factors' quantized indexes, we are able to study the long-range impact of potential policies or emergencies.

●Secondly, we formulate our global model by appealing to communication network models. The links

(population movement or diffusion of atmospheric pollution) and the signals can only be transmitted by signal channels. For each time of iteration, the signals received are regarded as protocol data unit (PDU), the potential policies and emergencies are viewed as service primitives, and the internal impact generated by factors' interaction acts just as the timer in the communication protocol. All of the three impacts (PDU, service primitives and timer) are protocol events which will cause protocol state change and carry out some protocol action in a communication model, which means giving out signals in this case. The health measurements of the global model are studied using network properties. We study the average shortest path of the global network and the giant component to measure the global Earth health and tipping point.

We also discuss in detail how our model work and the likely result of our model. Besides, how our model helps decision making and the critical nodes and relationship are also involved in this paper. In the last part of the paper, we discuss the advantages and disadvantage of our model and give some potential future work.

2. Assumption

Once a critical transition occurs, it is impossible for the system to return to its previous state. Critical transition means that once it occurs, there will be a reduction in biodiversity and severe impacts on much of what we depend on to sustain our quality of life. Such state change is extremely hard and even unable to go back to previous condition (Robert S., 2012). Here we assume that the critical transition is irreversible to simplify the model.

The population act as the dominate driver of all the factors. Of all the factors that impose impact on the environment, the human beings and the highly advanced civilization are the key driver and cataclysm of the environment degradation. It has been long studied that the global environment degradation has been accelerated heavily since the industrial evolution. The population factor has been the fundamental driving force for agriculture, industry, pollution, and impacts environment either directly or indirectly through the factors just mentioned.

The intrinsic trait of an area remains constant in a long period of time. The intrinsic trait of an area is some inherent properties of the certain area, including landform and monsoon, etc. The intrinsic trait is opposite concept of the drivers (or factors) which impact the Earth health. We assume that the intrinsic trait remains constant in that not only can we simplify the model but also help study the mutual effect of local conditions and global system. For example, we can apply Gaussian diffusion model (Donna B. S. et, al., 1999) to our global pollution diffusion analysis if the wind speed remain homogenous and stable in the entire space.

With a certain temporal and spatial resolution constraint, the health condition and impact are discrete. In a global network model, the node health condition could evolve because of its inner interaction of driving factors or outer impacts, and sometimes potential policy and emergency. In this paper, we assume that the health condition and impact are discrete. That is, the node has a pre-condition before inner or outer impact takes effect and a post-condition after it. For every iteration with a certain temporal and spatial resolution, the impacts it received in the last iteration actually take effect, and the node export impact to other nodes through the network while receiving impact from other nodes, then the node come to a post-condition. The outer impact received in this iteration, and the inner impacts generated by driving factors are stored until next iteration. This assumption will be explained as a communication network model later in 3.3.5. However, it must be aware that if the temporal and spatial resolution is extremely small, the health condition and impact are nearly continuous.

3. Two-tier Communication Network Model

Model Overview and Concepts Definition

We formulate the global network into a two-tier communication network model (TCNM). The important elements and concepts are defined as following:

Figure 3-1 Node definition in state level

●Factor: Inspired by Forrester’s World Model (Donella M., et, al., 2004), we define the factors that drive the

environmental evolution (or degradation) and thusly determine the Earth Health measurement in five dimensions, which are: population, environment, resource, industry and agriculture, in which the population acts as the key pressure of environment evolution.

●Health Measurement: We use a joint health measurement using the environment factors to measure earth

health, including species diversity index, air pollution index and landform change rate. We use an analytic hierarchy process (AHP) approach to determine the importance of each factor, which will be discussed in detail later.

●Node: The nodes of the HCNM are generally based on physical geographic division, where the area

represented by one node share analogous trait and the trait is assumed to be the same regardless of the nuance within the area. However, the resolution of a node can be modified according to the decision maker's jurisdiction level, such as county, state (as shown in Fig. 3-1) and nation, etc. The node has two states, which are active and inactive. The distinction between the two states is that the inactive nodes are those which the health measurement under the tipping point (threshold) and cannot be transferred to active again. People cannot live in inactive nodes, while those nodes are still able to release pollution to other nodes. This will be discussed later.

●Signal: The signals are the impacts that a node imposes on another node. The two types of the signal are

diffusion of atmospheric pollution and population movement. The signals can only be transmitted in signal channels.

●Link: The links between the nodes are defined as signal channels, that is, the signals (the air pollution and

population movement) can be transmitted from one node to another through signal channels. The signal channels represent physical adjacency relations between two nodes (the links in Fig. 3-1).

The advantage of using communication network models is that the different kinds of propagation flow are represented by similar entities (signal), and are conducted by the same link (signal channel) in the same network. All the impacts from outer or inner generators are regarded as communicational protocol event and will cause protocol state change and carry out protocol actions, which simply the dynamic global model and enable us to integrate multiple interactions within the same network to generate a more intuitionistic working process of the model.

Our model is also two-tiered. That is, while the global network defined as a communication network, the nodes of the global network are intrinsic Bayesian Belief Network within itself. This makes it possible to delve deeper into the interactions of different factors within the node and to quantify the outside (global) and policy impact on the node.

In this section, we define the nodes and then discuss the node-tier of the TCNM in detail, including the

communication protocol and the advantage of this analogy, the revised Bayesian Belief Network model which reveals how factors interact with each other, and the measurement of node level Earth health, etc.

3.1 Node

3.1.1 Node definition

● Node : An abstraction of geographic division on Earth, whose area share analogous trait, and whose trait is

assumed to be the same regardless of the nuance within the area.

Although we define the nodes regardless of whether the node represents ocean or land, there're two important issues that is worth mention. On the one hand, our model is people-oriented, and since people live on land, the human civilization will directly impose direct impact on land and offshore areas, and affect the ocean indirectly. On the other hand, the trait of lands and oceans are different, which make it difficult to measure the nodes’ health condition using a general quantized measurement. However, all these issues have been taken into consideration in the nodal health condition.

The resolution of a node can be modified according to the decision maker's jurisdiction level, such as county, state and nation, etc. The nodes have two distinct states, which are active and inactive, which will be explained in 3.1.5.

3.1.2 Driving factors of environment degradation

The node status is influenced by the interaction of five factors inspired by Forrester’s World Model (Natali H., et, al., 2006) shown in Fig. 3.1.2-1, which are environment, resources, population, industry, and agriculture. The arrows reflect the mutual interaction of those factors. The environmental factors are selected as the measurement of node’s health, which will be discussed in the next section. We use these five factors to represent the node’s condition.

Figure 3.1.2-1 Forrester’s World Model

Environment

To identify the index of environment, we consider the following metrics:

● Species Diversity : Species diversity studies the abundance of each species. The areas with high species

diversity are generally healthier than others. We uses Simpson's Diversity Index (SDI) to measure the species diversity (Simpson E. H., 1949), it can be defined as: SDI =1?1N (N ?1)

∑(N i (N i ?1))i

(1) Where SDI is Simpson’s diversity index (Simpson, 1949). N i is the number of species i; N is the number of all species.

● Air Quality: Air Quality Index (AQI) is used to indicate the extent of air pollution. The United States

Environmental Protection Agency (EPA) has developed an index which they use to report daily air quality.

represents hazardous air quality whereas a value below 50 means good. An AQI value of 100 generally corresponds to the mark for the pollution.

Table 3.1.2-1 Air Quality Index stages (EPA, 2012)

Air Quality Index (AQI)Levels of Health Concern Colors

0~50Good Green

51~100Moderate Yellow

101~150Unhealthy for Sensitive Group Orange

151~200Unhealthy Red

201~300Very Unhealthy Purple

301~500Hazardous Maroon

The AQI is based on the five pollutants regulated by the Clean Air Act: ground-level ozone, particulate matter, carbon monoxide, sulfur dioxide, and nitrogen dioxide. To integrate all the pollutants’ impacts, AQI

is represented by this formula:

I i=

C?C low

C?ig??C low

(I?ig??I low)+I low(2)

Where I i is the AQI of pollutant i, C is the pollutant concentration, C low is the concentration breakpoint that is lower than C, C?ig? represents the concentration breakpoint that higher than C, I low is the index breakpoint corresponding to C low;II?ig? represents the index breakpoint corresponding to C?ig?.

If multiple pollutants are measured at the same time, then the max AQI of pollutant is the area’s AQI value (David M., 2009).

AQI=*I i+,1 5- (3)●Landform Change Ratio (LCR): LCR can be directly calculated by using the area of landform changed to

divide the total area. LCR is useful because it studies the proportion of the land used by humankind (e.g.

urban area, farmland).

Resources: Our model chooses the sum value of four kinds of resources to quantify the resources. The four resources are land, forest, coal, and oil. We assume that the amount of a kind of resource before industrial evolution as the upper bound of the resource, so that we could index the amount of resources by ratio. The data of different resources should be normalized to its canonical form before summing process.

Agriculture:We appeal to Agricultural Production Indices (API) to study agriculture factor. API is based on the sum of price-weighted quantities of different agricultural commodities produced after deductions of quantities used as seed and feed weighted in a similar manner, which is introduced by the Food and Agriculture Organization of the United Nations (1995).

Industry: The industrial factor is quantized by Industrial Production Index (IPI), which is an economic indicator published by the Federal Reserve Board of the United States that measures the real production output of manufacturing, mining, and utilities.

Population: the population in a certain area, which will be quantized later.

3.1.3 Node characterizing with communication protocol

The nodes of the TCNM can be characterized using communication protocol because of their intrinsic similarity. Before further relationships are discussed between nodes of TCNM and communication protocol, some communication protocol element must be explained to better understand the advantage of this analogy. The communication protocol have several elements, including service primitive, protocol data unit, protocol variables, protocol status, protocol action, protocol event, etc.(Stefan B., 2000).

●Service primitive: the user of a communication gets services using service primitive, while the protocols

receive instruction through service primitive.

●Protocol data unit (PDU): the PDU act as the basic unit as peer entities transfer messages, which define the

content and form of messages. The PDU are generally known as signals.

●Protocol variables: the variables that a protocol uses.

●Protocol status: the status when the protocol entity protocol is waiting for input events.

●Protocol action: protocol action is the action a protocol entity performs when protocol status transfers from

the former to latter.

●Protocol event: protocol event is the trigger of protocol status transfer, including PDU from peer protocol

entities, service primitive from protocol users, and inner timer signals.

Protocol status transfers to another status when some certain protocol event occurs, during which the protocol will perform some actions. Relatively speaking, the former status is the latter's pre-condition, while the latter is the former's post-condition. The protocol transformations are represented by protocol status machine.

Now we can discuss how TCNM nodes can be characterized using communication protocol and the advantage of this analogy. As stated in 3.1, we have defined five driving factors of environment, which will change over time. These driving factors are similar to protocol variables in a communication model, because the node conditions of TCNM are quantified by these driving factors. Besides, the node condition's counterparts in the communication model are protocol statuses.

We have known that the TCNM node conditions change over time, either because of the mutual interaction of driving factors or result from outside impact on the node, or because of some emergencies or policies. One complex consideration of our modeling is how to combine all these impacts from three sources to predict the future condition of the node. By compare the outside impact to PDU, the emergencies and policies to service primitive, and the impact form inner node to inner timer of the protocol(although the third analogy is not very appropriate, the impact from inner node and timer share a key character in common, which is that they are both call up by the node/entity itself), we are able to treat all the impacts as protocol events of a communication model and integrate these impacts from three different sources to study the evolvement of node condition. Thusly, all the impacts can be simply represented and quantized by the numerical changes of driving factors of the nodes. Another important abstraction we benefit from communication protocol model is the introduction of pre and post conditions. We have known that the node condition changes temporally, but since the nodes are in a global TCNM network, the impact that the node imposes on other nodes will eventually affect itself sooner or later, slighter or heavier, which makes it difficult to study the node condition evolution over time. By intruding pre and post conditions, we can assume that all the impacts from three generators (node itself, outside and emergency) will not take effect immediately, but will be stored to next iteration. In this situation the node is in a condition (pre-condition) with some pending impacts. These impacts will work in next iteration, when this happens, the node condition changes and exports impacts towards other nodes, and transferred to next status (post-condition). This abstraction and division enables us to discuss the node condition evolution when each time of iteration occurs, which will be discussed in detail in section 3.3.5.

3.1.4 A Revised Bayesian Belief Network study of factors

In order to learn from the historical data about how mutual interaction of factors is guided by the potential impact of the local area’s intrinsic trait (landform, monsoon), and thusly influence the environment evolution of the area, we introduce a machine learning based approach using Bayesian Belief Network. As an extended version of Markov Chain, the Bayesian Belief Network can use a probabilistic transition based directed acyclic graph (DAG) to represent the complex interaction of factors. However, the size of the network is limited. In our model, the number of Bayesian Belief Network nodes is small so that we could use Bayesian Belief Network to train and predict the local area condition. In order to simulate the real-world problem and prevent our model from over-fitting, we revise the original Bayesian Belief Network superseding the general DAG by a directed bipartite graph. Thusly the outcome of the model fulfills our demand better.

Discretization of status

The original intention of using Bayesian Belief Network is to reveal the impact of interactions of status. In order to use the nodes of Bayesian Belief Network to represent the status, we must discretize the driving factors that constitute the status.

We initially normalize all the factors and then discretize each factor to five grades with respect to the limit of status numbers and the accuracy of results. As pre-defined in session 3.1, the environmental, resource, industrial, agricultural factors already have clear upper bounds, which make it possible to normalization. Here, as for population factor, considering the study of Department of Economic and Social Affairs of United Nations about world population prospects (Anthony D. B., et, al., 2012), we use the three times of current population as the upper bound of population factor. The population factor will stay at the upper bound if the population exceeds it. We use to represent the population of the current area.

The detailed quantitated factor’s normalization metrics are shown in table 3.1.4-1.

Table 3.1.4-1 Quantitated factor’s discretization metrics

Factor

Environment Resources Population Industry Agriculture Max Value

1.0 4.0 ?3.0

100.0 100.0 Level Gap 0.2 0.8 ?0.6 20.0 20.0

Construction of Bayesian Belief Network model

Now that we have discretized all the factors of the nodes’ conditions, we can build the key element of the Bayesian Belief Network, say, status and connection, and thusly explain the input and output of the model.

We use a directed bipartite graph to supersede the DAG in the original Bayesian Belief Network model. The Fig.

3.1.4-1, whose X set’s status are represented by a five-dimension vector of discretized factors (for example, if the

current five factors are respective 0.5, 1.5, 45.0, 30.0, then the status is represented as

? = (2 1 1 2 1)????????????????????????? in the network), and whose Y set is made up by the changes (contiguous value) of respective factors of the status after every iteration, could illustrate our bipartite Bayesian Belief Network model. From this definition the size Y set seems to be very large, but actually the output of the network is a continuous probabilistic distribution of Y set, so the size of Y size will not influence the efficiency of Bayesian Belief Network. Another crucial definition is the connection in the Bayesian Belief Network. Since the purpose of the Bayesian Belief Network is to get the probability of the connections of status, the connection in our bipartite Bayesian Belief Network is the one-way transfer from X set to Y set, which is also a DAG that fits the Bayesian Belief Network nature. Be aware that the in Y set could be change of either factor dimension.

Figure 3.1.4-1 a simple bipartite Bayesian Belief Network model

Considering the independent changes of either dimension (factors) of the status will impose a cascading impact on the model, including status transfer and actions carried out, we expect to reveal the probability distribution of changes of status’ factors respectively in a certain status. Note that the “status” mentioned here is a status vector consists of all the factors that are closely related to the targeted factor that we want to study. Hence, our expect outcome of the Bayesian Belief Network model could be represented by following set:

* ( | ?C )| S + (4)

Where S is the set of five factors, represents the targeted factor we want to study, ?

the distribution of the change of five factors respectively. When running the model, the five dimension of the status (factors) will be affected according to this distribution.

Training of Bayesian Belief Network model

Training the Bayesian Belief Network consists of two parts, which are structure training and parameter training.

For the structure of Bayesian Belief Network, the foremost issue arises from determining the factors that have mutual interactions. The result of the model is not in proportion with the number of connections between factors, which means that lesser and more appropriate connections will be more accurate and robust. In this paper, we still appeal to Forrester’s Global Model theory (Natali H., et, al., 2006) and only keep the direct connection between the factors shown in Fig. 3.1.2-1. The latent connection and impact will be involved in the Bayesian Belief Network through machine learning approach. Training of parameters of the network will be discussed later in parameter estimation section 4.2.

3.1.5 Nodal Health Measurement

In this section we discuss the nodal health measurement (NH) that could be used to quantify the status of the node of the HCNM. We assume that the health of an area is determined by the environmental factors mentioned in 3.1.4, including nodal species diversity, air quality and landform change. So the node’s health (NH) can be defined as following:

NH=ωs·NSDI+ωa·NAQI+ωl·NCLR (5)

Whereωs, ωa, ωl denotes the weight of species diversity, air quality, landform change respectively, and NSDI, NAQI, NCLR denotes the normalized value of SDI, (500?AQI), (1?CLR) in order.

Weights estimate using analytical hierarchy process

Having defined the elements, we are going to decide the weights of three factors. Analytical Hierarchy Process (AHP) is a proper way to find the weights.

Firstly, we need to construct a pairwise comparison matrix to present the relative importance of these three elements. As studied by Anthony D. B., et, al. (2012), worldwide shifts in species ranges, phenology and abundances are concordant with ongoing climate change and habitat transformation. So we judge species diversity is the main environmental factor with highest weight. What’s more, air quality also has a big influence on environment. And landform change ratio is also related to the environment, but slightly indifferent. So the pairwise comparison matrix can be defined as following:

A=(a ij) ·n(6) Where a ij is the relative importance of element i and element j.

The pairwise comparison matrix is expressed as following:

A=

[

153?52?

5

3?1

3

2?

2

5?

2

3?1]

(7)

Our next step is to get the largest eigenvalue and the largest eigenvector, the results areλ=3 and= ,0.5 0.3 0.2-where denotes the largest eigenvalue, and is the normalized value of the largest eigenvector according to the matrix A and A =.

Then, we check the consistency of the pairwise comparison matrix by using consistency ratio () (Saaty T L. 1980). If the is below 0.1, the degree of inconsistency in this matrix can be accepted. The CR of A equals to 0, which indicates the matrix A is acceptable. The equation to calculate the CR is as following:

I=λ?

(8)

=CI

RI

(9)

Where n is rank of A, CI is the index of consistency, RI denotes index of random consistency.

Thus, we can conclude that the parameters of NH are: ωs=0.5 ωa=0.3 ωl=0.2.

Nodal tipping point/threshold

Here we discuss the tipping point of nodal environment health with respect to the tipping point of the three factors of the environment health respectively.

The major biotic changes were the extinction of at least 75% of Earth’s species (Anthony D. B., et, al., 2012). Thus, the threshold of species diversity is 0.25. An empirical landscape-scale studies shows that the critical threshold of landform change may lie between 50 and 90% (Jackson, S. T., et, al., 2009), here we use an estimated value 0.68 as the threshold of LCR. Besides, 300 is an empirical value of the AQI’s tipping point. When any of the three thresholds have been reached, the nodal health is considered in danger, in which case should the model warn the decision makers.

A general nodal threshold (NH t) is set to a value, which could be computed as the weighted average value of all three thresholds, which is(NH t)=0.5×0.25+0.3×(1?0.68)+0.2×(1?0.6)=0.301. When NH is under 0.301, then the local node’s threshold has been reached.

Once the nodal health reaches it threshold, the node transfers form active condition to inactive condition, which is irreversible. Population will be reallocated to adjacent nodes and the inactive nodes can only transfer air pollution, which will be discussed later.

3.2 Link

The links of the global network represent the geographical adjacency of the nodes. In our two-tier communication network model, the signal channels are a good analogy of edges. The signal channels are defined as following:

Signal channel: the channel which enables signal pass through (B Sklar, 1988).

In a global network, one node can impose various kinds of impact on another node, which make it difficult to define the meaning of the link. In our model, we assume that the impact can be either diffusion of atmosphere pollution or population movement. It will be easy to construct two separate networks to represent the two kinds of impact as links respectively, which make it easy to explain and construct. However, this coarse isolation could make it difficult or even impossible to consider all the impacts and factors as a whole to study their interactions. Thus, we come up with a mechanism to combine the various outside impacts on the node by using the signal channels to represent the links.

Hence, the signals are pre-defined in session 3.1.3 as protocol data unit (PDU). There're two types of signals, which are diffusion of atmosphere pollution and population movement. The signal channels lay between two nodes if the two nodes are of adjacency relation. We will use this abstraction and analogy to explain the impacts from nodes to nodes, the critical nodes and relationships and the global health measurement later.

3.3 Topology

After the node and link definition, in this section we are going to discuss about the network topology, that is, how signals pass through the signal channel and thusly enable air pollution and population movement spread over the global network. Firstly we are going to study how the active or inactive condition of the node affects the signal transformation. Secondly the diffusion of atmosphere pollution and population movement is discussed in detail and policy and emergency are also mentioned. Finally we're going to use a communication protocol analogy to study during each time of iteration, how the impacts from all three sources (inner interaction of factors, outside impacts received and policy or emergency impacts) work together to transfer the node from pre-condition to post-condition, and how impacts will be exported to other nodes by the node itself.

3.3.1 Influence of active/inactive node condition

We have talked about the threshold or tipping point in section 3.1.5, that once the threshold has been reached, there will be a reduction in biodiversity and severe impacts on much of what we depend on to sustain our quality of life. We have also quantized the threshold. The distinction between the active and inactive condition is thusly defined. The nodes are initially all active. The inactive nodes are those which the health measurement exceeds the tipping point (threshold) and cannot be transferred to active again. People cannot live in inactive nodes, while those nodes are still able to release pollution to other nodes.

Population Reallocation

When a threshold has been reached, the area presented by the node is no longer livable. Hence, the first and foremost influence of change from active to inactive condition is population reallocation. We assume that the people will move to adjacent active nodes according to their health condition. The population reallocated to an adjacency active node n' are represented as following:

=NH

∑NH

· (10)

Where means the population of node n before it goes inactive, ∑NH means the sum of all the adjacent active nodes' health value defined in 3.1.5, NH represents the health value of adjacent active node n', and thusly

is the population reallocated to n'. An extreme case is that all the adjacent nodes have been inactive. If that happens, the population cannot escape from the current node. Sadly, that part of population will be excluded from our model.

Signal Transfer

Another impact of inactive state is on signal transfer. The atmosphere pollution and population can transfer through an active node, as we have known before. However, an inactive node can only transfer atmosphere pollution. Besides, population cannot move into it.

3.3.2 Migration Model

The aim of this model is to study how people move to adjacent area and thusly moves across the network. Like mentioned in 3.3.3, we assume that people movement, which is regarded as a kind of signal, can only move to the adjacent area through signal channel if people feel that area is addicted to them. So we develop a Migration model based on the I. S. Lowry regression model (Lowery, 1996). The original attraction for migration in Lowry’s model is job opportunity. However, the nodal health is better than job opportunity to represent the attraction to people. To calculate the extent of immigration between the two adjacent areas, we select the nodal health discussed before, including environmental factors of species diversity, air quality, landform change ratio and the population to

measure it. The expression is as following:

M ij=K(NH i

NH j

·i j

D ij

) (11)

Where M ij is the population move from area j to area i; NH i and NH j are the nodal health condition of the two areas respectively;i and j are the population of the two areas; D ij is the distance between the two areas.

The formula above studies the migration between two nodes. During a single iteration, the population would migrate through the network. Thus after the iteration, the final population of the node i i could be defined as following:

i

=i+∑M ij?∑M ji

j S

j S

(12)

Where i means the original population of node I before iteration, S denotes the set of all the adjacent nodes of node i, M ij is the population moves in from node j to node I, and M ij represents the population moves out from

3.3.3 Air Pollution Dispersion Model

We based our air pollution dispersion model on the steady-state Gaussian plume equation (Donna D., 1999) to analysis the diffusion of atmosphere pollution over network. To avoid over-fitting, our model use the air quality of node’s central zone to present the area’s air quality. The transmission of different air pollutants can be calculated individually, because the spread of these air pollutants are not related to others. However, we can describe the different air pollutants in the same model, in that the transmission pattern of air pollutants are similar. We assume that the air pollution signal can only transfer through the signal channel, which means if the areas are non-adjacent, one can’t affect another directly. The pollutants’ concentration of a certain node caused by the single transmission process can be expressed by Gaussian plume equation as following:

(x y z)=

Q

2πμσyσz

?

1

2(

y2

σy2

+

z2

σz2

)

(13)

Where (x y z)denotes the pollutants’ concentration of a certain node, which could be either pollution source node or adjacent receiver node; Q denotes the source concentration of air pollutants; x is distance between the center of source and receiver (Remember, the receiver could be the source itself) in the downwind direction, the y is distance in crosswind direction and the z is the vertical distance; μdenotes mean wind speed at release height;σyσz are standard deviation of lateral and vertical concentration distribution (m). The parameters such as μ σyσz can be measured. The method will be discussed in the other part.

Hence, we can get a series of the pollutants’ concentration of a certain node, as treating the node as either its neighbor node’s receivers or source of the pollutants respectively. All the computed value of pollutants’ concentration of a certain node is added up together in an iteration to obtain the final concentration, which can be expressed as following:

TC k=∑C ik

i S

(14)

Where TC k is the final concentration of a node with respect to both the impact form the node’s adjacent area and itself, S represents a set of nodes including the current node’s neighbors and itself, C ik is the concentration of the node respectively computed by treating the node as either source or receivers of pollutants.

3.3.4 Policy and emergency impact

The policy and emergency are viewed equally in the model, because they both can be explained by the change of value of driving factors. About this issue, we will discuss later in the model running section.

3.3.5 Node status evolve in every iteration

We have discussed impacts from all three sources in the model, including inner interactions of several factors, outside signals, and emergency or policy. These impacts can be represented by protocol events, which are originally defined as service primitive form users, PDU form peer protocol entities and inner time. In a general communication protocol model, the node status evolution process could be regarded as discrete, which means when the protocol is at a certain status, an event occurs so that the protocol transfer to next status, and during the transfer process, the protocol entity carry out some event. Similarly, in TCNM, the note status evolves as shown in Fig. 3.3.5-1 during each time of iteration.

Figure 3.3.5-1 Node status transfer

The similarity and mapping from TCNM and communication protocol have been discussed in 3.1.3. The impacts from all three sources are eventually reflected in the change of five factors of the status. Since we have trained the Bayesian Belief Network and get the distribution of the change of status’ five dimensions (five factors), we could compute the evolution of the node status. For each time of iteration, the node in its pre-condition status performs the impacts which are received in the last iteration, export new impact to outer network and then come to the post-condition. In the transfer process, the node will receive outside impact, policy and emergency impact, and inner interaction of factors, however, these factors will not take effect immediately. Actually, the pending impacts will perform effect in the next generation as just described.

3.4 Global Health Measurement

We have come up with a nodal health measurement (NH) in section 3.1.5, which is used to measure the health condition of the local nodes, and determine the tipping point of local area. In this section, a novel global health measurement (GH) is raised to measure the global health with lattice network structure. Moreover, the global tipping point is estimated using properties of the network structure. Before further extend our approach, some SNA concepts and properties must be explained.

●Average degree: the average value of the all node's degree.

●Giant component: a giant component is a connected component of a given random graph that contains a

constant fraction of the entire graph's vertices.

●Network Resilience: the ability of the network to provide and maintain an acceptable level of service in the

face of various faults and challenges to normal operation.

●Site percolation: Fill in each site with probability of p, which also means each site is removed with

probability of (1-p) (A. Rosowsky, 1999) as illustrated in Fig. 3.4-1.

Figure 3.4-1 A square lattice in 2-dimensions

There's also a critical theory about network resilience in a lattice network, as the p of site percolation reach a threshold Pc, a giant component suddenly appears. Edge removal is the opposite process – at some point the p drops below Pc and the network becomes disconnected (A. Rosowsky, 1999). Rosowsky have learned that the threshold of square, triangular, honeycomb and kagome are 0.5928716, 0.5, 0.765069, and 0.6546537. We use an average value of 0.628 as the threshold of a general lattice network, which also indicates that once 37.2 percent of all the nodes have been removed, the lattice network becomes disconnected.

Our TCNM is nearly a 2-dimension lattice network. This is because all the nodes initially share bond with its adjacent nodes and the bond number is within a little range around an average value. As more and more nodes become inactive, the number of remaining active nodes decreases. When the ratio of active nodes of all the nodes comes lower than 0.628, the size of giant component suddenly collapse. In reality, this mean the remaining active nodes are becoming isolated islands in the network, which reflect the tipping point of Earth Health. Thus, the global health measurement GH in a certain time can be represented as following, where N a means the number of

active nodes and N i means inactive nodes.

GH=

N a

N a+N i

(15)

Where GH is 1 if all the nodes are active and the tipping point (threshold) is 0.628.

4. Discussion and Simulation Run

In this section we are going to discuss some issues of our model, since TCNM has been developed and well explained in section 3. Although our model cannot actually be verified lacking of necessary data, the data requirements and collect suggestions are extended in detail in section 4.1. The methodology of estimate parameters of the model is discussed in section 4.2, including parameters of Bayesian Belief Network, air pollution and immigration. The impacts of policy and emergency are studies in section 4.3 and how the model can help decision making is also mentioned. In section 4.4 we talk about the sensitivity and robustness of the model. Finally in section 4.5, we evaluate how critical the nodes and links are by network structure analysis.

4.1 Dataset Requirements and Collect Suggestion

As a model considering many complex factors, our TCNM model need large amount of historical data to support it. Only with enough suitable data, can the output of our model meet the expectation. Considering the availability and cost of collecting data, we try to apply public data to most of our quantized indexes, which makes it possible to collect available and reliable data for prediction. However, although most of the historical data can be obtained someway, there is still part of necessary data that cannot be obtained without authority or decision makers’support. Table 4.1-1 lists all the needed historical data and their possible source.

Table 4.1-1 needed historical data and their possible source

Data Requirement Possible Source

Air Quality Index U.S. Environmental Protection Agency

Web Database

Species Diversity Index Require Decision makers’ Support

Landform Change Rate Require Decision makers’ Support

Resources Including land, forest, coal, and oil.

US Census Bureau.

Web Database

Industrial Production Index Federal Reserve Board.

Web Database

Agricultural Production Indices Food and Agriculture Organization of

the United Nations.

Web Database

According to the table, most of data can be obtained from the public database online, expect species diversity and the landform change ratio, which makes our model of wider application.

4.2 Parameters Estimate

Had we collected enough historical data, we could run our model to estimate parameter that determine from the dataset.

●Parameter of Bayesian Belief Network

Training for parameters of the Bayesian Belief Network is similar to normal parameter training in machine learning area. We use the pre-defined factors, namely, the historical data as the input, and train the model using gradient descent (S.Russell, et, al., 1995) to get the outcome of the model, that is the distribution of the change of status’ five dimensions (five factors). This could be used to represent the inner impact, which is compared to the inner timer in the communication protocol.

●Parameter of Air Pollution

We based our air pollution dispersion model on the steady-state Gaussian plume equation to analyze the diffusion of atmosphere pollution. As a classical point source emission model, the Gaussian’s equation has been widely used, such as Industrial Source Complex (ISC3) Dispersion Models (Donna B. S. et, al., 1999) used by U.S. Environmental Protection Agency. Our model can estimate the parameters by using the parameter estimate approach in ISC3 model.

●Parameter of immigration

For the parameter of immigration, we use the classic Laurie regression model. The model is based on the theory of gravity. It assumes that a gravitation parameter K is a constant rate in a certain condition just like the gravitational constant. Therefore, if we have enough data related to immigration, K can be directly calculated.

4.3 Impact of policy and decision support

Our model is expected to put forward some useful suggestions about the decision maker’s potential local policies. In this section, we will use three scenarios to explain how the TCNM model supports decision-making. Without enough necessary data to estimate parameters, the scenarios and outputs are expected data based on the general empirical situation, with respect to World2 model (Donella M., 2004), not for specific area or decision maker. Nevertheless, the outcome of our model would be more accurate with more training data, which is the intrinsic trait of the machine learning approach.

Consider a developing area located in the plain district, whose is adjacent to three areas of similar geographic, developmental and climate trait. The area has typical geographical and climate characteristics, which can be expressed and quantified in our model. Fig. 4.3-1 illustrate the quantitative indexes of the area, which indicates a beautiful area with enough resources and proper population and the agriculture development is slightly higher than the industry.

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

Environment Resources Population Industry Argiculture

Figure 4.3-1 a typical developing area’s quantitative indicators

In order to show the changes of factors in a figure, the five factors have been normalized (see section 3.1.2). Also, Needless to say, in order to compare nodal health condition with tipping point, we estimate general tipping point of the node using AHP approach (see section 3.1.5), which is 0.301 and can be used to judge whether a node have reached its tipping point or not. In reality, whether to warn the decision makers not only hinge on the general tipping point, but on the tipping point of each environmental factor respectively.

In situation 1, decision makers give the highest priority to the development of the industry and ignore the protection of the environment. The policies of the decision maker are considered as the external policy factors in out model. If the decision makers adopt these policies, the expected results are shown in Fig. 4.3-2.

Figure 4.3-2 the expected output in situation 1

As the figure illustrated, the development of the area is really fast in the first 50 years, but the index of resource and environment decrease rapidly at the same time. Even worse, the index of resources and environment are close to their tipping points in yr. 80, and the population, industry and agriculture reached their bottleneck due to the shortage of resource and the bad environment. In the last twenty years, the resource consumption obviously becomes slow and the environment begins to get better somehow due to purification ability of the environment itself. However the development of industry and agriculture begin to stagnant or even retrogress.

If the decision makers choose policies as we mentioned above, the index of environment will surely under the tipping point. Then our model is expected to warn them by pointing out that the policies may cause serious environmental pollution and may even lead to local state shift. As a result, the local ecological environment system will crash.

Situation 2

In situation 2, decision maker apply a compromise strategy, which slows down the development of the area, but still does not pay enough attention to avoid environment pollution and control resource consumption. In this case, the expected output of our model is shown in Fig. 4.3-3.

Figure 4.3-3 the expected output in situation 2 Although the index of environment does not below the general nodal tipping point yet, we can also observe that environment and resources are suffering from a remarkable decline, which suggests that the health condition of this are will eventually under tipping point in the long run. For this kind of policies, we will also send warning to decision makers. 00.10.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 years 20 years 40 years 60 years 80 years 100 years

Environment Resources Population Industry Argiculture 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 years 20 years 40 years 60 years 80 years 100 years

Environment Resources Population Industry Argiculture

In this scenario, we consider a positive policy. Decisions makers understand that coarse policies would impose negative impacts on the environment, so that they decide to choose sustainable policies, including protecting the environment and limit the resource consumption. By doing this, the expected output of our model is shown in Fig.

4.3-4.

Figure 4.3-4 the expected output in situation 3

As illustrated, the policies not only make it possible for the population and development of agricultural and industries increase in a considerate rate, but also control the pollution and resource consumption. Although the development of area is not as fast as in the first situation, the speed will not dramatically slow down. Compared other situations, we can conclude the sustainable development will be better than situation 1 and 2 in the next 100 years.

4.4 Sensitivity and Uncertainty Analysis

When applying the model to the real world, it is very difficult to guarantee the data that are completely “clean”, which means correct and complete. Except for the personal error arises when quantifying the real -world situation according to some artificial standard, how to deal with some controversial region adjacency and some depopulated zone may lead to unexpected result. Fortunately, the sensitivity of TCNM model is very robust. TCNM not only considers the graph structure based on area adjacent, but also analysis node’s inherent feature and potential factors of evolution using machine learning. The result is likely to take into consideration the region's long -term stability of the geographical and climate factors. So changing or missing a few connections would not have big influence on TCNM model.

Our model also considers some uncertain factors, such as the influence of the different policy, or other unexpected events. We have not included natural disasters, which seldom occur in our model, because prediction of natural disasters accurately is still a big challenge to human and thusly another complex problem. But some special natural disasters (such as Japan's frequent earthquake) are considered as indispensable factors in dynamic ecological system. In the future we may find a better method, and then we will take these factors into our model.

4.5 Critical nodes and links

One of the powerful elements of using network modeling is the ability to analyze the network structure, just as we have done in 3.4 to determine the global health measurement. In this sector, we are going to analyze the critical nodes and links of the TCNM network. Again, we first talk about some important concepts of network science.

● Degree Centrality : the number of links incident upon a node (i.e., the number of ties that a node has)

(Freeman et, al., 1979). The degree centrality is not very useful in our model, because the TCNM is nearly a lattice network, whose nodes’ degree lies in a small range. Moreover, the degree centrality represents little meaning in reality which makes it useless to be involved to measure the critical nodes.

● Betweenness Centrality : a measure of a node's centrality in a network equal to the number of shortest paths

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 years 20 years 40 years 60 years 80 years 100 years

Environment Resources Population Industry Argiculture

betweenness centrality does not have to be those with high degrees. However, those nodes are critical because the pollution and people are more likely to pass through it to influence the other part of the network. At the same time, these nodes thusly receive more impact from other nodes and are more likely to be affected. The betweenness centrality of node is defined as following:

C()=∑σst()σst

s i t

(16)

Where σst is the total number of shortest paths from node to node and σst() is the number of those paths that pass through node.

●Closeness Centrality: the natural distance metric between all pairs of nodes, defined by the length of their

shortest paths (Sabidussi, G., 1966). The closeness centrality of a node determines how easy the impacts exported by the node can spread all over the network, say, atmosphere pollution, etc. The closeness centrality of node is defined as following:

C()=∑ ( )

||

j 1

?C

C? C

(17)

Where ( ) is the length of the shortest path connecting nodes and . C and C ax are the minimum and maximum lengths of the shortest paths respectively.

●Edge Betweenness: the number of shortest paths between pairs of nodes that run along it (Girvan M. et, al.,

2002). The edges that connect communities in the network have high edge betweenness. In TCNM network, edges with high edge betweenness are the critical channels for the diffusion of atmosphere pollution and population movement.

After the definition and explanation of some important concepts, we are now able to study how to measure critical nodes and links. The closeness centrality shows how fast the impacts generated by the node can spread over the network, while the betweenness centrality reflect how easy the node be affected by other nodes. We assume that the two considerations are equally important, so the how critical the node is can be defined as:

I()=C()+C()

2

(18)

Where I() represents the importance of node , C() means betweenness centrality and C( ) means closeness centrality.

The importance of the link can be easily represented by edge betweenness E():

E()=∑σst()σst

s t

(19)

Where σst is the total number of shortest paths from node to node and σst() is the number of those paths that pass through link.

By computing I( ) of all the nodes and E() of all the links, we are able to rank the importance of all the nodes and links and decide which are critical. In addition, since our model is two-tiered, the critical node of our inner network (five factors with respect to Forrester’s World Model) is also worth mention. The population is the critical factor of the inner network.

5. Conclusion

Predicting the Earth's health condition is an intricate problem which should embrace the complexity of Earth's interrelated systems and the global impact on local conditions and vice versa. In our two-tiered communication network model, we firstly identify several driving factors of environment condition inspired by Forrester's World Model, and study the impact generated by interaction of these factors through machine-learning approach. The nodes and edges are originally derived from geographic division which can be modified according to the resolution. A novel analogy between nodal character and communication protocol is introduced to give a better

two levels since our model is generally two-tiered. The local health is measured by the joint index of environment factors, while the global health is creatively introduced by the properties of lattice network resilience. The tipping point of the health condition is also discussed.

Then we discuss the topology of our global network. We talked about the impact on the signal transmission of node's active or inactive status. Besides, the diffusion of atmosphere pollution and population movement over the network are studied and modeled respectively.

Finally, we discuss and simulation run sector, we discuss the requirement of necessary data in order to validate or run the model in reality. Furthermore, the methodology of parameter estimate is studied. We also explained how policy impact the model and vice versa. Sensitivity is also discussed in this section. Moreover, the critical nodes and edges are researched using network structure.

5.1 Strength

●We use a two-tier communication network model to study and predict the Earth's health condition. The

Bayesian Belief Network enable us to introduce the latent intrinsic trait of the area represented by the node, and study the interaction of driving factors of environment degradation through machine learning approach.

By representing outside impacts as signals, we are able to consider all the impacts in a network as a whole.

●Also, we could reduce the complexity of taking into consideration all the impacts all three sources by

compare the nodal health condition to communication protocol, in that the impacts are treated as protocol event and nodal health condition evolution can be discretized by pre and post conditions. This abstraction and discretization enables us to study nodal health condition in each time of iteration separately. We appeal to network structure to study the global health measurement and critical nodes or links, which is an efficient way to use the properties of network to study real-world problems.

5.2 Weakness

●Our model is sound, creative and exciting; however, we do not have the necessary data to verify our model

although we have discussed the data requirement. Hence, the correctness of our model remains to be verified.

Besides, we use the resource amount of an area before industrial evolution as the upper bound of the quantized resource value, which is not a very accurate measurement.

●Since the measurement of critical nodes and links are numeric values, we can only get a rank list of critical

nodes and links respectively without a clear criterion whether a node or link is critical or not. Also, in order to obtain a joint nodal health measurement considering different importance of three environmental factors, we use analytical hierarchy process, which introduces subjective impacts into the model.

湖州市教育学会关于开展2019年优秀教育论文评选.doc

湖教学会〔2019〕1号 湖州市教育学会关于开展2019年优秀教育论文评选 的通知 各区（县）教育学会，各市属学校（理事单位）： 为进一步提升广大教育工作者教育科研的参与度,反映学会群众性教育教学研究的最新成果,决定开展2019年度优秀教育论文评选活动,现将有关事项通知如下： 一、参评对象 全市各教育学会团体会员单位的教师、教育行政管理人员。 二、论文主题 以“核心素养、课程建设、教学改革”为主题，题目自拟。 三、参评要求 1.论文要紧扣主题，针对当前教育教学中的重点、难点和热点，做到观点新颖明确、论据准确充分、论述严谨完整，理论联系实际，具有较强的说服力。 不受理游离上述主题的论文。来源于课题研究的，须以论文形式呈现，不受理课题研究结题报告。 2.参评论文必须是未曾在省、市级评选中获过奖项的。 3.参评论文必须是本人（或合作）独立完成的。 四、名额分配 每个理事单位会员（学校）限送8篇，每个单位会员（学

校）限送5篇。 五、文本要求 1.篇幅以3500～5000字为宜，用A4纸双面打印，一式三份。 2. 3.正文前须有【内容摘要】。 4.字体字号：文章题目宋体小三号加粗；正文、“内容摘要”和“关键词”宋体小四号。 5.凡文中的“引用”文字标明数字序号（如①…），在文章最后页逐一注明书名、页码、作者姓名。 6.标题层次： 一、……（一级标题，宋体四号加粗） （一）……（二级标题，宋体小四号加粗） 1. ……（三级标题，宋体小四号） （1）……（四级标题，宋体小四号） ①…… 六、报送截止日期 2019年5月30日 七、评审办法 1.区县教育学会对本区县送评论文进行初审，按总数的60％择优报市参加复评。市属学校（理事单位）由市教育学会组织初评。 2.市教育学会对上报参加市级复评的论文进行匿名评

优秀论文范例

毕业论文（设计） 题目 指导老师 专业班级电子商务20061 姓名 学号 20052021138 2009年 5月 30 日

都存在着一定的问题。本文通过对学院毕业论文选题、论文指导以及论文备份等工作流程的深入了解及进行了可行性分析后，对其作了需求分析、功能模块划分、数据库的设计以及界面设计，并完成了毕业论文选题系统的开发。本系统采用简单灵活的ASP 语言，并结合简单灵活的Access 数据库，实现在线毕业论文的选题工作，提供学生一个完全公正、开放的选题平台，同时在一定程度上方便了学生与指导教师间的相互交流，同时提供了论文电子稿的保存手段，提高了教务人员

(1) 1.1系统现状研究................................. 错误！未定义书签。 1.2系统开发方法 (1) (2) 2.1系统调研 (2) 2.2系统总体目标 (2) 2.3技术可行性分析 (2) (3) 3.1系统需求分析 (3) 3.1.1功能需求 (3) 3.1.2性能需求 (3) 3.2系统流程图 (4) 3.2.1数据流程图 (4) (4) 4.1概要设计 (4) 4.1.1功能模块结构图 (4) 4.1.2功能模块描述 (5) 4.2数据库设计 (6) 4.2.1表设计 (6) 4.3详细设计 (6) 4.3.1用户登录模块设计 (8) 4.3.2论题管理模块设计 (9) 4.3.3学生选题模块设计 (10) 4.3.4筛选学生模块设计 (11) 4.3.5留言联系模块设计 (12) 4.3.6上传论文模块设计 (13) 4.3.7查看论文模块设计 (13)

09年美赛A题优秀论文翻译

A题设计一个交通环岛 在许多城市和社区都建立有交通环岛，既有多条行车道的大型环岛（例如巴黎的凯旋门和曼谷的胜利纪念碑路口），又有一至两条行车道的小型环岛。有些环岛在进入口设有“停车”标志或者让行标志，其目的是给已驶入环岛的车辆提供行车优先权；而在一些环岛的进入口的逆向一侧设立的让行标志是为了向即将驶入环岛的车辆提供行车优先权；还有一些环岛会在入口处设立交通灯（红灯会禁止车辆右转）；也可能会有其他的设计方案。 这一设计的目的在于利用一个模型来决定如何最优地控制环岛内部，周围以及外部的交通流。该设计的目的在于可利用模型做出最佳的方案选择以及分析影响选择的众多因素。解决方案中需要包括一个不超过两页纸，双倍行距打印的技术摘要，它可以指导交通工程师利用你们模型对任何特殊的环岛进行适当的流量控制。该模型可以总结出在何种情况之下运用哪一种交通控制法为最优。当考虑使用红绿灯的时候，给出一个绿灯的时长的控制方法（根据每日具体时间以及其他因素进行协调）。找一些特殊案例，展示你的模型的实用性。 标题：一个环来控制一切:优化交通圈。 安德里亚?利维亚伦 安德烈娅?利维 拉塞尔?梅里克 哈维姆德学院 顾问:苏珊 摘要 我们的目的是利用车辆动力学考虑在圆形交叉路口的道路情况。我们主要根据进入圆形道路的速度决定最好的方式来控制车流量。我们假设在一个车道通过圆形道路循环，这样交通输入量能够被调节。（也就是，不会有优先的交通输入量） 对于我们的模型，可改变的参数是排队等候进入的速率，进入圆形道路的速率（服务速率），这个圆形道路最大的容量和离开这个道路的速率。我们使用带有队列和交通圈的隔室模型作为隔间。来自外界的车辆首先进行排队等候，然后进入圆环交叉路口，最后离开到外界。我们把服务速率和离开速率作为在圆环交叉路口的车辆数量参考。 另外，我们利用计算机来拟态一个可见表示，发生在不同情形下的圆环交叉路口。允许我们检验不同的情况，例如不平等的交通流量由于不同的队列，一些十字路口比其他车辆有一个更高的概率。这个拟态模仿实施栩栩如生，例如如何当前面是空道路时进行加速，而当前面有其他车辆时进行减速。大多数情况下，我们发现：一个高服务效率能够保持交通顺畅的最佳方式，这意味着对于进入交通的效率是最有效的。然而，当交通变得拥堵时，较低的服务率更好的适应了交通，这指示应该使用一个红绿灯。所以，在不同时间段，依靠预测中的交通流量，一个信号灯应该被安装进行循环实现。

关于组织参加江苏省优秀教学论文评比活动的通知

关于组织参加江苏省优秀教学论文评比活动的 通知 Document number：BGCG-0857-BTDO-0089-2022

关于组织参加 “2011年江苏省优秀教学论文评比”活动的通知各学校： 根据苏教研［2011］19号文件精神，市教研室将组织列出学科（见附件）的教学论文竞赛，比赛要求同省教研室要求。请各乡镇、学校以学科为单位，将论文的三项材料（论文电子稿、纸质稿、诚信承诺书各1份）和汇总表（汇总表为EXCEL电子表格）及时交县教研室相关学科教研员。 所有材料报送的截止时间为：2011年9月20日。 宝应县教育局教研室 二○一一年五月 附件：关于组织“2011年江苏省优秀教学论文评比”活动的通知 各市教研室（教科院、教科研中心）： 为了提高我省中小学、幼儿园教师和教研员的教育教学理论水平，鼓励广大教师积极开展教育教学研究，总结、交流在课程教学实践中取得的宝贵经验，提升教育教学业务水平，进一步推动我省课程改革的深入实施，经研究决定，拟组织“2011年江苏省优秀教学论文评比”活动。现将有关事项通知如下： 一、参评人员 教育行政管理人员、教研人员和广大教师。 二、参评学科

小学：英语、音乐、科学、综合实践活动 初中：语文、英语、物理、化学、生物、德育、音乐、综合实践活动 高中：语文、英语、物理、化学、德育、音乐、通用技术、综合实践活动 其余学科的论文评比将在明年组织进行。 三、论文要求 1．论文应突出科学性、前瞻性、学科性、实践性；要立意新颖，观点明确，论证充分，给人启迪，对学科教学有较深刻的见解；能体现教育教学新理念、新探索、新成果；关注课改、评价改革的热点，关注学科教育教学的发展，对优化课堂教学，提高教学效益，有自己的思考与发现等。 2．参评论文题目自拟，内容自定，应注重理论思考和实践研究，属于工作总结或解题指导等文章不予评比。 3．引用他人原始资料的信息、观点、句子等应做标注，文责自负，坚决反对抄袭行为，一旦发现将取消参评资格，并通知该作者所在单位及所在市、县教研室。 4．论文字数在3000～5000字为宜。 四、报送要求 1．各大市在广泛发动的基础上，经评审后推荐小学每学科20篇，初中每学科20篇，高中每学科10篇论文（包括纸质稿一式1份和电子稿1份）送省参评。每位作者只能报送1篇论文。 2．送评论文以大市为单位汇总，并填写《2011年江苏省优秀教学论文评比各市汇总目录》，（见附件1），于2011年10月30日前，将本市的参评材料统一报送至我室。

美赛论文要点

摘要： 第一段：写论文解决什么问题 1．问题的重述 a. 介绍重点词开头： 例1：“Hand move” irrigation, a cheap but labor-intensive system used on small farms, consists of a movable pipe with sprinkler on top that can be attached to a stationary main. 例2:……is a real-life common phenomenon with many complexities. 例3：An (effective plan) is crucial to……… b. 直接指出问题： 例 1：We find the optimal number of tollbooths in a highway toll-plaza for a given number of highway lanes: the number of tollbooths that minimizes average delay experienced by cars. 例2：A brand-new university needs to balance the cost of information technology security measures with the potential cost of attacks on its systems. 例3：We determine the number of sprinklers to use by analyzing the energy and motion of water in the pipe and examining the engineering parameters of sprinklers available in the market. 例4: After mathematically analyzing the …… problem, our modeling group would like to pres ent our conclusions, strategies, (and recommendations )to the ……. 例5：Our goal is... that (minimizes the time )………. 2．解决这个问题的伟大意义 反面说明。如果没有…… Without implementing defensive measure, the university is exposed to an expected loss of $8.9 million per year. 3．总的解决概述 a．通过什么方法解决什么问题 例：We address the problem of optimizing amusement park enjoyment through distributing Quick Passes (QP), reservation slips that ideally allow an individual to spend less time waiting in line. b．实际问题转化为数学模型

数学建模国赛一等奖论文

电力市场输电阻塞管理模型 摘要 本文通过设计合理的阻塞费用计算规则，建立了电力市场的输电阻塞管理模型。 通过对各机组出力方案实验数据的分析，用最小二乘法进行拟合，得到了各线路上有功潮流关于各发电机组出力的近似表达式。按照电力市场规则，确定各机组的出力分配预案。如果执行该预案会发生输电阻塞，则调整方案，并对引起的部分序内容量和序外容量的收益损失，设计了阻塞费用计算规则。 通过引入危险因子来反映输电线路的安全性，根据安全且经济的原则，把输电阻塞管理问题归结为：以求解阻塞费用和危险因子最小值为目标的双目标规划问题。采用“两步走”的策略，把双目标规划转化为两次单目标规划：首先以危险因子为目标函数，得到其最小值；然后以其最小值为约束，找出使阻塞管理费用最小的机组出力分配方案。 当预报负荷为982.4MW时，分配预案的清算价为303元/MWh，购电成本为74416.8元，此时发生输电阻塞，经过调整后可以消除，阻塞费用为3264元。 当预报负荷为1052.8MW时，分配预案的清算价为356元/MWh，购电成本为93699.2元，此时发生输电阻塞，经过调整后可以使用线路的安全裕度输电，阻塞费用为1437.5元。 最后，本文分析了各线路的潮流限值调整对最大负荷的影响，据此给电网公司提出了建议；并提出了模型的改进方案。

一、问题的重述 我国电力系统的市场化改革正在积极、稳步地进行，随着用电紧张的缓解，电力市场化将进入新一轮的发展，这给有关产业和研究部门带来了可预期的机遇和挑战。 电网公司在组织电力的交易、调度和配送时，必须遵循电网“安全第一”的原则，同时按照购电费用最小的经济目标，制订如下电力市场交易规则： 1、以15分钟为一个时段组织交易，每台机组在当前时段开始时刻前给出下一个时段的报价。各机组将可用出力由低到高分成至多10段报价，每个段的长度称为段容量，每个段容量报一个段价，段价按段序数单调不减。 2、在当前时段内，市场交易-调度中心根据下一个时段的负荷预报、每台机组的报价、当前出力和出力改变速率，按段价从低到高选取各机组的段容量或其部分，直到它们之和等于预报的负荷，这时每个机组被选入的段容量或其部分之和形成该时段该机组的出力分配预案。最后一个被选入的段价称为该时段的清算价，该时段全部机组的所有出力均按清算价结算。 电网上的每条线路上有功潮流的绝对值有一安全限值，限值还具有一定的相对安全裕度。如果各机组出力分配方案使某条线路上的有功潮流的绝对值超出限值，称为输电阻塞。当发生输电阻塞时，需要按照以下原则进行调整： 1、调整各机组出力分配方案使得输电阻塞消除； 2、如果1做不到，可以使用线路的安全裕度输电，以避免拉闸限电，但要使每条 线路上潮流的绝对值超过限值的百分比尽量小； 3、如果无论怎样分配机组出力都无法使每条线路上的潮流绝对值超过限值的百分 比小于相对安全裕度，则必须在用电侧拉闸限电。 调整分配预案后，一些通过竞价取得发电权的发电容量不能出力；而一些在竞价中未取得发电权的发电容量要在低于对应报价的清算价上出力。因此，发电商和网方将产生经济利益冲突。网方应该为因输电阻塞而不能执行初始交易结果付出代价，网方在结算时应该适当地给发电商以经济补偿，由此引起的费用称之为阻塞费用。网方在电网安全运行的保证下应当同时考虑尽量减少阻塞费用。 现在需要完成的工作如下： 1、某电网有8台发电机组，6条主要线路，附件1中表1和表2的方案0给出了各机组的当前出力和各线路上对应的有功潮流值，方案1~32给出了围绕方案0的一些实验数据，试用这些数据确定各线路上有功潮流关于各发电机组出力的近似表达式。 2、设计一种简明、合理的阻塞费用计算规则，除考虑电力市场规则外，还需注意：在输电阻塞发生时公平地对待序内容量不能出力的部分和报价高于清算价的序外容量出力的部分。 3、假设下一个时段预报的负荷需求是982.4MW，附件1中的表3、表4和表5分别给出了各机组的段容量、段价和爬坡速率的数据，试按照电力市场规则给出下一个时段各机组的出力分配预案。 4、按照表6给出的潮流限值，检查得到的出力分配预案是否会引起输电阻塞，并在发生输电阻塞时，根据安全且经济的原则，调整各机组出力分配方案，并给出与该方案相应的阻塞费用。 5、假设下一个时段预报的负荷需求是1052.8MW，重复3~4的工作。 二、问题的分析

美赛论文模板(强烈推荐)

Titile Summary During cell division, mitotic spindles are assembled by microtubule-based motor proteins1, 2. The bipolar organization of spindles is essential for proper segregation of chromosomes, and requires plus-end-directed homotetrameric motor proteins of the widely conserved kinesin-5 (BimC) family3. Hypotheses for bipolar spindle formation include the 'push?pull mitotic muscle' model, in which kinesin-5 and opposing motor proteins act between overlapping microtubules2, 4, 5. However, the precise roles of kinesin-5 during this process are unknown. Here we show that the vertebrate kinesin-5 Eg5 drives the sliding of microtubules depending on their relative orientation. We found in controlled in vitro assays that Eg5 has the remarkable capability of simultaneously moving at 20 nm s-1 towards the plus-ends of each of the two microtubules it crosslinks. For anti-parallel microtubules, this results in relative sliding at 40 nm s-1, comparable to spindle pole separation rates in vivo6. Furthermore, we found that Eg5 can tether microtubule plus-ends, suggesting an additional microtubule-binding mode for Eg5. Our results demonstrate how members of the kinesin-5 family are likely to function in mitosis, pushing apart interpolar microtubules as well as recruiting microtubules into bundles that are subsequently polarized by relative sliding. We anticipate our assay to be a starting point for more sophisticated in vitro models of mitotic spindles. For example, the individual and combined action of multiple mitotic motors could be tested, including minus-end-directed motors opposing Eg5 motility. Furthermore, Eg5 inhibition is a major target of anti-cancer drug development, and a well-defined and quantitative assay for motor function will be relevant for such developments

全国数模竞赛优秀论文

一、基础知识 1.1 常见数学函数 如：输入x=[-4.85 -2.3 -0.2 1.3 4.56 6.75]，则： ceil(x)= -4 -2 0 2 5 7 fix(x) = -4 -2 0 1 4 6 floor(x) = -5 -3 -1 1 4 6 round(x) = -5 -2 0 1 5 7 1.2 系统的在线帮助 1 help 命令： 1.当不知系统有何帮助内容时，可直接输入help以寻求帮助: >>help（回车） 2.当想了解某一主题的内容时，如输入： >> help syntax（了解Matlab的语法规定） 3.当想了解某一具体的函数或命令的帮助信息时，如输入： >> help sqrt （了解函数sqrt的相关信息）

2 lookfor命令 现需要完成某一具体操作，不知有何命令或函数可以完成，如输入： >> lookfor line （查找与直线、线性问题有关的函数） 1.3 常量与变量 系统的变量命名规则：变量名区分字母大小写；变量名必须以字母打头，其后可以是任意字母，数字，或下划线的组合。此外，系统内部预先定义了几个有特殊意 1 数值型向量（矩阵）的输入 1．任何矩阵（向量），可以直接按行方式 ．．．输入每个元素：同一行中的元素用逗号（，）或者用空格符来分隔；行与行之间用分号（；）分隔。所有元素处于一方括号（[ ]）内； 例1： >> Time = [11 12 1 2 3 4 5 6 7 8 9 10] >> X_Data = [2.32 3.43；4.37 5.98] 2 上面函数的具体用法，可以用帮助命令help得到。如：meshgrid(x,y) 输入x=[1 2 3 4]; y=[1 0 5]; [X,Y]=meshgrid(x, y)，则 X = Y =

关于开展优秀调研成果(论文)评选工作的通知

关于开展优秀调研成果（论文）评选工作的通知 作者-县教育工会-教育来源-本站原创-点击数823-更新时间-2011-6-13 各基层教育工会： 为进一步提高全县教育工会调研工作水平，切实推动工会理论创新并指导实际工作，根据省、市教育工会的工作部署，决定开展全县教育工会优秀调研成果（论文）评选工作，现将有关事项通知如下： 一、评选范围 近年来我县教育系统围绕教育改革发展热点、难点问题形成的调研报告和理论研究文章。 二、评选标准 调研报告选题围绕教育工会当前的重点工作或涉及教职工切身利益的重点热点问题，调查深入扎实，反映情况准确客观，分析合理深刻，对策建议有针对性、创造性和可操作性，篇幅不宜过长。论文论题有价值，密切联系工会工作实际，观点鲜明正确、有创新，论述透彻、逻辑严密、结构合理、文字简练。 三、相关要求 1、各基层教育工会如有重大调研课题，可先将课题负责人、课题名称、课题实施方案等于6月25日前报县教育工会。 2、调研成果（论文）评选以基层教育工会为参评单位。各基层教育工会在初评的基础上，选择质量较高的1－3篇文章报送县教育工会参评。 3、报送稿件首页标明“调研报告”或“论文”。如已取得实际效果的,请附简短文字说明，于9月30日前发：176012495＠https://www.wendangku.net/doc/3f614402.html,。

4、调研报告、论文每篇字数控制在8000字以内，排版格式一律使用A4型版面，按照标题、单位名称、正文顺序行文。标题使用2号宋体，正文使用3号仿宋体排版。文章结尾须注明执笔人、工作单位、形成时间、详细通讯地址、邮编、主要联系人及联系电话。 四、表彰推荐 此次优秀调研成果（论文）评选将设一、二、三等奖和优秀奖。届时，县教育育工会将邀请专家进行评审，对获奖作者颁发证书，并择优向省教育工会推荐。 青田县教育工会 2011年6月13日附件：青田县教育工会优秀调研成果和论文评选参考课题 调研参考题目： 一、事业单位绩效工资实施情况、存在问题及对策建议； 二、农村教职工住房、医疗保障状况调查； 三、事业单位职工收入状况及问题分析； 四、影响教职工创新积极性的因素调查； 五、民办学校工会、教代会建设情况调查。 理论研究参考课题： 一、事业单位绩效考核制度研究； 二、工会维护职工发展权益的路径及作用研究； 三、其他关于教育系统改革发展、职工权益保护、工会自身建设等方面的问题研究。

SARS传播的数学模型 数学建模全国赛优秀论文

SARS传播的数学模型 (轩辕杨杰整理) 摘要 本文分析了题目所提供的早期SARS传播模型的合理性与实用性，认为该模型可以预测疫情发展的大致趋势，但是存在一定的不足.第一，混淆了累计患病人数与累计确诊人数的概念；第二，借助其他地区数据进行预测，后期预测结果不够准确；第三，模型的参数L、K的设定缺乏依据，具有一定的主观性. 针对早期模型的不足，在系统分析了SARS的传播机理后，把SARS的传播过程划分为：征兆期，爆发期，高峰期和衰退期4个阶段.将每个阶段影响SARS 传播的因素参数化，在传染病SIR模型的基础上，改进得到SARS传播模型.采用离散化的方法对本模型求数值解得到：北京SARS疫情的预测持续时间为106天，预测SARS患者累计2514人，与实际情况比较吻合. 应用SARS传播模型，对隔离时间及隔离措施强度的效果进行分析，得出结论：“早发现，早隔离”能有效减少累计患病人数；“严格隔离”能有效缩短疫情持续时间. 在建立模型的过程中发现，需要认清SARS传播机理，获得真实有效的数据.而题目所提供的累计确诊人数并不等于同期累计患病人数，这给模型的建立带来不小的困难. 本文分析了海外来京旅游人数受SARS的影响，建立时间序列半参数回归模型进行了预测，估算出SARS会对北京入境旅游业造成23.22亿元人民币损失，并预计北京海外旅游人数在10月以前能恢复正常. 最后给当地报刊写了一篇短文，介绍了建立传染病数学模型的重要性.

1．问题的重述 SARS （严重急性呼吸道综合症，俗称：非典型肺炎）的爆发和蔓延使我们认识到，定量地研究传染病的传播规律，为预测和控制传染病蔓延创造条件，具有很高的重要性.现需要做以下工作： （1） 对题目提供的一个早期模型，评价其合理性和实用性. （2） 建立自己的模型，说明优于早期模型的原因；说明怎样才能建立一个真正能够预测以及能为预防和控制提供可靠、足够信息的模型，并指出这样做的困难；评价卫生部门采取的措施，如：提前和延后5天采取严格的隔离措施，估计对疫情传播的影响. （3） 根据题目提供的数据建立相应的数学模型，预测SARS 对社会经济的影响. （4） 给当地报刊写一篇通俗短文，说明建立传染病数学模型的重要性. 2．早期模型的分析与评价 题目要求建立SARS 的传播模型，整个工作的关键是建立真正能够预测以及能为预防和控制提供可靠、足够的信息的模型.如何结合可靠、足够这两个要求评价一个模型的合理性和实用性，首先需要明确： 合理性定义 要求模型的建立有根据，预测结果切合实际. 实用性定义 要求模型能全面模拟真实情况，以量化指标指导实际. 所以合理的模型能为预防和控制提供可靠的信息；实用的模型能为预防和控制提供足够的信息. 2.1早期模型简述 早期模型是一个SARS 疫情分析及疫情走势预测的模型， 该模型假定初始时刻的病例数为0N ， 平均每病人每天可传染K 个人（K 一般为小数），K 代表某种社会环境下一个病人传染他人的平均概率，与全社会的警觉程度、政府和公众采取的各种措施有关.整个模型的K 值从开始到高峰期间保持不变，高峰期后 10天的范围内K 值逐步被调整到比较小的值，然后又保持不变. 平均每个病人可以直接感染他人的时间为L 天.整个模型的L 一直被定为20.则在L 天之内，病例数目的增长随时间t (单位天)的关系是： t k N t N )1()(0+?= 考虑传染期限L 的作用后，变化将显著偏离指数律，增长速度会放慢.采用半模拟循环计算的办法，把到达L 天的病例从可以引发直接传染的基数中去掉. 2.2早期模型合理性评价 根据早期模型对北京疫情的分析与预测，其先将北京的病例起点定在3月1日，经过大约59天在4月29日左右达到高峰，然后通过拟合起点和4月20日以后的数据定出高峰期以前的K =0.13913.高峰期后的K 值按香港情况变化，即10天范围内K 值逐步被调整到0.0273.L 恒为20.由此画出北京3月1日至5月7日疫情发展趋势拟合图像以及5月7日以后的疫情发展趋势预测图像，如图1.

(完整)美赛一等奖经验总结,推荐文档

当我谈数学建模时我谈些什么——美赛一等奖经验总结 作者：彭子未 前言：2012 年3月28号晚，我知道了美赛成绩，一等奖（Meritorus Winner），没有太多的喜悦，只是感觉释怀，一年以来的努力总算有了回报。从国赛遗憾丢掉国奖，到美赛一等，这一路走来太多的不易，感谢我的家人、队友以及朋友的支持，没有你们，我无以为继。 这篇文章在美赛结束后就已经写好了，算是对自己建模心得体会的一个总结。现在成绩尘埃落定，我也有足够的自信把它贴出来，希望能够帮到各位对数模感兴趣的同学。 欢迎大家批评指正，欢迎与我交流，这样我们才都能进步。 个人背景：我2010年入学，所在的学校是广东省一所普通大学，今年大二，学工商管理专业，没学过编程。 学校组织参加过几届美赛，之前唯一的一个一等奖是三年前拿到的，那一队的主力师兄凭借这一奖项去了北卡罗来纳大学教堂山分校，学运筹学。今年再次拿到一等奖，我创了两个校记录：一是第一个在大二拿到数模美赛一等奖，二是第一个在文科专业拿数模美赛一等奖。我的数模历程如下： 2011.4 校内赛三等奖 2011.8 通过选拔参加暑期国赛培训（学校之前不允许大一学生参加） 2011.9 国赛广东省二等奖 2011.11 电工杯三等奖 2012.2 美赛一等奖（Meritorious Winner） 动机：我参加数学建模的动机比较单纯，完全是出于兴趣。我的专业是工商管理，没有学过编程，觉得没必要学。我所感兴趣的是模型本身，它的思想，它的内涵，它的发展过程、它的适用问题等等。我希望通过学习模型，能够更好的去理解一些现象，了解其中蕴含的数学机理。数学模型中包含着一种简洁的哲学，深刻而迷人。 当然获得荣誉方面的动机可定也有，谁不想拿奖呢？ 模型：数学模型的功能大致有三种：评价、优化、预测。几乎所有模型都是围绕这三种功能来做的。比如，今年美赛A题树叶分类属于评价模型，B题漂流露营安排则属于优化模型。 对于不同功能的模型有不同的方法，例如评价模型方法有层次分析、模糊综合评价、熵值法等；优化模型方法有启发式算法（模拟退火、遗传算法等）、仿真方法（蒙特卡洛、元

数学建模全国赛07年A题一等奖论文

关于中国人口增长趋势的研究 【摘要】 本文从中国的实际情况和人口增长的特点出发，针对中国未来人口的老龄化、出生人口性别比以及乡村人口城镇化等，提出了Logistic、灰色预测、动态模拟等方法进行建模预测。 首先，本文建立了Logistic阻滞增长模型，在最简单的假设下，依照中国人口的历史数据，运用线形最小二乘法对其进行拟合，对2007至2020年的人口数目进行了预测，得出在2015年时，中国人口有13.59亿。在此模型中，由于并没有考虑人口的年龄、出生人数男女比例等因素，只是粗略的进行了预测，所以只对中短期人口做了预测，理论上很好，实用性不强，有一定的局限性。 然后，为了减少人口的出生和死亡这些随机事件对预测的影响，本文建立了GM(1,1) 灰色预测模型，对2007至2050年的人口数目进行了预测，同时还用1990至2005年的人口数据对模型进行了误差检验，结果表明，此模型的精度较高，适合中长期的预测，得出2030年时，中国人口有14.135亿。与阻滞增长模型相同，本模型也没有考虑年龄一类的因素，只是做出了人口总数的预测，没有进一步深入。 为了对人口结构、男女比例、人口老龄化等作深入研究，本文利用动态模拟的方法建立模型三，并对数据作了如下处理：取平均消除异常值、对死亡率拟合、求出2001年市镇乡男女各年龄人口数目、城镇化水平拟合。在此基础上，预测出人口的峰值，适婚年龄的男女数量的差值，人口老龄化程度，城镇化水平，人口抚养比以及我国“人口红利”时期。在模型求解的过程中，还对政府部门提出了一些有针对性的建议。此模型可以对未来人口做出细致的预测，但是需要处理的数据量较大，并且对初始数据的准确性要求较高。接着，我们对对模型三进行了改进，考虑人为因素的作用，加入控制因子，使得所预测的结果更具有实际意义。 在灵敏度分析中，首先针对死亡率发展因子θ进行了灵敏度分析，发现人口数量对于θ的灵敏度并不高，然后对男女出生比例进行灵敏度分析得出其灵敏度系数为0.8850，最后对妇女生育率进行了灵敏度分析，发现在生育率在由低到高的变化过程中，其灵敏度在不断增大。 最后，本文对模型进行了评价，特别指出了各个模型的优缺点，同时也对模型进行了合理性分析，针对我国的人口情况给政府提出了建议。 关键字：Logistic模型灰色预测动态模拟 Compertz函数

关于组织江苏省优秀教学论文评比的通知

关于组织江苏省优秀教学论文评比的通知 Document number：BGCG-0857-BTDO-0089-2022

苏教研[2011] 19号 关于组织“2011年江苏省优秀教学论文评比”活 动的通知 各市教研室（教科院、教科研中心）： 为了提高我省中小学、幼儿园教师和教研员的教育教学理论水平，鼓励广大教师积极开展教育教学研究，总结、交流在课程教学实践中取得的宝贵经验，提升教育教学业务水平，进一步推动我省课程改革的深入实施，经研究决

定，拟组织“2011年江苏省优秀教学论文评比”活动。现将有关事项通知如下： 一、参评人员 教育行政管理人员、教研人员和广大教师。 二、参评学科 小学：英语、音乐、科学、综合实践活动 初中：语文、英语、物理、化学、生物、德育、音乐、综合实践活动 高中：语文、英语、物理、化学、德育、音乐、通用技术、综合实践活动其余学科的论文评比将在明年组织进行。 三、论文要求 1．论文应突出科学性、前瞻性、学科性、实践性；要立意新颖，观点明确，论证充分，给人启迪，对学科教学有较深刻的见解；能体现教育教学新理念、新探索、新成果；关注课改、评价改革的热点，关注学科教育教学的发展，对优化课堂教学，提高教学效益，有自己的思考与发现等。 2．参评论文题目自拟，内容自定，应注重理论思考和实践研究，属于工作总结或解题指导等文章不予评比。 3．引用他人原始资料的信息、观点、句子等应做标注，文责自负，坚决反对抄袭行为，一旦发现将取消参评资格，并通知该作者所在单位及所在市、县教研室。 4．论文字数在3000～5000字为宜。 四、报送要求 1．各大市在广泛发动的基础上，经评审后推荐小学每学科20篇，初中每学科20篇，高中每学科10篇论文（包括纸质稿一式1份和电子稿1份）送省参评。每位作者只能报送1篇论文。

论文网站大全

高速公路路径诱导策略研究 1.维普VIP密码 https://www.wendangku.net/doc/3f614402.html, 账号:nm531密码:131420 2.维普 https://www.wendangku.net/doc/3f614402.html,/ 帐号：nm531 密码：131420 3.免费万方入口 http://218.69.114.37/wf/cddb/cddbft.htm 4.比较好的ibrary https://www.wendangku.net/doc/3f614402.html,/libweb/elib/do/l ogin User Name:68-13313 Password:bigchalk 5.高权限ezproxy期刊 https://https://www.wendangku.net/doc/3f614402.html,/login 21976000002515 6.万方硕博论文全文 从1977年到2004年,免费下载 http://218.69.114.37/wf/cddb/cddbft.htm 7.国外硕博论文全文下载（这个可是重量级的！)

快速检索地https://www.wendangku.net/doc/3f614402.html,/theses/etd-search.h tml 按作者名检https://www.wendangku.net/doc/3f614402.html,/theses/browse/by_a uthor/ 按系（专业）检索https://www.wendangku.net/doc/3f614402.html,/theses/browse/by_d epartment https://www.wendangku.net/doc/3f614402.html,ki全库，非常好用 https://www.wendangku.net/doc/3f614402.html,/ 用户名：dx0031 密码：lhtsjy 9.一些可用的cnki全文数据库 https://www.wendangku.net/doc/3f614402.html,/index.htm 用户名及密码 sypbxy/sypbxy bjyyys/bjyyys K10129/gyzyjs hljhd/hljhd hun/sr2015 nj0084b/zjswdx

关于评选优秀论文的通知.doc

关于评选2011～2012学年度深圳大学优秀硕士论文的通知 各学院（部）： 为激励硕士研究生及其导师的创新意识和进取精神，进一步提高硕士研究生的培养质量，我院现开展2011～2012学年度深圳大学优秀硕士学位论文评选工作。 一、评选范围 2011年9月至2012年7月在我校完成硕士学位论文并取得硕士学位的学历硕士。 二、评选条件 1.选题为本学科前沿，有重要理论意义或实践意义； 2.在理论或方法上有创新，具有较好的社会效益或应用前景，或取得重要成果； 3.体现硕士研究生在本学科及相关领域掌握坚实的理论基础和系统的专门知识，材料详实，推理严密，数据可靠，表达准确，格式规范； 4.论文符合学术规范，无学术不端行为。 以下硕士学位论文不得参加评选：1.论文作者在硕士学位论文答辩前已获得副高级以上职称（含）；2.涉密论文。 三、评选时间 即日起各学院（部）根据通知要求组织评选工作，11月28日下午5点前将推荐材料(双面打印的申请表1份、论文1本及相关证明材料)报送我院,逾期不再受理。 四、评选名额：全校评选优秀硕士学位论文共计26篇。

五、评选程序 1.学生申请，填写《深圳大学优秀硕士学位论文申请表》（见附件1），导师审核同意，向所在学院（部）递交学位论文和有关申请材料。 2.学院审议并限额推荐（各学院推荐名额参见附件2，可空缺），各学位评定分委员会 表决确定向我院推荐的候选论文名单。 3.研究生院（筹）根据各学院（部）推荐材料，组织评审委员会评选优秀硕士学位论文。 4.名单公示。优秀硕士学位论文入选名单及其申请表在校内公示，任何单位或个人如发 现入选论文有剽窃、造假或者结论不成立等问题可在公示期内向我院反映。 5.优秀硕士学位论文名单报学校审定并公布。 六、表彰与奖励 1.按《深圳大学优秀硕士学位论文评选与奖励办法》予以奖励。 2.我校对应年度的省级优秀硕士学位论文推荐名单从校级优秀硕士学位论文中遴选产 生。 未尽事宜，参见《深圳大学优秀硕士学位论文评选与奖励办法》。如有疑问请致电：2616-0142，程老师。推荐材料报送地址：办公楼438。 深圳大学研究生院（筹） 二〇一二年十一月十二日

优秀教育类论文题目参考大全

优秀教育类论文题目参考大全 学前教育论文题目 1、教育研究要为教育创新做出更大的贡献 2、儿童数概念发展研究的新进展 3、幼儿问题意识概念的建构 4、保护幼儿的学习生态 5、对幼儿游戏规则的探讨--兼谈幼儿规则游戏 6、幼教百年沉思录(一) 7、伴随幼儿教育30年的历程 8、幼儿教师开展课题研究之分析、表述方法 9、儿童阅读障碍的生成与诊治研究综述 10、上海幼儿园利用家庭、社区德育资源的调查与思考 11、符合多元智能理论的教学活动设计 12、培养幼儿视觉--空间智能初探 13、浅谈分享阅读的几种风格 14、“不想像你那样做” 15、自制小沙锤(中班玩沙活动) 16、好玩的沙子(中班科学活动) 17、细细的沙(中班科学活动) 18、细细的沙粒(中班主题教育活动) 19、论信息时代幼儿教师的角色 20、试析幼儿教师专业化的特征及其实现途径 21、从关注文本到关注儿童 22、关于幼儿艺术教育若干问题的对话 23、以开放的心胸开发自我 24、美国幼儿园的节日教育活动 25、幼儿园课程编制的基本原理 26、“学前双语教育师资培训研究”课题开题会隆重召开 27、幼儿园教师的社会地位从哪里来 28、脑科学的新进展带给学前教育的启示 29、让幼儿在宽广的语境中积累审美经验--一种看待幼儿园艺术综合教育的新视角 30、这本书“女”一点 31、从现代认知心理学角度重新解读蒙太梭利教学法 32、1889～1949中国学前儿童教育大事记 33、辛勤耕耘六十载献身幼教半世情--访卢乐山教授 34、我自豪--我的青春属于孩子们 35、长得一模一样 36、“我爱卡通”活动设计的思考--回归生活视野下幼儿园艺术课程内容的选择 37、幼儿探究性活动特质的三维视界 38、幼儿教师开展探索型主题活动应具备的能力 39、儿童工作室的构建 40、种树者必培其根种德者必养其心--谈幼儿阶段道德意识的培养 41、解读儿童画 42、好东西大家一起吃

国赛优秀论文

B甲004 目录 摘要 (3) 关键词 (3) 一、系统方案 (3) 1.1、方案比较与论证 (3) 1.1.1、控制器模块 (3) 1.1.2、电机及驱动模块 (3) 1.1.3、测速模块 (4) 1.1.4、音频产生模块 (4) 1.1.5、无线收发模块 (4) 1.1.6、声音采集处理模块 (4) 1.2、最终方案 (4) 二、电路设计 (5) 2.1、系统组成 (5) 2.2、电动机驱动电路 (5) 2.3、行程测量模块 (5) 2.4、声光报警模块 (6) 2.5、周期性音频脉冲信号产生模块 (6) 2.6、无线收发模块设计 (6) 2.7、声音采集计算系统 (6) 三、软件设计 (7) 3.1、电机驱动部分流程图 (7) 3.2、主程序流程图 (7) 3.3单片机控制MMC-1芯片的程序 (7) 3.4无线接收模块程序 (7) 四、系统测试 (8) 4.1、测试仪器 (8) 4.2、调试 (8) 4.2.1 速度调试 (8) 4.2.2 功率放大测试 (8) 4.2.3 声源频率测试 (8) 4.2.4 声音接收测试 (8) 五、总结 (9) 5.1、结论 (9) 5.2、结束语 (9) 六、参考文献 (9) 七、附录 (9) 附录一、部分电路原理图 (9) 附录二、主程序流程图 (11) 附录三、部分程序附录 (13)

摘要： 本课题设计制作小组本着简单、准确、可靠、稳定、通用、性价比低的原则，采用STC89C52作为声源系统的控制核心，使用凌阳SPCE061A作为音频信号分析处理系统核心，应用电机控制ASSP芯片MMC-1驱动电机。本系统电路分为声源移动模块,声音产生模块，声音采集处理模块，无线控制模块和显示报警模块。声音收发和无线传输模块测量声源与声音接收器之间的距离，控制声源移动。首先测量声源S距A、B的距离差，距离差为零表示小车已运动到OX线，然后测量S距A、C的距离差，距离差为零表示小车寻找到W点。小车在OX线上运动时，利用S距A、B的距离差校正路线，同时声光报警，LCD液晶显示屏显示小车运行路程和时间。 关键词：STC89C52；电机控制芯片MMC-1；PT2262/2272无线收发；周期性音频脉冲信号；TEA2025B音频放大 一、系统方案 1.1方案比较与论证 根据题目要求，本系统主要由控制器模块、直流电机及其驱动模块、声音产生模块，声音采集处理模块和无线控制模块、声光报警模块等构成。为较好的实现各模块的功能，我们分别设计了几种方案并分别进行了论证。 1.1.1控制器模块 方案一：采用大规模可编程逻辑器件（如FPGA）作为系统的控制中心，目前，大规模可编程逻辑器件容量不断增大，速度不断提高，且多具有ISP功能，也可以在不改变硬件电路的情况下改变功能，但在本系统中，它的高处理功能得不到从分利用，还考虑到VHDL语言描述也没有单片机语言那么方便，所以这个方案不采用。 方案二：采用单片机STC89C52作为中心控制器。STC89C52单片机算数运算功能强，软件编程灵活，自由度大，具有超低功耗，抗干扰能力强等特点。还具有ISP在线编程功能，在改写单片机存储内部的程序时不需要将单片机从工作环境中取出，方便快捷。在后来的实验中我们发现，STC89C52精确度和运算速度也都完全符合我们系统的要求。故采用STC89C52单片机为我们整个系统的控制核心。 1.1.2 电机及驱动模块 采用电机控制ASSP 芯片MMC-1驱动（实物图如图1）。MMC-1为多通道两相四线式步进电机/直流电机控制芯片，基于NEC 电子16 位通用MCU（ PD78F1203）固化专用程序实现，支持UART 和SPI 串行接口。MMC-1 共有三个通道电机控制单元，通过设置寄存器可分别设置工作模式，实现不同功能。可以用来驱动直流电机和步进电机。 方案一：采用步进电机。步进电机是数字控制电机，不但控制精度高，而且简单可靠，但价格过高，重量大，占用端口资源多且控制复杂，不予采用。 2