Extra-large city flu decision-making method based on time-space flow of crowds with different economic attributes

Abstract

The invention relates to the technical field of flu prevention and control, in particular to an extra-large city flu decision-making method based on space-time flow of crowds with different economic attributes. The method comprises the following steps: establishing a meta population flow network model based on individuals, groups and space, and fusing a classical SEIRD model to construct an extra-large city influenza evolution model; and constructing a multi-objective optimization model which takes the best influenza propagation control effect and the minimum economic loss as objectives and takes regional sealing control and current limiting of six types of crowds with different economic attributes among regions as main decisions. According to the extra-large city influenza decision-making method based on the space-time flow of the crowds with different economic attributes, the economic attributes of the different crowds in the extra-large city are considered according to the importance and complexity of extra-large city influenza prevention and control, and an influenza evolution model based on the space-time flow of the crowds with different economic attributes is constructed; and carrying out a regional sealing control and different economic attribute crowd flow limiting integrated decision based on double targets of influenza propagation control and economic influence.

Description

Technical Field

[0001] The present invention relates to the technical field of influenza prevention and control, and in particular to a decision-making method for influenza in a megacity based on the spatiotemporal flow of people with different economic attributes. Background Art

[0002] In recent years, the global public health system has faced unprecedented challenges, and major public health emergencies have had a serious impact on the global social economy and public health security. Megacities, with their dense populations and frequent economic activities, have become high-risk areas for influenza transmission. Due to high population mobility and complex social structures, influenza transmission in megacities is obviously dense, clustered, superimposed and rapidly diffusive, exacerbating the chain, compound and complexity of influenza impacts.

[0003] The spread of influenza among people and regions is a typical spatiotemporal evolution phenomenon and a human-land interaction process. Influenza prevention and control is essentially a spatiotemporal problem of human-virus confrontation. Population mobility in megacities has promoted economic prosperity and development while also becoming a major route for influenza transmission. The economic development of megacities is highly dependent on the continuous flow of populations in different industries. It can be seen that intervention measures related to population mobility in the process of influenza prevention and control will inevitably have a severe impact on the economy. Therefore, how to effectively control the spread of influenza while minimizing the negative impact on the economy is an important practical issue that needs to be urgently addressed in influenza prevention and control management.

[0004] In summary, influenza evolution analysis based on infectious disease models and research on influenza prevention and control and intervention measures have been research hotspots in recent years. However, in terms of serving influenza prevention and control in megacities and conducting emergency management of major influenza outbreaks from the perspective of balancing influenza prevention and control with economic development, current research still has the following shortcomings.

[0005] (1) Most current influenza evolution models focus on the national or regional macro-scale, and these models usually assume that the individuals involved in the evolution are homogeneous and equivalent. Although these studies provide important perspectives for understanding the macro-transmission of influenza, few studies focus on the impact of population mobility and the complexity of social structure on influenza evolution in megacities, and thus fail to provide a scientific decision-making basis for the precise prevention and control of influenza in megacities.

[0006] (2) Most studies explore influenza prevention and control strategies from the perspectives of infection scale and economic losses. However, the infection scale is directly related to social stability, while economic losses are related to economic development. These two are two key and conflicting goals in influenza prevention and control. Currently, few studies have attempted to comprehensively consider these two goals in order to seek prevention and control strategies that balance the control of influenza transmission and minimize economic impact.

[0007] Therefore, a decision-making method for influenza in megacities based on the spatio-temporal mobility of people with different economic attributes is designed to provide another technical solution to the above technical problems. Summary of the Invention

[0008] Based on this, it is necessary to provide a decision-making method for influenza in megacities based on the spatio-temporal mobility of people with different economic attributes to solve the technical problems raised in the above background technology.

[0009] To solve the above technical problems, the present invention adopts the following technical solutions:

[0010] A decision-making method for influenza in megacities based on the spatio-temporal mobility of people with different economic attributes is as follows:

[0011] Establish a meta-population mobility network model based on individuals, groups, and space, and fuse the classic SEIRD model to construct an influenza evolution model for megacities;

[0012] Construct a multi-objective optimization model with the best influenza transmission control effect and the least economic loss as the objectives, and the main decisions are regional lockdown and flow restriction of six types of people with different economic attributes between regions;

[0013] Solve the multi-objective optimization model through the NSGA-II algorithm.

[0014] As a preferred implementation manner of the decision-making method for influenza in megacities based on the spatio-temporal mobility of people with different economic attributes provided by the present invention, the steps of the population mobility network model are as follows:

[0015] The population mobility network model is represented by a directed spatio-temporal network graph G=(V, A t );

[0016] Among them, V represents the node set, referring to each region that constitutes the urban space, and A t ={(i, j) t : i, j ∈ V} is the set of directed arcs at time t ∈ T, T is the decision-making cycle duration, any node is connected by various types of people flowing between nodes daily (t ∈ T), and if there is a flow of people between any two nodes, there is an arc connection between them;

[0017] Since there are six types of people flowing on any arc, let R={1, 2,..., 6} represent the set of people with different economic attributes, namely management and administrative support, retail, production and manufacturing, entertainment and leisure, service, and consumer groups, then

[0018] Among them, represents the number of people of the r ∈ R type flowing on the arc (i, j) ∈ A at time t ∈ T t ;

[0019] As a preferred embodiment of the method for making decisions on influenza in megacities based on the spatio-temporal mobility of people with different economic attributes provided by the present invention, a megacity influenza evolution model is constructed by integrating the classical SEIRD model. The steps are as follows:

[0020] Individuals at each node are classified into susceptible, latent, infected, recovered, and deceased according to their health status;

[0021] Susceptibles within each node are converted into latents according to the infection probability, and latents are transformed into infected after the incubation period; infected individuals recover and acquire immunity after the infection period; some infected individuals die due to the lethality of influenza.

[0022] As a preferred embodiment of the method for making decisions on influenza in megacities based on the spatio-temporal mobility of people with different economic attributes provided by the present invention, a multi-objective optimization model with the goal of influenza transmission control and economic loss and mainly making decisions on regional lockdown and flow restriction of six types of people with different economic attributes between regions is constructed. The steps are as follows:

[0023] The goal of influenza transmission control is to use the cumulative number of infected people within the prevention and control period T as an indicator to measure the effect of influenza transmission control. The expression is as follows:

[0024]

[0025] The economic loss target conducts economic loss measurement based on three parts: health economic loss, employment economic loss, and consumption loss;

[0026] The flow of urban population is restricted through medical resource capacity constraints, new infections constraints, population flow balance constraints, and network connectivity constraints.

[0027] As a preferred embodiment of the method for making decisions on influenza in megacities based on the spatio-temporal mobility of people with different economic attributes provided by the present invention, the economic loss target conducts economic loss measurement based on three parts: health economic loss, employment economic loss, and consumption loss. The steps are as follows:

[0028] The health economic loss is the economic burden and loss caused by influenza. The expression is as follows:

[0029]

[0030] Among them, the economic loss directly brought by infection at time t Define the unit case rehabilitation treatment cost as θ 1 , the unit case death treatment cost as θ 2 , and the opportunity loss cost caused by death as θ 3 , and are the number of infected and deceased individuals belonging to node i at time t, respectively;

[0031] The employment economic loss consists of employment remuneration loss and production loss, and the expression is as follows:

[0032]

[0033] where and are the number of the r-th type of population traveling from node i to node j at the initial moment of influenza evolution and at time t, respectively, and w r is the average salary of industry r, and p r is the per capita industrial production value of industry r; the cumulative remuneration loss of the employed population at time t and the cumulative production loss of the industry

[0034] The employment economic loss at time t

[0035] The consumption loss mainly refers to the potential consumption loss caused by the reduction in supply due to the decrease in the employed population and the consumption volume that has not occurred due to the direct reduction of the consumer population The expression is as follows:

[0036]

[0037] where and are the number of the r-th type of population traveling from node i to node j at the initial moment of influenza evolution and at time t, respectively, and c pc is the per capita consumption volume, v c is the economic value generated per unit of consumption volume, and p r is the per capita industrial production value of industry r;

[0038] The consumption loss at time t

[0039] The objective of minimizing economic loss Z Economy has the following expression:

[0040]

[0041] where represents that the flow ratio of the r-th type of population on any arc at time t ∈ T is l (l ∈ L), represents the open or sealed state of node i ∈ V at time t ∈ T; represents that the node is open; otherwise, it is closed;

[0042] The objective function can be expressed as:

[0043] Z = min(Z Infect , Z Economy ).

[0044] As a preferred embodiment of the method for making decisions on influenza in megacities based on the spatio-temporal mobility of people with different economic attributes provided by the present invention, the multi-objective optimization model is solved by the NSGA-II algorithm, and the steps are as follows:

[0045] Generate an initial population through a three-stage construction algorithm;

[0046] Successively apply selection, crossover, and mutation genetic operations to continuously generate offspring populations;

[0047] Use a dynamic elite strategy based on non-dominated sorting and crowding distance to conduct selection of new populations to maintain population diversity and superiority, and obtain Pareto optimal solutions.

[0048] As a preferred embodiment of the method for making decisions on influenza in megacities based on the spatio-temporal mobility of people with different economic attributes provided by the present invention, an initial population is generated through a three-stage construction algorithm, and the steps are as follows:

[0049] S1: Open all nodes, randomly assign flow ratios from the set L of population flow ratios for 6 types of people between any two nodes, and thus obtain the seed solution S 0 ;

[0050] S2: Perturb the seed solution S by adjusting the node open / closed states and population flow ratios 0 to generate an enlarged initial population with 2N solutions;

[0051] When constructing each solution, first randomly select 0.1|V| nodes, change their states from open to closed; then, randomly select 0.5|V| population flow arcs and randomly perturb the population flow ratios on these flow arcs;

[0052] S3: Perform fast non-dominated sorting on the individuals in the population according to the fitness values and adopt an elite selection strategy to select the top N individuals to form the initial population.

[0053] As a preferred embodiment of the method for making decisions on influenza in megacities based on the spatio-temporal mobility of people with different economic attributes provided by the present invention, selection, crossover, and mutation genetic operations are successively applied to continuously generate offspring populations, and the steps are as follows:

[0054] Adopt a pairwise tournament selection method to select the parent individuals for crossover and mutation operations;

[0055] For two parental individuals undergoing crossover operation, single-point crossover is performed based on their two chromosome strings C and F respectively.

[0056] Perturb the values of randomly selected gene positions to form a mutation operation, which is used to increase the diversity of the population.

[0057] It can be seen without doubt that through the above technical solutions of this application, the technical problems to be solved by this application can surely be solved.

[0058] Meanwhile, through the above technical solutions, the present invention has at least the following beneficial effects:

[0059] 1. The influenza decision-making method for megacities based on the spatio-temporal mobility of people with different economic attributes provided by the present invention, aiming at the importance and complexity of influenza prevention and control in megacities, considering the economic attributes of different populations in megacities, constructs an influenza evolution model based on the spatio-temporal mobility of people with different economic attributes, and conducts integrated decision-making on regional lockdown and flow restriction of people with different economic attributes under the dual objectives of influenza transmission control and economic impact.

[0060] 2. The present invention aims at the impact of the spatio-temporal mobility of people with different economic attributes on influenza evolution and economic development in megacities, establishes a meta-population mobility network model based on individuals, groups and space, and constructs an influenza evolution model for megacities by integrating the classical SEIRD model; secondly, aiming at the conflict between the two objectives of influenza transmission and economic loss, proposes a multi-objective optimization model with the minimization of influenza transmission and economic loss as the objective to carry out research on the integrated decision-making optimization of regional lockdown and flow restriction of different populations, and designs an improved NSGA-II algorithm to solve the model in combination with the problem characteristics; finally, takes the influenza prevention and control in the main urban area of City A during the influenza period as a case background to verify the effectiveness of the influenza evolution model and the prevention and control decision-making model, and obtains several management implications beneficial to influenza prevention and control in megacities through the sensitivity analysis of key parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0062] Figure 1 It is a schematic diagram of influenza evolution based on the population mobility network with different economic attributes of the present invention;

[0063] Figure 2 It is the chromosome coding scheme of the present invention;

[0064] Figure 3 It is a schematic diagram of the selection of the new population of the present invention;

[0065] Figure 4 Schematic diagram of the crossover operation of the present invention;

[0066] Figure 5 Schematic diagram of the mutation operation of the present invention;

[0067] Figure 6 Schematic diagram for comparing the simulated values and statistical values during the evolution period of the present invention;

[0068] Figure 7 Thermal map of the distribution of infected cases in each region of the present invention;

[0069] Figure 8 Schematic diagram of the model operation result of the present invention;

[0070] Figure 9 Schematic diagram of the population flow network and the lockdown intensity of each node under different schemes of the present invention;

[0071] Figure 10 Schematic diagram for comparing the Pareto solution sets under different virus infection rates of the present invention;

[0072] Figure 11 Schematic diagram for comparing the Pareto solution sets under different decision-making periods of the present invention;

[0073] Figure 12 Schematic diagram of the changes in economic losses and the number of infected people during the evolution period of the present invention;

[0074] Figure 13 Schematic diagram of the sensitivity analysis of the proportion of different population types of the present invention. Detailed implementation manners

[0075] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0076] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0077] It should be noted that, without conflict, the embodiments in the present invention and the features and technical solutions in the embodiments can be combined with each other.

[0078] It should be noted that similar reference numerals and letters denote similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0079] Reference Figures 1 - 13 A decision-making method for influenza in megacities based on the spatio-temporal mobility of people with different economic attributes

[0080] 1. Model construction

[0081] 1.1 Problem-solving

[0082] The main industries that make up urban economic activities are divided into five categories: management and administrative support (scientific research, public management, etc.), retail (wholesale, retail, etc.), production and manufacturing (manufacturing, construction, etc.), entertainment and leisure (culture, sports and entertainment, etc.), and services (leasing, business, etc.). The above five types of productive economic activity populations, together with the consumer population, jointly constitute six types of populations in urban economic activities. Different types of groups contribute differently to the economy, mainly reflected in industry output, wage remuneration, and consumption expenditure, that is, they have different economic attributes

[0083] According to administrative divisions and industrial characteristics, the interior of a megacity is divided into a series of sub-regions (streets). The urban economic activities drive the flow of people with different economic attributes between different regions. In the influenza environment, individuals with different health states (susceptibles, infectives, and latents, etc.) within each region change their health states through contact behaviors under the drive of economic activities in the current cycle, and return to the destination through the "departure-destination" association to participate in the evolution of the next cycle. A schematic diagram of the influenza evolution based on the population mobility network with different economic attributes is as Figure 1 shown

[0084] During the influenza prevention and control process, regional lockdowns and restrictions on population movement between regions inevitably bring employment economic losses due to a reduction in the employed population, consumption losses due to a reduction in the consumer population, and health economic losses caused by infections while achieving influenza control

[0085] 1.2 Population mobility network model

[0086] Based on the above analysis, first construct a meta-population mobility network model based on individuals, groups, and space to capture the dynamic changes of different groups in different decision-making states. The population mobility network model can be expressed as a directed spatio-temporal network graph G=(V,A t ). Among them, V represents the set of nodes, referring to each region (street, township, or urban functional area) that makes up the urban space; A t ={(i,j) t :i,j∈V} is the set of directed arcs at time t∈T, T is the duration of the decision-making cycle, and any node is connected to each other through various types of people flowing between nodes daily (t∈T). If there is a flow of people between any two nodes, there is an arc connection between them There is a flow of people Since there are six types of population flows on any arc, let \(R = \{1, 2, \ldots, 6\}\) represent the set of populations with different economic attributes, namely management and administrative support, retail, production and manufacturing, entertainment and leisure, service, and consumer groups. Then we have Among them, represents the number of the \(r\in R\) - type population flow on the arc \((i, j)\in A\) at time \(t\in T\). t

[0087] 1.3 Influenza Evolution Model

[0088] Based on the classical SEIRD model, the influenza evolution within each node at time \(t\in T\) is carried out, and at time \(t + 1\), the reaction and diffusion of the infection are carried out through the population flow network, thereby realizing the spatio - temporal dynamic evolution of the whole - domain influenza within the period, providing a basic support for influenza control and economic loss measurement.

[0089] According to the SEIRD model, the individuals in each node are classified into susceptible (S), exposed (E), infected (I), recovered (R), and deceased (D) according to their health status. The susceptible individuals in each node are converted into exposed individuals according to the infection probability, and the exposed individuals turn into infected individuals after the incubation period; the infected individuals recover and acquire immunity after the infection period; some infected individuals die due to the lethality of influenza. Without loss of generality, the following assumptions are made for the evolution process:

[0090] (1) Once an infected individual is discovered, it is immediately controlled and does not follow the territorial management. Only exposed and susceptible individuals will flow between regions and participate in virus transmission;

[0091] (2) Since the influenza evolution cycle is relatively short, the impact of newly born and deceased populations on influenza evolution within the decision - making cycle is not considered.

[0092] (3) Influenza is only transmitted through direct contact, and other possible transmission routes are not considered.

[0093] Based on the above assumptions, for the influenza evolution within each node \(i\in V\), the following symbols, parameters, and state variables are defined: The number of susceptible individuals belonging to node \(i\) at time \(t\) and present at node \(j\);

[0094] The number of exposed individuals belonging to node \(i\) at time \(t\) and present at node \(j\);

[0095] The number of infected individuals belonging to node \(i\) at time \(t\) and present at node \(j\);

[0096] ​The number of recovered individuals who were at node j at time t and belong to node i;

[0097] The number of deceased individuals who belong to node i at time t;

[0098] The total number of individuals at node i at time t;

[0099] The total number of susceptible individuals at node i at time t;

[0100] The total number of latent individuals at node i at time t;

[0101] The total number of infected individuals at node i at time t;

[0102] The total number of recovered individuals at node i at time t;

[0103] The total number of deceased individuals at node i at time t;

[0104] The total number of susceptible individuals migrating into node i at time t;

[0105] The total number of susceptible individuals migrating out of node i at time t;

[0106] The total number of latent individuals migrating into node i at time t;

[0107] The total number of latent individuals migrating out of node i at time t;

[0108] β i : The infection probability of susceptible individuals at node i coming into contact with infected individuals;

[0109] δ i : The probability of latent individuals at node i converting to confirmed cases;

[0110] γ i : The recovery rate of infected individuals at node i;

[0111] μ i : The mortality rate of infected individuals at node i.

[0112] Based on the above parameter definitions, the state transition equation of the influenza evolution model SERID for node i in each region at time t + 1 is as follows:

[0113]

[0114] Among them, Formulas (1) and (2) indicate that the susceptible and latent populations participating in the evolution of each node are composed of the populations remaining locally and migrating from other nodes; Formulas (3) and (4) indicate that the susceptible and latent populations after the evolution of each node return to their respective nodes; Formula (5) represents the change in the number of susceptible individuals, where represents the number of infected individuals; Formula (6) represents the change in the number of latent individuals, where represents the number of confirmed infected individuals; Formula (7) represents the change in the number of infected individuals, where and respectively represent the number of individuals transitioning from the infected state to the recovered and death compartments; Formulas (8) and (9) respectively represent the changes in the number of individuals in the recovered and death compartments.

[0115] Applying the above node-oriented influenza evolution model to the entire population mobility network model within the evolution period T, the global spatio-temporal evolution state of influenza can be obtained.

[0116] 1.4 Multi-objective optimization model

[0117] Based on the above analysis, comprehensively considering the main realistic constraints of influenza prevention and control management and ensuring the normal operation of society in the influenza environment, a multi-objective optimization model is constructed with the best influenza transmission control effect and the minimum economic loss as the objectives, and regional lockdown and flow restriction of six types of populations with different economic attributes between regions as the main decisions.

[0118] 1.4.1 Decision variables

[0119] There are mainly two types of decisions: regional lockdown and population flow restriction decisions. The regional lockdown decision refers to regulating the closed and open states of regions by selectively closing or opening some nodes in the population mobility network. The flow restriction decision refers to restricting the flow quantity of six types of populations between any two nodes in the population mobility network. Based on the requirements of population flow restriction at different risk levels in the actual emergency management process, the flow restriction is reflected in different levels of population flow ratios, and the set of population flow ratios is denoted as L. Therefore, the two types of decision variables are defined as follows:

[0120] The open and closed state of node i ∈ V at time t ∈ T. Indicates that the node is open; otherwise, the node is closed.

[0121] At time t ∈ T, for any arc the flow ratio of the r-th type of population is l (l ∈ L),

[0122] 1.4.2 Objective functions

[0123] The two objectives are the best influenza transmission control effect and the minimum economic loss.

[0124] (1) Influenza transmission control objective

[0125] Taking the cumulative number of infected people within the prevention and control period T as an index to measure the effect of influenza transmission control, the calculation formula is shown in formula (10).

[0126]

[0127] Among them, is the total number of infected people of node ii at time t.

[0128] (2) Economic loss objective

[0129] Drawing on the economic loss quantification method in the influenza environment proposed in the literature, the economic loss measurement is carried out from three parts: health economic loss, employment economic loss and consumption loss. The measurement methods of each part are as follows.

[0130] ① Health economic loss. Health economic loss refers to the economic burden and loss caused by influenza, including not only the direct costs incurred for treating and managing diseases, but also the indirect economic losses caused by diseases or deaths. The health economic loss of the present invention mainly includes the treatment costs of patients, the treatment costs of dead patients and the opportunity losses brought about by individual deaths. Since the costs of treating and recovering from infections and treating and dying from infections are different, the unit case recovery treatment cost is defined as θ 1 , the unit case death treatment cost is θ 2 , and the opportunity loss cost caused by death is θ 3 . It can be obtained that the economic loss directly brought about by infection at time t The calculation formula is as follows:

[0131]

[0132] Among them, the economic loss directly brought about by infection at time t The unit case recovery treatment cost is defined as θ 1 , the unit case death treatment cost is θ 2 , and the opportunity loss cost caused by death is θ 3 , and are the number of infected people and dead people belonging to node i at time t, respectively.

[0133] ② Employment economic loss. Employment economic loss mainly includes two aspects. One is the loss of employment remuneration caused by the reduction of the employed population due to the lockdown and flow restriction policies; the other is the production loss caused by the lack of personnel in the work positions. Define as the population quantity at the initial moment of influenza evolution, w r is the average salary of industry r, pr is the per capita industrial production value of industry r. Then, the cumulative remuneration loss of the employed population at time t and the cumulative production loss of the industry are calculated as follows:

[0134]

[0135] where and are the numbers of the r-th type of people traveling from node i to node j at the initial moment of influenza evolution and at time t, respectively. w r is the average salary of industry r, and p r is the per capita industrial production value of industry r; the cumulative remuneration loss of the employed population at time t and the cumulative production loss of the industry

[0136] Therefore, the employment economic loss at time t

[0137] ③ Consumption loss. In the present invention, the consumption loss mainly refers to the potential consumption loss caused by the reduction in supply due to the decrease in the employed population and the consumption volume that does not occur due to the direct reduction of the consumer population Define as the consumption volume of node i at time t, c pc as the per capita consumption volume, and v c as the economic value generated by unit consumption volume. Then and are calculated as follows:

[0138]

[0139] where and are the numbers of the r-th type of people traveling from node i to node j at the initial moment of influenza evolution and at time t, respectively, is the consumption volume of node i at time t, c pc is the per capita consumption volume, v c is the economic value generated by unit consumption volume, and p r is the per capita industrial production value of industry r;

[0140] Therefore, the consumption loss at time t

[0141] Based on the above analysis, the economic loss minimization objective Z Economy can be expressed as:

[0142]

[0143] where Denote the proportion of the flow of the r-th type of population on any arc at time t∈T as l (l∈L). Denote the opening and lockdown status of node i∈V at time t∈T; Denote that the node is open, otherwise the node is closed;

[0144] The objective function of the research problem can be expressed as:

[0145] Z = min(Z Infect , Z Economy ) (17)

[0146] 3. Constraints

[0147] (1) Medical resource capacity constraint

[0148] In the actual influenza prevention and control process, the medical resource capacity is limited, and the severity of the infected patients is different. Ensuring that the treatment needs of patients are met is a necessary condition for maintaining social stability in influenza prevention and control. Therefore, the model considers the constraint based on the limited medical resource admission capacity, which is mainly reflected in the number of hospital beds and the number of ICUs in each node (region).

[0149]

[0150] Among them, is the total number of infected people at node i at time t, and ξ represents the proportion of severe patients among the infected. Constraint formulas (18) and (19) respectively represent that the number of mild patients and severe patients in each region at any time t∈T of the infected does not exceed the total number of hospital beds and the number of ICUs.

[0151] (2) Newly added infected person constraint

[0152] To effectively contain the spread of influenza, it is necessary to control the newly added number of infections within a reasonable and controllable range. That is, the number of infected people is net decreased compared with the previous time period within a certain time period. Therefore, the following constraint is defined:

[0153]

[0154] Among them, is the total number of infected people at node i at time t, and k is the evolution time period. It should be noted that formula (20) does not constrain that the number of infected people in all nodes must achieve a net decrease synchronously, but allows some nodes to maintain a certain infection level under the overall downward state of infection. By allowing a certain infection rate in local areas within the overall control framework, the relationship between influenza prevention and control and economic activities can be balanced more flexibly, and a more comprehensive and sustainable influenza management can be achieved.​

[0155] (3) Population flow balance constraint

[0156] At any time t, the population flow of a node should satisfy the in - out balance. Based on assumption (1), that is, once an infected person is diagnosed, they will be quarantined on the spot, the population flow of a node should satisfy the following constraint: the number of susceptible and latent individuals flowing into the node at the beginning should be equal to the number of susceptible and latent individuals flowing out at the beginning, plus the total number of diagnosed individuals staying in this node. Thus, the following balance constraint is defined:

[0157]

[0158] where I ji is the number of infected individuals moving from node j to node i.

[0159] (4) Network connectivity constraint

[0160] Since lockdown and traffic restriction affect the normal operation of the city and will cause social instability due to public resistance. Therefore, social stability needs to be considered during the influenza prevention and control process, and the operation of the core functions of the city should be ensured. To this end, network connectivity constraints are introduced to support the basic operation of key urban services.

[0161] First, define the connectivity matrix of the population movement network G=(V, A t ) at time t as M t , and its element is expressed as follows:

[0162]

[0163] When , it means the two nodes are connected, then otherwise they are not connected. Define |M t | as the node scale of the largest connected sub - graph of network G at time t, and S as the set of nodes of the connected sub - graph. The specific calculation formula is:

[0164]

[0165] During the influenza prevention and control management process, the population movement network at any time t should have a certain degree of fluidity to support the basic operation of the city. The following constraint is made:

[0166] |M t |≥ε|V| (24)

[0167] where ε is the minimum connectivity requirement to maintain the operation of the city, and |V| is the number of nodes in the population movement network.

[0168] 2. Algorithm solution design

[0169] 2.1 Algorithm framework

[0170] Since there is an obvious conflict between the two goals of influenza transmission control and economic loss, the proposed optimization model belongs to a typical multi-objective optimization problem. Deb et al. [Deb K,Pratap A,Agarwal S,et al.A fast and elitist multiobjective genetic algorithm:NSGA-II[J].IEEE transactions on evolutionary computation,2002,6(2):182-197] The NSGA-II (Nondominated Sorting Genetic Algorithm II) proposed by Deb et al. in 2002 is an effective algorithm for solving multi-objective optimization problems. With its fast non-dominated sorting method and crowding degree calculation strategy, it performs excellently in dealing with multi-objective optimization problems and has received wide attention and application. Therefore, the present invention designs an improved non-dominated sorting genetic algorithm (NSGA-II) based on the characteristics of the research problem to solve the model.

[0171] The overall process of the algorithm is as follows: Generate an initial population through a designed three-stage construction algorithm, continuously generate offspring populations by applying genetic operations such as selection, crossover, and mutation in sequence, and use a dynamic elite strategy based on non-dominated sorting and crowding distance to conduct new population selection to maintain population diversity and superiority, so as to gradually approach the Pareto optimal solution. During the algorithm evolution process, to improve the optimization efficiency, after each crossover and mutation operation, a local search program based on the perturbation adaptation of the crossover and mutation solution space is designed to enhance the quality of the offspring solutions.

[0172] 2.2 Encoding Scheme

[0173] Encoding the solution is the first and crucial step of the genetic algorithm. The effectiveness of the encoding scheme directly affects the solving complexity and efficiency of the algorithm. According to the problem characteristics, an encoding scheme consisting of double chromosome strings is designed.

[0174] The chromosome string C is a two-layer integer list with a length of |V|, representing the node lockdown decision. The first layer stores the area index, and each gene value in the second layer represents the lockdown state of the corresponding node. 1 indicates that the node is open, and 0 indicates that the node is locked down. The chromosome string F is a floating-point number list with a length of |R|×|V'|×|V'|, and each gene bit stores the flow ratio of the corresponding population category between any two nodes, representing the flow-limiting decision of various populations between regions. Where V'∈V represents the set of open nodes, obtained from the chromosome string C. Figure 1 The chromosome expression corresponding to the scheme is as Figure 2 shown. It can be seen that regions 1 to 6 are open, and the lockdown state in the chromosome string C is 1. Regions 7 to 9 are closed, and the corresponding lockdown state is 0. Since regions 1 to 6 are open. For any node, there are |R| types of population flows to 6 nodes (including the node itself), that is, any node corresponds to |R|×6 elements to store the flow-limiting decision. Therefore, the length of the chromosome string F is |R|×6×6.

[0175] 2.3 Initializing the population

[0176] Considering the influence of the quality of the initial population on the convergence and global search ability of the genetic algorithm, the present invention designs a three-stage initial population construction algorithm of "first constructing the seed solution - then constructing the enlarged initial population - and finally selecting the population" to construct an initial population with a population size of N. The specific steps are as follows:

[0177] Stage 1: Constructing the seed solution. Open all nodes, and randomly assign the flow ratios for the 6 types of people between any two nodes from the set L of population flow ratios, thereby obtaining the seed solution S 0 .

[0178] Stage 2: Constructing the enlarged initial population. Perturb the seed solution S by adjusting the node open / closed states and the population flow ratios 0 to generate an enlarged initial population with 2N solutions. When constructing each solution, first randomly select 0.1|V| nodes and change their states from open to closed; then, randomly select 0.5|V| population flow arcs and randomly perturb the population flow ratios on these flow arcs.

[0179] Stage 3: Population selection. Perform fast non-dominated sorting on the individuals in the population according to the fitness values and adopt the elitist selection strategy to select the first N individuals to form the initial population.

[0180] 2.4 Fitness evaluation

[0181] Since both objectives of the decision-making model are minimum functions, the objective function is used as the fitness function to evaluate the quality of each individual. According to the multi-objective optimization model, the objective function values Z Infect and Z Economy of each individual are used as the individual fitness values.

[0182] 2.5 Fast non-dominated sorting

[0183] For the combined population composed of the parent population and the offspring population during the algorithm evolution process, according to the individual fitness values Z Infect and Z Economy , the individuals in the population are graded according to non-dominated sorting. The specific steps are as follows:

[0184] Step 1: Calculating the domination relationship. For any pair of individuals (p i , p j ) in the population, compare the fitness values. If for all objectives, f(p i ) ≤ f(p j ) and at least for one objective f(p i ) < f(p j ), then p i dominates p j denoted as pi <p j And add p j to the dominating set S i of p i , increasing the domination number n j of p j .

[0185] Step 2: Divide the fronts. Starting from the first front set F 1 , gradually divide the subsequent front sets. For each individual p i in the first front, for each individual p j in its dominating set, reduce n j by 1. Put each individual with n j reduced to 0 into the next front.

[0186] Step 3: Repeat the above operations until all individuals are ranked.

[0187] 2.6 Crowding degree calculation

[0188] After obtaining the non-dominated ranks of the population individuals, further calculate the crowding degree of each individual j in each non-dominated rank F First, sort the individuals in each rank separately in ascending order according to the fitness values Z Infect and Z Economy . Secondly, accumulate the crowding degrees under each objective respectively. The specific calculation formula is as follows:

[0189]

[0190] In the formula, f k i+1 and f k i-1 respectively represent the objective values of the adjacent individuals of individual i sorted according to the k-th objective in the non-dominated rank of F j , f k max and f k min respectively represent the maximum and minimum values in the non-dominated rank of F j . |F j | is the number of individuals in this non-dominated rank. For the boundary solutions, set the crowding degree to the maximum value M.

[0191] 2.7 Dynamic elite population selection

[0192] Considering that non-optimal solutions help to expand the optimization space of the algorithm, and in order to better balance the concentration and diversity of the algorithm search, a dynamic elite population selection strategy based on the Pareto front is designed to effectively manage the population to improve the algorithm search efficiency. Specifically, for the combined population composed of the parent population and the offspring population, when the new population selection condition is met, first, all individuals in the population are non-dominated sorted to obtain the Pareto front set F = {F 1 , F 2 , …, F m}, then the individuals within any Pareto front set F i ∈F are sorted by crowding degree, and finally, the top n i individuals are dynamically selected within this front set to form a new population according to the elite strategy, where the selection schematic diagram is as shown in Figure 3 .

[0193] Initially, the number of individuals selected from each Pareto front set into the new population is equal, that is As the number of algorithm iterations increases, within a certain range, the number of individuals selected from the first front is continuously increased by the step size τ, and at the same time, the number of individuals selected from other fronts is equivalently reduced to ensure the consistency of the number of individuals in the new population. The calculation formula for the number of individuals selected within each Pareto front is as follows:

[0194]

[0195] where iter is the number of iterations, is the maximum number of individuals selected from the first Pareto front.

[0196] It can be seen that different from the traditional elite strategy that preferentially selects elite solutions from the front fronts, the dynamic elite strategy proposed by the present invention enables the algorithm to include individuals from different Pareto fronts during the search process, enriching the diversity of solutions; at the same time, as the evolutionary search deepens, better individuals have a higher selection probability, enabling the algorithm to have better concentration in the later stage. Thus, the unity of algorithm diversity and concentration is achieved through this strategy.

[0197] 2.8 Selection, Crossover, and Mutation Operations

[0198] (1) Selection Operation

[0199] The pairwise tournament selection method is adopted to select the parent individuals for crossover and mutation operations. The individuals in the tournament group are compared based on the non-dominated sorting rank and the crowding degree distance. When the non-dominated ranks are different, the individual with the lower rank is retained; when the non-dominated ranks are the same, the individual with the larger crowding degree distance is retained.

[0200] (2) Crossover operation

[0201] For the two parent individuals undergoing the crossover operation, single-point crossover is performed based on their two chromosome strings C and F respectively. During the operation, a crossover point is randomly selected. Offspring 1 inherits the first half of Parent 1 and the second half of Parent 2, while Offspring 2 does the opposite, thus forming new gene combinations. Since chromosome strings C and F correspond to node lockdown and population flow ratio decisions respectively, there is a significant dependency between the two decisions. Therefore, during crossover, first crossover chromosome string C, and then update the nodes with a population flow limit value of 0 in chromosome string F according to the crossover result, and then perform the crossover operation on chromosome string F. The schematic diagram of the crossover operation is as Figure 4 shown;

[0202] (3) Mutation operation

[0203] The mutation operation increases the diversity of the population by perturbing the values of randomly selected gene positions. During the operation, the mutation operation is performed on chromosome strings C and F of the individuals participating in the mutation in turn. Specifically, for chromosome string C, 1 - 3 gene positions are randomly selected, and the values of the gene positions are swapped between 0 and 1, that is, changed from 0 to 1 or from 1 to 0; then, for chromosome string F, 1 - 3 gene positions are randomly selected, and the population flow level is reselected from the population flow level set L. The schematic diagram of the mutation operation is as Figure 5 shown;

[0204] 2.9 Local search algorithm

[0205] To improve the optimization performance of the algorithm, the local search algorithm is used to perform post-optimization on the new solutions generated by each crossover and mutation to improve the search efficiency and solution quality of the algorithm. The core idea of the local search algorithm is to conduct a fine-grained exploration of the population flow level changes in the solution space perturbed by crossover and mutation, thereby improving the local quality of the new individuals and enhancing the operation efficiency of the algorithm by reducing the randomness of crossover and mutation. The pseudo-code of the algorithm is as follows:

[0206]

[0207]

[0208] 3 Experimental simulation

[0209] To verify the effectiveness of the proposed model, the present invention takes the influenza prevention and control in 156 townships and sub-districts of 9 districts and counties in the main city of City A as the case background. Taking November 1, 2022 as the initial moment of model operation and December 10, 2022 as the end moment, a simulation analysis is carried out. The experimental environment is as follows: The model is programmed and implemented using Python 3.11.2, with a computer processor of 12th Gen Intel(R) Core(TM) i7-12700, a CPU main frequency of 2.10 GHz, 16.0 G of RAM, and Windows 10 Professional Edition.

[0210] 3.1 Parameter Settings

[0211] The relevant parameters involved in the experimental simulation mainly include three parts: influenza evolution model parameters, optimization model parameters, and algorithm parameters.

[0212] For the influenza evolution model parameters, according to relevant research [周林,黄鹏,代应,等.区域互救与外部救助耦合的突发性疫情初期应急物资协同调度研究[J].中国管理科学,2024,32(08):297-307] The settings are as follows: β = 0.41, δ = 0.4, γ = 0.01, μ = 0.004, ξ = 0.1, k = 10, ε = 0.9.

[0213] For the decision model parameters, according to the economic statistics bulletin of City A in 2021 and the statistical yearbook in 2022, the average monthly salary of five types of people and the industrial production value are classified and statistically analyzed. The specific values are as follows: The average monthly salary per person corresponding to management and administrative support, retail, production and manufacturing, entertainment and leisure, and service industries are 11,142.25 yuan, 6,249.66 yuan, 8,022.42 yuan, 8,913.8 yuan, and 6,685.35 yuan respectively, and the corresponding industrial production values are 16,713.38 yuan, 9,374.49 yuan, 12,033.63 yuan, 13,370.7 yuan, and 10,028.03 yuan respectively. The average monthly consumption economic value per person is 2,487.5 yuan. Set the treatment cost and opportunity loss cost for each infected person and deceased person, with θ 1 = 20,016 yuan, θ 2 = 300,000 yuan, θ 3 = 438,599.1 yuan. Considering the actual strictness of lockdown and flow restriction, the population flow level is set as L = {0, 0.25, 0.5, 0.75, 1}.

[0214] For the algorithm parameters, through a large number of preliminary experiments, the parameter values are: population size N = 100, the crossover rate and mutation rate are 0.8 and 0.05 respectively, Iter_LS = 1.5|V|, τ = 0.05, n 1 f = 0.85N, and the algorithm termination condition is that the number of times the best solution does not improve is Iter_noimp = 20.

[0215] 3.2 Influenza Evolution Simulation

[0216] To verify the applicability and effectiveness of the proposed influenza evolution model based on the population mobility network, first, under the scenario of no lockdown and flow restriction decisions, a simulation analysis of influenza evolution in the main 9 districts of City A was carried out. During the evolution period, the comparison between the simulated values and the actual statistical values of the cumulative number of infected people in the main 9 districts of City A is as Figure 6 shown.

[0217] From Figure 6 it can be seen that during the evolution period, the simulated values obtained by the evolution model generally show the same growth trend as the statistical values. From the perspective of the cumulative number of infected people, the number of infected people obtained by the evolution model is 140,130 cases, and the officially reported confirmed cases are 132,587, with an error of 5.6%. Considering that in the influenza environment at that time, there were certain missed detections and some mild or asymptomatic infected people were not detected due to technical reasons, the actual number of infected people would be higher than the reported confirmed number. Therefore, it shows that the influenza evolution model proposed by the present invention can better simulate the overall development of influenza.

[0218] Furthermore, from Figure 7 the heat map of the distribution of infected cases given, it can be seen that there is also good consistency between the simulated values and the statistical values of the number of infected people in most regions, indicating that the proposed influenza evolution model also has good accuracy in the influenza evolution of each region of the city. Among the 9 regions, the areas with the most severe influenza are Yuzhong District and Yubei District (the simulated values are 22,334 cases and 23,414 cases respectively, and the statistical values are 22,271 cases and 23,093 cases respectively). These two regions are densely populated and have frequent traffic, with characteristics of high population aggregation and frequent population spatio-temporal mobility. It shows that high population density and population mobility play an important role in the spread of influenza.

[0219] 3.3 Results of multi-objective influenza control

[0220] Based on the effectiveness of the evolution model, the NSGA-II algorithm was run to solve the multi-objective optimization model, and the obtained Pareto front is as Figure 8 shown. The Pareto front shows the influence relationship between the two groups of objectives. It can be seen that the two objectives of the number of infected people and economic loss conflict with each other. The plan with a smaller cumulative number of infected people corresponds to a larger economic loss, and vice versa. The economic loss is between 44 billion yuan and 65 billion yuan, and the number of infected people varies between 94,800 cases and 99,000 cases. Overall, it shows a trend from a high number of infected people and low economic loss to a low number of infected people and high economic loss.

[0221] 3.4 Comparative analysis of different prevention and control strategies

[0222] To analyze the effect of the multi-objective optimization model on influenza control, three solutions were selected from the Pareto optimal solution set (denoted as Pareto Solution 1, 2, and 3 respectively), and compared with the baseline scenario without control measures (no control solution) and the solutions under two single-objective optimization strategies (the solution with minimum economic loss and the solution with minimum number of infections). The comparison of the above solutions in terms of five key indicators, namely economic loss, number of infections, reduction in the number of workers, reduction in the number of consumers, and the number of lockdown nodes, is shown in Table 1. To visually display the lockdown and flow-limiting situations under different solutions, the lockdown intensity is defined as the ratio of the flow-limiting number of people from this node to other nodes to the normal state. Figure 9 The urban population flow network and the lockdown intensity of each node under Pareto Solution 1, the solution with minimum economic loss, and the solution with minimum number of infections are given.

[0223] Table 1 Comparison of solutions under different control strategies

[0224]

[0225] As can be seen from Table 1, Pareto Solution 1 significantly outperforms the no-control solution in terms of the two key indicators of economic loss and number of infections, indicating that the multi-objective control strategy can obtain a decision-making solution that is comprehensively better in both objectives. Further, compared with the single-objective solutions, compared with the extremely high number of infections caused by minimizing economic loss and the sharp increase in economic loss caused by the solution with minimum number of infections, the three Pareto solutions achieve a more balanced trade-off between the two objectives. At the same time, since the Pareto frontier solutions have a certain non-dominated range in the two objective values, it provides a flexible decision-making space for decision-makers, facilitating decision-makers to select the most suitable decision-making solution according to the actual situation and preferences.

[0226] 3.5 Parameter sensitivity analysis

[0227] (1) Sensitivity analysis of infection rate

[0228] To reflect the impact of the virus infection rate on influenza prevention and control, sensitivity analysis of the virus infection rate was carried out by taking β = 0.41, 0.51, and 0.61 respectively. The simulation results of the multi-objective control model and the no-control scenario under the three infection rates are compared as Figure 10 shown.

[0229] From Figure 10 it can be clearly seen that the infection scale and economic loss increase significantly with the increase of the virus infection rate, and there are significant differences in the influenza control effects under different infection rates. As the virus infection rate increases, the control effect of the multi-objective optimization model is more prominent than that of the no-control scenario. It shows that the more severe the influenza, the more significant the effect of the multi-objective optimization model.

[0230] (2) Sensitivity analysis of decision-making cycle

[0231] To explore the impact of different lockdown and flow-limiting decision-making cycles on influenza control, with other parameters remaining unchanged, the decision-making cycles for lockdown and flow-limiting are set to 7 days, 14 days, 28 days, and 40 days respectively. The operation results of the multi-objective optimization model under the four decision-making cycles are as Figure 11 shown.

[0232] By comparing the influenza control effects under different decision-making cycles, it can be seen that shorter decision-making cycles help significantly reduce economic losses and the number of infected people, indicating that refined lockdown and flow-limiting decisions can respond promptly to the changes in influenza, thus bringing better influenza prevention and control effects. In addition, to further visually display the changes in the two objectives under different decision-making cycles, Figure 12 the changes in the two objectives within the evolution cycle are compared for two scenarios with decision-making cycles of 7 days and 40 days. It can be seen that compared with a decision-making cycle of 40 days, the effect is more significant after the 20th day with a decision-making cycle of 7 days.

[0233] (3) Sensitivity analysis of the proportion of population types

[0234] To explore the impact of the changes in the number of people with different economic attributes on influenza prevention and control and economic losses, the number of five types of working populations is adjusted step by step by 2% within the range of [0, 14%], and the reduction amount of this type of population is increased proportionally to the other four types of populations. Since the total number of infected people remains unchanged under the condition that the total number of people and the population activity rules remain unchanged, the economic loss results under different population quantity changes are as Figure 13 shown.

[0235] It can be seen that there are significant differences in the economic impact of the reduction of personnel in different industries, and the impact effect has a strong positive correlation with the per capita industrial output value of each industry. The higher the per capita output value, the greater the economic loss.

[0236] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the present invention to the specific embodiments described. Obviously, according to the content of this specification, many modifications and changes can be made. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the present invention, so that those skilled in the art in the relevant technical field can understand and utilize the present invention well. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A decision-making method for influenza in megacities based on the spatiotemporal flow of people with different economic attributes, characterized by: Here are the steps: Establish a meta-population flow network model based on individuals, groups and space, and integrate the classic SEIRD model to construct a megacity influenza evolution model; A multi-objective optimization model was constructed with the goal of achieving the best influenza transmission control effect and the minimum economic loss, with regional lockdown and flow control of six groups of people with different economic attributes between regions as the main decision-making; The multi-objective optimization model is solved by NSGA-II algorithm.

2. The megacity influenza decision-making method based on the spatiotemporal flow of people with different economic attributes according to claim 1 is characterized in that: The population flow network model has the following steps: Through the directed space-time network graph G = (V, A t ) represents the population mobility network model; Among them, V represents the node set, which refers to the various areas that constitute the urban space, and A t ={(i,j) t :i,j∈V} is a directed arc set at time t∈T, T is the duration of the decision cycle, and any nodes are connected to each other through various types of people flowing between nodes every day (t∈T). If there is a flow of people between any two nodes, there is an arc connection between them; Since there are six types of crowd flows on any arc, R = {1, 2, ..., 6} represents the collection of people with different economic attributes, namely management and administrative support industry, retail industry, production and manufacturing industry, entertainment and leisure industry, service industry and consumer industry, then there is in, Denotes arc (i, j)∈A at time t∈T t The number of people flowing in the r∈R category above.

3. The megacity influenza decision-making method based on the spatiotemporal flow of people with different economic attributes according to claim 2 is characterized in that: The classic SEIRD model is integrated to construct a megacity influenza evolution model. The steps are as follows: The individuals at each node are divided into susceptible, latent, infected, recovered and dead according to their health status; The susceptible persons in each node are converted into latent persons according to the infection probability, and the latent persons are converted into infected persons after the incubation period; the infected persons recover and gain immunity after the infection period; some infected persons die due to the mortality rate of influenza.

4. The megacity influenza decision-making method based on the spatiotemporal flow of people with different economic attributes according to claim 1 is characterized in that: A multi-objective optimization model was constructed with the goal of controlling influenza transmission and economic losses, and with regional lockdown and flow control of six groups of people with different economic attributes between regions as the main decision-making. The steps are as follows: The influenza transmission control target is to use the cumulative number of infections within the prevention and control period T as an indicator to measure the effectiveness of influenza transmission control. The expression is as follows: The economic loss target is calculated based on three parts: health economic loss, employment economic loss and consumption loss; The flow of urban population is limited through constraints on medical resource capacity, new infections, population flow balance and network connectivity.

5. The megacity influenza decision-making method based on the spatiotemporal flow of people with different economic attributes according to claim 4 is characterized in that: The economic loss target is calculated based on the economic loss of health, economic loss of employment and consumption loss. The steps are as follows: Health economic loss refers to the economic burden and loss caused by influenza, expressed as follows: Among them, the economic loss directly caused by infection at time t Define the cost of rehabilitation treatment per case as θ1, the cost of death treatment per case as θ2, and the opportunity loss cost caused by death as θ3. and are the number of infected and dead people belonging to node i at time t, respectively; The economic loss of employment is composed of the loss of employment compensation and the loss of production, and the expression is as follows: in, and are the number of people in the rth category going from node i to node j at the initial moment of influenza evolution and at time t, respectively, and w r is the average salary of industry r, p r is the per capita industry production value of industry r; the cumulative compensation loss of employed population at time t and industry cumulative production losses Employment economic loss at time t Consumption loss mainly refers to the potential consumption loss caused by supply reduction due to the reduction of employment population. and the amount of consumption that did not occur due to a direct reduction in the number of consumers The expression is as follows: in, and are the number of people of type r going from node i to node j at the initial moment of influenza evolution and at time t, is the consumption of node i at time t, c pc is the per capita consumption, v c is the economic value generated per unit consumption, p r is the per capita industry production value of industry r; Consumption loss at time t Minimum economic loss target Z Economy The expression is as follows: in, Denotes any arc at time t∈T The flow ratio of the rth group of people is l(l∈L), Indicates the open and closed state of node i∈V at time t∈T; Indicates that the node is open; otherwise, the node is closed; The objective function is expressed as: Z=min(Z Infect ,WITH Economy )。 6. The megacity influenza decision-making method based on the spatiotemporal flow of people with different economic attributes according to claim 1 is characterized in that: The multi-objective optimization model is solved by the NSGA-II algorithm. The steps are as follows: The initial population is generated through a three-stage construction algorithm; The selection, crossover and mutation genetic operations are applied sequentially to continuously generate the offspring population; A dynamic elite strategy based on non-dominated sorting and crowding distance is used to select new populations to maintain population diversity and superiority, and the Pareto optimal solution is obtained.

7. The megacity influenza decision-making method based on the spatiotemporal flow of people with different economic attributes according to claim 6 is characterized in that: The initial population is generated through a three-stage construction algorithm. The steps are as follows: S1: Open all nodes and randomly assign flow ratios from the flow ratio set L to the six types of people between any two nodes, thereby obtaining the seed solution S0; S2: By adjusting the node on / off status and the crowd flow ratio, the seed solution S0 is disturbed to generate an expanded initial population with 2N solutions; When constructing each solution, we first randomly select 0.1|V| nodes and change their states from open to closed; then, we randomly select 0.5|V| crowd flow arcs and randomly perturb the crowd flow ratio on the flow arcs; S3: Perform fast non-dominated sorting on the individuals in the population according to their fitness values ​​and adopt an elite selection strategy to select the first N individuals to form the initial population.

8. The megacity influenza decision-making method based on the spatiotemporal flow of people with different economic attributes according to claim 6 is characterized in that: The selection, crossover and mutation genetic operations are applied in sequence to continuously generate the offspring population. The steps are as follows: The parent individuals selection for crossover and mutation operations is carried out using the pairwise tournament selection method; For the two parent individuals undergoing the crossover operation, single-point crossover is performed based on their two chromosome strings C and F respectively; The randomly selected gene bit values ​​are disturbed to form a mutation operation to increase the diversity of the population.

Images