Public facility reachability evaluation method and system based on geographical multi-agent

Abstract

The invention discloses a public facility reachability evaluation method and system based on geographic multiple agents, and belongs to the field of urban calculation and geographic information. According to the method, a multi-agent system comprising a user agent and a facility agent is firstly constructed, and then a dynamic utility function combined with an evolutionary game theory is introduced to simulate the dynamic decision and congestion avoidance behavior of a user; and user strategies are balanced through multiple rounds of iterative evolution, so that a macroscopic reachability pattern emerging from a microcosmic game is obtained. According to the method, starting from the microscopic individual level, by simulating the dynamic game process of a large number of agents, the defects of a static calculation model are overcome, and dynamic evaluation and predictive simulation of reachability of public facilities are achieved. Compared with the prior art, the evaluation result is more accurate, dynamic feedback of the real world can be reflected, and a scientific decision support tool is provided for urban planning.

Description

Technical Field

[0001] This invention belongs to the field of geographic information and urban computing, and particularly relates to a method and system for assessing the accessibility of public facilities based on geographic multi-agent systems. Background Technology

[0002] Accessibility of public facilities is a core indicator for measuring the rationality of urban and rural planning and assessing the level of equalization of public services. It is of great significance for optimizing the allocation of urban resources and improving the quality of life of residents.

[0003] Currently, the most widely used method in the field of public facility accessibility assessment is the Two-Step Floating Catchment Area (2SFCA) method and its various improved methods, such as Enhanced Two-Step Floating Catchment Area (E2SFCA) and Three-Step Floating Catchment Area (3SFCA). These methods typically use a geographical unit (such as a community or transit zone) as the basic unit and measure accessibility by calculating the ratio between the supply capacity of facilities and the demand of the population within a specific search radius (or time threshold).

[0004] However, the assessment methods for accessibility of public facilities still have the following shortcomings: (1) Existing methods homogenize all users within a geographical unit and treat them as a whole without individual differences for macro-scale assessment calculation. This ignores the individual heterogeneity of users and cannot reflect the autonomous decision-making behavior of micro-individuals in the real world, resulting in deviations between the assessment results and the actual situation, and generating aggregation errors; (2) Existing methods are based on a static supply and demand model and cannot reflect that the public service system is a complex evolutionary system formed by the dynamic choices of a large number of users. For example, it cannot simulate the dynamic feedback process when a popular facility generates a "crowding effect" due to too many users choosing it, and other users will turn to other facilities; (3) In areas where the service range of facilities overlaps, the demand of users in the overlapping area will be repeatedly calculated by all facilities within their reach. This will artificially exaggerate the total demand, resulting in a systematic underestimation of the accessibility level; (4) It lacks scenario simulation and decision support capabilities. Due to its static calculation nature, existing methods are difficult to use as "deductive" planning tools. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing public facility accessibility assessment methods, which only focus on the macro level, employ static calculation models, and fail to simulate the dynamic decision-making behavior of individual users. This invention provides a geographic multi-agent-based public facility accessibility assessment method and system. By constructing a multi-agent system comprising user agents and facility agents, this method simulates and extrapolates the macro-accessibility patterns emerging from the micro-level choices of numerous users under specific planning scenarios. It can dynamically assess public facility accessibility from the individual level by simulating the dynamic game process of a large number of users.

[0006] To achieve the above-mentioned objectives, the present invention specifically adopts the following technical solution:

[0007] In a first aspect, the present invention provides a method for assessing the accessibility of public facilities based on geographic multi-agent systems, which includes the following steps:

[0008] S1. Obtain user distribution data and public facility data within the research area. After presetting the service demand rate corresponding to the public facilities under the target category, calculate the number of users with service demand based on the user distribution data. Then, define users as user intelligent agents with independent decision-making ability and preset individual preference attributes, so that the distribution of user intelligent agents corresponds to the number of users with service demand. Define public facilities as facility intelligent agents with service quality level, maximum service capacity and geographical location attributes, so that the distribution of facility intelligent agents corresponds to the public facility data. In this way, a multi-agent system for simulating user behavior in choosing public facilities is constructed.

[0009] S2. For each user agent in the multi-agent system, construct a set of its neighboring interaction objects as the neighboring interaction object set of that user agent.

[0010] S3. For the user agent set in S1, construct a dynamic utility function based on evolutionary game theory to evaluate the expected benefits after choosing different facility agents.

[0011] S4. Within the preset iteration period, each iteration uses the dynamic utility function as the decision basis, designs the neighbor interaction rules based on the set of neighbor interaction objects, and follows the principle of evolutionary game theory. By simulating the exploration behavior of user agents and the learning behavior based on the neighbor interaction rules, the probability distribution of each user agent choosing different facility agents is dynamically updated.

[0012] S5. Continuously iterate S4 until the maximum number of iterations required for the iteration cycle is reached, then terminate the iteration and consider the multi-agent system to have reached a convergence state.

[0013] S6. After the state of the multi-agent system converges, obtain the final probability distribution of each user agent choosing different facility agents. Calculate the total service demand carried by each facility agent by summarizing the final probability distribution, and calculate the supply-demand ratio by combining the maximum service capacity attribute of each facility agent.

[0014] S7. For each user agent, based on the supply-demand ratio and the spatial attenuation relationship between the user agent and each facility agent, generate the public facility accessibility evaluation value for that user agent. Summarize the public facility accessibility evaluation values ​​of all user agents to form the public facility accessibility evaluation value of the final user group and visualize it, generating an accessibility distribution heat map covering the entire study area.

[0015] Based on the above scheme, each step can be implemented in the following preferred manner.

[0016] As a preferred embodiment of the first aspect mentioned above, in step S1, the acquired user distribution data includes the user community name, the number of user communities, the number of users in each user community, and the latitude and longitude coordinates representing the user's location; the acquired public facility data includes the public facility name, the latitude and longitude coordinates representing the public facility's location, the service quality level, and the maximum service capacity.

[0017] As a preferred option in the first aspect mentioned above, the method for calculating the number of users required for the service in each user community is as follows:

[0018] S11. Calculate the number of users in each user community by dividing the number of users in the study area by the number of user communities in the study area.

[0019] S12. Multiply the number of users in each user community by the service demand rate to calculate the number of users in each user community who require service.

[0020] As a preferred embodiment of the first aspect, in step S1, the individual preference attributes of the user agent are divided into multiple levels, each corresponding to a different preference coefficient value. The preference coefficient value is used to influence the service quality score of the user agent's perception facility agent when constructing the dynamic utility function in step S3.

[0021] As a preferred embodiment of the first aspect mentioned above, the specific method for constructing a set of neighboring interactive objects for the user's intelligent agent in step S2 is as follows:

[0022] S21. Taking the user community where the user agent is located as the center, select the user community closest to the center as the candidate community. The number of candidate communities is the same as the preset number of neighboring communities, and the candidate communities include the user community where the user agent is located.

[0023] S22. Take the user community where the user agent is located as the target community, and select all user communities whose distance from the target community is no greater than a preset distance threshold from the candidate communities obtained in S21, to form a neighboring community set.

[0024] S23. Other user agents located within the neighboring community set are considered as the neighboring interaction objects of this user agent, and all neighboring interaction objects constitute the neighboring interaction object set of this user agent.

[0025] As a preferred embodiment of the first aspect mentioned above, in step S3, the dynamic utility function comprehensively considers the service quality score of the facility agent, the geographical distance cost, and the dynamic congestion cost; wherein, the service quality score is obtained by multiplying the facility agent's level and the user agent's preference coefficient value, and the dynamic congestion cost is proportional to the sum of the probabilities of all user agents who select the facility agent.

[0026] Furthermore, in step S3, the first Dynamic utility function of a user agent The specific calculation method is as follows:

[0027]

[0028]

[0029]

[0030]

[0031] In the formula, Indicates the number of facility agents. Indicates the first The user agent selects the first... The probability of a facility's intelligent agent; , , These are the preset service quality weighting coefficient, geographical distance cost weighting coefficient, and dynamic congestion cost weighting coefficient; Indicates the first The first user intelligent agent perceived the first Service quality score of each public facility intelligent agent; Indicates the first The level of each facility's intelligent agent, Indicates the first The preference coefficient values ​​of each user agent; Indicates by the first The user agent to the first The geographical distance cost resulting from the commuting distance of each facility agent; Indicates dynamic congestion cost; This indicates that all user agents choose the first... The sum of probabilities of each facility agent; Indicates the first The maximum reception capacity of each facility's intelligent agent is determined by the first... The sum of user intelligent agents that a facility intelligent agent can receive is calculated; This indicates the number of user agents.

[0032] As a preferred embodiment of the first aspect mentioned above, in step S4, when the number of iterations is... When the user agent's exploration behavior is triggered by a hyperparameter that decays exponentially with the number of iterations, the probability distribution of the user agent's current selection of different facility agents will be replaced with a newly generated random probability distribution.

[0033] Furthermore, in step S4, the user agent's exploratory behavior is determined by hyperparameters. Triggered, its number of iterations increases The increase is exponentially decaying, and the specific calculation method is as follows:

[0034]

[0035] In the formula, The initial exploration probability, The attenuation rate, This represents the number of iteration rounds.

[0036] As a preferred approach to the first aspect mentioned above, the design method for the learning behavior of a user intelligent agent is as follows:

[0037] S41. Randomly select a user agent as the imitation target from the set of neighboring interaction objects of the user agent, calculate the dynamic utility function of the user agent and the imitation target in the current iteration, and take the difference between the two dynamic utility functions as the expected benefit utility difference.

[0038] S42. Calculate the imitation probability based on the obtained difference in expected benefit utility, and have the user agent apply the imitation probability to adopt the probability distribution of the imitation target in order to update its own probability distribution; wherein, the imitation probability is used to represent the proportion of the user agent adopting the probability distribution of the imitation target, and the imitation probability value is negatively correlated with the absolute value of the difference in expected benefit utility.

[0039] Furthermore, in step S42, the imitation probability The calculation method is as follows:

[0040]

[0041] In the formula, Indicates parameters of individual irrationality; This represents an exponential function with the natural constant as its base.

[0042] As a preferred embodiment of the first aspect above, in step S6, the supply-demand ratio of a facility agent is calculated as follows: the sum of the probabilities of all user agents choosing the facility agent is obtained from the final probability distribution and used as the total service demand of the facility agent. The total service demand of the facility agent is divided by the maximum reception capacity to obtain the supply-demand ratio of the facility agent.

[0043] As a preferred embodiment of the first aspect above, in step S7, the specific steps for generating the corresponding public facility accessibility evaluation value for a user agent are as follows: using the Gaussian distance decay function as a spatial decay relationship model, the supply and demand ratios of all facility agents within the user agent's interaction range are weighted and summed to obtain the public facility accessibility evaluation value for the user agent.

[0044] As a preferred embodiment of the first aspect mentioned above, in step S7, the accessibility evaluation value of public facilities for each user agent is spatially interpolated and visualized to generate an accessibility distribution heatmap covering the entire study area.

[0045] Secondly, the present invention provides a public facility accessibility assessment system based on geographic multi-agent systems, comprising:

[0046] The data acquisition module is used to acquire user distribution data and public facility data within the area to be evaluated;

[0047] The assessment module is used to generate an accessibility distribution heat map covering the area to be assessed, based on user distribution data and public facility data within the area to be assessed, according to the public facility accessibility assessment method described in any of the schemes of the first aspect above.

[0048] Thirdly, the present invention provides a computer electronic device, which includes a memory and a processor;

[0049] The memory is used to store computer programs;

[0050] The processor is configured to, when executing the computer program, implement the public facility accessibility assessment method as described in any of the embodiments of the first aspect above.

[0051] Compared with the prior art, the present invention has the following advantages when using the method of the present invention to assess the accessibility of public facilities:

[0052] This invention constructs each user as a user agent with independent decision-making capabilities and individual preference attributes, simulating differentiated individual choice behaviors in the real world. Its evaluation results represent the macro-emergence of micro-games, effectively avoiding aggregation errors and making the evaluation results closer to the realities of the physical world. By introducing a dynamic utility function and iterative update mechanism based on evolutionary game theory, this invention treats the public service system as a dynamically evolving complex system, simulating the dynamic feedback process where users actively avoid and switch to other public facilities after a facility experiences a "crowding effect" due to excessive user selection. Through an innovative probabilistic demand calculation method, this invention ensures that the total demand of each user is calculated exactly once throughout the system, solving the problem of duplicate demand calculation and making the evaluation results more fair and reasonable. Furthermore, this invention supports scenario simulation and verification, allowing planners to quantitatively extrapolate the potential impact of different policies by adjusting model parameters, providing scientific decision support for urban planning and the optimal allocation of public resources. Attached Figure Description

[0053] Figure 1 This is a flowchart of the steps of the method of the present invention;

[0054] Figure 2 A schematic diagram of the user intelligent agent interaction relationship designed for the method of the present invention;

[0055] Figure 3 This is a system block diagram of the present invention;

[0056] Figure 4 This is a schematic diagram of a computer electronic device provided by the present invention. Detailed Implementation

[0057] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be practiced in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below. Technical features in the various embodiments of the present invention can be combined accordingly without mutual conflict.

[0058] In the description of this invention, it should be understood that the terms "first" and "second" are used only for descriptive purposes and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined with "first" and "second" may explicitly or implicitly include at least one of those features.

[0059] like Figure 1As shown, in a preferred embodiment of the present invention, the above-mentioned geographical multi-agent-based public facility accessibility assessment method includes the following steps S1 to S7. The specific implementation process of each step will be described in detail below.

[0060] S1. Obtain user distribution data and public facility data within the research area. After presetting the service demand rate corresponding to the public facilities under the target category, calculate the number of users with service demand based on the user distribution data. Then, define users as user intelligent agents with independent decision-making ability and preset individual preference attributes, so that the distribution of user intelligent agents corresponds to the number of users with service demand. Define public facilities as facility intelligent agents with service quality level, maximum service capacity and geographical location attributes, so that the distribution of facility intelligent agents corresponds to the public facility data. In this way, a multi-agent system for simulating user behavior in choosing public facilities is constructed.

[0061] It should be noted that in step S1 of this invention, the individual preference attribute refers to the heterogeneous decision-making patterns exhibited by different individuals in the real world when choosing public facilities due to differences in their values, risk perception, and cost sensitivity. The individual preference attribute is the core attribute used to quantify this inherent, subjective decision-making tendency in the model. Service quality level is a quantitative indicator used to characterize the service level that the public facility agent can provide; it is related to real-world attributes such as the scale, level, and professional capabilities of the public facility. Maximum service capacity is an attribute used to characterize the maximum number of users that the public facility agent can serve per unit of time. Geographic location attribute describes the location of the public facility in geographic space, usually represented by latitude and longitude coordinates; this attribute is the basis for calculating the geographical distance cost between users and public facilities.

[0062] It should be noted that in step S1 of this invention, the user distribution data obtained includes the user community name, the number of user communities, the number of users in each user community, and the latitude and longitude coordinates representing the user's location; the public facility data obtained includes the public facility name, the latitude and longitude coordinates representing the public facility's location, the service quality level, and the maximum service capacity.

[0063] It should be noted that, in step S1 of this invention, the method for calculating the number of users requiring service in each user community is as follows:

[0064] S11. Calculate the number of users in each user community by dividing the number of users in the study area by the number of user communities in the study area.

[0065] S12. Multiply the number of users in each user community by the service demand rate to calculate the number of users in each user community who require service.

[0066] It should be noted that in step S1 of this invention, the individual preference attributes of the user agent are divided into multiple levels, each corresponding to a different preference coefficient value. The preference coefficient value is used to affect the service quality score of the user agent's perception facility agent when constructing the dynamic utility function in S3.

[0067] It should be noted that in step S1 of this embodiment, the individual preference attributes of the user agent are divided into three levels: low, medium, and high, and the proportion of users at each level to the total number of users is respectively... , and And assign different preference coefficient values ​​to them respectively. , and ;in, , and These represent the proportions of users at the low, medium, and high levels out of the total number of users; , and These represent the preference coefficient values ​​for the low, medium, and high levels, respectively.

[0068] S2. For each user agent in the multi-agent system, construct a set of its neighboring interaction objects as the neighboring interaction object set of that user agent.

[0069] It should be noted that, in step S2 of this invention, the specific method for constructing a set of neighboring interactive objects for a user intelligent agent is as follows:

[0070] S21. Taking the user community where the user agent is located as the center, select the user community closest to the center as the candidate community. The number of candidate communities is the same as the preset number of neighboring communities, and the candidate communities include the user community where the user agent is located.

[0071] S22. Take the user community where the user agent is located as the target community, and filter out the candidate communities obtained in S21 that are no more than a preset distance threshold from the target community. All user communities form a set of neighboring communities;

[0072] S23. Other user agents located within the neighboring community set are considered as the neighboring interaction objects of this user agent, and all neighboring interaction objects constitute the neighboring interaction object set of this user agent.

[0073] S3. For the user agent set in S1, construct a dynamic utility function based on evolutionary game theory to evaluate the expected benefits after choosing different facility agents.

[0074] It should be noted that in step S3 of this invention, the dynamic utility function comprehensively considers the service quality score of the facility agent, the geographical distance cost, and the dynamic congestion cost; wherein, the service quality score is obtained by multiplying the level of the facility agent and the preference coefficient value of the user agent, and the dynamic congestion cost is proportional to the sum of the probabilities of all user agents who select the facility agent.

[0075] It should be noted that in step S3 of this invention, the first... Dynamic utility function of a user agent The specific calculation method is as follows:

[0076]

[0077]

[0078]

[0079]

[0080] In the formula, Indicates the number of facility agents. Indicates the first The user agent selects the first... The probability of a facility's intelligent agent; , , These are the preset service quality weighting coefficient, geographical distance cost weighting coefficient, and dynamic congestion cost weighting coefficient; Indicates the first The first user intelligent agent perceived the first Service quality score of each public facility intelligent agent; Indicates the first The level of each facility's intelligent agent, Indicates the first The preference coefficient values ​​of each user agent; Indicates by the first The user agent to the first The geographical distance cost resulting from the commuting distance of each facility agent; Indicates dynamic congestion cost; This indicates that all user agents choose the first... The sum of probabilities of each facility agent; Indicates the first The maximum reception capacity of each facility's intelligent agent is determined by the first... The sum of user intelligent agents that a facility intelligent agent can receive is calculated; This indicates the number of user agents.

[0081] In step S3 of this embodiment, the construction process of the dynamic utility function is as follows: First, a policy space is defined for each user agent. The policy space represents the set of policies that the user agent can decide autonomously. In this embodiment, it is defined as a policy vector representing the user's behavior of choosing public facilities. Furthermore, the sum of all elements in this vector is always equal to 1. The specific calculation method is as follows:

[0082]

[0083]

[0084] In the formula, Representing the first The user agent selects the first... The probability of a facility agent.

[0085] Furthermore, this embodiment constructs the aforementioned dynamic utility function based on the policy vector. Used to evaluate the user agent in the policy vector The expected total return is shown in the formula above, and will not be repeated here.

[0086] It should be noted that the dynamic congestion cost in the above dynamic utility function reflects the first... The overall load of each facility agent is affected by the choices made by all user agents within the system, and is determined by the choices made by all user agents. The sum of probabilities of each facility agent and the The sum of user intelligent bodies received by each facility's intelligent body Decision. This invention uses the policy selection of all agents as... The input established a dynamic feedback mechanism of "crowding-avoidance" to reflect the dynamic facility selection behavior of intelligent agents.

[0087] S4. Within the preset iteration period, each iteration uses the dynamic utility function as the decision basis, designs the neighbor interaction rules based on the set of neighbor interaction objects, and follows the principles of evolutionary game theory. By simulating the exploration behavior of user agents and the learning behavior based on the neighbor interaction rules, the probability distribution of each user agent choosing different facility agents is dynamically updated.

[0088] It should be noted that in step S4 of this invention, when the number of iterations is... When the user agent's exploration behavior is triggered by a hyperparameter that decays exponentially with the number of iterations, the probability distribution of the user agent's current selection of different facility agents will be replaced with a newly generated random probability distribution.

[0089] In step S4 of this embodiment, the exploratory behavior of the user agent is determined by hyperparameters. Triggered, its number of iterations increases The increase is exponentially decaying, and the specific calculation method is as follows:

[0090]

[0091] In the formula, The initial exploration probability, The attenuation rate, This represents the number of iteration rounds.

[0092] In the above formula, hyperparameters The number of iteration rounds is determined The study explores the trade-off between exploratory and learning behaviors, and encourages agents to conduct extensive global policy space searches in the early stages of iteration through an exponential decay function, thus avoiding premature entrapment in local optima.

[0093] It should be noted that in step S4 of this invention, when the number of iterations is... At times, such as Figure 2 As shown, the design method for the learning behavior of a user agent is as follows:

[0094] S41. Randomly select a user agent from the set of neighboring interactive objects of the user agent. As a target for imitation, the user agent in the current iteration is calculated respectively. and the target of imitation The dynamic utility function is used, and the difference between the two dynamic utility functions is taken as the expected return utility difference. ;in, Indicates the number of iteration rounds. Imitation target The dynamic utility function, Indicates the number of iteration rounds. Time User Intelligent Agent The dynamic utility function;

[0095] S42. Calculate the imitation probability based on the obtained difference in expected benefit utility, and have the user agent apply the imitation probability to adopt the probability distribution of the imitation target in order to update its own probability distribution; wherein, the imitation probability is used to represent the proportion of the user agent adopting the probability distribution of the imitation target, and the imitation probability value is negatively correlated with the absolute value of the difference in expected benefit utility.

[0096] It should be noted that in this embodiment S42, the imitation probability The calculation method is as follows:

[0097]

[0098] In the formula, the individual irrational parameter Used to indicate the degree of irrationality of an individual; This represents an exponential function with the natural constant as its base.

[0099] In the above formula, the individual irrationality parameter The smaller the value, the more rational the user agent, and the less likely it is to imitate a user agent with worse performance. The larger the dynamic utility function, the higher the probability that the user agent will imitate a strategy with lower returns. For example, when the user agent's dynamic utility function is greater than the target's dynamic utility function, ,if And since its value is small, the imitation probability is... The dynamic utility function of the user agent will be very small, and the user agent will not learn from the experience of the target. When the dynamic utility function of the user agent is less than the dynamic utility function of the target, ,if And since its value is small, the imitation probability is... The user's intelligent agent will learn and imitate the target's experience if the target is large. And when... At that time, regardless of whether the difference in expected benefit utility is positive or negative, The smaller the value, the greater the probability of imitation.

[0100] S5. Iterate through S4 until the maximum number of iterations required for the iteration cycle is reached, then terminate the iteration and consider the multi-agent system to have reached a convergence state.

[0101] S6. After the multi-agent system state converges, obtain the final probability distribution of each user agent choosing different facility agents. Calculate the total service demand carried by each facility agent by summarizing the final probability distribution, and calculate the supply-demand ratio by combining the maximum service capacity attribute of each facility agent.

[0102] It should be noted that in step S6 of this invention, the supply-demand ratio of a facility intelligence agent is calculated as follows: the sum of the probabilities of all user intelligence agents choosing the facility intelligence agent is obtained from the final probability distribution and used as the total service demand of the facility intelligence agent. The total service demand of the facility intelligence agent is divided by the maximum reception capacity to obtain the supply-demand ratio of the facility intelligence agent.

[0103] In step S6 of this embodiment, the first Supply and demand ratio of individual facility intelligent agents The calculation method is as follows:

[0104]

[0105] In the formula, Indicates the first The total service demand of each facility's intelligent agent.

[0106] Based on this, the sum of the total selection probabilities of each user agent is constant at 1, ensuring that the total demand of each user agent is calculated only once in the entire system, avoiding the problems of demand inflation and inaccurate accessibility assessment caused by overlapping service scopes.

[0107] S7. For each user agent, based on the supply-demand ratio and the spatial attenuation relationship between the user agent and each facility agent, generate the public facility accessibility evaluation value for that user agent. Summarize the public facility accessibility evaluation values ​​of all user agents to form the public facility accessibility evaluation value of the final user group and visualize it, generating an accessibility distribution heat map covering the entire study area.

[0108] It should be noted that in step S7 of this invention, the specific steps for generating the corresponding public facility accessibility evaluation value for a user agent are as follows: using the Gaussian distance decay function as a spatial decay relationship model, the supply and demand ratios of all facility agents within the interaction range of the user agent are weighted and summed to obtain the public facility accessibility evaluation value of the user agent.

[0109] In step S7 of this embodiment, for the first Each user agent uses the Gaussian distance decay function as shown in the following formula. As a spatial attenuation relationship model, the supply-demand ratio of all facility agents within the user agent's interaction range is calculated and then weighted and summed to obtain the public facility accessibility evaluation value for the user agent. :

[0110]

[0111] In the formula, This is the spatial attenuation parameter used in reachability calculations.

[0112] It should be noted that in step S7 of this invention, the accessibility evaluation value of public facilities for each user agent is spatially interpolated and visualized to generate an accessibility distribution heat map covering the entire study area.

[0113] To better demonstrate the specific implementation and technical effects of the present invention, the public facility accessibility assessment method based on geographical multi-agent shown in steps S1 to S7 of the above preferred implementation is applied to a specific example below.

[0114] Example

[0115] The implementation process of the public facility accessibility assessment method based on geographic multi-agents used in this embodiment is as described above and will not be repeated here.

[0116] In one specific embodiment of the present invention, the public facility is a hospital at various levels in a city, and the user is a resident of the city. This embodiment takes the assessment of the accessibility of medical public facilities in a certain area as an example to explain in detail the implementation process of the method of the present invention.

[0117] In step S1 of this embodiment, the names, addresses, and latitude and longitude coordinates of all 14,090 residential communities in the target area are obtained through the map service API interface. Based on the main population data bulletin of the target area, the permanent resident population of the 13 districts in the area is obtained, totaling 12,498,000 people. Based on the national average, it is assumed that 2% of the residents in each residential community have medical needs. The number of residents in each community is then determined. The calculation method is as follows:

[0118]

[0119] in, That is, the total population of the region. This represents the number of residential communities within the administrative division of this region on the map.

[0120] The specific details of the number of residents and patients seeking medical treatment in each residential community are shown in Table 1:

[0121] Table 1. Statistics on the number of residents in residential communities across 13 administrative divisions in the target area

[0122] Furthermore, the geographical location, grade (46 Grade III Class A, 21 Grade III Class B, 21 Grade II Class A, and 52 Grade II Class B), and number of beds of 140 hospitals in the target area were obtained and screened through the Medical Security Bureau, and these were defined as facility intelligent entities. Service quality level was quantified according to hospital grade, and maximum service capacity was calculated based on the officially published number of beds. Specifically, based on publicly available statistical bulletins on the development of health and wellness, the outpatient-to-inpatient ratio was 34:1, the average length of stay was 9.2 days, and at maximum efficiency, each bed could accommodate 40 × 34 = 1360 patients per year. The maximum daily outpatient capacity was calculated as the total number of hospital beds × 1360 / 365. The data summary is shown in Table 2.

[0123] Table 2. Hospital Information

[0124] In this embodiment, the individual preference attributes set by the user agent (i.e., the resident agent) are concretized as health beliefs. Their values ​​are directly used in the calculation of the dynamic utility function in step S3, influencing the perceived intensity of service quality by different resident agents, thereby affecting their final medical treatment decisions. Health beliefs are a psychological state reflecting an individual's cognition of health and disease. Their theory originates from the Health Belief Model (HBM), typically encompassing a comprehensive understanding of multiple dimensions, including an individual's judgment of disease severity, their own susceptibility, the effectiveness of preventative measures, and potential obstacles during the medical treatment process. This embodiment further divides health beliefs into three levels: low, medium, and high, with the number of users at each level representing a percentage of the total number of users. , , Each of these is then assigned a preference coefficient value (i.e., a health belief value). , , .

[0125] In step S2 of this embodiment, the parameter for setting the proximity interaction rule is specifically: the preset number of neighboring communities. Distance threshold Each resident agent will select from its 50 nearest neighboring communities that are no more than 5 kilometers away, and will use all resident agents in these communities as its learning and imitation models.

[0126] In step S3 of this embodiment, the constructed dynamic utility function is as described above and will not be repeated. The service quality weight coefficient, distance cost weight coefficient, and congestion cost weight coefficient are set to 5, 1, and 5, respectively. In the dynamic utility function of this embodiment, the first step in the above calculation formula... The utility of a resident intelligent agent This indicates that the residents chose the first option. The probability of a hospital; Then it means the first The first resident perceived Service quality score of each hospital; Then it means the first Service quality at different hospital levels Indicates the first Health belief value of individual user agents; geographical distance cost Then specifically represents the first The residents to the The commute distance to each hospital.

[0127] In step S4 of this embodiment, for the exploration behavior, the initial exploration probability is... attenuation rate Regarding learning behavior, individual irrational parameters .

[0128] In step S5 of this embodiment, the iteration cycle during the system evolution process That is, the maximum number of iterations required for each iteration cycle is 500.

[0129] In step S6 of this embodiment, after the system state described in step S5 converges, the total service demand carried by each facility agent is calculated based on the final policy probability distribution, and the supply-demand ratio is calculated in combination with its maximum service capacity attribute.

[0130] In step S7 of this embodiment, a Gaussian distance decay function is used as the spatial decay relationship model, wherein the spatial decay parameter in the reachability calculation... Set to 5; use the Kriging interpolation method based on ArcGIS software to interpolate the study area, and use the geometric interval method in ArcGIS software to visualize the public facility accessibility evaluation value of the end user group. In the final result map, the higher the accessibility value, the richer the available medical resources in the area.

[0131] Furthermore, this invention utilizes a multi-agent system constructed based on steps S1 to S7. The system adjusts the attributes or number of agents according to preset planning assumptions, and repeats steps S1 to S7 after adjustment to obtain a public facility accessibility evaluation value under the assumed scenario. Finally, the effectiveness of the planning assumption is verified by comparing the evaluation values ​​before and after adjustment.

[0132] Specifically, by analyzing the basic scenario results obtained from the initial runs of S1 to S7, this invention identifies a significant high-level medical service blind spot in area H of the target region, where the initial average accessibility rating for users is 2.1. Based on this, this embodiment proposes a specific planning hypothesis: within this service blind spot, an existing Grade II Class B hospital is selected and upgraded to a Grade III Class A hospital. Experiments are then conducted to confirm whether this hypothesis can effectively improve the level of medical services in the area.

[0133] Based on the above planning assumptions, this embodiment adjusts the system model constructed in step S1, modifying the service quality level attribute of the facility agent by increasing the quantified value corresponding to service quality from 3 to 6. The attributes and system parameters of all other agents remain unchanged. Steps S4 to S7 are repeated on the adjusted system for iterative evolution until the system reaches a new equilibrium state under new conditions. The accessibility evaluation values ​​of all residents are then recalculated. Comparing the accessibility evaluation values ​​obtained under the new scenario with those in the initial state, the analysis shows that the average accessibility evaluation value for users in the original service blind spot and surrounding areas has significantly improved, with an average increase of 0.59. This result demonstrates that the method of this invention can propose policy recommendations based on a multi-agent system and conduct extended verification through scenario simulation, thereby providing a scientific basis for optimizing the layout of medical resources.

[0134] Furthermore, this invention can also be used to evaluate other types of planning assumptions. For example, to simulate the impact of improved transportation infrastructure on residents' travel choices, the spatial decay parameter in the Gaussian distance decay function can be adjusted. In this embodiment, when the spatial decay parameter increases from 5 to 15, the simulation results show that the accessibility centers of the previously dispersed L and M areas have merged, indicating that improved transportation can effectively expand the service radius of high-level facilities and promote regional integration. As another example, to simulate the changing trend of residents' health awareness increasing and their focus on the medical experience rather than distance costs, the weighting coefficients in the dynamic utility function can be adjusted, increasing the weighting coefficient of dynamic congestion costs. The number of patients at Hospital 9, located in the central urban area, fluctuated between 2,300 and 2,700. This was because the hospital became crowded when many residents chose it. In subsequent iterations, the resident agent observed this crowding and prioritized other nearby hospitals with better service quality and lower crowding levels. Meanwhile, the number of patients at Hospital 0, located in Zone M and far from the central urban area, decreased and eventually converged to around 500. This result clearly demonstrates that the present invention can simulate and quantify the dynamic impact of changes in resident preferences on the overall healthcare system structure.

[0135] To demonstrate the geographic multi-agent-based public facility accessibility assessment method shown in S1-S7 above, this embodiment compares the method of the present invention with existing macro-static methods (specifically including 2SFCA, E2SFCA, and 3SFCA) represented by the two-step mobile search method on the same medical dataset described in the preceding embodiments. The comparison results clearly demonstrate that the present invention has better performance in the following core technical dimensions, as shown in Table 3:

[0136] Table 3. Comparison Results

[0137] Specifically, in this embodiment, the beneficial effects of the present invention are reflected in:

[0138] 1) This invention overcomes the strong dependence of existing methods on preset parameters such as distance thresholds. 2SFCA and its improved methods rely on defining the service range of facilities through preset fixed distance thresholds. This rigid boundary setting makes the calculation results extremely sensitive to the specific setting of this threshold; a user whose distance is exactly outside the threshold will be completely excluded, while a user whose distance is exactly within the threshold will be fully included, which does not conform to the gradual decay of distance's influence in the real world. In contrast, this invention introduces a continuous spatial decay relationship model defined by the Gaussian distance decay function when constructing the dynamic utility function. This makes distance cost, as a core factor influencing user decisions, act smoothly and continuously, which is more in line with real-world decision-making behavior, thus avoiding calculation biases introduced by improper preset threshold division.

[0139] 2) It solves the problem of discrepancies between the macroscopic model and reality. In the dataset of this embodiment, the 2SFCA method calculated the highest accessibility value in a certain area in the southwest of region M. However, this area lacks high-level medical services in reality, and its economic development level is incomparable to that of the central urban area. The reason why the 2SFCA method yields results that are seriously inconsistent with reality is precisely because it only considers the macroscopic factor of "proximity". In contrast, the user agent of this invention can comprehensively consider multiple microscopic factors such as service quality and congestion to make dynamic decisions. Its simulation results show that the accessibility value of this area is low, which is completely consistent with the real situation. The results of the E2SFCA and 3SFCA methods are also roughly similar to those of the 2SFCA method, thus jointly proving the accuracy and superiority of the microscopic simulation of this invention.

[0140] 3) It provides more realistic verification of micro-level behavior. By comparing the simulated number of patients with the actual daily average number of patients at the top ten hospitals in the target area, it was found that the Pearson correlation coefficient between the simulation results and the real data of the method of this invention is as high as 0.84, while the correlation coefficient between the simulated ranking and the real ranking is also as high as 0.72. This result strongly proves that the simulation of micro-level game behavior of this invention can effectively capture and reflect the complex laws of the real world through macro-level emergence results, providing a verification capability higher than that of macro-level static models.

[0141] The above comparison clearly demonstrates that, compared with the prior art, the present invention can provide a more accurate and realistic evaluation method in multiple dimensions such as evaluation scale, calculation model and application capability.

[0142] It should also be noted that the public facility accessibility assessment method based on geographic multi-agent technology in the above embodiments can essentially be executed by a computer program or module. Therefore, similarly, based on the same inventive concept, another preferred embodiment of the present invention also provides a public facility accessibility assessment system based on geographic multi-agent technology, corresponding to the public facility accessibility assessment method based on geographic multi-agent technology provided in the above embodiments, such as... Figure 3 As shown, it includes:

[0143] The data acquisition module is used to acquire user distribution data and public facility data within the area to be evaluated;

[0144] The assessment module is used to generate an accessibility distribution heat map covering the area to be assessed based on user distribution data and public facility data within the area to be assessed, according to the public facility accessibility assessment method described in the above embodiments.

[0145] Similarly, based on the same inventive concept, another preferred embodiment of the present invention also provides a computer electronic device corresponding to the geographically multi-agent-based public facility accessibility assessment method provided in the above embodiments, such as... Figure 4 As shown, it includes a memory and a processor;

[0146] The memory is used to store computer programs;

[0147] The processor is configured to implement the geographic multi-agent-based public facility accessibility assessment method in the above embodiments when executing the computer program.

[0148] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0149] It is understood that the processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0150] It should also be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the system described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here. In the embodiments provided in this application, the division of steps or modules in the system and method is merely a logical functional division, and there may be other division methods in actual implementation. For example, multiple modules or steps may be combined or integrated together, and a module or step may also be split.

[0151] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the invention. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the invention. Therefore, all technical solutions obtained through equivalent substitution or transformation fall within the protection scope of the present invention.

Claims

1. A method for assessing the accessibility of public facilities based on geographic multi-agent systems, characterized in that, Includes the following steps: S1. Obtain user distribution data and public facility data within the research area. After presetting the service demand rate corresponding to the public facilities under the target category, calculate the number of users with service demand based on the user distribution data. Then, define users as user intelligent agents with independent decision-making ability and preset individual preference attributes, so that the distribution of user intelligent agents corresponds to the number of users with service demand. Define public facilities as facility intelligent agents with service quality level, maximum service capacity and geographical location attributes, so that the distribution of facility intelligent agents corresponds to the public facility data. In this way, a multi-agent system for simulating user behavior in choosing public facilities is constructed. S2. For each user agent in the multi-agent system, construct a set of its neighboring interaction objects as the neighboring interaction object set of that user agent. S3. For the user agent set in S1, construct a dynamic utility function based on evolutionary game theory to evaluate the expected benefits after choosing different facility agents. S4. Within the preset iteration period, each iteration uses the dynamic utility function as the decision basis, designs the neighbor interaction rules based on the set of neighbor interaction objects, and follows the principle of evolutionary game theory. By simulating the exploration behavior of user agents and the learning behavior based on the neighbor interaction rules, the probability distribution of each user agent choosing different facility agents is dynamically updated. S5. Continuously iterate S4 until the maximum number of iterations required for the iteration cycle is reached, then terminate the iteration and consider the multi-agent system to have reached a convergence state. S6. After the state of the multi-agent system converges, obtain the final probability distribution of each user agent choosing different facility agents. Calculate the total service demand carried by each facility agent by summarizing the final probability distribution, and calculate the supply-demand ratio by combining the maximum service capacity attribute of each facility agent. S7. For each user agent, based on the supply-demand ratio and the spatial attenuation relationship between the user agent and each facility agent, generate the public facility accessibility evaluation value for that user agent. Summarize the public facility accessibility evaluation values ​​of all user agents to form the public facility accessibility evaluation value of the final user group and visualize it, generating an accessibility distribution heat map covering the entire study area.

2. The public facility accessibility assessment method based on geographic multi-agent as described in claim 1, characterized in that, In step S1, the user distribution data obtained includes the user community name, the number of user communities, the number of users in each user community, and the latitude and longitude coordinates representing the user's location; the public facility data obtained includes the public facility name, the latitude and longitude coordinates representing the public facility's location, the service quality level, and the maximum service capacity. The method for calculating the number of users required to provide services in each user community is as follows: S11. Calculate the number of users in each user community by dividing the number of users in the study area by the number of user communities in the study area. S12. Multiply the number of users in each user community by the service demand rate to calculate the number of users in each user community who require service.

3. The public facility accessibility assessment method based on geographic multi-agent as described in claim 1, characterized in that, In step S1, the individual preference attributes of the user agent are divided into multiple levels, each corresponding to a different preference coefficient value. The preference coefficient value is used to influence the service quality score of the user agent's perception facility agent when constructing the dynamic utility function in step S3.

4. The public facility accessibility assessment method based on geographic multi-agent as described in claim 1, characterized in that, In step S2, the specific method for constructing a set of neighboring interaction objects for the user's intelligent agent is as follows: S21. Taking the user community where the user agent is located as the center, select the user community closest to the center as the candidate community. The number of candidate communities is the same as the preset number of neighboring communities, and the candidate communities include the user community where the user agent is located. S22. Take the user community where the user agent is located as the target community, and select all user communities whose distance from the target community is no greater than a preset distance threshold from the candidate communities obtained in S21, to form a neighboring community set. S23. Other user agents located within the neighboring community set are considered as the neighboring interaction objects of this user agent, and all neighboring interaction objects constitute the neighboring interaction object set of this user agent.

5. The public facility accessibility assessment method based on geographic multi-agent as described in claim 3, characterized in that, In step S3, the dynamic utility function comprehensively considers the service quality score of the facility agent, the geographical distance cost, and the dynamic congestion cost. The service quality score is obtained by multiplying the facility agent's level and the user agent's preference coefficient. The dynamic congestion cost is proportional to the sum of the probabilities of all user agents who choose the facility agent.

6. The public facility accessibility assessment method based on geographic multi-agent as described in claim 1, characterized in that, In step S4, when the number of iterations is At that time, the user agent's exploration behavior will be triggered by a hyperparameter that decays exponentially with the number of iterations. When triggered, the probability distribution of the user agent's current selection of different facility agents will be replaced with a newly generated random probability distribution. The design method for the learning behavior of a user agent is as follows: S41. Randomly select a user agent as the imitation target from the set of neighboring interaction objects of the user agent, calculate the dynamic utility function of the user agent and the imitation target in the current iteration, and take the difference between the two dynamic utility functions as the expected benefit utility difference. S42. Calculate the imitation probability based on the obtained difference in expected benefit utility, and have the user agent apply the imitation probability to adopt the probability distribution of the imitation target in order to update its own probability distribution; wherein, the imitation probability is used to represent the proportion of the user agent adopting the probability distribution of the imitation target, and the imitation probability value is negatively correlated with the absolute value of the difference in expected benefit utility.

7. The public facility accessibility assessment method based on geographic multi-agent as described in claim 1, characterized in that, In step S6, the supply-demand ratio of a facility agent is calculated as follows: obtain the sum of the probabilities of all user agents choosing the facility agent from the final probability distribution and use it as the total service demand of the facility agent. Divide the total service demand of the facility agent by the maximum reception capacity to obtain the supply-demand ratio of the facility agent.

8. The public facility accessibility assessment method based on geographic multi-agent as described in claim 1, characterized in that, In step S7, the specific steps for generating the corresponding public facility accessibility evaluation value for a user agent are as follows: using the Gaussian distance decay function as a spatial decay relationship model, the supply and demand ratios of all facility agents within the user agent's interaction range are weighted and summed to obtain the public facility accessibility evaluation value for the user agent.

9. A public facility accessibility assessment system based on geographic multi-agent systems, characterized in that, include: The data acquisition module is used to acquire user distribution data and public facility data within the area to be evaluated; An assessment module is used to generate an accessibility distribution heat map covering the area to be assessed based on user distribution data and public facility data within the area to be assessed, according to the public facility accessibility assessment method according to any one of claims 1 to 8.

10. A computer electronic device, characterized in that, Including memory and processor; The memory is used to store computer programs; The processor is configured to implement the public facility accessibility assessment method as described in any one of claims 1 to 8 when executing the computer program.

Images