A smart decision-making method and system for assessing the suitability of urban industrial layout

Abstract

This invention discloses an intelligent decision-making method and system for assessing the suitability of urban industrial layout, relating to the field of urban planning and regional development. The method includes: collecting multi-source heterogeneous data of the target urban area to establish a standardized spatiotemporal database; constructing an initial assessment factor system based on the standardized spatiotemporal database; loading the standardized spatiotemporal database into three-dimensional geographic information units to construct an urban multi-agent simulation environment; using a causal inference engine, calculating the causal net effect of candidate industrial layout schemes on preset urban indicators based on historical data in the standardized spatiotemporal database, and generating a causal net effect assessment report; and fusing the real-time data of the initial assessment factor system, the current state of the urban multi-agent simulation environment, and the causal net effect assessment report to generate a dynamic decision state vector. This achieves a leap from static location assessment to dynamic path planning, improving the foresight of industrial layout decisions.

Description

Technical Field

[0001] This invention relates to the field of urban planning and regional development, and in particular to an intelligent decision-making method and system for assessing the suitability of urban industrial layout. Background Technology

[0002] In the field of urban planning and regional development, suitability assessment of industrial layout is a key technology supporting scientific decision-making. The mainstream approach is based on a fusion framework of geographic information systems and multi-criteria decision analysis. It constructs an assessment index system that includes economic, environmental, and social dimensions, and uses methods such as the analytic hierarchy process (AHP) to determine the weights of the indicators. Through spatial overlay analysis, it generates a comprehensive suitability score for the region. With technological advancements, big data analysis has been introduced to incorporate real-time and dynamic data sources, enhancing the timeliness and data dimensions of the assessment and providing important static reference for decision-makers.

[0003] The existing technological paradigm is essentially a static evaluation model. Its core limitation lies in its inability to cope with the dynamic evolution of complex urban systems. The method outputs an optimal layout point based on current or historical data, which can trigger a chain of socio-economic and environmental changes. This static snapshot-style evaluation lacks the ability to simulate and predict the long-term consequences of decisions and the ability to dynamically optimize development paths, resulting in planning schemes that may lack resilience in dealing with future uncertainties. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an intelligent decision-making method for assessing the suitability of urban industrial layout, addressing the key problems of existing static assessment models that cannot simulate the long-term dynamic interactions of multiple stakeholders, lack foresight regarding the long-term consequences of industrial layout, and fail to generate adaptive development paths.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides an intelligent decision-making method for assessing the suitability of urban industrial layout, which includes collecting multi-source heterogeneous data of the target urban area, establishing a standardized spatiotemporal database, and constructing an initial assessment factor system based on the standardized spatiotemporal database.

[0008] A standardized spatiotemporal database is loaded into a three-dimensional geographic information unit to construct a multi-agent simulation environment for the city.

[0009] Using a causal inference engine, the net causal effect of candidate industrial layout schemes on preset urban indicators is calculated based on historical data in a standardized spatiotemporal database, and a net causal effect assessment report is generated.

[0010] The real-time data of the initial evaluation factor system, the current state of the urban multi-agent simulation environment, and the causal net effect evaluation report are integrated to generate a dynamic decision state vector.

[0011] The layout decision-making agents are divided into groups focusing on different objectives. The grouped layout decision-making agents are then placed in a city multi-agent simulation environment. The grouped layout decision-making agents are trained using a reinforcement learning algorithm based on the dynamic decision state vector, and output a dynamic path map of industrial layout.

[0012] As a preferred embodiment of the intelligent decision-making method for assessing the suitability of urban industrial layout as described in this invention, the method includes the following steps: collecting multi-source heterogeneous data of the target urban area and establishing a standardized spatiotemporal database.

[0013] Acquire multi-source heterogeneous data of the target city area, and perform coordinate correction, unit unification and outlier removal on the multi-source heterogeneous data of the target city area.

[0014] The processed multi-source heterogeneous data of the target urban area are integrated according to a unified spatiotemporal benchmark to establish a standardized spatiotemporal database.

[0015] As a preferred embodiment of the intelligent decision-making method for assessing the suitability of urban industrial layout as described in this invention, the method includes the following steps: Constructing an initial assessment factor system based on a standardized spatiotemporal database.

[0016] Based on a standardized spatiotemporal database, evaluation factors are selected from urban planning principles and industrial layout standards to form a candidate set of evaluation factors.

[0017] Based on a standardized spatiotemporal database and a candidate set of evaluation factors, the availability and representativeness of each factor in the candidate set of evaluation factors are verified. Based on the standardized spatiotemporal database and the verified candidate set of evaluation factors, an initial evaluation factor system is constructed.

[0018] As a preferred embodiment of the intelligent decision-making method for assessing the suitability of urban industrial layout as described in this invention, the method involves loading a standardized spatiotemporal database into a three-dimensional geographic information unit to construct an urban multi-agent simulation environment, including the following steps:

[0019] Spatial geographic data and infrastructure data from standardized spatiotemporal databases are loaded into three-dimensional geographic information units to form a basic three-dimensional scene;

[0020] Based on population and industry distribution data in a standardized spatiotemporal database, define the initial attributes and behavioral rules of planning management agents, industrial production agents, and community agents. Place these agents, with their defined attributes and behavioral rules, into a basic 3D scene, establish the interaction logic between agents and the environment, and between agents themselves, and construct a multi-agent simulation environment for the city.

[0021] As a preferred embodiment of the intelligent decision-making method for assessing the suitability of urban industrial layout as described in this invention, the method includes the following steps: Utilizing a causal inference engine, it calculates the net causal effect of candidate industrial layout schemes on preset urban indicators based on historical data in a standardized spatiotemporal database, and generates a net causal effect assessment report.

[0022] Regions with similar industrial layouts were selected from historical data in a standardized spatiotemporal database as the treatment group regions, and control group regions that did not implement similar industrial layouts were matched.

[0023] Based on the matched treatment group and control group regions, the difference-in-differences algorithm is used to calculate the estimated net causal effect of candidate industrial layout schemes on urban indicators through the selection of parameters reflecting industrial benefits, resource consumption and community impact.

[0024] The estimated net causal effects are tested, and the tested net causal effects of the candidate industrial layout schemes for each urban indicator are summarized to generate a net causal effect assessment report.

[0025] As a preferred embodiment of the intelligent decision-making method for assessing the suitability of urban industrial layout as described in this invention, the method involves fusing real-time data from the initial assessment factor system, the current state of the urban multi-agent simulation environment, and the causal net effect assessment report to generate a dynamic decision state vector, including the following steps:

[0026] By accessing the data interface of the initial evaluation factor system, the current value of each factor in the real-time data of the initial evaluation factor system is extracted to form a real-time data value sequence of the initial evaluation factor system.

[0027] Based on the runtime memory of the city multi-agent simulation environment, the attribute values ​​and relationship states of the planning and management agent, industrial production agent, and community agent in the current state of the city multi-agent simulation environment are extracted to form a snapshot of the current state of the city multi-agent simulation environment.

[0028] By analyzing the net causal effect assessment report, we extract each net causal effect estimate from the report to form a set of net causal effect assessment report estimates.

[0029] The real-time data sequence of the initial evaluation factor system, the current state snapshot of the urban multi-agent simulation environment, and the estimated value set of the causal net effect evaluation report are vectorized and concatenated to generate a dynamic decision state vector.

[0030] As a preferred embodiment of the intelligent decision-making method for assessing the suitability of urban industrial layout as described in this invention, the method includes: dividing the layout decision-making agents into groups focusing on different objectives; placing the grouped layout decision-making agents in an urban multi-agent simulation environment; and training the grouped layout decision-making agents based on dynamic decision state vectors using a reinforcement learning algorithm to output a dynamic path map of industrial layout, comprising the following steps:

[0031] Based on different optimization objectives such as prioritizing industrial benefits, controlling resource consumption, and balancing community impact, the layout decision-making agents are divided into three groups: those prioritizing industrial benefits, those controlling resource consumption, and those balancing community impact. These groups are then placed into a multi-agent simulation environment for the city.

[0032] The decision-making agents for prioritizing industry benefits, controlling resource consumption, and balancing community impact interact with the planning and management agent, the industrial production agent, and the community agent, respectively, receiving dynamic decision-making state vectors. The policy networks of these agents are updated using a deep Q-learning algorithm. After the deep Q-learning algorithm is trained, the policy sequence generated by the decision-making agent group that obtains the highest cumulative reward during the simulation period is selected as the dynamic path map for industrial layout.

[0033] Secondly, the present invention provides an intelligent decision-making system for assessing the suitability of urban industrial layout, including a factor construction module, which collects multi-source heterogeneous data of the target urban area, establishes a standardized spatiotemporal database, and constructs an initial assessment factor system based on the standardized spatiotemporal database.

[0034] The environment simulation module loads a standardized spatiotemporal database into a three-dimensional geographic information unit to construct a multi-agent simulation environment for the city.

[0035] The analysis module uses a causal inference engine to calculate the net causal effect of candidate industrial layout schemes on preset urban indicators based on historical data in a standardized spatiotemporal database, and generates a net causal effect assessment report.

[0036] The fusion module integrates real-time data from the initial evaluation factor system, the current state of the urban multi-agent simulation environment, and the causal net effect evaluation report to generate a dynamic decision state vector.

[0037] The path generation module divides the layout decision-making agents into groups that focus on different objectives. The grouped layout decision-making agents are then placed in a city multi-agent simulation environment. The grouped layout decision-making agents are trained using a reinforcement learning algorithm based on the dynamic decision state vector, and output a dynamic path map of industrial layout.

[0038] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the intelligent decision-making method for urban industrial layout suitability assessment as described in the first aspect of the present invention.

[0039] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the intelligent decision-making method for urban industrial layout suitability assessment as described in the first aspect of the present invention.

[0040] The beneficial effects of this invention are as follows: By constructing a standardized spatiotemporal database and a multi-agent simulation environment, a real data foundation and interactive computing platform are provided for dynamic evaluation. By using a causal inference engine to calculate the net effect of layout schemes, the evaluation dimension is elevated from correlation analysis to causal inference, thereby accurately quantifying the real impact of different industrial layout strategies on various urban indicators. This effectively overcomes the evaluation bias caused by confounding variables in traditional methods. By introducing a multi-objective reinforcement learning mechanism to train group decision-making agents, they can autonomously learn and optimize layout paths in the simulation environment. This not only dynamically balances multiple objectives such as industrial benefits, resource consumption, and community impact, but also generates optimal sequence schemes that adapt to long-term development. Thus, it realizes a leap from static site evaluation to dynamic path planning, improving the foresight of industrial layout decisions. Attached Figure Description

[0041] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 A flowchart of an intelligent decision-making method for assessing the suitability of urban industrial layout.

[0043] Figure 2 A schematic diagram of an intelligent decision-making system for assessing the suitability of urban industrial layout.

[0044] Figure 3 A schematic diagram illustrating the construction of a multi-agent simulation environment for a city.

[0045] Figure 4 This is a schematic diagram for generating dynamic decision-making state vectors. Detailed Implementation

[0046] 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.

[0047] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0048] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0049] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides an intelligent decision-making method for assessing the suitability of urban industrial layout, comprising the following steps:

[0050] S1. Collect multi-source heterogeneous data of the target urban area and establish a standardized spatiotemporal database.

[0051] S1.1 Obtain multi-source heterogeneous data of the target city area, and perform coordinate correction, unit unification and outlier removal on the multi-source heterogeneous data of the target city area.

[0052] S1.2 Integrate the processed multi-source heterogeneous data of the target urban area according to a unified spatiotemporal benchmark to establish a standardized spatiotemporal database.

[0053] Furthermore, using geospatial latitude and longitude and standard timestamps as index keys, the multi-source heterogeneous data of the target urban area after coordinate correction, unit unification and outlier removal are associated with a unified grid unit or administrative boundary; establishing a standardized spatiotemporal database involves storing the multi-source heterogeneous data of the target urban area after integration according to a unified spatiotemporal benchmark in a relational database or spatiotemporal database, forming a standardized spatiotemporal database containing spatial information, temporal information and attribute information.

[0054] S2. Construct an initial evaluation factor system based on a standardized spatiotemporal database.

[0055] S2.1 Based on a standardized spatiotemporal database, evaluation factors are selected from urban planning principles and industrial layout specifications to form a candidate set of evaluation factors.

[0056] S2.2 Based on the standardized spatiotemporal database and the candidate set of evaluation factors, the availability and representativeness of each factor in the candidate set of evaluation factors are verified. Based on the standardized spatiotemporal database and the verified candidate set of evaluation factors, an initial evaluation factor system is constructed.

[0057] Furthermore, based on a standardized spatiotemporal database and a candidate set of evaluation factors, the availability and representativeness of each factor in the candidate set are verified. Data availability verification involves checking whether the data required for each factor in the candidate set exists in the standardized spatiotemporal database or can be derived from existing data in the standardized spatiotemporal database. Representativeness verification assesses whether the completeness and spatial coverage of the data for each factor in the time series can effectively represent the corresponding characteristics of the target urban area. Based on the standardized spatiotemporal database and the verified candidate set of evaluation factors, the initial evaluation factor system is constructed by formally identifying the factors in the candidate set that have passed the data availability and representativeness verification as the constituent items of the initial evaluation factor system, and specifying the specific data source fields and update frequency of each factor in the standardized spatiotemporal database, thus completing the construction of the initial evaluation factor system.

[0058] S3. Load the standardized spatiotemporal database into the three-dimensional geographic information unit to construct a multi-agent simulation environment for the city.

[0059] S3.1 Load the spatial geographic data and infrastructure data from the standardized spatiotemporal database into the three-dimensional geographic information unit to form a basic three-dimensional scene.

[0060] Furthermore, the system reads digital elevation models, land use type vector boundaries, road network line data, and power substation location data stored in a standardized spatiotemporal database. To form a basic 3D scene, the system renders the digital elevation model as a terrain surface within a 3D geographic information unit. Corresponding surface textures are then overlaid on the terrain surface according to the land use type vector boundaries. The road network line data and power substation location data are visualized in their respective 3D spatial locations, thus constructing a basic 3D scene that includes terrain, land cover, and infrastructure.

[0061] S3.2. Based on the population distribution and industry distribution data in the standardized spatiotemporal database, define the initial attributes and behavioral rules of the planning management intelligent agent, industrial production intelligent agent, and community intelligent agent. Place the planning management intelligent agent, industrial production intelligent agent, community intelligent agent, etc., with defined attributes and behavioral rules into the basic three-dimensional scene, establish the interaction logic between the intelligent agent and the environment and between the intelligent agents, and construct a city multi-agent simulation environment.

[0062] S4. Using a causal inference engine, calculate the net causal effect of candidate industrial layout schemes on preset urban indicators based on historical data in a standardized spatiotemporal database, and generate a net causal effect assessment report.

[0063] S4.1 Select regions with similar industrial layouts from the historical data of the standardized spatiotemporal database as the treatment group regions and match them with the control group regions that have not implemented similar industrial layouts.

[0064] Furthermore, regions with similar industrial layouts are selected from historical data in a standardized spatiotemporal database as treatment group regions. Specifically, historical industrial project registration records are queried to identify administrative regions or grid units where similar industries (such as the chemical industry) have been established within a specific time period (e.g., the past ten years). These regions or grid units are listed as candidates for treatment group regions. Control group regions that have not implemented similar industrial layouts are matched using a propensity score matching method. Among the candidate treatment group regions that did not have such industrial layouts during the same period, the regions that are closest in terms of covariates such as population size, economic development level, and environmental background value before the industrial layout are established are selected, forming control group regions that correspond one-to-one with the treatment group regions.

[0065] S4.2 Based on the matched treatment group region and control group region, the double difference algorithm is used to calculate the estimated net causal effect of the candidate industrial layout scheme on the urban indicators through the selected parameter settings that reflect industrial benefits, resource consumption and community impact.

[0066] Furthermore, based on the matched treatment group and control group regions, a difference-in-differences algorithm is used to calculate the estimated net causal effect of candidate industrial layout schemes on urban indicators through selected parameter settings reflecting industrial benefits, resource consumption, and community impacts. The calculation process involves extracting the mean values ​​of specific urban indicators for the treatment group region in the year before (early stage) and three years after (late stage) of the industrial layout from historical data in a standardized spatiotemporal database as the late-stage and early-stage values ​​of the treatment group, respectively. Simultaneously, the mean values ​​of the same urban indicators for the matched control group region in the same period are extracted as the late-stage and early-stage values ​​of the control group. These values ​​are then substituted into the expression. Calculations were performed, and the results were obtained. This is the estimated net causal effect of the candidate industrial layout scheme on the city's indicators.

[0067] The expression for the estimated net causal effect of urban indicators is as follows:

[0068] ;

[0069] in, This is an estimate of the net causal effect of urban indicators. To process the previous values ​​of the group, To process the later values ​​of the group, This is the value of the control group in the later period. This represents the initial value of the control group.

[0070] S4.3. Test the estimated net causal effect values, summarize the tested net causal effect values ​​of the candidate industrial layout schemes for each urban indicator, and generate a net causal effect assessment report.

[0071] Furthermore, a t-test is used to analyze whether the estimated net causal effect is not caused by chance. The probability of chance in measuring the observed effect is calculated by obtaining the observed current sample result. A probability threshold of 0.05 is set. When the observed current sample result is less than 0.05, the estimated net causal effect is considered not to be caused by chance. The tested net causal effect estimates for each city indicator of the candidate industrial layout schemes are compiled. This involves organizing the tested net causal effect estimates for all preset city indicators corresponding to each candidate industrial layout scheme and the observed current sample results. The net causal effect assessment report is generated by outputting the compiled data according to a standardized report format. The report content clearly lists the net causal effect estimates for each candidate industrial layout scheme and each city indicator, as well as the observed current sample results, forming a structured net causal effect assessment report.

[0072] The statistical expression is:

[0073] ;

[0074] This is the standard error in units of the difference between the estimated and hypothesized net causal effect. These are assumed values.

[0075] S5. Integrate the real-time data of the initial evaluation factor system, the current state of the urban multi-agent simulation environment, and the causal net effect evaluation report to generate a dynamic decision state vector.

[0076] S5.1 By accessing the data interface of the initial evaluation factor system, extract the current value of each factor in the real-time data of the initial evaluation factor system to form a real-time data value sequence of the initial evaluation factor system.

[0077] Furthermore, by accessing the data interface of the initial evaluation factor system, the current value of each factor in the real-time data of the initial evaluation factor system is extracted. Specifically, this involves calling the application programming interface defined in the initial evaluation factor system or directly reading the corresponding database fields. The real-time data value sequence of the initial evaluation factor system is formed by arranging the current value of each extracted factor according to the factor order defined in the initial evaluation factor system, thus forming an ordered list of values, which is the real-time data value sequence of the initial evaluation factor system.

[0078] S5.2 Based on the runtime memory of the city multi-agent simulation environment, extract the attribute values ​​and relationship states of the planning and management agent, industrial production agent, and community agent in the current state of the city multi-agent simulation environment to form a snapshot of the current state of the city multi-agent simulation environment.

[0079] Furthermore, based on querying the runtime memory of the urban multi-agent simulation environment, the attribute values ​​and relationship states of the planning management agent, industrial production agent, and community agent in the current state of the urban multi-agent simulation environment are extracted. Specifically, the memory data of the urban multi-agent simulation environment program at runtime is read to obtain the current land reserve attribute of the planning management agent, the current capacity utilization rate attribute of the industrial production agent, the current satisfaction attribute of the community agent, and the relationship state between agents, such as supply chain links or geographical adjacency. The current state snapshot of the urban multi-agent simulation environment is formed by encapsulating and saving the extracted attribute values ​​and relationship states of the planning management agent, industrial production agent, and community agent in a specific data format, forming a complete record of the state of the urban multi-agent simulation environment at a certain moment, i.e., the current state snapshot of the urban multi-agent simulation environment.

[0080] S5.3 By analyzing the net causal effect assessment report, extract each net causal effect estimate from the report to form a set of net causal effect assessment report estimates.

[0081] Furthermore, by analyzing the net causal effect assessment report, the estimated value of each net causal effect in the report is extracted. Specifically, the structured file of the net causal effect assessment report is read, and the descriptive fields of the estimated values ​​of the net causal effects of each candidate industrial layout scheme for each preset city indicator are identified. The set of estimated values ​​of the net causal effect assessment report is formed by organizing all the extracted net causal effect estimates according to the dual dimensions of candidate industrial layout schemes and preset city indicators, forming a set containing all the estimates, namely the set of estimated values ​​of the net causal effect assessment report.

[0082] S5.4. Vectorize and concatenate the real-time data numerical sequence of the initial evaluation factor system, the current state snapshot of the urban multi-agent simulation environment, and the estimated value set of the causal net effect evaluation report to generate a dynamic decision state vector.

[0083] Furthermore, the real-time data sequence of the initial evaluation factor system, the current state snapshot of the urban multi-agent simulation environment, and the estimated value set of the causal net effect assessment report are vectorized and concatenated. Specifically, each value in the real-time data sequence of the initial evaluation factor system, each attribute value and relational state value deconstructed from the current state snapshot of the urban multi-agent simulation environment, and each estimated value in the estimated value set of the causal net effect assessment report are sequentially concatenated end to end, transforming them into a one-dimensional numerical array. The dynamic decision state vector is generated by using this one-dimensional numerical array as the final output dynamic decision state vector. The dimension of the dynamic decision state vector is jointly determined by the number of factors in the initial evaluation factor system, the number of state variables to be recorded in the urban multi-agent simulation environment, and the number of estimated values ​​in the causal net effect assessment report.

[0084] S6. Divide the layout decision-making agent into groups of decision-making agents focusing on different objectives. Place the grouped layout decision-making agents in a city multi-agent simulation environment. The grouped layout decision-making agents are trained by reinforcement learning algorithm based on dynamic decision state vectors and output dynamic path map of industrial layout.

[0085] S6.1 Based on the different optimization objectives of prioritizing industrial benefits, controlling resource consumption, and balancing community impact, the layout decision-making agents are divided into groups: those prioritizing industrial benefits, those controlling resource consumption, and those balancing community impact, and then placed into a city multi-agent simulation environment.

[0086] Furthermore, based on different optimization objectives—prioritizing industrial benefits, controlling resource consumption, and balancing community impact—the layout decision-making agents are divided into three groups: those prioritizing industrial benefits, those controlling resource consumption, and those balancing community impact. Specifically, a reward function is set for the group prioritizing industrial benefits, with the core objective of maximizing regional GDP growth; for the group controlling resource consumption, a reward function is set for minimizing energy consumption per unit of output; and for the group balancing community impact, a reward function is set for balancing economic growth and resident satisfaction. Within the urban multi-agent simulation environment, these groups are instantiated as interactive agents and deployed to coexist with the planning management agent, the industrial production agent, and the community agent.

[0087] S6.2 The decision-making agents for prioritizing industry benefits, controlling resource consumption, and balancing community impact interact with the planning and management agent, the industrial production agent, and the community agent, respectively, receiving dynamic decision-making state vectors. The policy networks of the decision-making agents for prioritizing industry benefits, controlling resource consumption, and balancing community impact are updated using a deep Q-learning algorithm. After the deep Q-learning algorithm is trained, the policy sequence generated by the decision-making agent group that obtains the highest cumulative reward during the simulation period is selected as the dynamic path map for industrial layout.

[0088] Furthermore, the decision-making agents prioritizing industry benefits, controlling resource consumption, and balancing community impact interact with the planning and management agent, the industrial production agent, and the community agent. Each agent receives a dynamic decision state vector and updates its policy network using a deep Q-learning algorithm. Specifically, each decision-making agent receives its current dynamic decision state vector at each decision step, selects and executes a layout action based on its own policy network, and then obtains an immediate reward calculated based on its group-specific reward function from the city multi-agent simulation environment. It also observes new dynamic decision state vectors, stores transition samples using an experience replay mechanism, and periodically samples data from the experience replay pool. The parameters of the group's policy network are updated according to the deep Q-learning algorithm expression. After the deep Q-learning algorithm training is complete, the policy sequence generated by the decision-making agent group that obtains the highest cumulative reward during the simulation period is selected as the dynamic path map for industrial layout.

[0089] The expression for the deep Q-learning algorithm is:

[0090] ;

[0091] in, For dynamic decision state vectors Execute layout action below Long-term expected cumulative rewards, For dynamic decision state vectors Execute layout action below The output value, For the new dynamic decision state vector Execute new layout actions Long-term expected cumulative rewards, For dynamic decision-making state vectors, For layout actions, For learning rate, As a discount factor, This represents the new dynamic decision state vector. This is part of a new strategic plan.

[0092] Specifically, the total discounted cumulative reward obtained by the decision-making agents for prioritizing industry benefits, controlling resource consumption, and balancing community impact during the entire simulation period is calculated separately. The total cumulative reward values ​​of these three decision-making agents are compared, and the series of layout actions taken by the decision-making agent group with the highest total cumulative reward value during the simulation are arranged in chronological order to form a dynamic path map of industrial layout.

[0093] The expression for the total return obtained within the simulation period is:

[0094] ;

[0095] in, Grouping of decision-making agents The total return obtained over the entire simulation period, For the total decision-making cycle, Grouping decision-making agents In time When the environment is in a dynamic decision-making state vector Execute layout decision actions in time The immediate rewards obtained from the environment afterwards Grouping and indexing decision-making agents;

[0096] This embodiment also provides an intelligent decision-making system for assessing the suitability of urban industrial layout, including: a factor construction module, which collects multi-source heterogeneous data of the target urban area, establishes a standardized spatiotemporal database, and constructs an initial assessment factor system based on the standardized spatiotemporal database;

[0097] The environment simulation module loads a standardized spatiotemporal database into a three-dimensional geographic information unit to construct a multi-agent simulation environment for the city.

[0098] The analysis module uses a causal inference engine to calculate the net causal effect of candidate industrial layout schemes on preset urban indicators based on historical data in a standardized spatiotemporal database, and generates a net causal effect assessment report.

[0099] The fusion module integrates real-time data from the initial evaluation factor system, the current state of the urban multi-agent simulation environment, and the causal net effect evaluation report to generate a dynamic decision state vector.

[0100] The path generation module divides the layout decision-making agents into groups that focus on different objectives. The grouped layout decision-making agents are then placed in a city multi-agent simulation environment. The grouped layout decision-making agents are trained using a reinforcement learning algorithm based on the dynamic decision state vector, and output a dynamic path map of industrial layout.

[0101] This embodiment also provides a computer device applicable to the intelligent decision-making method for urban industrial layout suitability assessment, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the intelligent decision-making method for urban industrial layout suitability assessment as proposed in the above embodiment.

[0102] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0103] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the intelligent decision-making method for assessing the suitability of urban industrial layout as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0104] In summary, this invention provides a realistic data foundation and interactive computation platform for dynamic evaluation by constructing a standardized spatiotemporal database and a multi-agent simulation environment. By utilizing a causal inference engine to calculate the net effect of layout schemes, the evaluation dimension is elevated from correlation analysis to causal inference, thereby accurately quantifying the real impact of different industrial layout strategies on various urban indicators. This effectively overcomes the evaluation bias caused by confounding variables in traditional methods. By introducing a multi-objective reinforcement learning mechanism to train group decision-making agents, these agents can autonomously learn and optimize layout paths in the simulation environment. This not only dynamically balances multiple objectives such as industrial benefits, resource consumption, and community impact, but also generates optimal sequence schemes that adapt to long-term development. Thus, it achieves a leap from static site evaluation to dynamic path planning, enhancing the foresight of industrial layout decisions.

[0105] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An intelligent decision-making method for assessing the suitability of urban industrial layout, characterized in that: This includes collecting multi-source heterogeneous data from the target urban area, establishing a standardized spatiotemporal database, and constructing an initial evaluation factor system based on the standardized spatiotemporal database. A standardized spatiotemporal database is loaded into a three-dimensional geographic information unit to construct a multi-agent simulation environment for the city. Using a causal inference engine, the net causal effect of candidate industrial layout schemes on preset urban indicators is calculated based on historical data in a standardized spatiotemporal database, generating a net causal effect assessment report, including the following steps: Regions with similar industrial layouts were selected from historical data in a standardized spatiotemporal database as the treatment group regions, and control group regions that did not implement similar industrial layouts were matched. Based on the matched treatment group and control group regions, the difference-in-differences algorithm is used to calculate the estimated net causal effect of candidate industrial layout schemes on urban indicators through the selection of parameters reflecting industrial benefits, resource consumption and community impact. The net causal effect estimates are tested, and the tested net causal effect estimates for each urban indicator of the candidate industrial layout schemes are summarized to generate a net causal effect assessment report. The dynamic decision-making state vector is generated by fusing real-time data from the initial evaluation factor system, the current state of the urban multi-agent simulation environment, and the causal net effect evaluation report, including the following steps: By accessing the data interface of the initial evaluation factor system, the current value of each factor in the real-time data of the initial evaluation factor system is extracted to form a real-time data value sequence of the initial evaluation factor system. Based on the runtime memory of the city multi-agent simulation environment, the attribute values ​​and relationship states of the planning and management agent, industrial production agent, and community agent in the current state of the city multi-agent simulation environment are extracted to form a snapshot of the current state of the city multi-agent simulation environment. By analyzing the net causal effect assessment report, we extract each net causal effect estimate from the report to form a set of net causal effect assessment report estimates. The real-time data numerical sequence of the initial evaluation factor system, the current state snapshot of the urban multi-agent simulation environment, and the estimated value set of the causal net effect evaluation report are vectorized and concatenated to generate a dynamic decision state vector. The layout decision-making agents are divided into groups focusing on different objectives. These grouped agents are then placed in a multi-agent urban simulation environment. The grouped agents are trained using a reinforcement learning algorithm based on dynamic decision state vectors to output a dynamic path map for industrial layout. This process includes the following steps: Based on different optimization objectives such as prioritizing industrial benefits, controlling resource consumption, and balancing community impact, the layout decision-making agents are divided into three groups: those prioritizing industrial benefits, those controlling resource consumption, and those balancing community impact. These groups are then placed into a multi-agent simulation environment for the city. The decision-making agents for prioritizing industry benefits, controlling resource consumption, and balancing community impact interact with the planning and management agent, the industrial production agent, and the community agent, respectively, receiving dynamic decision-making state vectors. The policy networks of these agents are updated using a deep Q-learning algorithm. After the deep Q-learning algorithm is trained, the policy sequence generated by the decision-making agent group that obtains the highest cumulative reward during the simulation period is selected as the dynamic path for industrial layout.

2. The intelligent decision-making method for assessing the suitability of urban industrial layout as described in claim 1, characterized in that: Collect multi-source heterogeneous data of the target urban area and establish a standardized spatiotemporal database, including the following steps: Acquire multi-source heterogeneous data of the target city area, and perform coordinate correction, unit unification and outlier removal on the multi-source heterogeneous data of the target city area. The processed multi-source heterogeneous data of the target urban area are integrated according to a unified spatiotemporal benchmark to establish a standardized spatiotemporal database.

3. The intelligent decision-making method for assessing the suitability of urban industrial layout as described in claim 2, characterized in that: The initial evaluation factor system is constructed based on a standardized spatiotemporal database, including the following steps: Based on a standardized spatiotemporal database, evaluation factors are selected from urban planning principles and industrial layout standards to form a candidate set of evaluation factors. Based on a standardized spatiotemporal database and a candidate set of evaluation factors, the availability and representativeness of each factor in the candidate set of evaluation factors are verified. Based on the standardized spatiotemporal database and the verified candidate set of evaluation factors, an initial evaluation factor system is constructed.

4. The intelligent decision-making method for assessing the suitability of urban industrial layout as described in claim 3, characterized in that: The standardized spatiotemporal database is loaded into 3D geographic information units to construct a multi-agent simulation environment for the city, including the following steps: Spatial geographic data and infrastructure data from standardized spatiotemporal databases are loaded into three-dimensional geographic information units to form a basic three-dimensional scene; Based on the population and industry distribution data in the standardized spatiotemporal database, define the initial attributes and behavioral rules of the planning management agent, industrial production agent, and community agent. Then, place the planning management agent, industrial production agent, and community agent with defined attributes and behavioral rules into the basic three-dimensional scene, establish the interaction logic between the agent and the environment and between the agents, and construct a city multi-agent simulation environment.

5. An intelligent decision-making system for assessing the suitability of urban industrial layout, based on the intelligent decision-making method for assessing the suitability of urban industrial layout as described in any one of claims 1 to 4, characterized in that: This includes a factor construction module, which collects multi-source heterogeneous data from the target urban area, establishes a standardized spatiotemporal database, and constructs an initial evaluation factor system based on the standardized spatiotemporal database. The environment simulation module loads a standardized spatiotemporal database into a three-dimensional geographic information unit to construct a multi-agent simulation environment for the city. The analysis module uses a causal inference engine to calculate the net causal effect of candidate industrial layout schemes on preset urban indicators based on historical data in a standardized spatiotemporal database, and generates a net causal effect assessment report. The fusion module integrates real-time data from the initial evaluation factor system, the current state of the urban multi-agent simulation environment, and the causal net effect evaluation report to generate a dynamic decision state vector. The path generation module divides the layout decision-making agents into groups that focus on different objectives. The grouped layout decision-making agents are then placed in a city multi-agent simulation environment. The grouped layout decision-making agents are trained using a reinforcement learning algorithm based on the dynamic decision state vector, and output a dynamic path map of industrial layout.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the intelligent decision-making method for urban industrial layout suitability assessment as described in any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the intelligent decision-making method for assessing the suitability of urban industrial layout as described in any one of claims 1 to 4.

Images