Industrial transfer spatial decision-making method and system based on ANN-GIS-CA architecture

Abstract

The present invention provides an industrial transfer spatial decision-making method and system based on an ANN-GIS-CA architecture, belonging to the field of land and spatial resource utilization. The method first divides the industrial transfer recipient areas into several grids based on GIS and determines the sub-areas to which the grids belong, then constructs an evaluation index system for the industrial transfer reception capacity and calculates the reception capacity score, constructs the evaluation index system structure for the reception effect and calculates the social and economic effect score and the ecological environment effect score; then calculates the six state probabilities of all grids, whether they belong to prohibited development areas and the Euclidean distance from the existing industrial grids; uses the above calculation results as attributes to construct a partitioned asynchronous cellular automaton model, substitutes the grid data of the base year into the cellular automaton model for multiple iterations, and after the accuracy reaches the threshold, divides the future industrial transfer into five different scenarios for simulation, and then makes a decision based on the simulation results. The present invention improves the accuracy and effectiveness of industrial spatial decision-making during industrial transfer.

Description

Technical Field

[0001] The present invention belongs to the field of land and space resource utilization, and particularly relates to an industrial transfer space decision-making method and system based on an ANN-GIS-CA architecture. Background Art

[0002] Spatial layout is one of the fundamental issues that geographers have long been concerned about, involving how to reasonably arrange elements in geographical space to achieve specific social, economic, and environmental goals; a reasonable spatial layout can promote the efficient utilization of resources, improve the quality of life of residents, protect the natural environment, and promote regional sustainable development. Spatial optimization decision-making is a key issue in the field of geographic information systems and is usually defined as multi-objective optimization with constraints. Industrial transfer is the change and migration of industries in spatial layout, and in essence, it is a process of matching the types of industries to be transferred to specific geographical units to achieve the optimal resource and environmental effects. There is an interdependent and mutually promoting relationship among spatial layout, spatial optimization decision-making, and industrial transfer. Industrial transfer is an embodiment of the dynamic adjustment of spatial layout, spatial layout is the basis for realizing spatial optimization decision-making, and spatial optimization decision-making is the key means to ensure that industrial transfer and spatial layout achieve the expected social, economic, and environmental goals. Since the spatial optimization method was first proposed in 1959, through continuous development and improvement, it has formed a rich theoretical system and application results in the field of geography.

[0003] At the same time, the progress of Geographic Information System (GIS) technology has provided a powerful tool and platform for geography. It not only provides rich spatial data analysis tools for geographical research but also can visualize, exploratory analyze, and build models for geographical data, enabling the wide application and in-depth development of spatial optimization methods. Through GIS technology, researchers can quickly and accurately process and analyze spatial data, providing basic data and decision-making support for subsequent spatial optimization. In addition, GIS can provide decision-makers with intuitive map displays and spatial analysis results to help them better understand problems and make decisions.

[0004] Currently, GIS mainly combines two optimization algorithms to solve the spatial layout optimization problem. One is the numerical optimization model, and the other is the intelligent optimization model. The former is prone to problems such as large computational volume and low accuracy when solving complex non-linear combinatorial optimization problems of spatial optimization decision-making, while intelligent algorithms can obtain approximate optimal solutions within an acceptable time. Although these studies have laid a foundation for the spatial optimization of industrial transfer, they have not incorporated the influencing factors and their spatial differences into the model establishment. At the same time, the dynamic changes of the human and natural environments are ignored.

[0005] In contrast, the Cellular Automata (CA) model can be used to simulate the impact of the dynamic changes in the human and natural environments on the research object. The CA model has powerful spatial computing capabilities and is widely used in the study of the evolution process of self-organizing systems. The three core characteristics of this model - the discreteness of time, space, and state, and the local grid dynamics characteristics - make it an ideal tool for simulating the spatio-temporal evolution process of complex systems. In recent years, scholars have achieved many meaningful research results in urban system simulation using the CA model. These research results show that complex urban spatial structures can be simulated through simple local transformation rules, reflecting the essence of complexity science that "complex systems result from the interaction of simple subsystems" and providing a reliable basis for theoretical research in geography and other fields.

[0006] Many phenomena in geography belong to the category of dynamic complex systems, which are mainly characterized by openness, dynamics, self-organization, and non-equilibrium dissipative structures. Taking the urban system as an example, its development and evolution are jointly influenced by many factors such as natural conditions, social structure, economic level, cultural elements, political background, and legal framework, presenting a highly complex behavior pattern. Given this complexity, urban CA models must comprehensively consider these factors. Although CA has the potential to simulate certain characteristics of urban systems, a single-form CA model is difficult to comprehensively simulate the diverse characteristics of cities. Therefore, the academic community has proposed various CA models to simulate the extensive characteristics of cities from multiple perspectives.

[0007] In the early CA models, the utilization of spatial information was quite limited, which hindered their effective integration with GIS. However, since the 1990s, scholars have begun to attempt to combine CA with GIS, and this integration has significantly improved the accuracy and reliability of CA in simulating actual urban situations. When dealing with complex spatial relationships, the functions of existing GIS have certain limitations. To better study the complex spatio-temporal dynamic change characteristics of geographical systems, it is necessary to couple dynamic models in GIS. CA has powerful spatial modeling and computing capabilities and can simulate complex dynamic systems with spatio-temporal characteristics, just making up for the deficiencies of GIS.

[0008] However, there are still some limitations in the CA model when simulating the dynamic changes of geographical phenomena. On the one hand, many CA models use unified transformation rules to drive the evolution of all cells, ignoring the spatial heterogeneity of geographical phenomena and their related factors, thus limiting the simulation accuracy of the geographical CA model. On the other hand, these models usually assume that all cells evolve at the same rate, which does not conform to the dynamic change law of actual geographical phenomena and also limits the improvement of simulation accuracy. When the CA model is used to simulate real urban development, especially when simulating the changes of different land use types, the complexity of the model increases significantly, and the accuracy and reality of the simulation effect cannot be guaranteed. At the same time, in the existing technology, there are many studies on the optimization of industrial spatial layout with resource environment as the constraint condition, and the research results generally only include qualitative descriptions, without involving operable planning strategies for spatial optimization. Although there are a small number of evaluation methods for ecological carrying capacity combined with GIS zoning, they only use the concept of ecological carrying capacity to guide industrial layout, and no regulatory model for the interaction between ecological carrying capacity and industrial layout has been formed. Summary of the Invention

[0009] In view of the above defects or deficiencies in the prior art, the present invention aims to provide an industrial transfer space decision-making method and system based on the ANN-GIS-CA architecture. On the basis of fully considering the industrial transfer bearing capacity and the social, economic and ecological environment effects of industrial transfer, the spatio-temporal impact process of industrial bearing capacity on industrial transfer and the spatio-temporal process of industrial transfer on resource environment effects are mapped to the geographical information environment GIS, and the spatio-temporal information mechanism of industrial transfer is constructed. Starting from the spatial differences of industrial transfer, the constrained CA and zonal asynchronous CA methods are integrated to construct an optimization model for industrial spatial layout, so as to improve the accuracy and effectiveness of spatial layout optimization.

[0010] To achieve the above object, the embodiments of the present invention adopt the following technical solutions:

[0011] In a first aspect, an embodiment of the present invention provides an industrial transfer space decision-making method based on the ANN-GIS-CA architecture, and the method includes the following steps:

[0012] Step S1, dividing the industrial transfer recipient area into a number of grids based on the geographical information system GIS to form cells, and using manual interpretation to mark the existing industrial transfer situation in the grids, generating grid data for the base year and verification year based on GIS; the grid data includes spatial information and attribute information; at the same time, judging the sub-region to which the grid belongs according to a preset rule, and taking the attribution result as the attribute information of the grid data;

[0013] Step S2, obtaining the original data of the industrial transfer bearing capacity of all grids based on the geographical information system GIS, and constructing an evaluation index system for the industrial transfer bearing capacity;

[0014] Step S3: Based on the constructed evaluation index system for the industrial transfer acceptance capacity, calculate the industrial transfer acceptance capacity scores of all grids, and use the Gaussian mixture model to identify the industrial types suitable for acceptance in each grid. Moreover, take the scores and the suitable industrial types for acceptance as the attribute information of the grid data;

[0015] Step S4: Based on GIS, obtain the original data of the socioeconomic effects and ecological environment effects of all grids in undertaking industrial transfer, and construct an evaluation index system for the socioeconomic effects and ecological environment effects of undertaking industrial transfer. The structure of the constructed acceptance effect evaluation index system includes primary indicators, secondary indicators, and the source and direction attributes of the secondary indicators. The direction attributes include positive indicators and negative indicators. Among them, a positive indicator refers to an indicator with the better the attribute value, while a negative indicator refers to an indicator with the smaller the attribute value;

[0016] Step S5: According to the secondary indicators and the direction attributes of the secondary indicators of the socioeconomic effects and ecological environment effects in the acceptance effects respectively, calculate the socioeconomic effect scores and ecological environment effect scores of all grids, and take the scores as the attribute information of the grid data;

[0017] Step S6: Select all grids with existing industrial transfer in the base year, and randomly select a predetermined number of grids without existing industrial transfer in the base year. Extract the socioeconomic effect scores and ecological environment effect scores of these grids as samples, construct an artificial neural network ANN model, calculate the state probabilities of all grids evolving into six states: resource-intensive, labor-intensive, technology-intensive, capital-intensive, comprehensive, and without existing industrial transfer respectively, and take the six states and the state probabilities as the attribute information of the grid data;

[0018] Step S7: Use GIS to determine whether each grid belongs to a prohibited development area, and take the determination result as the attribute information of the grid data;

[0019] Step S8: According to the grid spatial information in the base year, use GIS to calculate the Euclidean distance of each grid from the existing industrial grids, and take the Euclidean distance as the attribute information of the grid data;

[0020] Step S9: Take the grid as a cell, take the six states of resource-intensive, labor-intensive, technology-intensive, capital-intensive, comprehensive, and without existing industry as the states of the cells, and take the industrial type suitable for acceptance in each grid, the six state probabilities, whether each grid is a prohibited development area, and the Euclidean distance of each grid from the existing industrial grids as the attributes of the cells, and construct a partitioned asynchronous cellular automaton model;

[0021] Step S10: Calculate the asynchronous evolution rate of the sub-region from the base year to the validation year based on the grid data of the base year and the validation year; substitute the grid data of the base year and iterate the cellular automaton model multiple times. When the cellular automaton model iterates, it traverses all cells, and a cell will be visited only when the iteration count is an integer multiple of the evolution rate and the cell state is no existing industrial transfer. When the total number of cells in the states of resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive types is greater than or equal to the proportion threshold of the total number of grids in the states of resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive types in the validation year, the iteration of the cellular automaton model pauses.

[0022] Step S11: Set the accuracy threshold and verify the accuracy of the iteration result of the cellular automaton model. If the accuracy reaches the accuracy threshold, go to step S12; if not, return to S10.

[0023] Step S12: Divide the future industrial transfer scenarios into five different simulation scenarios; then use the Markov chain method to predict the demand quantities of the five industrial types of resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive types under each future scenario, and use the demand quantities as the termination conditions of the model to simulate the spatio-temporal processes of future industrial transfer under the five different scenarios respectively. Among them, the five different simulation scenarios include the baseline scenario, the major infrastructure construction scenario, the ecological protection priority scenario, the economic growth priority scenario, and the coordinated development scenario of ecological protection and economic growth.

[0024] As a preferred embodiment of the present invention, the structure of the bearing capacity evaluation index system constructed in step S2 includes first-level indicators, second-level indicators, third-level indicators, and the sources and direction attributes of the third-level indicators; the direction attributes include positive indicators and negative indicators; among them, a positive indicator refers to an indicator with the larger the attribute value, the better, while a negative indicator refers to an indicator with the smaller the attribute value, the better.

[0025] When calculating the industrial transfer bearing capacity scores of all grids, use the global principal component analysis (GPCA) method according to the third-level indicators and the direction attributes of the third-level indicators.

[0026] As a preferred embodiment of the present invention, in step S5, calculate the social and economic effect scores and the ecological and environmental effect scores based on the second-level indicators and their direction attributes. The calculation process specifically includes the following steps:

[0027] Step S51: Respectively perform dimensionless normalization processing on the positive indicators and negative indicators in the second-level indicators using different range normalization formulas to obtain the normalized second-level indicators.

[0028] Step S52: Determine the preliminary weights of the secondary indicators through the Generalized Principal Component Analysis (GPCA) method, the Entropy Weight Method (EWM), and the Analytic Hierarchy Process (AHP) respectively, and integrate the three preliminary weights using the Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS) method and the weighted average method to obtain two integrated weights. Then calculate the average of the two integrated weights as the comprehensive weight of each finally obtained secondary indicator.

[0029] Step S53: According to the normalized secondary indicators and the comprehensive weights of each secondary indicator, calculate the social - economic effect score and the ecological - environmental effect score of each grid respectively, and quantify the social - economic effect and the ecological - environmental effect of each grid through the scores.

[0030] As a preferred embodiment of the present invention, Step S52 further includes:

[0031] Step S521: Calculate the first preliminary weight of the secondary indicators through the GPCA method.

[0032] Step S522: Calculate the second preliminary weight of the secondary indicators using the EWM method.

[0033] Step S523: Calculate the third preliminary weight of the secondary indicators using the AHP method.

[0034] Step S524: Use the TOPSIS method to integrate the first preliminary weight, the second preliminary weight, and the third preliminary weight to obtain the first integrated weight.

[0035] Step S525: Use the weighted average weight method to integrate the first preliminary weight, the second preliminary weight, and the third preliminary weight to obtain the second integrated weight.

[0036] Step S526: Synthesize the first integrated weight and the second integrated weight to obtain the comprehensive weight of each secondary indicator. The first integrated weight represents the weight obtained by the objective weighting method, and the second integrated weight represents the weight obtained by the subjective weighting method.

[0037] As a preferred embodiment of the present invention, in Step S7, the prohibited development areas include three situations: the water area in the grid is greater than 50% of the total area of the grid; the grid is located within a protected area where any form of development is prohibited according to relevant policies; the average slope of the grid is greater than 15°.

[0038] As a preferred embodiment of the present invention, in Step S9, the partitioned asynchronous cellular automaton model consists of cells, a cell space, a neighborhood, and a transition rule, where the transition rule adopts the roulette selection rule.

[0039] In the described partitioned asynchronous cellular automaton model, socioeconomic effects, ecological and environmental effects, and Euclidean distance are introduced as external factor constraint conditions. At the same time, grid data based on the GIS data platform is adopted to enhance the ability to express complex geographical phenomena.

[0040] As a preferred embodiment of the present invention, the specific steps of the roulette wheel selection rule are as follows:

[0041] Step S91, calculate the total fitness: First, calculate the total fitness of all individuals in the population;

[0042] Step S92, calculate the selection probability: The selection probability of each individual is equal to its fitness divided by the total fitness;

[0043] Step S93, construct the roulette wheel: Map the selection probability of each individual onto a roulette wheel to form a cumulative probability distribution;

[0044] Step S94, select an individual: By generating a random number between 0 and 1, determine the position on the roulette wheel, and thus select the corresponding individual.

[0045] As a preferred embodiment of the present invention, in step S10, when a cell is accessed, the probabilities of the cell evolving into six states of resource-intensive, labor-intensive, technology-intensive, capital-intensive, comprehensive, and no existing industrial transfer are substituted into the roulette wheel selection model to randomly select an industrial type; then determine whether the cell is a prohibited development area, and whether the Euclidean distance between the cell and the existing industrial transfer grid is less than or equal to the average Euclidean distance between the existing industrial grids in the administrative region where the cell is located; after meeting these two conditions, if the industrial type randomly selected by the roulette wheel selection model is compatible with the industrial type suitable for the cell to undertake, the state of the cell changes to the industrial type randomly selected by the roulette wheel selection rule, otherwise the state of the cell remains no existing industrial transfer; every time an iteration is performed, the model calculates the total number of cells in the states of resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive.

[0046] As a preferred embodiment of the present invention, in step S11, when running the five different scenarios simulated, the simulation processes are as follows:

[0047] Run the baseline scenario, simulate the future based on the change rules of industrial transfer in the base year and the verification year, that is, without making any modifications to the cellular automaton model, continue to run the cellular automaton suspended in S10 until the quantity requirement is met;

[0048] Run the scenario of major transportation infrastructure construction. Based on the baseline scenario, in this scenario, the area within the predetermined range of the newly added transportation infrastructure that may be newly built in the future in the industrial transfer recipient area is also set as the transferable area. That is, a new rule is added to the model in S9, and then the cellular automaton suspended in S10 is continued to run until the quantity requirement is met.

[0049] Run the scenario of giving priority to ecological protection. In this scenario, the ecological environment effects of industrial transfer of cells are classified according to the standard deviation classification method. The areas with higher ecological environment effects are regarded as the areas where industrial transfer will occur in the future. That is, in the model described in S9, change "whether the Euclidean distance between the cell and the existing industrial transfer grid is less than or equal to the average Euclidean distance between the existing industrial grids in the administrative region where the cell is located described in S8" to "the ecological environment effect value of the cell is greater than 1.5 standard deviations of all ecological environment effect values", and then continue to run the cellular automaton suspended in S10 until the quantity requirement is met.

[0050] Run the scenario of giving priority to economic growth. In this scenario, the social and economic effects of industrial transfer of cells are classified according to the standard deviation classification method. The areas with higher social and economic effects are regarded as the areas where industrial transfer will occur in the future. That is, in the model described in S9, change "whether the Euclidean distance between the cell and the existing industrial transfer grid is less than or equal to the average Euclidean distance between the existing industrial grids in the administrative region where the cell is located described in S8" to "the social and economic effect value of the cell is greater than 1.5 standard deviations of all social and economic effect values", and then continue to run the cellular automaton suspended in S10 until the quantity requirement is met.

[0051] Run the scenario of coordinated development of ecological protection and economic growth. The areas with higher ecological environment effects and social and economic effects are regarded as the areas where industrial transfer will occur in the future, which is equivalent to the overlapping part of the areas described in the ecological protection priority scenario and the economic growth priority scenario. In the model described in S9, change "whether the Euclidean distance between the cell and the existing industrial transfer grid is less than or equal to the average Euclidean distance between the existing industrial grids in the administrative region where the cell is located described in S8" to "the ecological environment effect value of the cell is greater than 1.5 standard deviations of all ecological environment effect values, and the social and economic effect value of the cell is greater than 1.5 standard deviations of all social and economic effect values", and then continue to run the cellular automaton suspended in S10 until the quantity requirement is met.

[0052] In the second aspect, the embodiment of the present invention further provides an industrial transfer spatial decision-making system based on the ANN-GIS-CA architecture, the system comprising: a grid division module, a carrying capacity evaluation index construction module, a carrying capacity calculation module, a carrying effect evaluation index construction module, a carrying effect evaluation module, a state probability calculation module, a forbidden area determination module, a Euclidean distance calculation module, a CA model construction module, a CA model iteration module, an accuracy verification module and a scenario simulation module; wherein,

[0053] The grid division module is used to divide the industrial transfer receiving area into several grids based on the geographic information system GIS to form cells, and use manual interpretation to mark the existing industrial transfer situation in the grid, and generate grid data of the basic year and the verification year based on GIS; the grid data includes spatial information and attribute information; it is also used to determine the sub-area to which the grid belongs according to preset rules, and use the attribution result as the attribute information of the grid data;

[0054] The module for constructing the evaluation index of industrial transfer capacity is used to obtain the original data of industrial transfer capacity of all grids based on the geographic information system GIS, and to construct an evaluation index system of industrial transfer capacity;

[0055] The undertaking capacity calculation module is used to calculate the industrial transfer undertaking capacity scores of all grids based on the constructed industrial transfer undertaking capacity evaluation index system, and use the Gaussian mixture model to identify the industry type that each grid is suitable for undertaking, and use the scores and the industry type that is suitable for undertaking as the attribute information of the grid data;

[0056] The construction module of the evaluation index of the undertaking effect is used to obtain the original data of the social and economic effects and the ecological and environmental effects of all grids undertaking industrial transfer based on GIS, and to construct the evaluation index system of the social and economic effects and the ecological and environmental effects of undertaking industrial transfer; the constructed evaluation index system structure of the undertaking effect includes the primary index, the secondary index, and the source and direction attributes of the secondary index; the direction attributes include positive index and negative index; wherein the positive index refers to the index whose attribute value is larger, the better, and the negative index refers to the index whose attribute value is smaller, the better;

[0057] The undertaking effect evaluation module is used to calculate the social and economic effect scores and the ecological and environmental effect scores of all grids according to the secondary indicators of the social and economic effect and the ecological and environmental effect in the undertaking effect and the directional attributes of the secondary indicators, and use the scores as the attribute information of the grid data;

[0058] The state probability calculation module is used to select all the grids with existing industrial transfers in the base years, randomly select a predetermined number of grids without existing industrial transfers in the base years, extract the socio-economic effect scores and ecological environment effect scores of these grids as samples, construct an artificial neural network ANN model, calculate the state probabilities of all grids evolving into six states: resource-intensive, labor-intensive, technology-intensive, capital-intensive, comprehensive, and without existing industrial transfers, and use the six states and state probabilities as the attribute information of the grid data;

[0059] The prohibited development area determination module is used to use GIS to determine whether each grid belongs to a prohibited development area, and use the determination result as the attribute information of the grid data;

[0060] The Euclidean distance calculation module is used to calculate the Euclidean distance between each grid and the existing industrial grids based on the grid spatial information in the base years using GIS, and use the Euclidean distance as the attribute information of the grid data;

[0061] The CA model construction module is used to construct a partitioned asynchronous cellular automaton model with grids as cells, six states of cells being resource-intensive, labor-intensive, technology-intensive, capital-intensive, comprehensive, and without existing industries, and the suitable industrial types for each grid to undertake, six state probabilities, whether each grid is a prohibited development area, and the Euclidean distance between each grid and the existing industrial grids as the attributes of the cells;

[0062] The CA model iteration module is used to calculate the sub-region asynchronous evolution rate from the base year to the verification year based on the grid data of the base year and the verification year; substitute the grid data of the base year and iterate the cellular automaton model multiple times; when iterating, the cellular automaton model will traverse all cells, and only when the iteration times are an integer multiple of the evolution rate and the cell state is without existing industrial transfer will the cell be accessed; when the total number of cells in the resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive states is greater than or equal to the proportion threshold of the total number of grids in the resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive states in the verification year, the cellular automaton model pauses iteration;

[0063] The accuracy verification module is used to set an accuracy threshold and verify the accuracy of the iteration results of the cellular automaton model; if the accuracy reaches the accuracy threshold, the scenario simulation module is started, and if it does not reach the accuracy threshold, the CA model iteration module is started;

[0064] The scenario simulation module is used to divide the future industrial transfer scenarios into five different simulation scenarios; then, the Markov chain method is used to predict the demand quantities of five industrial types, namely resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive types, under each future scenario, and the demand quantities are used as the termination conditions of the model to respectively simulate the spatio-temporal processes of future industrial transfer under the five different scenarios; among them, the five different simulation scenarios include a baseline scenario, a major infrastructure construction scenario, an ecological protection priority scenario, an economic growth priority scenario, and a scenario of coordinated development of ecological protection and economic growth.

[0065] The technical solutions provided by the embodiments of the present invention have the following beneficial effects:

[0066] The industrial transfer spatial decision-making method and system based on the ANN-GIS-CA architecture provided by the embodiments of the present invention constructs an industrial space layout optimization model for industrial transfer by integrating ANN, GIS, and CA, and predicts the spatio-temporal patterns of future industrial transfer under different scenarios; uses the CA model to simulate industrial transfer; and more accurately simulates the spatio-temporal optimization process of industrial transfer through a combination of qualitative, quantitative, and positioning methods, and subjective and objective methods, predicts the development trends of industrial transfer under different development modes, improves the optimization effect of the industrial space layout during industrial transfer, rationally arranges the industrial transfer space, and realizes the sustainable development of regional industries.

[0067] Of course, it is not necessary for any product or method implementing the present invention to simultaneously achieve all the above-mentioned advantages. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0069] Figure 1 is the flowchart of the industrial transfer spatial decision-making method based on the ANN-GIS-CA architecture provided by the embodiments of the present invention;

[0070] Figure 2 is the schematic diagram of the structure of the cellular automaton model adopted by the embodiments of the present invention;

[0071] Figure 3 is the schematic diagram of the structure of the geographical cellular automaton model constructed by the embodiments of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0072] After discovering the above problems, the inventors of this application conducted a detailed study on the optimization of the industrial spatial layout brought about by the existing industrial transfer. The study found that when coupling the powerful spatial modeling and computing capabilities of CA with the GIS database platform to improve the optimization of the industrial spatial layout, more external factors need to be considered and integrated as the constraint conditions of the model in the CA model to improve the accuracy and realism of the simulation.

[0073] All the cell state transition rules of the traditional CA model are the same and only consider the neighborhood effect, and cannot depict the spatial differentiation law of geographical objects and the influence of other factors. Especially in the context of industrial transfer, the application of CA can significantly depict the impact of industrial migration on land use patterns. In each iteration process, the change of land use is determined by the combined action of all transfer functions. By integrating the planning objectives (including the trend and goal of industrial transfer) into the transfer function, the change of land use can be effectively guided and controlled.

[0074] In addition, in the past few decades, satellite remote sensing technology has become an important data source for many geographical studies, providing a large amount of valuable information about surface features. Satellite remote sensing images provide key land use information for the CA model, which is crucial for model initialization. Since remote sensing data is essentially in a raster structure, it can be seamlessly integrated into the input data of the CA model, thus greatly improving the accuracy and practicality of the CA model in simulating complex geographical phenomena.

[0075] However, for the industrial transfer in different regions, different constraint conditions need to be considered due to the specific land resource utilization situation, geographical, economic, and humanistic environments at that time and place. For example, when a series of severe environmental problems such as water pollution, land degradation, biodiversity loss, and frequent natural disasters are faced within a region, these problems become rigid conditions restricting economic development. As an important aspect of economic development, industrial transfer is also restricted by the resource environment. Therefore, how to optimize the spatial layout of industrial transfer is an urgent problem to be solved for the sustainable development of regional industries. And how to establish a multi-scenario spatial optimization model of industrial transfer based on the constrained CA model under the influence of the resource environment is the key problem that needs to be solved for spatial optimization decision support.

[0076] It should be noted that the defects existing in the above solutions of the prior art are all the results obtained by the inventors through practice and careful research. Therefore, the process of discovering the above problems and the solutions proposed by the embodiments of the present invention below for the above problems should both be the contributions made by the inventors to the present invention during the process of the present invention.

[0077] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention described and illustrated in the accompanying drawings here can be arranged and designed in various different configurations. It should be noted that, without conflict, the embodiments and features in the embodiments of the present invention can also be combined with each other.

[0078] It should be noted that: similar reference numerals and letters indicate similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. In the description of the present invention, the terms "first", "second", "third", "fourth", etc. are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.

[0079] After the above in-depth analysis, the embodiments of the present invention provide an industrial transfer space decision-making method and system based on the ANN-GIS-CA architecture. The method includes: First, based on the industrial transfer acceptance capacity and the social and economic effects and ecological environment effects of industrial transfer, the spatio-temporal impact process of industrial acceptance capacity on industrial transfer and the spatio-temporal process of industrial transfer on resource and environmental effects are mapped to the geographic information environment, and the spatio-temporal information mechanism for analyzing industrial transfer is constructed. Second, starting from the spatial differences of industrial transfer, the constrained CA and zonal asynchronous CA methods are integrated, and the constraint effects of industrial acceptance capacity, social and economic effects and ecological environment effects of industrial transfer, political policies, etc. on industrial transfer are comprehensively considered to construct an industrial transfer space optimization decision support model. Finally, based on the decision support model, different industrial transfer scenarios are set, the advantages and existing problems of different industrial transfers in different scenarios are analyzed, and on this basis, the model parameters are continuously adjusted to realize the spatial optimization configuration analysis of industrial transfer.

[0080] Specifically, as Figure 1 shown, the industrial transfer space decision-making method based on the ANN-GIS-CA architecture includes the following steps:

[0081] Step S1, divide the industrial transfer recipient areas into several grids based on the geographic information system GIS to form cells, and use manual interpretation to mark the existing industrial transfer situations in the grids to generate grid data for the base year and verification year based on GIS; the grid data includes spatial information and attribute information; at the same time, judge the sub-regions to which the grids belong according to preset rules, and use the attribution results as the attribute information of the grid data.

[0082] In this step, the preset rules include division by administrative attributes, division by geographic coordinates, etc. Finally, the receiving area will be divided into several square grids. The location information and transaction information of enterprises are found using the enterprise list. The grid where the enterprise is located is determined based on the location information, and the enterprise type is determined based on the transaction information. Then, all grids are marked with six states: resource-intensive, labor-intensive, technology-intensive, capital-intensive, comprehensive, or no existing industrial transfer. The data for the base year and the verification year are marked separately.

[0083] Supported by computer hardware and software systems, the GIS is used to collect, store, manage, calculate, analyze, display, and describe the geographical distribution data in the space of the earth's surface (including the atmosphere). Many geographical phenomena belong to typical dynamic complex systems and have the characteristics of dissipative structures such as openness, dynamics, self-organization, and non-equilibrium.

[0084] The grid data is a type of vector data based on the GIS, including spatial information and attribute information. The spatial information includes coordinates, etc., and the attribute information includes sub-region attribution, etc. At the same time, the attribute information will increase the attribute items and corresponding assignments with the calculation of the grid information. For example, the corresponding receiving capacity score, six states, etc. Usually, the calculated values of the grid will be listed as attribute items in the attribute information of the grid data to ensure the integrity and application of the grid data.

[0085] Step S2: Obtain the original data of the industrial transfer receiving capacity of all grids based on the Geographic Information System (GIS), and construct an evaluation index system for the industrial transfer receiving capacity; the structure of the constructed receiving capacity evaluation index system includes first-level indicators, second-level indicators, third-level indicators, and the source and direction attributes of the third-level indicators; the direction attributes include positive indicators and negative indicators; among them, a positive indicator refers to an indicator with the better the attribute value, and a negative indicator refers to an indicator with the smaller the attribute value.

[0086] In this step, the space to be optimized is the destination of industrial transfer, that is, the receiving area. Transferring industries to the receiving area involves changes and optimizations in the spatial layout of the receiving area. The first thing to do is to evaluate the ability of this area to receive the transferred industries, that is, the receiving capacity. This receiving capacity is related to the natural, historical, and technical and organizational conditions of this area. For example, for the manufacturing industry as the primary industry, having the ability to receive the transfer of the manufacturing industry means that the receiving area should have a certain attraction for the relevant manufacturing industry to "draw it in", and at the same time, it should have certain basic conditions to enable enterprises to survive and have the potential for sustainable development.

[0087] Based on this, the industrial transfer acceptance capacity proposed in this embodiment includes three aspects: attraction, support and development. Among them, the political factors, natural resource endowment, economic development level, market attractiveness, labor quality and cost input of the receiving area determine the attractiveness of the transferred manufacturing industry and can bring in related industries; support is the ability of the receiving industry to survive. To introduce the transferred industry into the region, integrate it into the regional economy and survive, it is necessary to fully consider the urbanization level, transportation infrastructure, communication infrastructure and environmental acceptance of the receiving area; development is the ability of the transferred industry to integrate with the original industry in the region to promote the optimization and upgrading of the industrial structure, and to continuously lengthen the regional characteristic industrial chain, thereby forming a competitive advantage. It is generally determined by the receiving area's ability to attract foreign investment, economic growth rate and information development potential.

[0088] In a specific embodiment, the raw data of the receiving capacity at least includes the following data of the receiving area: greenhouse gas, land use, population density, night lights, traffic network, PM2.5, communication base stations, mineral resources, panel data, etc. As shown in Table 1, each type of data is further divided into specific detailed types and their related sources. For example, the panel data further includes: total water resources, number of labor, total number of middle school enrollment, employee remuneration, cost of business start-up process, natural resource rent, GDP growth rate, per capita GNI, net inflow of foreign direct investment, electricity access rate, number of secure Internet servers, and urbanization rate, etc.

[0089] Table 1 Original data types and sources of the carrying capacity of the receiving area

[0090]

[0091] Based on the acquired original data of carrying capacity, the constructed carrying capacity evaluation index system structure includes primary indicators, secondary indicators, tertiary indicators and indicator directions. The primary indicators include several secondary indicators, the secondary indicators include several tertiary indicators, and the indicator direction is set for each type of tertiary indicator, that is, the consistency between the indicator value change and the direction of target achievement. The directional attributes of the indicators include positive indicators or negative indicators; among them, positive indicators refer to the higher the value, the better, while negative indicators refer to the lower the value, the better.

[0092] In a specific embodiment, as shown in Table 2, the primary indicators include: attraction, support and development. Attraction refers to the degree of attraction of a region or country to external enterprises and industries during the industrial transfer process, which is one of the key factors affecting the industrial transfer acceptance capacity; support refers to the various support conditions and guarantee capabilities that a region has when receiving and stably developing newly transferred industries; development is the ability of a region to effectively promote the sustainable development of new industries and continuously improve its competitiveness after receiving and integrating foreign industrial transfers.

[0093] The attractiveness index includes at least the following secondary indicators: political factors, natural resource endowment, economic development level, market attractiveness, labor quality and cost input; the supporting capacity index includes at least the following secondary indicators: urbanization level, transportation infrastructure, communication infrastructure and environmental carrying capacity; the development capacity index includes at least the following secondary indicators: ability to attract foreign investment, economic growth rate and information development potential.

[0094] Table 2 Industrial transfer acceptance index system

[0095]

[0096]

[0097] Step S3, based on the constructed industrial transfer acceptance capacity evaluation index system, calculate the industrial transfer acceptance capacity scores of all grids, and use the Gaussian mixture model to identify the industry type that each grid is suitable for accepting, and use the scores and the industry types suitable for accepting as attribute information of the grid data.

[0098] In this step, the calculation of the industrial transfer acceptance capacity of the receiving area adopts the time series global principal component analysis (GPCA) method. The specific calculation process is as follows:

[0099] Step S31, respectively, dimensionlessly normalize the positive indicators and negative indicators in the three-level indicators using different range standardization formulas to obtain standard three-level indicators.

[0100] Step S32, based on the standard three-level indicators and according to the time series global principal component analysis method, construct a time series space subject carrying capacity data table; and solve the matrix of the data table.

[0101] Step S33, converting the data table into a matrix form, and calculating the covariance matrix V of the matrix, then solving the eigenvalues ​​and corresponding eigenvectors of the data table matrix through the covariance matrix, and calculating the contribution rate and cumulative contribution rate of the eigenvector.

[0102] When calculating the cumulative contribution rate, the contribution rates of all eigenvectors are first sorted from large to small; based on the contribution rates of the sorted eigenvectors, the cumulative contribution rate is calculated.

[0103] Step S34, taking the standard three-level indicators whose cumulative contribution rate is greater than a preset threshold (for example, 80%) as evaluation indicators; and calculating the weights of all the standard three-level indicators used as evaluation indicators according to the contribution rate of the feature vector.

[0104] Step S35, constructing a scoring calculation function for the current city's industrial transfer acceptance capacity, substituting all the standard third-level indicators and corresponding weights used as evaluation indicators into the scoring calculation function, and calculating the acceptance capacity scores of all grids.

[0105] Step S35, based on the acceptance score of each grid, a Gaussian mixture model (GMM) is used to calculate the industry type that each grid is suitable for accepting.

[0106] In this step, the GMM is a multivariate Gaussian distribution function composed of a linear combination of multiple independent Gaussian distribution functions, which can fit any type of distribution. It is usually used to solve the situation where the data in the same set contains multiple different distributions, and uses the maximum expected algorithm to estimate the parameters of the model, which can provide strong descriptive ability and can accurately quantify the research object. Through the industrial layout optimization model constructed based on the Gaussian mixture model, the receiving area is clustered according to the first indicator type (attractiveness, support and development power) of the administrative region, so that the clustering results are more in line with objective reality, helping all walks of life to carry out industrial transfer scientifically and rationally. At the same time, because it uses a combination of multiple Gaussian distributions to fit the data, it can adapt to data distributions of various shapes and sizes.

[0107] The specific operations for constructing a Gaussian mixture model are as follows:

[0108] Suppose there are n observation data x1, x2, …, x n , divided into k categories, the Gaussian mixture model can be expressed as:

[0109]

[0110] In formula (1), φ(x|θ k ) is the probability density of mixed Gaussian distribution, θ k =(μ k ,σ k 2 ), μ k and σ k 2 are the mean and variance of the kth Gaussian distribution, α j is the mixing coefficient of Gaussian distribution, that is, the weight of single Gaussian distribution, 0<α j <1,∑α j =1; p(x|θ) is the probability that x is classified into the kth class, and θ is the model parameter to be estimated, which is μ here. k , σ k 2 , α j 3 in total.

[0111] In this step, θ is estimated using the maximum likelihood method (Expectation Maximization, EM), that is, the obtained parameter θ should maximize the value of p(x|θ). The set optimization objective function is as follows:

[0112]

[0113] In Equation (2), N is the total number of samples, and N k is the number of samples in the k-th category.

[0114] The specific calculation steps of the EM algorithm are as follows: Step S351, determine the number of classification categories: one category corresponds to one Gaussian distribution, and thus a multivariate mixture Gaussian distribution function is formed; Step S352, determine the initial values of the function parameters: randomly assign the mean and variance of each Gaussian distribution function, as well as their mixing coefficients for combining into the mixture Gaussian distribution function; Step S353, calculate the probability of each sample under each Gaussian distribution; Step S354, find the maximum likelihood function of the average value of the k-th Gaussian distribution; Step S355, find the maximum likelihood function of the variance of the k-th Gaussian distribution; Step S356, find the maximum likelihood function of the mixing coefficient; Step S357, repeat Steps S352 to S355 until the model converges. The above EM algorithm is a prior art, and the specific parameters are set according to the needs of this embodiment and will not be elaborated here.

[0115] From the steps of the EM algorithm, it can be seen that its core is, after giving the initial values of the variables, updating the parameters according to the likelihood values, iteratively optimizing step by step to gradually increase the likelihood function value, and finally finding the optimal parameters. The one with the largest likelihood function value of the model is its value. By using the EM algorithm to fit the data, it is based on the observed variables to calculate the posterior probability of each component density.

[0116] Step S4, based on GIS, obtain the original data of the social and economic effects and ecological environment effects of all grid cells in undertaking industrial transfer, and construct an evaluation index system for the social and economic effects and ecological environment effects of undertaking industrial transfer; the structure of the constructed evaluation index system for undertaking effects includes first-level indicators, second-level indicators, and the source and direction attributes of the second-level indicators; the direction attributes include positive indicators and negative indicators; among them, a positive indicator refers to an indicator with the larger the attribute value, the better, while a negative indicator refers to an indicator with the smaller the attribute value, the better.

[0117] In this step, when evaluating the social and economic effects and ecological environment effects of all grid cells, the social and economic effects and ecological environment effects are used as two first-level indicators to construct an evaluation index system for the undertaking effects of industrial transfer in the receiving area.

[0118] As shown in Table 3, the socio-economic effect evaluation includes several secondary evaluation indicators such as changes in regional employed population, population density, industrial employed population density, per capita GDP, growth of industrial employed personnel, industrial added value, labor productivity, and education expenditure.

[0119] The production factor costs, market potential, and social development brought about by industrial transfer will change the regional employment situation. Industrial transfer can promote the frame skipping of the regional industrial structure, shorten the time of industrial upgrading, accelerate the industrialization process, create new employment opportunities for the receiving areas, promote the transfer of rural surplus labor to urban areas, and the transfer of employees in the primary industry to the secondary and tertiary industries, driving the growth of the employment rate. It also changes the spatial distribution of population density and employment rate. GDP represents the total market value of all final goods and services produced by a country within a certain period of time. Per capita GDP is usually used as an indicator to measure the average living standard of residents, and its change can reflect the economic structure adjustment and the economic benefits brought about by industrial upgrading. Input-production efficiency refers to the ratio of output (usually products or services) generated for each unit of input (such as labor, capital, raw materials, etc.) during the production process, and it is an important indicator for evaluating the economic benefits of industrial transfer. High input-production efficiency means that higher output can be obtained with a certain amount of resource input, thus enhancing the overall economic benefits. The fixed asset investment brought about by industrial transfer and the mergers and acquisitions of enterprises in the receiving areas increase the capital stock in the receiving areas and improve the quality of capital; in addition, industrial transfer also brings technical support and talent transfer to the receiving areas, realizing the optimization and upgrading of the industrial structure by enhancing the scale economic benefits and resource utilization efficiency. Industrial added value refers to the total value added in the production process of each industrial sector in a country or region within a certain period of time, which is equal to the total output of the industry minus the intermediate consumption (i.e., the cost of raw materials and services used in the production process). According to different industrial sectors, industrial added value can be divided into: primary industry added value, secondary industry added value, and tertiary industry added value. By analyzing the added value of different industries, the economic structure of a country or region can be understood. Industries with high added value usually have higher added value, reflecting the quality and sustainability of economic growth. Industrial transfer can drive the flow of technology from developed regions to underdeveloped regions, promoting the improvement of the local technical level. By promoting technology spillover, the innovation ability of enterprises is enhanced. Labor productivity can be expressed by the quantity of a certain product produced by the same labor in unit time. The more products produced in unit time, the higher the labor productivity; it can also be expressed by the labor time consumed in producing a unit product. The less labor time required to produce a unit product, the higher the labor productivity. Total factor productivity can reflect the production efficiency of an enterprise and the economic benefits of labor input. The total amount of public education expenditure generally refers to the total expenditure of the government on education within a specific period (usually a fiscal year). These expenditure costs mainly include: basic education expenditure, higher education expenditure, teachers' salaries and benefits, education expenditure facilities, education scientific research, etc.Education expenditure is constrained by a country's economic development. The following four indicators are usually used internationally to measure a country's education expenditure: the proportion of education expenditure in national income, the proportion of education expenditure in gross domestic product, the proportion of education expenditure in total national fiscal expenditure, and the proportion of education infrastructure expenditure in total infrastructure expenditure. Education expenditure can reflect the degree of importance a country or region places on education. A high proportion of education expenditure usually indicates that the region regards education as a priority area for development.

[0120] As shown in Table 3, the ecological and environmental effect evaluation includes several secondary evaluation indicators such as water quality, air pollutants, carbon storage changes, changes in land use type area, and changes in terrestrial ecosystem productivity.

[0121] During the industrial transfer process, the transfer of different types of industries such as mining, heavy industry, and agricultural development may lead to deforestation and land degradation, directly affecting the ecosystem in the basin and threatening biodiversity. And with the rapid development of industry, the number of enterprises discharging sewage and waste has increased, and the water quality of rivers may be seriously polluted, affecting the drinking water safety and agricultural water use of local residents. The transfer of some polluting industries may lead to increased local air pollution, especially industries that rely on fossil fuels, which may release a large amount of PM2.5, SO2 and other atmospheric pollutants, affecting air quality and public health. The population mobility caused by rapid industrial transfer may lead to a decline in the environmental carrying capacity of local communities, resulting in over-exploitation of resources and conflicts between local communities and enterprises.

[0122] Table 3 Indicator system and data sources for socio-economic and ecological environmental effects

[0123]

[0124]

[0125] Step S5, respectively calculating the socio-economic effect scores and eco-environmental effect scores of all grids according to the secondary indicators of socio-economic effect and eco-environmental effect in the undertaking effect and the directional attributes of the secondary indicators, and using the scores as the attribute information of the grid data.

[0126] In this step, the socio-economic effect score and the ecological environmental effect score are calculated based on the secondary indicators and their directional attributes. The calculation process specifically includes the following steps:

[0127] Step S51, respectively, dimensionlessly normalize the positive index and the negative index in the secondary index using different range normalization formulas to obtain a standardized secondary index.

[0128] In this step, the dimensionless normalization process is the same as the existing data normalization process.

[0129] In step S52, the preliminary weights of the secondary indicators are determined by the GPCA method, the Entropy Weight Method (EWM), and the Analytic Hierarchy Process (AHP) respectively, and the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) and the weighted average method are used to integrate the three preliminary weights to obtain two integrated weights. Then, the average value of the two integrated weights is calculated as the comprehensive weight of each secondary indicator finally obtained.

[0130] Specifically, it includes:

[0131] In step S521, the first preliminary weight of the secondary indicator is calculated by the GPCA method.

[0132] In this step, the GPCA method is an existing method and will not be elaborated here.

[0133] In step S522, the second preliminary weight of the secondary indicator is calculated by using the EWM method.

[0134] In this step, the EWM belongs to a kind of objective weighting method. The smaller the entropy value of the indicator, the greater the degree of variation of the indicator value, the greater the amount of information provided, and the greater the weight of this indicator.

[0135] In step S523, the third preliminary weight of the secondary indicator is calculated by using the AHP method.

[0136] In this step, the AHP method does not require prior probability estimation and is a systematic analysis method combining qualitative and quantitative methods. The steps for the AHP to determine the weights of evaluation indicators are as follows: First, establish a hierarchical structure; second, construct a pairwise judgment matrix, and use a scale method of 1 to 7 and its reciprocal to quantify the relative importance degree of pairwise elements; third, calculate the subjective weight w' i ; fourth, conduct a consistency test. When the Consistency Ratio (CR) < 0.1, it indicates that the consistency of the judgment matrix is reasonable; when CR ≥ 0.1, it means that the judgment matrix is unreasonable and needs to be re-tested for consistency.

[0137] In step S524, the TOPSIS method is used to integrate the first preliminary weight, the second preliminary weight, and the third preliminary weight to obtain the first integrated weight.

[0138] In this step, the TOPSIS method is a multi-objective decision analysis method, which selects a certain number of evaluation indicators according to different evaluation objects. The ideal value of each indicator is selected, and the distance between each scheme and the ideal value (i.e., the proximity) is calculated. Therefore, the TOPSIS method can be used to calculate the comprehensive weight of the subjective and objective weighting methods. The specific steps include:

[0139] Standardize the weight matrix;

[0140] Determine the ideal solution and the negative ideal solution;

[0141] Calculate the distances between each index value and the positive and negative ideal solutions;

[0142] Calculate the relative proximity, and calculate the relative proximity according to the distances from the ideal solution and the negative ideal solution;

[0143] Integrate the first preliminary weight, the second preliminary weight, and the third preliminary weight according to the relative proximity and the weight matrix to obtain the first integrated weight.

[0144] In step S525, adopt the weighted average weight method to integrate the first preliminary weight, the second preliminary weight, and the third preliminary weight to obtain the second integrated weight.

[0145] In this step, in addition to the TOPSIS model, in order to make the weight result closer to the actual situation, the preliminary weights obtained by the GPCA, EMW, and AHP methods can also be weighted and then the weighted average weight is calculated. Preferably, the weight of GPCA is 0.5, the weight of EMW is 0.3, and the weight of the AHP method is 0.2, and the weighted average weight is calculated:

[0146] w i = 0.5w GPCA + 0.3w EMW + 0.2w AHP (3)

[0147] In formula (3), w i is the weighted average weight of index i.

[0148] In step S526, integrate the first integrated weight and the second integrated weight to obtain the comprehensive weight of each secondary index. The first integrated weight represents the weight obtained by the objective weighting method, and the second integrated weight represents the weight obtained by the subjective weighting method. Therefore, the obtained comprehensive weight can comprehensively consider the importance of the indicators from both subjective and objective aspects.

[0149] In this step, combine the index weights calculated by TOPSIS and take the average of the two as the final weight:

[0150]

[0151] In formula (4), W i is the comprehensive weight of index i; w 1 i is the weight of index i obtained by the TOPSIS model, and w 2 i is the weighted average weight of index i.

[0152] Step S53: According to the normalized secondary indicators and the comprehensive weights of each secondary indicator, calculate the social and economic effect scores and ecological and environmental effect scores of each grid respectively, and quantify the social and economic effects and ecological and environmental effects of each grid through the scores.

[0153] In this step, after obtaining the comprehensive weights, the secondary indicators are weighted and summed to finally obtain the social and economic effects and ecological and environmental effects of each grid.

[0154] Step S6: Select all grids with existing industrial transfers in the base year, and randomly select a predetermined number of grids without existing industrial transfers in the base year. Extract the social and economic effect scores and ecological and environmental effect scores of these grids as samples to construct an artificial neural network ANN model, and calculate the state probabilities of all grids evolving into six states: resource-intensive, labor-intensive, technology-intensive, capital-intensive, comprehensive, and without existing industrial transfers respectively.

[0155] In this step, the input values of the ANN model are the social and economic effect scores and ecological and environmental effect scores of the sample grids, and the output values are the probability values of all grids evolving into six states: resource-intensive, labor-intensive, technology-intensive, capital-intensive, comprehensive, and without existing industrial transfers. In this step, the ANN performs operations and simulations by imitating the functions of the human brain. The neural network is composed of a series of neurons and has the powerful ability to perform complex parallel operations. The neural network cellular automaton consists of simple networks. The model contains two relatively independent modules: model training and simulation. The same neural network is used for these two modules. In the model training module, the parameters of the model are automatically obtained using the training data; then these parameters are input into the simulation module for simulation operations.

[0156] Step S7: Use GIS to determine whether each grid belongs to a prohibited development area, and use the determination result as the attribute information of the grid data.

[0157] The prohibited development areas in this step include three situations: the water area in the grid is greater than 50% of the total grid area, mainly considering the protection of water sources; the grid is located within a protected area, and any form of development is prohibited according to relevant policies; the average slope of the grid is greater than 15°, and it is generally difficult to carry out large-scale infrastructure construction in these areas.

[0158] Step S8: According to the grid spatial information of the base year, use GIS to calculate the Euclidean distance between each grid and the existing industrial grids, and use the Euclidean distance as the attribute information of the grid data.

[0159] Step S9: Using the grid as a cell, with six states of resource-intensive, labor-intensive, technology-intensive, capital-intensive, comprehensive, and no existing industry as the states of the cells, and with the industrial type suitable for each grid to undertake, the probabilities of the six states, whether each grid is a prohibited development area, and the Euclidean distance between each grid and the existing industrial grids as the attributes of the cells, construct a partitioned asynchronous cellular automaton model.

[0160] In this step, the cellular automaton is a grid dynamics model that divides space into discrete cells, each cell is assigned a discrete state and evolves within discrete time steps. The cellular automaton model can better express local temporal causal relationships and spatial interactions. The cellular automaton model can well simulate the evolution process of self-organizing systems, and at the same time it also has strong spatial computing capabilities. The CA model can simulate the spatio-temporal evolution process of complex systems through local interactions and temporal causal relationships.

[0161] As Figure 2 shown, the cellular automaton model consists of four basic parts: cells, cell space, neighborhood, and transition rules. The cellular automaton model is composed of a discrete cell space and cells with discrete states located within this cell space, neighborhoods, and their transition rules.

[0162] The cellular automaton model can express the concepts of time and space simultaneously, and express and simulate the spatio-temporal evolution process of complex systems through discrete cell space division, discrete state division, and discrete evolution steps. The geographical system is a complex giant system, and the spatio-temporal dynamic evolution process of complex geographical phenomena can be well expressed and simulated by the cellular automaton model, which is the geographical cellular automaton model.

[0163] As Figure 3As shown, the Geo - Cellular Automata (Geo - CA) model is a cellular automata model used to solve the dynamic simulation and expression of complex geographical phenomena. Therefore, like a general cellular automata model, the Geo - CA model is not a specific model but just a model framework. Under the framework of the Geo - CA model, the model can be specified according to the specific geographical problems that need to be analyzed and simulated, so as to meet the needs of geographical analysis. In fact, the Geo - CA model is an extension and specification of the general cellular automata model and is an effective means to simulate and analyze the spatio - temporal evolution laws of complex geographical phenomena. By integrating and integrating the cellular automata model with various theories and methods such as geographic information system technology, multi - agent model, data mining method, probabilistic reasoning method, and fuzzy logic, the spatio - temporal dynamic process of complex geographical phenomena can be scientifically analyzed and objectively simulated. Therefore, the Geo - CA model can explore the spatio - temporal evolution laws of complex geographical phenomena, predict the future development trends and patterns of geographical phenomena, and can also be incorporated into the research framework of the "Digital Earth". In the zonal asynchronous CA model of this embodiment, various external factors such as socio - economic effects, ecological environment effects, and Euclidean distance are introduced as constraint conditions, and combined with the GIS data platform, which enhances the expression ability of the zonal asynchronous CA model for complex geographical phenomena.

[0164] In order to enable the cellular automata model to be used to simulate complex geographical phenomena, it is necessary to expand and improve the cellular automata model. At the same time, it is necessary to integrate the theory of geographical concepts into the cellular automata model to make it meet the requirements of geographical simulation, so as to obtain geographical cells. As Figure 3 shown, the geographical cells partition the cell space composed of geographical space through spatial data mining, and determine the transformation rules for each partition respectively, and combine different evolution rates to simulate geographical phenomena.

[0165] The transformation rule adopts the roulette selection mechanism. The roulette selection is a selection mechanism commonly used in genetic algorithms, which simulates the process of roulette gambling and is used to select individuals from a population for reproduction. The probability of an individual being selected is determined according to its fitness, and the higher the fitness of an individual, the greater the probability of being selected. The specific steps of the roulette model selection are as follows:

[0166] Step S91, calculate the total fitness: First, calculate the total fitness of all individuals in the population.

[0167] Step S92, calculate the selection probability: The selection probability of each individual is equal to its fitness divided by the total fitness.

[0168] Step S93, construct the roulette: Map the selection probability of each individual onto a roulette to form a cumulative probability distribution.

[0169] Step S94, Select an individual: Determine the position on the roulette wheel by generating a random number between 0 and 1, and thus select the corresponding individual.

[0170] Through roulette wheel selection, it is ensured that individuals with high fitness have a greater chance of being selected, but it also gives individuals with lower fitness a certain chance, maintaining the diversity of the population.

[0171] Step S10, Based on the grid data of the base year and the verification year, calculate the asynchronous evolution rate of the sub-region from the base year to the verification year; substitute the grid data of the base year and iterate the cellular automaton model multiple times; when the cellular automaton model iterates, it will traverse all cells, and a cell will be visited only when the iteration number is an integer multiple of the evolution rate and the cell state is no existing industrial transfer; when the total number of cells belonging to the resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive states is greater than or equal to the proportion threshold of the total number of grids belonging to the resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive states in the verification year, the cellular automaton model pauses iteration.

[0172] In this step, when a cell is visited, the probabilities of the cell evolving into six states of resource-intensive, labor-intensive, technology-intensive, capital-intensive, comprehensive, and no existing industrial transfer will be substituted into the roulette wheel selection model to randomly select an industrial type; next, it is judged whether the cell is a prohibited development area, and whether the Euclidean distance between the cell and the existing industrial transfer grid is less than or equal to the average Euclidean distance between the existing industrial grids in the administrative region where the cell is located; after meeting these two conditions, if the industrial type randomly selected by the roulette wheel selection rule is compatible with the industrial type suitable for the cell to undertake, the state of the cell will change to the industrial type randomly selected by the roulette wheel selection model, otherwise the state of the cell remains no existing industrial transfer. Each time an iteration is performed, the model will calculate the total number of cells belonging to the resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive states once.

[0173] In this step, the proportion threshold is selected according to the actual situation. Preferably, the proportion threshold in this step is 90%.

[0174] Step S11, Set the accuracy threshold and verify the accuracy of the iteration result of the cellular automaton model; if the accuracy reaches the accuracy threshold, go to step S12, otherwise return to S10.

[0175] In this step, the data of the base year is used to simulate the data of the verification year and verify the model accuracy. When the model accuracy meets the standard, the model can be used to simulate the future industrial transfer distribution. The model accuracy is judged whether it meets the standard by setting the accuracy threshold.

[0176] Step S12: Divide the future industrial transfer scenarios into five different simulation scenarios; then use the Markov chain method to predict the demand quantities of five industrial types, namely resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive types, under each future scenario, and use the demand quantity as the termination condition of the model to simulate the spatio-temporal process of future industrial transfer under the five different scenarios respectively; among them, the five different simulation scenarios include the baseline scenario, the major infrastructure construction scenario, the ecological protection priority scenario, the economic growth priority scenario, and the coordinated development scenario of ecological protection and economic growth.

[0177] In this step, the evolution rate refers to the ratio of the number of grids converted from grids without existing industrial transfer to grids of any type of industry within a certain administrative region (i.e., sub-region) during the period from the base year to the verification year, to the total number of grids in this administrative region.

[0178] When running the five different scenarios simulated, the simulation processes are as follows:

[0179] Run the baseline scenario, simulate the future based on the change rules of industrial transfer between the base year and the verification year, that is, without making any modifications to the cellular automaton model, and continue to run the cellular automaton suspended in S10 until the quantity requirement is met;

[0180] Run the major transportation infrastructure construction scenario. Based on the baseline scenario, in this scenario, consider that the transportation infrastructure (railways, canals, etc.) that may be newly built in the future in the industrial transfer recipient area is also set as a transferable area within the predetermined range of the new transportation infrastructure, that is, add a new rule to the model in S9, and then continue to run the cellular automaton suspended in S10 until the quantity requirement is met;

[0181] Run the ecological protection priority scenario. In this scenario, classify the ecological environment effects of industrial transfer of cells according to the standard deviation classification method, and the areas with higher ecological environment effects are regarded as the areas where industrial transfer will occur in the future. That is, in the model in S9, change "whether the Euclidean distance between the cell and the existing industrial transfer grid is less than or equal to the average Euclidean distance between the existing industrial grids in the administrative region where the cell is located in S8" to "the ecological environment effect value of the cell is greater than 1.5 standard deviations of all ecological environment effect values", and then continue to run the cellular automaton suspended in S10 until the quantity requirement is met;

[0182] Run the economic growth priority scenario. In this scenario, the socio-economic effects of industrial transfer in cells are classified according to the standard deviation classification method. Areas with higher socio-economic effects are regarded as areas where industrial transfer will occur in the future. That is, on the basis of the model described in S9, change "whether the Euclidean distance between the cell and the existing industrial transfer grid is less than or equal to the average Euclidean distance between the existing industrial grids in the administrative region where the cell is located described in S8" to "the socio-economic effect value of the cell is greater than 1.5 standard deviations of all socio-economic effect values", and then continue to run the cellular automaton suspended in S10 until the quantity requirement is met;

[0183] Run the scenario of coordinated development of ecological protection and economic growth. In this scenario, areas with higher ecological environment effects and socio-economic effects are regarded as areas where industrial transfer will occur in the future, that is, the overlapping part of the areas described in the ecological protection priority scenario and the economic growth priority scenario. That is, on the basis of the CA model described in S9, change "whether the Euclidean distance between the cell and the existing industrial transfer grid is less than or equal to the average Euclidean distance between the existing industrial grids in the administrative region where the cell is located described in S8" to "the ecological environment effect value of the cell is greater than 1.5 standard deviations of all ecological environment effect values, and the socio-economic effect value of the cell is greater than 1.5 standard deviations of all socio-economic effect values", and then continue to run the cellular automaton suspended in S10 until the quantity requirement is met.

[0184] The five scenario settings (resource-intensive, labor-intensive, technology-intensive, capital-intensive, comprehensive) proposed in this step combine socio-economic effects and ecological environment effects, and can comprehensively reflect the diversity and complexity of industrial transfer. This scenario setting not only pays attention to the spatial distribution of industrial types, but also considers the spatio-temporal evolution laws under different scenarios. In addition, the method of combining qualitative, quantitative and positioning, as well as the idea of combining subjective and objective, is adopted in the scenario setting process, which can more accurately simulate and predict the spatio-temporal optimization process of industrial transfer.

[0185] For the above five scenarios, run one of the scenarios each time and obtain the future prediction results under the current scenario; according to actual needs, select one of the scenarios as the decision, so as to obtain the spatial decision of industrial transfer that meets the requirements.

[0186] Based on the same idea, the embodiment of the present invention also provides an industrial transfer spatial decision-making system based on the ANN-GIS-CA architecture. The system includes: a grid division module, a bearing capacity evaluation index construction module, a bearing capacity calculation module, a bearing effect evaluation index construction module, a bearing effect evaluation module, a state probability calculation module, a prohibited opening area determination module, an Euclidean distance calculation module, a CA model construction module, a CA model iteration module, an accuracy verification module and a scenario simulation module; among them,

[0187] The grid division module is used to divide the industrial transfer receiving area into several grids based on the geographic information system GIS to form cells, and use manual interpretation to mark the existing industrial transfer situation in the grid, and generate grid data of the basic year and the verification year based on GIS; the grid data includes spatial information and attribute information; it is also used to determine the sub-area to which the grid belongs according to preset rules, and use the attribution result as the attribute information of the grid data;

[0188] The module for constructing the evaluation index of industrial transfer capacity is used to obtain the original data of industrial transfer capacity of all grids based on the geographic information system GIS, and to construct an evaluation index system of industrial transfer capacity;

[0189] The undertaking capacity calculation module is used to calculate the industrial transfer undertaking capacity scores of all grids based on the constructed industrial transfer undertaking capacity evaluation index system, and use the Gaussian mixture model to identify the industry type that each grid is suitable for undertaking, and use the scores and the industry type that is suitable for undertaking as the attribute information of the grid data;

[0190] The construction module of the evaluation index of the undertaking effect is used to obtain the original data of the social and economic effects and the ecological and environmental effects of all grids undertaking industrial transfer based on GIS, and to construct the evaluation index system of the social and economic effects and the ecological and environmental effects of undertaking industrial transfer; the constructed evaluation index system structure of the undertaking effect includes the primary index, the secondary index, and the source and direction attributes of the secondary index; the direction attributes include positive index and negative index; wherein the positive index refers to the index whose attribute value is larger, the better, and the negative index refers to the index whose attribute value is smaller, the better;

[0191] The undertaking effect evaluation module is used to calculate the social and economic effect scores and the ecological and environmental effect scores of all grids according to the secondary indicators of the social and economic effect and the ecological and environmental effect in the undertaking effect and the directional attributes of the secondary indicators, and use the scores as the attribute information of the grid data;

[0192] The state probability calculation module is used to select grids with existing industrial transfers in all base years, and randomly select a predetermined number of grids without existing industrial transfers in the base years, extract the social and economic effect scores and ecological and environmental effect scores of these grids as samples, construct an artificial neural network ANN model, and respectively calculate the state probabilities of all grids evolving into six states: resource-intensive, labor-intensive, technology-intensive, capital-intensive, comprehensive, and no existing industrial transfer, and use the six states and state probabilities as attribute information of the grid data;

[0193] The prohibited development area determination module is used to use GIS to determine whether each grid belongs to a prohibited development area, and use the determination result as the attribute information of the grid data;

[0194] The Euclidean distance calculation module is used to calculate the Euclidean distance between each grid and the existing industrial grids using GIS according to the grid spatial information of the base year, and use the Euclidean distance as the attribute information of the grid data;

[0195] The CA model construction module is used to construct a partitioned asynchronous cellular automaton model with grids as cells, six states of resource-intensive, labor-intensive, technology-intensive, capital-intensive, comprehensive, and no existing industries as cells, and the industrial types suitable for each grid to undertake, the probabilities of the six states, whether each grid is a prohibited development area, and the Euclidean distance between each grid and the existing industrial grids as the attributes of the cells;

[0196] The CA model iteration module is used to calculate the sub-region asynchronous evolution rate from the base year to the verification year based on the grid data of the base year and the verification year; substitute the grid data of the base year and iterate the cellular automaton model multiple times; when the cellular automaton model iterates, it will traverse all cells, and only when the iteration times are an integer multiple of the evolution rate and the cell state is no existing industry transfer will the cell be accessed; when the total number of cells in the resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive states is greater than or equal to the proportion threshold of the total number of grids in the resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive states in the verification year, the cellular automaton model pauses iteration;

[0197] The accuracy verification module is used to set an accuracy threshold and verify the accuracy of the iteration results of the cellular automaton model; if the accuracy reaches the accuracy threshold, the scenario simulation module is started, and if it does not reach the accuracy threshold, the CA model iteration module is started;

[0198] The scenario simulation module is used to divide the future industrial transfer scenarios into five different simulation scenarios; then use the Markov chain method to predict the demand quantities of the five industrial types of resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive in each future scenario, and use the demand quantities as the termination conditions of the model to simulate the spatio-temporal processes of future industrial transfer in the five different scenarios respectively; among them, the five different simulation scenarios include the baseline scenario, the major infrastructure construction scenario, the ecological protection priority scenario, the economic growth priority scenario, and the coordinated development scenario of ecological protection and economic growth.

[0199] In this embodiment, each module is implemented by a processor, and a memory is appropriately added when storage is required. Among them, the processor may be, but is not limited to, a microprocessor MPU, a central processing unit (CPU), a network processor (NP), a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), other programmable logic devices, discrete gate, transistor logic devices, discrete hardware components, etc. The memory may include a random access memory (RAM), or may also include a non-volatile memory (NVM).

[0200] In the above embodiment, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions.

[0201] In addition, it should be noted that the industrial transfer spatial decision-making system based on the ANN-GIS-CA architecture described in this embodiment corresponds to the industrial transfer spatial decision-making method based on the ANN-GIS-CA architecture. The description and limitation of the method also apply to the system, and will not be elaborated here.

[0202] From the above technical solutions, it can be seen that the industrial transfer spatial decision-making method and system provided by the embodiments of the present invention integrate ANN, GIS, and CA to construct an optimization model for the industrial spatial layout during industrial transfer, and predict the spatio-temporal patterns of future industrial transfer under different scenarios; use the CA model to simulate industrial transfer; accurately simulate the spatio-temporal optimization process of industrial transfer through a combination of qualitative, quantitative, and positioning methods, and subjective and objective methods, predict the development trends of industrial transfer under different development models, improve the optimization effect of the industrial spatial layout during industrial transfer, rationally layout the industrial transfer space, and achieve sustainable regional industrial development.

[0203] The above description is only a preferred embodiment of the present invention and an explanation of the applied technical principles, and is not intended to limit the scope of the present invention claimed, but merely represents the preferred embodiments of the present invention. Those skilled in the art should understand that the scope of the invention involved in the present invention is not limited to the technical solutions formed by the specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the inventive concept. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts fall within the scope of protection of the present invention.

Claims

1. An industrial transfer space decision-making method based on the ANN-GIS-CA architecture, characterized in that The method includes the following steps: Step S1: Based on the Geographic Information System (GIS), divide the industrial transfer recipient areas into several grids to form cells, and use manual interpretation to mark the existing industrial transfer situations in the grids, generating grid data for the base year and verification year based on GIS; the grid data includes spatial information and attribute information; at the same time, judge the sub-areas to which the grids belong according to preset rules, and use the attribution results as the attribute information of the grid data; Step S2: Based on the Geographic Information System (GIS), obtain the original data of the industrial transfer acceptance capacity of all grids, and construct an evaluation index system for the industrial transfer acceptance capacity; Step S3: Based on the constructed evaluation index system for the industrial transfer acceptance capacity, calculate the industrial transfer acceptance capacity scores of all grids, and use the Gaussian mixture model to identify the industrial types suitable for each grid to undertake, and use the scores and the suitable industrial types to be undertaken as the attribute information of the grid data; Step S4: Based on GIS, obtain the original data of the social and economic effects and ecological environment effects of all grids undertaking industrial transfer, and construct an evaluation index system for the social and economic effects and ecological environment effects of undertaking industrial transfer; the structure of the constructed evaluation index system for the undertaking effects includes first-level indicators, second-level indicators, and the source and direction attributes of the second-level indicators; the direction attributes include positive indicators and negative indicators; among them, a positive indicator refers to an indicator with the better the attribute value, and a negative indicator refers to an indicator with the smaller the attribute value; Step S5: Calculate the social and economic effect scores and ecological environment effect scores of all grids respectively according to the second-level indicators and the direction attributes of the second-level indicators in the undertaking effects, and use the scores as the attribute information of the grid data; Step S6: Select the grids with existing industrial transfers in all base years, and randomly select a predetermined number of grids without existing industrial transfers in the base years. Extract the social and economic effect scores and ecological environment effect scores of these grids as samples, construct an Artificial Neural Network (ANN) model, and calculate the state probabilities of all grids evolving into six states: resource-intensive, labor-intensive, technology-intensive, capital-intensive, comprehensive, and no existing industrial transfer respectively, and use the six states and the state probabilities as the attribute information of the grid data; Step S7: Use GIS to determine whether each grid belongs to a prohibited development area, and use the determination result as the attribute information of the grid data; Step S8: According to the grid spatial information in the base year, use GIS to calculate the Euclidean distance of each grid from the existing industrial grids, and use the Euclidean distance as the attribute information of the grid data; Step S9: Construct a partitioned asynchronous cellular automata model with grids as cells, six states of resource-intensive, labor-intensive, technology-intensive, capital-intensive, comprehensive, and no existing industry as cells, and the industrial types suitable for each grid to undertake, the six state probabilities, whether each grid is a prohibited development area, and the Euclidean distance of each grid from the existing industrial grids as the attributes of the cells; Step S10: Calculate the asynchronous evolution rate of the sub-region from the base year to the verification year based on the grid data of the base year and the verification year; substitute the grid data of the base year and iterate the cellular automaton model multiple times. When the cellular automaton model iterates, it traverses all cells, and a cell is only accessed when the iteration count is an integer multiple of the evolution rate and the cell state is without existing industrial transfer. When the total number of cells belonging to the resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive types is greater than or equal to the proportion threshold of the total number of grids belonging to the resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive types in the verification year, the cellular automaton model pauses the iteration. Step S11: Set the accuracy threshold and verify the accuracy of the iteration result of the cellular automaton model. If the accuracy reaches the accuracy threshold, go to Step S12; if not, return to S10. Step S12: Divide the future industrial transfer scenarios into five different simulation scenarios; then use the Markov chain method to predict the demand quantities of the five industrial types of resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive types under each future scenario, and use the demand quantities as the termination conditions of the model to simulate the spatio-temporal processes of future industrial transfer under the five different scenarios respectively. Among them, the five different simulation scenarios include the baseline scenario, the major infrastructure construction scenario, the ecological protection priority scenario, the economic growth priority scenario, and the coordinated development scenario of ecological protection and economic growth.

2. The method according to claim 1, characterized in that, The structure of the carrying capacity evaluation index system constructed in Step S2 includes first-level indicators, second-level indicators, third-level indicators, and the source and direction attributes of the third-level indicators; the direction attributes include positive indicators and negative indicators; among them, a positive indicator refers to an indicator with a better attribute value when it is larger, while a negative indicator refers to an indicator with a better attribute value when it is smaller. When calculating the industrial transfer carrying capacity scores of all grids, use the global principal component analysis (GPCA) method according to the third-level indicators and the direction attributes of the third-level indicators.

3. The method according to claim 1, characterized in that In Step S5, calculate the socio-economic effect score and the eco-environmental effect score based on the second-level indicators and their direction attributes. The calculation process specifically includes the following steps: Step S51: Respectively perform dimensionless normalization processing on the positive indicators and negative indicators in the second-level indicators using different range standardization formulas to obtain the standardized second-level indicators. Step S52: Respectively determine the preliminary weights of the second-level indicators through the principal component analysis (GPCA) method, the entropy weight method (EWM), and the analytic hierarchy process (AHP), and use the technique for order preference by similarity to ideal solution (TOPSIS) method and the weighted average method to integrate the three preliminary weights to obtain two integrated weights; then calculate the average value of the two integrated weights as the comprehensive weight of each finally obtained second-level indicator. Step S53: According to the normalized second-level indicators and the comprehensive weights of each second-level indicator, calculate the socio-economic effect score and the eco-environmental effect score of each grid respectively, and quantify the socio-economic effect and the eco-environmental effect of each grid through the scores.

4. The method according to claim 3, characterized in that, Step S52 further includes: Step S521: Calculate the first preliminary weight of the secondary indicators by the GPCA method; Step S522: Calculate the second preliminary weight of the secondary indicators by the EWM method; Step S523: Calculate the third preliminary weight of the secondary indicators by the AHP method; Step S524: Integrate the first preliminary weight, the second preliminary weight, and the third preliminary weight by the TOPSIS method to obtain the first integrated weight; Step S525: Integrate the first preliminary weight, the second preliminary weight, and the third preliminary weight by the weighted average weight method to obtain the second integrated weight; Step S526: Synthesize the first integrated weight and the second integrated weight to obtain the comprehensive weight of each secondary indicator; the first integrated weight represents the weight obtained by the objective weighting method, and the second integrated weight represents the weight obtained by the subjective weighting method.

5. The method according to claim 1, wherein In Step S7, the prohibited development areas include three situations: the water area in the grid is greater than 50% of the total grid area; the grid is located within a protected area, and any form of development is prohibited in these places according to relevant policies; the average slope of the grid is greater than 15°.

6. The method according to claim 1, wherein In Step S9, the partitioned asynchronous cellular automaton model consists of cells, a cell space, a neighborhood, and a transition rule, where the transition rule adopts the roulette selection rule; In the partitioned asynchronous cellular automaton model, social and economic effects, ecological and environmental effects, and Euclidean distance are introduced as external factor constraint conditions, and at the same time, grid data based on the GIS data platform is used to enhance the expression ability of complex geographical phenomena.

7. The method according to claim 6, wherein The specific steps of the roulette selection rule are as follows: Step S91: Calculate the total fitness: First, calculate the total fitness of all individuals in the population; Step S92: Calculate the selection probability: The selection probability of each individual is equal to its fitness divided by the total fitness; Step S93: Construct the roulette: Map the selection probability of each individual onto a roulette to form a cumulative probability distribution; Step S94: Select an individual: Generate a random number between 0 and 1 to determine the position on the roulette, thereby selecting the corresponding individual.

8. The method according to claim 1, wherein In Step S10, when a cell is visited, the probabilities of the cell evolving into six states of resource-intensive, labor-intensive, technology-intensive, capital-intensive, comprehensive, and no existing industrial transfer are substituted into the roulette selection model to randomly select an industrial type; then, it is judged whether the cell is a prohibited development area and whether the Euclidean distance between the cell and the existing industrial transfer grid is less than or equal to the average Euclidean distance between the existing industrial grids in the administrative region where the cell is located; after meeting these two conditions, if the industrial type randomly selected by the roulette selection model is compatible with the industrial type suitable for the cell to undertake, the state of the cell changes to the industrial type randomly selected by the roulette selection rule, otherwise the state of the cell remains no existing industrial transfer; for each iteration, the model calculates the total number of cells in the states of resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive.

9. The method according to claim 8, wherein In Step S11, when running the five different scenarios simulated, the simulation processes are as follows: Run the baseline scenario, simulate the future based on the change rules of industrial transfer in the base year and verification year, that is, without making any modifications to the cellular automata model, continue to run the cellular automata suspended in S10 until the quantity requirement is met; Run the major transportation infrastructure construction scenario. Based on the baseline scenario, in this scenario, consider that the areas within the predetermined range of the newly built transportation infrastructure that may be constructed in the future in the industrial transfer recipient areas are also set as transferable areas. That is, add a new rule to the model in S9, and then continue to run the cellular automata suspended in S10 until the quantity requirement is met; Run the ecological protection priority scenario. In this scenario, classify the ecological environment effects of industrial transfer of cells according to the standard deviation classification method. The areas with higher ecological environment effects are regarded as the areas where industrial transfer will occur in the future. That is, in the model described in S10, change "whether the Euclidean distance between the cell and the existing industrial transfer grid is less than or equal to the average Euclidean distance between the existing industrial grids in the administrative region where the cell is located" to "the ecological environment effect value of the cell is greater than 1.5 standard deviations of all ecological environment effect values", and then continue to run the cellular automata suspended in S10 until the quantity requirement is met; Run the economic growth priority scenario. In this scenario, classify the social and economic effects of industrial transfer of cells according to the standard deviation classification method. The areas with higher social and economic effects are regarded as the areas where industrial transfer will occur in the future. That is, in the model described in S10, change "whether the Euclidean distance between the cell and the existing industrial transfer grid is less than or equal to the average Euclidean distance between the existing industrial grids in the administrative region where the cell is located" to "the social and economic effect value of the cell is greater than 1.5 standard deviations of all social and economic effect values", and then continue to run the cellular automata suspended in S10 until the quantity requirement is met; Run the scenario of coordinated development of ecological protection and economic growth. The areas with higher ecological environment effects and social and economic effects are regarded as the areas where industrial transfer will occur in the future, which is equivalent to the overlapping part of the areas described in the ecological protection priority scenario and the economic growth priority scenario. In the model described in S10, change "whether the Euclidean distance between the cell and the existing industrial transfer grid is less than or equal to the average Euclidean distance between the existing industrial grids in the administrative region where the cell is located" to "the ecological environment effect value of the cell is greater than 1.5 standard deviations of all ecological environment effect values, and the social and economic effect value of the cell is greater than 1.5 standard deviations of all social and economic effect values", and then continue to run the cellular automata suspended in S10 until the quantity requirement is met.

10. An industrial transfer spatial decision-making system based on the ANN-GIS-CA architecture, characterized in that, The system includes: a grid division module, a construction module for the evaluation index of the bearing capacity, a calculation module for the bearing capacity, a construction module for the evaluation index of the bearing effect, an evaluation module for the bearing effect, a calculation module for the state probability, a determination module for the prohibited and open areas, a calculation module for the Euclidean distance, a construction module for the CA model, an iteration module for the CA model, an accuracy verification module, and a scenario simulation module; among them, The grid division module is used to divide the industrial transfer receiving area into several grids based on the geographic information system GIS to form cells, and use manual interpretation to mark the existing industrial transfer situation in the grid, and generate grid data of the basic year and the verification year based on GIS; the grid data includes spatial information and attribute information; it is also used to determine the sub-area to which the grid belongs according to preset rules, and use the attribution result as the attribute information of the grid data; The module for constructing the evaluation index of industrial transfer capacity is used to obtain the original data of industrial transfer capacity of all grids based on the geographic information system GIS, and to construct an evaluation index system of industrial transfer capacity; The undertaking capacity calculation module is used to calculate the industrial transfer undertaking capacity scores of all grids based on the constructed industrial transfer undertaking capacity evaluation index system, and use the Gaussian mixture model to identify the industry type that each grid is suitable for undertaking, and use the scores and the industry type that is suitable for undertaking as the attribute information of the grid data; The construction module of the evaluation index of the undertaking effect is used to obtain the original data of the social and economic effects and the ecological and environmental effects of all grids undertaking industrial transfer based on GIS, and to construct the evaluation index system of the social and economic effects and the ecological and environmental effects of undertaking industrial transfer; the constructed evaluation index system structure of the undertaking effect includes the primary index, the secondary index, and the source and direction attributes of the secondary index; the direction attributes include positive index and negative index; wherein the positive index refers to the index whose attribute value is larger, the better, and the negative index refers to the index whose attribute value is smaller, the better; The undertaking effect evaluation module is used to calculate the social and economic effect scores and the ecological and environmental effect scores of all grids according to the secondary indicators of the social and economic effect and the ecological and environmental effect in the undertaking effect and the directional attributes of the secondary indicators, and use the scores as the attribute information of the grid data; The state probability calculation module is used to select grids with existing industrial transfers in all base years, and randomly select a predetermined number of grids without existing industrial transfers in the base years, extract the social and economic effect scores and ecological and environmental effect scores of these grids as samples, construct an artificial neural network ANN model, and respectively calculate the state probabilities of all grids evolving into six states: resource-intensive, labor-intensive, technology-intensive, capital-intensive, comprehensive, and no existing industrial transfer, and use the six states and state probabilities as attribute information of the grid data; The prohibited development area determination module is used to use GIS to determine whether each grid belongs to a prohibited development area, and use the determination result as attribute information of the grid data; The Euclidean distance calculation module is used to calculate the Euclidean distance of each grid from the existing industrial grid using GIS based on the grid spatial information of the basic year, and use the Euclidean distance as the attribute information of the grid data; The CA model construction module is used to construct a partitioned asynchronous cellular automaton model with grids as cells, six states of cells being resource-intensive, labor-intensive, technology-intensive, capital-intensive, comprehensive, and no existing industries, and the industrial type suitable for each grid to undertake, the probabilities of the six states, whether each grid is a prohibited development area, and the Euclidean distance between each grid and the existing industrial grid as the attributes of the cells; The CA model iteration module is used to calculate the asynchronous evolution rate of the sub-region from the base year to the verification year based on the grid data of the base year and the verification year; substitute the grid data of the base year and iterate the cellular automaton model multiple times; when the cellular automaton model iterates, it will traverse all cells, and only when the iteration times are an integer multiple of the evolution rate and the cell state is no existing industry transfer, the cell will be accessed; when the total number of cells in the resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive states is greater than or equal to the proportion threshold of the total number of grids in the resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive states in the verification year, the cellular automaton model pauses iteration; The accuracy verification module is used to set an accuracy threshold and verify the accuracy of the iteration results of the cellular automaton model; if the accuracy reaches the accuracy threshold, the scenario simulation module is started, and if it does not reach the accuracy threshold, the CA model iteration module is started; The scenario simulation module is used to divide the future industrial transfer scenarios into five different simulation scenarios; then use the Markov chain method to predict the demand quantities of the five industrial types of resource-intensive, labor-intensive, technology-intensive, capital-intensive, and comprehensive in each future scenario, and use the demand quantities as the termination conditions of the model to simulate the spatio-temporal processes of future industrial transfer in the five different scenarios respectively; among them, the five different simulation scenarios include the baseline scenario, the major infrastructure construction scenario, the ecological protection priority scenario, the economic growth priority scenario, and the coordinated development scenario of ecological protection and economic growth.

Images