Urban land utilization planning aided decision-making method and system based on big data

Abstract

The invention discloses an urban land utilization planning aided decision-making method and system based on big data, and the method comprises the steps: collecting land data and monitoring data of a plurality of preset cities, carrying out the semantic segmentation of a high-resolution remote sensing image of the land data according to deep learning, and obtaining a current land use type; setting initial weights of the partitions by adopting a multi-objective optimization algorithm; performing dynamic planning analysis according to the monitoring data to obtain a land utilization planning index, and when the land utilization planning index is greater than an index threshold value, introducing a dynamic weight adjustment mechanism to adjust the initial weight and refine partitions to obtain a grid region marked with the dynamic weight; step-by-step mask fuzzy C-means clustering is adopted to carry out secondary clustering on the grid region to obtain a local fine weight region, and a satisfactory decision-making mechanism is constructed according to the local fine weight region and urban development planning; and according to the satisfaction decision-making mechanism, constructing an urban land utilization planning auxiliary decision-making model, and carrying out multi-objective optimization on the urban land utilization planning auxiliary decision-making model.

Description

Technical Field

[0001] This invention relates to the field of urban land use planning, and in particular to a method and system for assisting decision-making in urban land use planning based on big data. Background Technology

[0002] With the acceleration of urbanization, the contradiction between supply and demand of urban land resources has become increasingly prominent. Traditional land use planning relies on experience and static data, which makes it difficult to adapt to the dynamic needs of multiple dimensions such as population flow, functional transformation, and ecological protection, resulting in insufficient scientificity and poor implementation of planning schemes.

[0003] Currently, urban land use faces multiple challenges: uneven population distribution leads to an imbalance between jobs and housing, and increased commuting pressure; conflicts between spatial development and ecological protection are intensifying, and problems such as damage to ecologically sensitive areas and decline in carbon sequestration capacity are becoming more prominent; spatial function indicators such as road network accessibility and public service coverage lack precise quantitative support, and planning decisions are highly subjective.

[0004] Meanwhile, the rapid development of big data technology has provided an opportunity for planning upgrades. High-resolution remote sensing data, mobile signaling, POIs, and road network data are emerging from multiple sources. However, existing planning methods have failed to fully integrate these data resources and lack efficient analytical tools such as dynamic weight adjustment and refined zoning and clustering. This makes it difficult to achieve synergistic optimization of economic, social, and ecological goals and to meet the demands of modern urban high-quality development for precise, intelligent, and dynamic land use planning. Therefore, it is urgent to construct big data-based auxiliary decision-making methods to break through the bottlenecks of traditional planning and improve the scientific and forward-looking nature of decision-making. Summary of the Invention

[0005] The purpose of this invention is to provide a big data-based method for assisting decision-making in urban land use planning.

[0006] To achieve the above objectives, the present invention is implemented according to the following technical solution: This invention includes the following steps: Land data and monitoring data from multiple preset cities are collected, and the land data and monitoring data are preprocessed. The land data includes high-resolution remote sensing classification data and land survey data. The monitoring data includes mobile phone signaling data, population census and sampling survey data, POI and road network data, building morphology data, remote sensing ecological parameters, and environmental monitoring data. The current land use type is obtained by semantic segmentation of high-resolution remote sensing images of the land data using deep learning. A multi-objective optimization algorithm is then used to set the initial weights of the partitions based on the urban development strategy and the current land use type. Based on the monitoring data, dynamic planning analysis is performed to obtain land use planning indicators. When the land use planning indicators are greater than the indicator threshold, a dynamic weight adjustment mechanism is introduced to adjust the initial weights and refine the partitions to obtain grid areas labeled with dynamic weights. Otherwise, the initial weights are maintained. A secondary clustering method using progressive masking fuzzy C-means clustering is employed to obtain local fine-weight regions. A satisfactory decision-making mechanism is then constructed based on these local fine-weight regions and the urban development plan. Based on the aforementioned satisfactory decision-making mechanism, an auxiliary decision-making model for urban land use planning is constructed. The auxiliary decision-making model for urban land use planning is then optimized using multiple objectives. The data to be decided is input into the auxiliary decision-making model for urban land use planning, and the auxiliary decision-making results are output.

[0007] Furthermore, a method for setting initial weights for zoning based on urban development strategies and the current land use types using a multi-objective optimization algorithm includes: With urban development strategy and current land use type as dual constraints, a multi-objective optimization function is established. The constraints of the multi-objective optimization function are that the total area of ​​all land use types is equal to the total area of ​​the planning area, and the area of ​​a single type does not exceed the resource and environmental carrying capacity threshold. The multi-objective optimization function includes maximizing the economic benefits of land use and minimizing ecological disturbance. The land use type within the zone is used as the decision variable. The analytic hierarchy process (AHP) is used to determine the initial weights. The driving factors of GDP contribution, employment creation, and ecological protection are selected as criteria. The land use type is used as the scheme, and a judgment matrix is ​​constructed through expert scoring. The land use types include cultivated land, forest land, grassland, water area, construction land, and unused land. Initial solutions are randomly generated. Solutions are classified according to Pareto dominance. Non-dominated solutions that simultaneously optimize economic and ecological objectives are selected. Crowding distance is calculated for solutions at the same level. Solutions with uniform distribution are retained to maintain population diversity. Tournament selection is used to retain high-quality solutions. New solutions are generated by simulating binary crossover and polynomial mutation. The optimization is iteratively performed until convergence. From the Pareto optimal solution set, the final solution is selected based on the city's development strategy preferences. According to the spatial distribution of the current land use type, the weight values ​​are allocated to specific zones to obtain the initial weights of the zones.

[0008] Furthermore, the method for obtaining land use planning indicators through dynamic planning analysis based on the monitoring data includes: Urban land use planning indicators are obtained based on monitoring data, and these indicators are divided into dimensions to obtain indicator dimensions. Urban land use planning indicators include population density, job-housing ratio, commuting distance, POI core density, road network accessibility, plot ratio, ecological sensitivity index, and carbon sequestration capacity. The indicator dimensions include population, space, and ecology. The urban land use planning indicators are standardized, and the population dynamics index, spatial function index, and ecological constraint index are obtained by weighted summation of the urban land use planning indicators. Process analysis employs state transition and result feedback, defining states and setting boundaries; State definition: using mesh cells as the basic unit, defining... Let t be the spatial coordinates. Land use planning index values; setting boundaries: ecological protection red lines, basic farmland, and other areas. The initial value is fixed at 0; the initial values ​​for other areas are determined by the current land use type. The urban utilization index value is updated based on changes in monitoring data, expressed as follows:

[0009] in Historical weighting coefficients To monitor data-driven metric increments, , , , The contribution coefficients of the monitoring data in each dimension, For population change, For changes in urban functions, This refers to changes in the ecological environment. Let be the urban land use planning index value at time t+1. Let be the urban land use planning index value at time t; When calculated When the value exceeds the preset threshold, dynamic weight adjustment is triggered; dynamic weight adjustment includes population-driven adjustment and ecological constraint adjustment; population-driven adjustment: the weight of residential land in the population inflow area is increased by 0.1-0.15, and the weight of industrial land is reduced by 0.05-0.1 at the same time; ecological constraint adjustment: when the ecological sensitivity index exceeds the threshold, the development intensity weight is reduced by 0.2, and the calculation of the ecological compensation index of the surrounding area is triggered. Remote sensing imagery is used to verify the accuracy of semantic segmentation of land use types. The impact of index adjustment on GDP and carbon emissions is simulated through a system dynamics model. If the simulation error is greater than 10%, the contribution coefficients in the state transition equation are retrained.

[0010] Furthermore, a method for adjusting the initial weights and refining the partitions by introducing a dynamic weight adjustment mechanism includes: Using both urbanization rate and GDP per capita as indicators, cities are divided into three stages: nascent, growth, and mature. Thresholds for land use planning indicators are then calculated.

[0011] in The threshold value for the land use planning index of category k in year s is... The threshold value for the k-th category of land use planning indicators from the previous year. This represents the average of the land use planning indicators for category k over the past three years. The historical threshold weights for the k-th type of land use planning indicators are... For the development stage coefficient; When the land use planning index is greater than When an emergency adjustment is triggered; when the land use planning index is greater than or equal to and less than When the land use planning index is less than the specified value, a routine adjustment is triggered. At that time, maintain the initial weights; The NSGA-III multi-objective optimization algorithm is adopted, with urban strategic objectives as constraints, and an initial weight vector is generated based on the current land use type, urban development strategy, adaptive threshold and land use planning indicators. The weights are dynamically adjusted, and the expressions for regular adjustment and emergency adjustment are as follows:

[0012]

[0013] in Adjust the step size based on the base. For symbolic functions, For the k-th land use planning indicator, As the initial weights, The dynamic weights are the result of regular adjustments. The dynamic weights were adjusted in an emergency. Based on the Gaussian projection coordinate system, the space within the city's administrative boundary is equally divided into grid cells. Inverse distance weighted interpolation is used to allocate block-level weights to the grid cells, as expressed in the following expression:

[0014] in For the first The weight of each grid, The adjusted weight for the p-th block. For the first The distance between each grid and the center of block p The number of blocks; The weights of the grid cells are graded, and the weights of the hotspot networks are increased by an additional 0.1 for the POI hotspot areas. Generate a GIS grid layer with weighted labels, merge grids with similar weights into fine grid areas, and output the boundary vector data and weights of the grid areas.

[0015] Furthermore, a method for obtaining local fine-weight regions by performing secondary clustering on the grid region using progressive masking fuzzy C-means clustering includes: Randomly generate a membership matrix and update the cluster centers based on the weighted Euclidean distance, as expressed by:

[0016] in Let y be the index vector of the y-th grid region. Ambiguity factor Let be the membership degree of the y-th grid region belonging to the b-th class. The number of grid areas, It is the cluster center of the b-th cluster; The membership degree is adjusted based on the distance from the sample to the cluster center, and the expression is:

[0017] in For weighted Euclidean distance, It is the cluster center of the c-th cluster. The number of cluster centers; Calculate the objective function:

[0018] in Let m be the objective function under the ambiguity factor; when If the iteration stops, then stop; otherwise, update the cluster centers. Based on the results of the first round of clustering, grid cells with the highest membership degree belonging to the development potential area and a development probability greater than 0.5 are extracted to form sub-region masks; if the area of ​​a sub-region is less than the minimum threshold, it is merged into an adjacent region. The sub-regions extracted by the mask are used as new datasets. The cluster centers, membership matrices and objective function values ​​are updated and recalculated. The number of clusters is dynamically adjusted according to the characteristics of the sub-regions. The index weights are recalculated according to the characteristics of the sub-regions. Clustering stops when the area of ​​the sub-region extracted by the mask is less than 5% of the total area of ​​the initial grid, and the local fine weight region is output.

[0019] Furthermore, the method for constructing a satisfactory decision-making mechanism based on the aforementioned local fine-grained weighted area and urban development plan includes: Based on urban development planning and the characteristics of local refined weighted areas, the decision-making objectives are decomposed into a quantifiable three-level indicator system, which includes the dimensions of economic vitality, social equity, and ecological security. The target weights are determined by combining the analytic hierarchy process with the urban development strategy. The target set is set according to the three-level indicator system, and the target weights of the local fine weight area are corrected by multipliers. Based on the three lines of the national land space plan and laws and regulations, inviolable boundary conditions are set; the three lines are the ecological protection red line, permanent basic farmland, and urban development boundary, and the boundary types include ecological protection red line, permanent basic farmland, and historical and cultural blocks; Satisfaction thresholds are set for secondary indicators by regression analysis of historical data and back-inference of planning objectives, replacing the global optimal solution. The satisfaction thresholds are dynamically adjusted according to the stage of urban development. The quantile method is used to determine the range of satisfaction thresholds based on data from the past five years. When the land use planning indicators do not meet the threshold but the fuzzy comprehensive evaluation is greater than or equal to 4, the weight is increased by the correction coefficient; conversely, if the land use planning indicators meet the threshold but the fuzzy comprehensive evaluation is less than or equal to 2, the weight is decreased by the correction coefficient. For each grid cell in the local fine weighting region, the deviation of each target index from the threshold is calculated, and the deviation is weighted according to priority to obtain the comprehensive deviation. Solutions with a deviation of less than or equal to 0.2 are retained as potential satisfactory solutions. The weights of the potential satisfactory solutions are adjusted using a correction coefficient, and the output is the satisfactory decision mechanism.

[0020] Furthermore, the method for constructing an auxiliary decision-making model for urban land use planning based on the aforementioned satisfactory decision-making mechanism includes: Based on urban development strategies and current land use types, a dual-branch multi-objective optimization network is used to output macro-level initial weights. Branch 1: Input the urban master plan text, extract strategic keyword vectors through the BERT model, and map them to economic, ecological, and social objective weights. Branch 2: Input the semantic segmentation results, and convert the land use type encoding into basic weights through a fully connected layer. The initial weights are obtained by combining the objective weights and the basic weights. If any indicator in the planning features exceeds the threshold, an adaptive threshold and feedback correction mechanism is triggered: the threshold is dynamically determined using the quantile method, and the weights are dynamically adjusted when a land use planning indicator exceeds the threshold. The expression is as follows:

[0021] in For correction factor, Let k be the value of the land use planning index. For the k-th land use planning indicator threshold, As the initial weights for land use planning indicators, These are the dynamically adjusted initial weights; The study area is divided into sub-regions, and a sliding window mask is generated for each sub-region. Clustering iteration is performed based on the dynamically adjusted initial weights and grid feature vectors. The high-weight regions are further refined to generate fine grid weights. A satisfactory decision-making mechanism based on multi-objective threshold constraints is embedded. Given the loss function of the urban land use planning auxiliary decision-making model, the model is validated at the micro, meso, and macro levels. The loss function is the sum of semantic segmentation loss, weight prediction loss, and constraint violation loss.

[0022] Furthermore, the method for multi-objective optimization of the urban land use planning auxiliary decision-making model includes: A multi-objective optimization framework is constructed based on the third-generation non-dominated sorting genetic algorithm, with economic development, ecological protection, and spatial efficiency as the core optimization objectives; the multi-objectives include economic objectives, ecological objectives, and spatial objectives. A uniform distribution of the Pareto optimal solution set is achieved through a reference point mechanism: Latin hypercube sampling is used to generate the initial weight scheme, the hyperplane intercept is calculated through ideal points and extreme points, the objective function value is normalized, dominant individuals are selected based on the vertical distance of the reference points, and iteration stops when the rate of change of the hypervolume index of the population for 50 consecutive generations is less than 1%, ensuring convergence to a stable solution set.

[0023] Secondly, a big data-based urban land use planning auxiliary decision-making system includes: Data acquisition and processing module: used to collect land data and monitoring data from multiple preset cities, and to preprocess the land data and monitoring data; the land data includes high-resolution remote sensing classification data and land survey data; the monitoring data includes mobile phone signaling data, population census and sampling survey data, POI and road network data, building morphology data, remote sensing ecological parameters and environmental monitoring data; Semantic segmentation and weight setting module: used to perform semantic segmentation on the high-resolution remote sensing image of the land data based on deep learning to obtain the current land use type, and to set the initial weight of the partition according to the urban development strategy and the current land use type using a multi-objective optimization algorithm; Planning Analysis and Grid Weight Partitioning Module: This module is used to perform dynamic planning analysis based on the monitoring data to obtain land use planning indicators. When the land use planning indicators are greater than the indicator threshold, a dynamic weight adjustment mechanism is introduced to adjust the initial weights and refine the partitions to obtain grid areas labeled with dynamic weights. Otherwise, the initial weights are maintained. Secondary Clustering and Satisfactory Decision Module: This module is used to perform secondary clustering on the grid area using progressive mask fuzzy C-means clustering to obtain local fine weight regions, and to construct a satisfactory decision mechanism based on the local fine weight regions and the urban development plan. Modeling and optimization module: This module is used to construct an auxiliary decision-making model for urban land use planning based on the aforementioned satisfactory decision-making mechanism, perform multi-objective optimization on the auxiliary decision-making model for urban land use planning, input the data to be decided into the auxiliary decision-making model for urban land use planning, and output the auxiliary decision-making results.

[0024] The beneficial effects of this invention are: This invention is a method and system for assisting decision-making in urban land use planning based on big data. Compared with existing technologies, this invention has the following technical advantages: This invention integrates multi-source big data and combines deep learning and multi-objective optimization algorithms to improve the scientific nature and accuracy of land use planning decisions; it introduces dynamic weight adjustment and secondary clustering mechanisms to achieve refined zoning and dynamic adaptation of decisions, adapting to the needs of different stages of urban development; and it constructs a multi-dimensional satisfactory decision-making mechanism to balance economic, social, and ecological goals, ensure the feasibility and synergy of planning schemes, and improve decision-making efficiency. Attached Figure Description

[0025] Figure 1 This is a flowchart illustrating the steps of a big data-based urban land use planning auxiliary decision-making method according to the present invention. Detailed Implementation

[0026] The present invention will be further described below through specific embodiments. The illustrative embodiments and descriptions herein are used to explain the present invention, but are not intended to limit the present invention.

[0027] The present invention discloses a method and system for assisting decision-making in urban land use planning based on big data, comprising the following steps: like Figure 1 As shown, this embodiment includes the following steps: Land data and monitoring data from multiple preset cities are collected, and the land data and monitoring data are preprocessed. The land data includes high-resolution remote sensing classification data and land survey data. The monitoring data includes mobile phone signaling data, population census and sampling survey data, POI and road network data, building morphology data, remote sensing ecological parameters, and environmental monitoring data. In the actual assessment, we take a prefecture-level city, City A, as the research object. City A is in the growth stage, with a total planned area of ​​500 square kilometers, covering 6 types of land use, including cultivated land, forest land, and construction land. We need to use this method to formulate a land use planning auxiliary scheme for 2018-2023. The growth stage is characterized by an urbanization rate of 45% and a per capita GDP of 80,000 yuan. Land data includes: high-resolution remote sensing classification data: 1-meter resolution remote sensing imagery, identifying 180 km² of arable land. 2 150km of forest land 2 30km of waterway2 110 km² of construction land 2 20km of grassland 2 10km of unused land 2 Land survey data: Records of various land use conversions over the past three years show an average annual increase of 3 km² in construction land. 2 The average annual reduction in arable land is 2 km². 2 .

[0028] The monitoring data includes: Population data: Mobile signaling data shows that the population density in the central urban area is 12,000 people / km². 2 Average annual inflow of 0.08 million people / km 2 The census data shows a job-to-residence ratio of 0.85 and an average commuting distance of 12km; spatial data shows a POI density of 35 kernels / km in the central urban area. 2 Road network density 8km / km 2 The average road network accessibility score is 0.75 (out of 1), and the average plot ratio for construction land is 1.8. Ecological data: Remote sensing ecological parameters indicate an ecologically sensitive area of ​​80 km². 2 The average carbon sequestration capacity is 2.8 t / (hm). 2 •a) The annual average PM2.5 concentration in environmental monitoring data is 32 μg / m³. 3 ; The monitoring data were standardized, and the population dynamics index was calculated to be 0.68, the spatial function index to be 0.72, and the ecological constraint index to be 0.65. 5% of the noise data in the remote sensing image was removed to ensure that the land use type identification accuracy was over 95%. The current land use type is obtained by semantic segmentation of high-resolution remote sensing images of the land data using deep learning. A multi-objective optimization algorithm is then used to set the initial weights of the partitions based on the urban development strategy and the current land use type. Based on the monitoring data, dynamic planning analysis is performed to obtain land use planning indicators. When the land use planning indicators are greater than the indicator threshold, a dynamic weight adjustment mechanism is introduced to adjust the initial weights and refine the partitions to obtain grid areas labeled with dynamic weights. Otherwise, the initial weights are maintained. A secondary clustering method using progressive masking fuzzy C-means clustering is employed to obtain local fine-weight regions. A satisfactory decision-making mechanism is then constructed based on these local fine-weight regions and the urban development plan. Based on the aforementioned satisfactory decision-making mechanism, an auxiliary decision-making model for urban land use planning is constructed. The auxiliary decision-making model for urban land use planning is then optimized using multiple objectives. The data to be decided is input into the auxiliary decision-making model for urban land use planning, and the auxiliary decision-making results are output.

[0029] In this embodiment, a method for setting initial weights for zoning based on urban development strategy and the current land use type using a multi-objective optimization algorithm includes: With urban development strategy and current land use type as dual constraints, a multi-objective optimization function is established. The constraints of the multi-objective optimization function are that the total area of ​​all land use types is equal to the total area of ​​the planning area, and the area of ​​a single type does not exceed the resource and environmental carrying capacity threshold. The multi-objective optimization function includes maximizing the economic benefits of land use and minimizing ecological disturbance. The land use type within the zone is used as the decision variable. The analytic hierarchy process (AHP) is used to determine the initial weights. The driving factors of GDP contribution, employment creation, and ecological protection are selected as criteria. The land use type is used as the scheme, and a judgment matrix is ​​constructed through expert scoring. The land use types include cultivated land, forest land, grassland, water area, construction land, and unused land. Initial solutions are randomly generated. Solutions are classified according to Pareto dominance. Non-dominated solutions that simultaneously optimize economic and ecological objectives are selected. Crowding distance is calculated for solutions at the same level. Solutions with uniform distribution are retained to maintain population diversity. Tournament selection is used to retain high-quality solutions. New solutions are generated by simulating binary crossover and polynomial mutation. The optimization is iteratively performed until convergence. From the Pareto optimal solution set, the final solution is selected based on the city's development strategy preferences. According to the spatial distribution of the current land use type, the weight values ​​are allocated to specific zones to obtain the initial weights of the zones. In actual assessments, the resource and environmental carrying capacity threshold is based on the regional territorial spatial planning outline, the ecological protection red line delineation standards, and resource consumption data from the past five years. The resource and environmental carrying capacity threshold is a weighted sum of the land resource threshold, water resource threshold, and ecological capacity threshold. Multi-objective optimization function: The objective is to maximize GDP contribution and minimize ecological disturbance, constraining the total area of ​​various land uses to 500 km². 2 Construction land shall not exceed the resource and environmental carrying capacity threshold of 130 km². 2 2. The analytic hierarchy process (AHP) was used to construct a judgment matrix. After expert scoring, the weights of cultivated land, forest land, and construction land were 0.25, 0.2, 0.3, 0.1, 0.1, and 0.05, respectively. 3. After 50 iterations of Pareto optimization, the results converged, and the initial weights of the partitions were determined. The weight of construction land in the central urban area was 0.35, and the weight of forest land in the ecological protection area was 0.4.

[0030] In this embodiment, the method for obtaining land use planning indicators through dynamic planning analysis based on the monitoring data includes: Urban land use planning indicators are obtained based on monitoring data, and these indicators are divided into dimensions to obtain indicator dimensions. Urban land use planning indicators include population density, job-housing ratio, commuting distance, POI core density, road network accessibility, plot ratio, ecological sensitivity index, and carbon sequestration capacity. The indicator dimensions include population, space, and ecology. The urban land use planning indicators are standardized, and the population dynamics index, spatial function index, and ecological constraint index are obtained by weighted summation of the urban land use planning indicators. Process analysis employs state transition and result feedback, defining states and setting boundaries; State definition: using mesh cells as the basic unit, defining... Let t be the spatial coordinates. Land use planning index values; setting boundaries: ecological protection red lines, basic farmland, and other areas. The initial value is fixed at 0; the initial values ​​for other areas are determined by the current land use type. The urban utilization index value is updated based on changes in monitoring data, expressed as follows:

[0031] in Historical weighting coefficients To monitor data-driven metric increments, , , , The contribution coefficients of the monitoring data in each dimension, For population change, For changes in urban functions, This refers to changes in the ecological environment. Let be the urban land use planning index value at time t+1. Let be the urban land use planning index value at time t; When calculated When the value exceeds the preset threshold, dynamic weight adjustment is triggered; dynamic weight adjustment includes population-driven adjustment and ecological constraint adjustment; population-driven adjustment: the weight of residential land in the population inflow area is increased by 0.1-0.15, and the weight of industrial land is reduced by 0.05-0.1 at the same time; ecological constraint adjustment: when the ecological sensitivity index exceeds the threshold, the development intensity weight is reduced by 0.2, and the calculation of the ecological compensation index of the surrounding area is triggered. Remote sensing imagery is used to verify the accuracy of semantic segmentation of land use types. The impact of index adjustment on GDP and carbon emissions is simulated through a system dynamics model. If the simulation error is greater than 10%, the contribution coefficients in the state transition equation are retrained. In the actual assessment, the remote sensing imagery was at a resolution of 1 meter, and the population density threshold for 2025 was calculated to be 15,000 people / km² using a formula.2 The ecological sensitivity index threshold is 0.7; Indicator Update: The historical weighting coefficient γ is set to 0.6, and the indicator increment driven by the 2017 monitoring data is 0.12. It is 0.35. It is 0.08; It is 0.4. It is 0.15; It is 0.25. The result is 0.05. It is 0.71.

[0032] In this embodiment, a method for adjusting the initial weights and refining the partitions by introducing a dynamic weight adjustment mechanism includes: Using both urbanization rate and GDP per capita as indicators, cities are divided into three stages: nascent, growth, and mature. Thresholds for land use planning indicators are then calculated.

[0033] in The threshold value for the land use planning index of category k in year s is... The threshold value for the k-th category of land use planning indicators from the previous year. This represents the average of the land use planning indicators for category k over the past three years. The historical threshold weights for the k-th type of land use planning indicators are... For the development stage coefficient; When the land use planning index is greater than When an emergency adjustment is triggered; when the land use planning index is greater than or equal to and less than When the land use planning index is less than the specified value, a routine adjustment is triggered. At that time, maintain the initial weights; The NSGA-III multi-objective optimization algorithm is adopted, with urban strategic objectives as constraints, and an initial weight vector is generated based on the current land use type, urban development strategy, adaptive threshold and land use planning indicators. The weights are dynamically adjusted, and the expressions for regular adjustment and emergency adjustment are as follows:

[0034]

[0035] in Adjust the step size based on the base. For symbolic functions, For the k-th land use planning indicator, As the initial weights, The dynamic weights are the result of regular adjustments. The dynamic weights were adjusted in an emergency. Based on the Gaussian projection coordinate system, the space within the city's administrative boundary is equally divided into grid cells. Inverse distance weighted interpolation is used to allocate block-level weights to the grid cells, as expressed in the following expression:

[0036] in For the first The weight of each grid, The adjusted weight for the p-th block. For the first The distance between each grid and the center of block p The number of blocks; The weights of the grid cells are graded, and the weights of the hotspot networks are increased by an additional 0.1 for the POI hotspot areas. Generate a GIS grid layer with weighted labels, merge grids with similar weights into fine grid areas, and output the boundary vector data and weights of the grid areas; In actual assessments, the nascent stage is defined as: urbanization rate less than 30% and per capita GDP less than 50,000 yuan; the growth stage is defined as: urbanization rate greater than or equal to 30% and less than 60% and per capita GDP greater than or equal to 50,000 yuan and less than 100,000 yuan; and the mature stage is defined as: urbanization rate greater than or equal to 60% and per capita GDP greater than or equal to 100,000 yuan. The space within the city's administrative boundaries is divided into 200m x 200m grids; the basic adjustment step size is dynamically adapted according to the city's development stage, with a value ranging from 0.02 to 0.05. The population density in the central urban area is 12,000 people / km². 2 (0.1×1.5≤1.2<1.2×1.5) triggers a regular adjustment, with a basic adjustment step size of 0.03. The weight of residential land is increased by 0.03, and the weight of industrial land is decreased by 0.02. The ecological sensitivity index of the ecologically sensitive area is 0.72>0.7, triggering an ecological constraint adjustment, with the weight of development intensity decreased by 0.2.

[0037] In this embodiment, a method for obtaining local fine-weight regions by performing secondary clustering on the grid region using progressive masking fuzzy C-means clustering includes: Randomly generate a membership matrix and update the cluster centers based on the weighted Euclidean distance, as expressed by:

[0038] in Let y be the index vector of the y-th grid region. Ambiguity factor Let be the membership degree of the y-th grid region belonging to the b-th class. The number of grid areas, It is the cluster center of the b-th cluster; The membership degree is adjusted based on the distance from the sample to the cluster center, and the expression is:

[0039] in For weighted Euclidean distance, It is the cluster center of the c-th cluster. The number of cluster centers; Calculate the objective function:

[0040] in Let m be the objective function under the ambiguity factor; when If the iteration stops, then stop; otherwise, update the cluster centers. Based on the results of the first round of clustering, grid cells with the highest membership degree belonging to the development potential area and a development probability greater than 0.5 are extracted to form sub-region masks; if the area of ​​a sub-region is less than the minimum threshold, it is merged into an adjacent region. The sub-regions extracted by the mask are used as new datasets. The cluster centers, membership matrices and objective function values ​​are updated and recalculated. The number of clusters is dynamically adjusted according to the characteristics of the sub-regions. The index weights are recalculated according to the characteristics of the sub-regions. In actual evaluation, the local fine-grained weighting region: each grid cell corresponds to the category with the highest final membership degree and its weight value; When the area of ​​the sub-region extracted by the mask is less than 5% of the total area of ​​the initial grid, clustering stops and the local fine weight region is output. Grid division: Using the Gaussian projection coordinate system, the planning area is divided into 12,500 200m × 200m grids, and the block weights are allocated by inverse distance weighted interpolation; Stepwise masking fuzzy C-means clustering: The fuzziness factor is 2, and the iteration stops when the difference in the objective function is <0.001. 3200 grids of development potential areas are extracted and merged to form 12 local fine weight areas, of which the development weight of the core area of ​​the central urban area is 0.42 and the development weight of the ecological protection area is 0.08.

[0041] In this embodiment, the method for constructing a satisfactory decision-making mechanism based on the local fine-grained weight region and urban development planning includes: Based on urban development planning and the characteristics of local refined weighted areas, the decision-making objectives are decomposed into a quantifiable three-level indicator system, which includes the dimensions of economic vitality, social equity, and ecological security. The target weights are determined by combining the analytic hierarchy process with the urban development strategy. The target set is set according to the three-level indicator system, and the target weights of the local fine weight area are corrected by multipliers. Based on the three lines of the national land space plan and laws and regulations, inviolable boundary conditions are set; the three lines are the ecological protection red line, permanent basic farmland, and urban development boundary, and the boundary types include ecological protection red line, permanent basic farmland, and historical and cultural blocks; Satisfaction thresholds are set for secondary indicators by regression analysis of historical data and back-inference of planning objectives, replacing the global optimal solution. The satisfaction thresholds are dynamically adjusted according to the stage of urban development. The quantile method is used to determine the range of satisfaction thresholds based on data from the past five years. When the land use planning indicators do not meet the threshold but the fuzzy comprehensive evaluation is greater than or equal to 4, the weight is increased by the correction coefficient; conversely, if the land use planning indicators meet the threshold but the fuzzy comprehensive evaluation is less than or equal to 2, the weight is decreased by the correction coefficient. For each grid cell in the local fine weighting region, the deviation of each target index from the threshold is calculated, and the deviation is weighted according to priority to obtain the comprehensive deviation. Solutions with a deviation of less than or equal to 0.2 are retained as potential satisfactory solutions. The weights of the potential satisfactory solutions are adjusted using a correction coefficient, and the output is used as a satisfactory decision mechanism. In actual assessments, the primary indicator for the economic vitality dimension is development efficiency, and the secondary indicators are GDP per unit area and the proportion of industrial land. The primary indicator for the social equity dimension is public service coverage, and the secondary indicators are accessibility to education / medical facilities and the job-housing ratio. The primary indicator for the ecological security dimension is satisfaction with ecological constraints, and the secondary indicators are the proportion of ecologically sensitive areas and carbon sequestration capacity. The constraints of the ecological protection red line are that the development intensity weight within the red line is less than or equal to 0.1; the three-dimensional constraints of permanent basic farmland are that the weight of converting farmland into construction land is 0; and the constraints of historical and cultural blocks are that the weight of building height is less than or equal to 0.3. If the land use planning indicators do not meet the threshold but the fuzzy comprehensive evaluation is greater than or equal to 4, it is considered acceptable by the planner; if the land use planning indicators meet the standards but the fuzzy comprehensive evaluation is less than or equal to 2, it is considered by the planner to have hidden problems. Micro-level validation: Grid weights match actual land use demand ≥ 85%; Meso-level validation: Deviation of zoning indicators ≤ 10%; Macro-level validation: Simulation error of model output scheme for GDP and carbon emissions ≤ 10%; Three-tier indicator system: Economic vitality dimension, GDP target per unit area: 80 million yuan / km² 2 In the social equity dimension, the accessibility target for educational facilities is 0.8, and in the ecological security dimension, the proportion of ecologically sensitive areas is controlled at 16%.

[0042] In this embodiment, the method for constructing an auxiliary decision-making model for urban land use planning based on the satisfactory decision-making mechanism includes: Based on urban development strategies and current land use types, a dual-branch multi-objective optimization network is used to output macro-level initial weights. Branch 1: Input the urban master plan text, extract strategic keyword vectors through the BERT model, and map them to economic, ecological, and social objective weights. Branch 2: Input the semantic segmentation results, and convert the land use type encoding into basic weights through a fully connected layer. The initial weights are obtained by combining the objective weights and the basic weights. If any indicator in the planning features exceeds the threshold, an adaptive threshold and feedback correction mechanism is triggered: the threshold is dynamically determined using the quantile method, and the weights are dynamically adjusted when a land use planning indicator exceeds the threshold. The expression is as follows:

[0043] in For correction factor, Let k be the value of the land use planning index. For the k-th land use planning indicator threshold, As the initial weights for land use planning indicators, These are the dynamically adjusted initial weights; The study area is divided into sub-regions, and a sliding window mask is generated for each sub-region. Clustering iteration is performed based on the dynamically adjusted initial weights and grid feature vectors. The high-weight regions are further refined to generate fine grid weights. A satisfactory decision-making mechanism based on multi-objective threshold constraints is embedded. Given the loss function of the urban land use planning auxiliary decision-making model, the model is validated at the micro, meso, and macro levels. The loss function is the sum of semantic segmentation loss, weight prediction loss, and constraint violation loss. In the actual assessment, the sub-region was 500m x 500m, the sliding window was 3 x 3, and the fine grid was 100m x 100m. 8. The urban land use planning auxiliary decision-making method based on big data according to claim 1, characterized in that the method for multi-objective optimization of the urban land use planning auxiliary decision-making model includes: A multi-objective optimization framework is constructed based on the third-generation non-dominated sorting genetic algorithm, with economic development, ecological protection, and spatial efficiency as the core optimization objectives; the multi-objectives include economic objectives, ecological objectives, and spatial objectives. A uniform distribution of the Pareto optimal solution set is achieved through a reference point mechanism: Latin hypercube sampling is used to generate the initial weight scheme, the hyperplane intercept is calculated through ideal points and extreme points, the objective function value is normalized, dominant individuals are selected based on the vertical distance of the reference points, and iteration stops when the rate of change of the hypervolume index of the population for 50 consecutive generations is less than 1%, ensuring convergence to a stable solution set. In the actual evaluation, the comprehensive deviation calculation is as follows: three potential satisfactory solutions with a deviation of ≤0.2 are selected, and after adjustment by correction coefficients, the optimal solution is determined. The planned construction land area has increased to 128 km². 2 175 km² of arable land will be preserved. 2 5km of new woodland 2 Ecologically sensitive areas are strictly controlled according to the red line. The planning scheme has an error of 8.5% in the simulation of GDP growth and a reduction of 12% in carbon emissions. Construction land is concentrated in the central urban area and industrial parks.

[0044] Secondly, a big data-based urban land use planning auxiliary decision-making system includes: Data acquisition and processing module: used to collect land data and monitoring data from multiple preset cities, and to preprocess the land data and monitoring data; the land data includes high-resolution remote sensing classification data and land survey data; the monitoring data includes mobile phone signaling data, population census and sampling survey data, POI and road network data, building morphology data, remote sensing ecological parameters and environmental monitoring data; Semantic segmentation and weight setting module: used to perform semantic segmentation on the high-resolution remote sensing image of the land data based on deep learning to obtain the current land use type, and to set the initial weight of the partition according to the urban development strategy and the current land use type using a multi-objective optimization algorithm; Planning Analysis and Grid Weight Partitioning Module: This module is used to perform dynamic planning analysis based on the monitoring data to obtain land use planning indicators. When the land use planning indicators are greater than the indicator threshold, a dynamic weight adjustment mechanism is introduced to adjust the initial weights and refine the partitions to obtain grid areas labeled with dynamic weights. Otherwise, the initial weights are maintained. Secondary Clustering and Satisfactory Decision Module: This module is used to perform secondary clustering on the grid area using progressive mask fuzzy C-means clustering to obtain local fine weight regions, and to construct a satisfactory decision mechanism based on the local fine weight regions and the urban development plan. Modeling and optimization module: This module is used to construct an auxiliary decision-making model for urban land use planning based on the aforementioned satisfactory decision-making mechanism, perform multi-objective optimization on the auxiliary decision-making model for urban land use planning, input the data to be decided into the auxiliary decision-making model for urban land use planning, and output the auxiliary decision-making results.

[0045] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A big data-based method for supporting decision-making in urban land use planning, characterized in that, Includes the following steps: Land data and monitoring data from multiple preset cities are collected, and the land data and monitoring data are preprocessed. The land data includes high-resolution remote sensing classification data and land survey data. The monitoring data includes mobile phone signaling data, population census and sampling survey data, POI and road network data, building morphology data, remote sensing ecological parameters, and environmental monitoring data. The current land use type is obtained by semantic segmentation of high-resolution remote sensing images of the land data using deep learning. A multi-objective optimization algorithm is then used to set the initial weights of the partitions based on the urban development strategy and the current land use type. Based on the monitoring data, dynamic planning analysis is performed to obtain land use planning indicators. When the land use planning indicators are greater than the indicator threshold, a dynamic weight adjustment mechanism is introduced to adjust the initial weights and refine the partitions to obtain grid areas labeled with dynamic weights. Otherwise, the initial weights are maintained. A secondary clustering method using progressive masking fuzzy C-means clustering is employed to obtain local fine-weight regions. A satisfactory decision-making mechanism is then constructed based on these local fine-weight regions and the urban development plan. Based on the aforementioned satisfactory decision-making mechanism, an auxiliary decision-making model for urban land use planning is constructed. The auxiliary decision-making model for urban land use planning is then optimized using multiple objectives. The data to be decided is input into the auxiliary decision-making model for urban land use planning, and the auxiliary decision-making results are output.

2. The urban land use planning auxiliary decision-making method based on big data according to claim 1, characterized in that, A method for setting initial weights for zoning based on urban development strategy and the current land use type using a multi-objective optimization algorithm includes: With urban development strategy and current land use type as dual constraints, a multi-objective optimization function is established. The constraints of the multi-objective optimization function are that the total area of ​​all land use types is equal to the total area of ​​the planning area, and the area of ​​a single type does not exceed the resource and environmental carrying capacity threshold. The multi-objective optimization function includes maximizing the economic benefits of land use and minimizing ecological disturbance. The land use type within the zone is used as the decision variable. The analytic hierarchy process (AHP) is used to determine the initial weights. The driving factors of GDP contribution, employment creation, and ecological protection are selected as criteria. The land use type is used as the scheme, and a judgment matrix is ​​constructed through expert scoring. The land use types include cultivated land, forest land, grassland, water area, construction land, and unused land. Initial solutions are randomly generated. Solutions are classified according to Pareto dominance. Non-dominated solutions that simultaneously optimize economic and ecological objectives are selected. Crowding distance is calculated for solutions at the same level. Solutions with uniform distribution are retained to maintain population diversity. Tournament selection is used to retain high-quality solutions. New solutions are generated by simulating binary crossover and polynomial mutation. The optimization is iteratively performed until convergence. From the Pareto optimal solution set, the final solution is selected based on the city's development strategy preferences. According to the spatial distribution of the current land use type, the weight values ​​are allocated to specific zones to obtain the initial weights of the zones.

3. The urban land use planning auxiliary decision-making method based on big data according to claim 1, characterized in that, Methods for obtaining land use planning indicators through dynamic planning analysis based on the monitoring data include: Urban land use planning indicators are obtained based on monitoring data, and these indicators are divided into dimensions to obtain indicator dimensions. Urban land use planning indicators include population density, job-housing ratio, commuting distance, POI core density, road network accessibility, plot ratio, ecological sensitivity index, and carbon sequestration capacity. The indicator dimensions include population, space, and ecology. The urban land use planning indicators are standardized, and the population dynamics index, spatial function index, and ecological constraint index are obtained by weighted summation of the urban land use planning indicators. Process analysis employs state transition and result feedback, defining states and setting boundaries; State definition: using mesh cells as the basic unit, defining... Let t be the spatial coordinates. Land use planning index values; setting boundaries: ecological protection red lines, basic farmland, and other areas. The initial value is fixed at 0; the initial values ​​for other areas are determined by the current land use type. The urban utilization index value is updated based on changes in monitoring data, expressed as follows: ； in Historical weighting coefficients To monitor data-driven metric increments, , , , The contribution coefficients of the monitoring data in each dimension, For population change, For changes in urban functions, This refers to changes in the ecological environment. Let be the urban land use planning index value at time t+1. Let be the urban land use planning index value at time t; When calculated When the value exceeds the preset threshold, dynamic weight adjustment is triggered; dynamic weight adjustment includes population-driven adjustment and ecological constraint adjustment; population-driven adjustment: the weight of residential land in the population inflow area is increased by 0.1-0.15, and the weight of industrial land is reduced by 0.05-0.1 at the same time; ecological constraint adjustment: when the ecological sensitivity index exceeds the threshold, the development intensity weight is reduced by 0.2, and the calculation of the ecological compensation index of the surrounding area is triggered. Remote sensing imagery is used to verify the accuracy of semantic segmentation of land use types. The impact of index adjustment on GDP and carbon emissions is simulated through a system dynamics model. If the simulation error is greater than 10%, the contribution coefficients in the state transition equation are retrained.

4. The urban land use planning auxiliary decision-making method based on big data according to claim 1, characterized in that, A method for adjusting the initial weights and refining the partitions by introducing a dynamic weight adjustment mechanism includes: Using both urbanization rate and GDP per capita as indicators, cities are divided into three stages: nascent, growth, and mature. Thresholds for land use planning indicators are then calculated. ； in The threshold value for the land use planning index of category k in year s is... The threshold value for the k-th category of land use planning indicators from the previous year. This represents the average of the land use planning indicators for category k over the past three years. The historical threshold weights for the k-th type of land use planning indicators are... For the development stage coefficient; When the land use planning index is greater than When an emergency adjustment is triggered; when the land use planning index is greater than or equal to and less than When the land use planning index is less than the specified value, a routine adjustment is triggered. At that time, maintain the initial weights; The NSGA-III multi-objective optimization algorithm is adopted, with urban strategic objectives as constraints, and an initial weight vector is generated based on the current land use type, urban development strategy, adaptive threshold and land use planning indicators. The weights are dynamically adjusted, and the expressions for regular adjustment and emergency adjustment are as follows: ； ； in Adjust the step size based on the base. For symbolic functions, For the k-th land use planning indicator, As the initial weights, The dynamic weights are the result of regular adjustments. The dynamic weights were adjusted in an emergency. Based on the Gaussian projection coordinate system, the space within the city's administrative boundary is equally divided into grid cells. Inverse distance weighted interpolation is used to allocate block-level weights to the grid cells, as expressed in the following expression: ； in For the first The weight of each grid, The adjusted weight for the p-th block. For the first The distance between each grid and the center of block p The number of blocks; The weights of the grid cells are graded, and the weights of the hotspot networks are increased by an additional 0.1 for the POI hotspot areas. Generate a GIS grid layer with weighted labels, merge grids with similar weights into fine grid areas, and output the boundary vector data and weights of the grid areas.

5. The urban land use planning auxiliary decision-making method based on big data according to claim 1, characterized in that, A method for obtaining local fine-weight regions by performing secondary clustering on the grid region using progressive masking fuzzy C-means clustering includes: Randomly generate a membership matrix and update the cluster centers based on the weighted Euclidean distance, as expressed by: ； in Let y be the index vector of the y-th grid region. Ambiguity factor Let be the membership degree of the y-th grid region belonging to the b-th class. The number of grid areas, It is the cluster center of the b-th cluster; The membership degree is adjusted based on the distance from the sample to the cluster center, and the expression is: ； in For weighted Euclidean distance, It is the cluster center of the c-th cluster. The number of cluster centers; Calculate the objective function: ； in Let m be the objective function under the ambiguity factor; when If the iteration stops, then stop; otherwise, update the cluster centers. Based on the results of the first round of clustering, grid cells with the highest membership degree belonging to the development potential area and a development probability greater than 0.5 are extracted to form sub-region masks; if the area of ​​a sub-region is less than the minimum threshold, it is merged into an adjacent region. The sub-regions extracted by the mask are used as new datasets. The cluster centers, membership matrices and objective function values ​​are updated and recalculated. The number of clusters is dynamically adjusted according to the characteristics of the sub-regions. The index weights are recalculated according to the characteristics of the sub-regions. Clustering stops when the area of ​​the sub-region extracted by the mask is less than 5% of the total area of ​​the initial grid, and the local fine weight region is output.

6. The urban land use planning auxiliary decision-making method based on big data according to claim 1, characterized in that, The method for constructing a satisfactory decision-making mechanism based on the aforementioned local fine-weighted regions and urban development planning includes: Based on urban development planning and the characteristics of local refined weighted areas, the decision-making objectives are decomposed into a quantifiable three-level indicator system, which includes the dimensions of economic vitality, social equity, and ecological security. The target weights are determined by combining the analytic hierarchy process with the urban development strategy. The target set is set according to the three-level indicator system, and the target weights of the local fine weight area are corrected by multipliers. Based on the three lines of the national land space plan and laws and regulations, inviolable boundary conditions are set; the three lines are the ecological protection red line, permanent basic farmland, and urban development boundary, and the boundary types include ecological protection red line, permanent basic farmland, and historical and cultural blocks; Satisfaction thresholds are set for secondary indicators by regression analysis of historical data and back-inference of planning objectives, replacing the global optimal solution. The satisfaction thresholds are dynamically adjusted according to the stage of urban development. The quantile method is used to determine the range of satisfaction thresholds based on data from the past five years. When the land use planning indicators do not meet the threshold but the fuzzy comprehensive evaluation is greater than or equal to 4, the weight is increased by the correction coefficient; conversely, if the land use planning indicators meet the threshold but the fuzzy comprehensive evaluation is less than or equal to 2, the weight is decreased by the correction coefficient. For each grid cell in the local fine weighting region, the deviation of each target index from the threshold is calculated, and the deviation is weighted according to priority to obtain the comprehensive deviation. Solutions with a deviation of less than or equal to 0.2 are retained as potential satisfactory solutions. The weights of the potential satisfactory solutions are adjusted using a correction coefficient, and the output is the satisfactory decision mechanism.

7. The urban land use planning auxiliary decision-making method based on big data according to claim 1, characterized in that, The method for constructing an auxiliary decision-making model for urban land use planning based on the aforementioned satisfactory decision-making mechanism includes: Based on urban development strategies and current land use types, a dual-branch multi-objective optimization network is used to output macro-level initial weights. Branch 1: Input the urban master plan text, extract strategic keyword vectors through the BERT model, and map them to economic, ecological, and social objective weights. Branch 2: Input the semantic segmentation results, and convert the land use type encoding into basic weights through a fully connected layer. The initial weights are obtained by combining the objective weights and the basic weights. If any indicator in the planning features exceeds the threshold, an adaptive threshold and feedback correction mechanism is triggered: the threshold is dynamically determined using the quantile method, and the weights are dynamically adjusted when a land use planning indicator exceeds the threshold. The expression is as follows: ； in For correction factor, Let k be the value of the land use planning index. For the k-th land use planning indicator threshold, As the initial weights for land use planning indicators, These are the dynamically adjusted initial weights; The study area is divided into sub-regions, and a sliding window mask is generated for each sub-region. Clustering iteration is performed based on the dynamically adjusted initial weights and grid feature vectors. The high-weight regions are further refined to generate fine grid weights. A satisfactory decision-making mechanism based on multi-objective threshold constraints is embedded. Given the loss function of the urban land use planning auxiliary decision-making model, the model is validated at the micro, meso, and macro levels. The loss function is the sum of semantic segmentation loss, weight prediction loss, and constraint violation loss.

8. The urban land use planning auxiliary decision-making method based on big data according to claim 1, characterized in that, A method for multi-objective optimization of the urban land use planning auxiliary decision-making model includes: A multi-objective optimization framework is constructed based on the third-generation non-dominated sorting genetic algorithm, with economic development, ecological protection, and spatial efficiency as the core optimization objectives; the multi-objectives include economic objectives, ecological objectives, and spatial objectives. A uniform distribution of the Pareto optimal solution set is achieved through a reference point mechanism: Latin hypercube sampling is used to generate the initial weight scheme, the hyperplane intercept is calculated through ideal points and extreme points, the objective function value is normalized, dominant individuals are selected based on the vertical distance of the reference points, and iteration stops when the rate of change of the hypervolume index of the population for 50 consecutive generations is less than 1%, ensuring convergence to a stable solution set.

9. A big data-based urban land use planning auxiliary decision-making system, used to execute the method described in any one of claims 1-7, characterized in that, include: Data acquisition and processing module: used to collect land data and monitoring data from multiple preset cities, and to preprocess the land data and monitoring data; the land data includes high-resolution remote sensing classification data and land survey data; the monitoring data includes mobile phone signaling data, population census and sampling survey data, POI and road network data, building morphology data, remote sensing ecological parameters and environmental monitoring data; Semantic segmentation and weight setting module: used to perform semantic segmentation on the high-resolution remote sensing image of the land data based on deep learning to obtain the current land use type, and to set the initial weight of the partition according to the urban development strategy and the current land use type using a multi-objective optimization algorithm; Planning Analysis and Grid Weight Partitioning Module: This module is used to perform dynamic planning analysis based on the monitoring data to obtain land use planning indicators. When the land use planning indicators are greater than the indicator threshold, a dynamic weight adjustment mechanism is introduced to adjust the initial weights and refine the partitions to obtain grid areas labeled with dynamic weights. Otherwise, the initial weights are maintained. Secondary Clustering and Satisfactory Decision Module: This module is used to perform secondary clustering on the grid area using progressive mask fuzzy C-means clustering to obtain local fine weight regions, and to construct a satisfactory decision mechanism based on the local fine weight regions and the urban development plan. Modeling and optimization module: This module is used to construct an auxiliary decision-making model for urban land use planning based on the aforementioned satisfactory decision-making mechanism, perform multi-objective optimization on the auxiliary decision-making model for urban land use planning, input the data to be decided into the auxiliary decision-making model for urban land use planning, and output the auxiliary decision-making results.

Images