Land space planning management method and system based on big data analysis

Abstract

The invention relates to the technical field of big data analysis, in particular to a territorial space planning management method based on big data analysis, which comprises the following steps: collecting multi-source heterogeneous territorial space data, and establishing a standardized territorial space data set according to the multi-source heterogeneous territorial space data; according to the method, by means of multi-source data acquisition modes such as real-time access of satellite remote sensing data flow and the like, the problem of hysteresis of data acquisition of a traditional method can be solved, and by means of a multi-target optimization model, the multi-source data in the standardized territorial space data set can be optimized. An effective quantitative analysis means is provided for multi-objective balance of territorial space planning, optimal balance can be found among complex multiple objectives, conflicts between ecological protection and economic development objectives are avoided, resource utilization efficiency is improved, scientificity and practicability of a planning scheme are improved through a dynamic verification optimization module, and the method is suitable for popularization and application. The problem that a traditional planning model is difficult to adapt to dynamic changes is effectively solved.

Description

Technical Field

[0001] This invention relates to the field of big data analytics, and in particular to a method and system for land spatial planning and management based on big data analytics. Background Technology

[0002] Territorial spatial planning and management is an important means of rationally allocating, utilizing, and protecting land resources, and it plays a crucial role in ensuring ecological security, promoting economic development, and improving social well-being. Traditional territorial spatial planning and management methods mainly rely on experience-based judgment and simple statistical analysis. In terms of data acquisition, they are often limited to limited field survey data and a small amount of monitoring data, making it difficult to comprehensively and timely reflect the dynamic changes in territorial space.

[0003] Traditional methods suffer from data lag in acquiring information such as land use change and ecological environment evolution, failing to grasp the latest state of national land space in real time. Furthermore, national land space planning needs to balance multiple objectives, including ecological protection, economic development, and resource utilization. Traditional methods lack effective quantitative analysis tools, making it difficult to find the optimal balance among complex multi-objectives. In practical planning, conflicts between ecological protection and economic development goals, as well as low resource utilization efficiency, frequently arise. Moreover, traditional planning models are mostly based on static assumptions and simple rules, making them ill-suited to the dynamic changes and uncertainties of national land space. When faced with new development demands or changes in the external environment, the models cannot be adjusted and optimized in a timely manner, leading to a reduction in the scientific validity and practicality of the planning schemes. Summary of the Invention

[0004] The main objective of this invention is to provide a land spatial planning management method and system based on big data analysis, aiming to solve the technical problems mentioned in the background.

[0005] This invention proposes a land spatial planning management method based on big data analysis, comprising: Collect multi-source heterogeneous land space data, including remote sensing image data, geographic information system data, socio-economic statistical data and environmental monitoring data, and establish a standardized land space dataset based on the multi-source heterogeneous land space data; Based on the entropy weight method, weights are assigned to the multi-source data in the standardized territorial spatial dataset to construct a fusion weight matrix, and a multi-dimensional territorial spatial feature matrix is ​​generated by weighted fusion. A hybrid analysis model is constructed, which includes a spatial feature extraction module and a classification prediction module. The multidimensional land spatial feature matrix is ​​input into the hybrid analysis model, and the spatial functional zoning prediction results are output. A classification model for territorial spatial planning schemes is established. The spatial functional zoning prediction results are input into the territorial spatial planning scheme classification model for cluster analysis, and the cluster effectiveness index is calculated. A preliminary territorial spatial planning scheme is generated based on the cluster effectiveness index. Obtain the preset planning objectives, and based on the preset planning objectives, use a multi-objective optimization algorithm to solve for the optimal solution set to generate an optimized land spatial planning scheme; A historical dataset is established based on the optimized territorial spatial planning scheme. New data is collected in real time and compared with the historical dataset to calculate the planning deviation rate. The parameters of the territorial spatial planning scheme classification model are updated based on the planning deviation rate.

[0006] In one embodiment of the present invention, the step of establishing a standardized land space dataset based on the multi-source heterogeneous land space data includes: An outlier detection algorithm is used to calculate the outlier threshold for pixel brightness in the remote sensing image data, and pixels whose brightness exceeds the outlier threshold in the remote sensing image data are removed. The process involves acquiring vector data, raster data, and coordinate system from the geographic information system data; converting the vector data into a standard geographic format; resampling the raster data to a uniform resolution; and unifying the coordinate system to a global geographic coordinate system. Based on the spatial distribution relationship of neighboring observation points, a weighted average method is used to complete the missing data of the environmental monitoring data, wherein the weights are inversely proportional to the distance between the observation point and the missing point. A random forest regression model is trained using population density, economic growth rate, and transportation network density as feature variables. The model is then used to predict and fill in missing fields in the socioeconomic statistics.

[0007] In one embodiment of the present invention, the step of assigning weights to the multi-source data in the standardized territorial spatial dataset based on the entropy weight method, constructing a fusion weight matrix, and generating a multidimensional territorial spatial feature matrix through weighted fusion includes: Evaluation indicators are set for the remote sensing image data, the geographic information system data, the socio-economic statistical data, and the environmental monitoring data. The evaluation indicators include data completeness, timeliness, and spatial resolution. The entropy value of each evaluation indicator is calculated based on the information entropy theory, wherein the entropy value is negatively correlated with data uncertainty; Each evaluation indicator is assigned a weight based on its entropy value, wherein the lower the entropy value, the higher the weight. The multi-source heterogeneous land space data are superimposed and fused according to the weights to generate a multi-dimensional land space feature matrix containing the number of spatial units and the number of fused features.

[0008] In one embodiment of the present invention, the step of constructing the hybrid analysis model includes: A deep convolutional network architecture is adopted. The multidimensional land space feature matrix is ​​input, and spatial texture features are extracted through multi-layer convolutional operations to output a high-dimensional feature map. The high-dimensional feature map is input into a random forest classifier, and several functional partition categories are classified based on the decision tree voting mechanism; An adaptive learning rate optimization algorithm is adopted, with classification accuracy as the training objective, and the parameters of the deep convolutional network architecture are adjusted through cross-validation. The classification accuracy is calculated using a confusion matrix, and the consistency coefficient is used to evaluate the consistency between the prediction results and the manually labeled results.

[0009] In one embodiment of the present invention, the step of inputting the spatial functional zoning prediction results into a land spatial planning scheme classification model for cluster analysis, calculating a clustering effectiveness index, and generating a preliminary land spatial planning scheme based on the clustering effectiveness index includes: The K-means++ algorithm is used to select the initial centers; Calculate the Euclidean distance from each spatial unit to each cluster center. After calculating the Euclidean distance, reassign each spatial unit to the cluster to which the nearest cluster center belongs. The silhouette coefficient is used to calculate the clustering effectiveness index, wherein the closer the clustering effectiveness index is to 1, the better the clustering effect. If the clustering effectiveness index is not greater than a preset threshold, the number of cluster centers is increased and the process is iterated again. If the clustering effectiveness index is greater than the preset threshold, a preliminary land spatial planning scheme is generated based on the corresponding clustering results.

[0010] In one embodiment of the present invention, the step of generating an optimized land spatial planning scheme by solving for the optimal solution set using a multi-objective optimization algorithm based on the preset planning objective includes: A multi-objective optimization model is established, which includes setting the ecological protection objective as the objective function of minimizing the development intensity of ecologically sensitive areas, the economic development objective as the objective function of maximizing the regional economic growth rate, and the resource utilization objective as the objective function of minimizing energy consumption per unit of output. The constraints of the multi-objective optimization model are set, including the lower limit of the cultivated land protection area, the threshold of the proportion of ecological protection areas, and the infrastructure coverage requirements. The multi-objective optimization model is optimized using a multi-objective evolutionary algorithm, and a Pareto front solution set is generated through population iteration. Based on the entropy weight method and the approximation ideal solution ranking method, the comprehensive optimal solution is selected from the Pareto front solution set as the optimized land spatial planning scheme.

[0011] In one embodiment of the present invention, the step of collecting new data in real time and comparing it with the historical dataset, calculating the planning deviation rate, and updating the parameters of the land spatial planning scheme classification model based on the planning deviation rate includes: New data is acquired in real time using the remote sensing image data, including information on newly added construction land, changes in vegetation cover, and distribution of pollution sources. Compare the newly added data with the data in the historical dataset, and calculate the percentage of the sum of the absolute values ​​of the deviation rates of each indicator to the planned value; Determine whether the deviation rate exceeds a preset threshold. If the deviation rate exceeds the preset threshold, readjust the cluster center weights and update the parameters of the land spatial planning scheme classification model.

[0012] This invention also discloses a land spatial planning management system based on big data analysis, comprising: The data acquisition and preprocessing module is used to acquire multi-source heterogeneous land space data, including remote sensing image data, geographic information system data, socio-economic statistical data and environmental monitoring data, and to establish a standardized land space dataset based on the multi-source heterogeneous land space data. The fusion analysis module is used to assign weights to the multi-source data in the standardized territorial spatial dataset based on the entropy weight method, construct a fusion weight matrix, and generate a multi-dimensional territorial spatial feature matrix through weighted fusion. The functional zoning module is used to construct a hybrid analysis model, which includes a spatial feature extraction module and a classification prediction module. The multidimensional land space feature matrix is ​​input into the hybrid analysis model, and the spatial functional zoning prediction results are output. The classification module is used to establish a classification model for land spatial planning schemes, input the spatial functional zoning prediction results into the land spatial planning scheme classification model for cluster analysis, calculate the cluster effectiveness index, and generate a preliminary land spatial planning scheme based on the cluster effectiveness index. The decision-making module is used to obtain preset planning objectives and, based on the preset planning objectives, solve for the optimal solution set through a multi-objective optimization algorithm to generate an optimized land spatial planning scheme. The dynamic verification and optimization module is used to establish a historical dataset based on the optimized land spatial planning scheme, collect new data in real time and compare it with the historical dataset, calculate the planning deviation rate, and update the parameters of the land spatial planning scheme classification model based on the planning deviation rate.

[0013] The present invention also discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of a land spatial planning and management method based on big data analysis.

[0014] The present invention also discloses a computer-readable storage medium storing a computer program thereon, characterized in that the computer program, when executed by a processor, implements the steps of a land spatial planning and management method based on big data analysis.

[0015] The beneficial effects of this invention are as follows: By acquiring multi-source data such as real-time access to satellite remote sensing data streams, this invention can promptly obtain information on newly added construction land, changes in vegetation cover, and the distribution of pollution sources, effectively solving the problem of data lag in traditional methods. It allows for real-time monitoring of the latest state of national land space, providing timely and comprehensive data support for planning decisions. Furthermore, by utilizing a multi-objective optimization model, it sets clear objective functions for ecological protection, economic development, and resource utilization, and combines constraints such as arable land red line constraints, lower limits for ecological protection zone areas, and infrastructure coverage requirements, employing a multi-objective evolutionary algorithm for optimization. This approach provides an effective quantitative analysis method for balancing multiple objectives in national land space planning, finding the optimal balance among complex objectives, avoiding conflicts between ecological protection and economic development goals, and improving resource utilization efficiency. Additionally, through a dynamic verification and optimization module, it monitors the implementation of the plan in real time and calculates the planning deviation rate. When the deviation rate exceeds a preset threshold, a feedback mechanism is triggered, readjusting the cluster center weights and iteratively optimizing the model parameters. This enables the model to adjust and optimize in a timely manner according to the dynamic changes and uncertainties of national land space, improving the scientific rigor and practicality of the planning scheme and effectively solving the problem that traditional planning models are unable to adapt to dynamic changes. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of a method flow according to an embodiment of this application.

[0017] Figure 2 This is a schematic diagram of the system structure according to an embodiment of this application.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0020] like Figure 1 As shown, this application provides a land spatial planning management method based on big data analysis, including: S1. Collect multi-source heterogeneous land space data, including remote sensing image data, geographic information system data, socio-economic statistical data and environmental monitoring data, and establish a standardized land space dataset based on the multi-source heterogeneous land space data; S2, Based on the entropy weight method, weights are assigned to the multi-source data in the standardized territorial spatial dataset to construct a fusion weight matrix, and a multi-dimensional territorial spatial feature matrix is ​​generated by weighted fusion. S3, construct a hybrid analysis model, input the multidimensional land space feature matrix into the hybrid analysis model, and output the spatial functional zoning prediction results; S4. Establish a classification model for land spatial planning schemes, input the spatial functional zoning prediction results into the classification model for land spatial planning schemes for cluster analysis, and generate a preliminary land spatial planning scheme based on the clustering results. S5, obtain the preset planning objectives, and based on the preset planning objectives, solve the optimal solution set through a multi-objective optimization algorithm to generate an optimized land space planning scheme; S6. Establish a historical dataset, collect new data in real time and compare it with the historical dataset, calculate the planning deviation rate, and update the parameters of the land spatial planning scheme classification model based on the planning deviation rate.

[0021] As described in steps S1-S6 above, traditional land spatial planning management methods suffer from problems such as difficulty in data integration, lack of dynamic adaptability in planning schemes, difficulty in coordinating multiple objectives, and unintuitive presentation of planning results, making it difficult to meet the requirements of modern society for the scientific, accurate, dynamic, and participatory nature of land spatial planning.

[0022] This invention proposes a land spatial planning and management method based on big data analysis. By collecting and preprocessing multi-source heterogeneous data, it can integrate and mine massive amounts of multi-source data. By using the entropy weight method to fuse data, data weights can be objectively determined. A hybrid analysis model is constructed to predict spatial functional zoning. Combining the spatial feature extraction capabilities of convolutional neural networks and the classification prediction advantages of random forest classifiers, cluster analysis and multi-objective optimization models are used to generate planning schemes. This method can balance objectives such as ecological protection, economic development, and resource utilization, realizing the intelligence, dynamism, and scientification of land spatial planning. It improves the accuracy, rationality, and operability of planning schemes, enhances the transparency and credibility of the planning process, and solves the shortcomings of traditional land spatial planning methods in terms of data utilization, dynamic adaptability, multi-objective balancing, and result presentation. It provides a new and efficient solution for land spatial planning and management.

[0023] In one embodiment of the present invention, the step of establishing a standardized land space dataset based on the multi-source heterogeneous land space data includes: S11, using an outlier detection algorithm, calculate the outlier threshold for the brightness of pixels in the remote sensing image data, and remove pixels in the remote sensing image data whose brightness exceeds the outlier threshold; S12, acquire vector data, raster data and coordinate system from the geographic information system data, convert the vector data into a standard geographic format, resample the raster data to a uniform resolution, and unify the coordinate system to a global geographic coordinate system; S13, based on the spatial distribution relationship of neighboring observation points, the missing data of the environmental monitoring data is supplemented by a weighted average method, wherein the weights are inversely proportional to the distance between the observation point and the missing point; S14. A random forest regression model is trained based on population density, economic growth rate, and transportation network density as feature variables. The random forest regression model is then used to predict and fill in the missing fields in the socio-economic statistics.

[0024] As described in steps S11-S14 above, data and socio-economic statistical data are often incomplete, affecting data integrity and planning accuracy. To address this issue, this invention proposes the following technical solution: An outlier detection algorithm is used to calculate the outlier threshold for pixel brightness in remote sensing image data, and pixels with brightness exceeding this threshold are removed. Since the outlier detection algorithm can identify abnormal pixels deviating from the normal range based on statistical principles, removing these outliers can improve the quality of remote sensing image data, ensure data accuracy, and prevent outliers from interfering with subsequent land spatial analysis. Vector data, raster data, and coordinate systems from geographic information system (GIS) data are acquired; vector data is converted to a standard geographic format; raster data is resampled to a uniform resolution; and the coordinate system is unified to the global geographic coordinate system. Differences in format, resolution, and coordinate systems between GIS data from different sources can hinder data fusion and analysis. By converting formats, unifying resolutions, and standardizing coordinate systems, data discrepancies can be eliminated, making GIS data easier to integrate and analyze, and providing a standardized data foundation for subsequent land spatial planning. Based on the spatial distribution relationships of neighboring observation points, a weighted average method is used to complete missing data in environmental monitoring data, where weights are inversely proportional to the distance between the observation point and the missing point. This utilizes the principle of spatial data correlation; data from neighboring observation points are valuable for reference regarding missing points, and observation points closer to the missing point have greater weights. Completing missing data in this way ensures the integrity of environmental monitoring data, providing data support for accurately assessing the state of the land spatial environment. A random forest regression model is trained using population density, economic growth rate, and transportation network density as feature variables. This model predicts and fills in missing fields in socioeconomic statistics. Because the random forest regression model has strong nonlinear fitting capabilities, training the model with these closely related socioeconomic features allows it to learn the potential relationships between data, thereby accurately predicting and filling in missing fields, improving socioeconomic statistics, and providing complete data for economic analysis in land spatial planning. In summary, through this series of steps, the present invention can improve the quality and integrity of multi-source heterogeneous land and space data, eliminate data discrepancies, and provide accurate, standardized, and complete data for subsequent land and space planning and management, thereby solving the problems existing in traditional data processing methods.

[0025] In one embodiment of the present invention, the step of assigning weights to the multi-source data in the standardized territorial spatial dataset based on the entropy weight method, constructing a fusion weight matrix, and generating a multidimensional territorial spatial feature matrix through weighted fusion includes: S21, set evaluation indicators for the remote sensing image data, the geographic information system data, the socio-economic statistical data and the environmental monitoring data, the evaluation indicators including data integrity, timeliness and spatial resolution; S22, calculate the entropy value of each of the evaluation indicators based on the information entropy theory, wherein the entropy value is negatively correlated with data uncertainty; S23, assign weights to each of the evaluation indicators according to the entropy value, wherein the lower the entropy value, the higher the weight; S24, the multi-source heterogeneous land space data are superimposed and fused according to the weights to generate a multi-dimensional land space feature matrix containing the number of spatial units and the number of fused features.

[0026] As described in steps S21-S24 above, multi-source data fusion is a crucial step in territorial spatial planning management. However, traditional data fusion methods have significant drawbacks. On the one hand, there is a lack of a scientific and reasonable evaluation index system to measure the value of different data sources, making it impossible to accurately determine the importance of each data source in territorial spatial planning. On the other hand, it is difficult to objectively determine the weight of each data source, often resorting to subjective assignment or simple averaging, which prevents data fusion from fully leveraging the advantages of each data source and affects the accuracy and scientific nature of planning analysis. To address these problems, this invention proposes the following technical solution: setting evaluation indicators for remote sensing image data, geographic information system data, socio-economic statistical data, and environmental monitoring data, including data integrity, timeliness, and spatial resolution. Since different types of data exhibit varying performance in these aspects, and these indicators are crucial for territorial spatial planning, data integrity affects the comprehensiveness of information, timeliness concerns whether the data can reflect the current situation, and spatial resolution determines the data's ability to depict spatial details. By setting these evaluation indicators, the quality and value of each data source can be comprehensively measured; the entropy value of each evaluation indicator is calculated based on information entropy theory, and the entropy value is negatively correlated with data uncertainty. Information entropy is an important tool for measuring information uncertainty. In data fusion, the lower the data uncertainty, the more stable and reliable the data, and the more valuable the information it contains. By calculating the entropy value, the degree of uncertainty of each evaluation indicator can be quantified, providing a basis for subsequent weight allocation. The weight of each evaluation indicator is allocated according to the entropy value, with a higher weight for a lower entropy value. Since a lower entropy value indicates lower data uncertainty, it more accurately reflects the actual situation of national land space and has greater value for planning, thus assigning it a higher weight. This weight allocation method can objectively reflect the importance of different indicators from various data sources, making data fusion more scientific and reasonable. By superimposing and fusing multi-source heterogeneous national land space data according to weights, a multi-dimensional national land space feature matrix containing the number of spatial units and the number of fused features is generated. Through weighted fusion, the advantages of various data sources can be integrated, fully leveraging the value of the data. The resulting multi-dimensional national land space feature matrix covers rich national land space information, providing a comprehensive and valuable data foundation for subsequent spatial analysis and planning. In summary, this invention achieves weighted fusion of multi-source data by scientifically setting evaluation indicators, calculating entropy values ​​based on information entropy theory, and rationally allocating weights. This generates a high-quality multidimensional territorial spatial feature matrix, solving the problems of unreasonable evaluation indicators and subjective weight determination in traditional data fusion methods. It provides more accurate and scientific data support for territorial spatial planning and management.

[0027] In one embodiment of the present invention, the step of constructing the hybrid analysis model includes: S31, adopting a deep convolutional network architecture, inputting the multidimensional land space feature matrix, extracting spatial texture features through multi-layer convolutional operations, and outputting a high-dimensional feature map; S32, input the high-dimensional feature map into a random forest classifier, and classify several functional partition categories based on the decision tree voting mechanism; S33 employs an adaptive learning rate optimization algorithm, with classification accuracy as the training objective, and adjusts the parameters of the deep convolutional network architecture through cross-validation; S34 calculates the classification accuracy using the confusion matrix and uses the consistency coefficient to evaluate the consistency between the prediction results and the manually labeled results.

[0028] As described in steps S31-S34 above, accurate spatial functional zoning prediction is a key task in land spatial planning and management. However, traditional analytical models have many shortcomings. On the one hand, a single model is insufficient to comprehensively and accurately extract the complex features of land space. For example, simple statistical models cannot effectively handle land spatial data with high nonlinearity and spatial correlation. On the other hand, traditional models have poor stability and accuracy in classification prediction, and are easily affected by data noise and outliers, resulting in large deviations in functional zoning prediction results, which are difficult to meet the actual needs of land spatial planning. To address the above problems, this invention proposes a technical solution for constructing a hybrid analytical model: adopting a deep convolutional network architecture, inputting a multi-dimensional land spatial feature matrix, extracting spatial texture features through multi-layer convolutional operations, and outputting a high-dimensional feature map. The deep convolutional network architecture has powerful automatic feature extraction capabilities, and its multi-layer convolutional operations can learn and extract spatial texture features at different scales and levels in land spatial data, such as land use patterns and topographic features. This approach fully leverages the potential information within national land spatial data, providing rich and effective feature representations for subsequent functional zoning prediction, and improving the accuracy and comprehensiveness of feature extraction. High-dimensional feature maps are input into a random forest classifier, which classifies several functional zoning categories based on a decision tree voting mechanism. The random forest classifier is an ensemble learning model composed of multiple decision trees, and the decision tree voting mechanism allows it to integrate the classification results of multiple decision trees. Since different decision trees exhibit randomness in data sampling and feature selection during training, the voting mechanism effectively reduces the error of individual decision trees, improving classification stability and accuracy, thus enabling more reliable functional zoning of national land space. An adaptive learning rate optimization algorithm is employed, using classification accuracy as the training objective, and adjusting the parameters of the deep convolutional network architecture through cross-validation. This algorithm automatically adjusts the learning rate based on gradient information during model training, using a larger learning rate in the early stages of training to accelerate convergence and a smaller learning rate in the later stages to avoid missing the optimal solution. By targeting classification accuracy and incorporating cross-validation, model performance can be evaluated under different parameter combinations to find the parameter settings that maximize classification accuracy, thereby improving training effectiveness and prediction accuracy. Classification accuracy is calculated using a confusion matrix, and the consistency coefficient is used to evaluate the consistency between the prediction results and the manually labeled results. The confusion matrix visually displays the model's classification performance across different functional partitions, and calculating the classification accuracy quantifies the model's overall classification performance. The consistency coefficient measures the degree of agreement between the model's predictions and the manually labeled results, providing another perspective on the model's reliability and accuracy. These two metrics allow for a comprehensive and objective evaluation of the performance of the hybrid analysis model, providing a basis for further optimization and application.In summary, this invention, by constructing a hybrid analysis model that comprehensively utilizes the advantages of deep convolutional networks and random forest classifiers, combined with an adaptive learning rate optimization algorithm and scientific evaluation metrics, effectively solves the problems of insufficient feature extraction, poor classification accuracy, and poor stability in the prediction of land spatial functional zoning by traditional models. It can perform land spatial functional zoning prediction more accurately and reliably, providing strong technical support for land spatial planning and management.

[0029] In one embodiment of the present invention, the step of inputting the spatial functional zoning prediction results into the land spatial planning scheme classification model for cluster analysis, calculating the cluster effectiveness index, and generating a preliminary land spatial planning scheme based on the cluster effectiveness index includes the following steps:

[0030] S41, the K-means++ algorithm is used to select the initial centers; S42, calculate the Euclidean distance from each spatial unit to each cluster center, and after calculating the Euclidean distance, reassign each spatial unit to the cluster to which the nearest cluster center belongs; S43, The silhouette coefficient is used to calculate the clustering effectiveness index, wherein the closer the clustering effectiveness index is to 1, the better the clustering effect; S44, determine whether the clustering effectiveness index is greater than a preset threshold. If the clustering effectiveness index is not greater than the preset threshold, increase the number of cluster centers and iterate again. If the clustering effectiveness index is greater than the preset threshold, generate a preliminary land space planning scheme based on the corresponding clustering results.

[0031] As described in steps S41-S44 above, in land spatial planning, the reasonable clustering of spatial functional zoning prediction results to generate planning schemes is a crucial step. However, traditional clustering methods have many problems. On the one hand, the traditional K-means algorithm randomly selects initial cluster centers, which easily gets trapped in local optima, resulting in clustering results that cannot accurately reflect the actual distribution of spatial functions. On the other hand, the lack of effective clustering effect evaluation indicators and adjustment mechanisms makes it difficult to determine the appropriate number of cluster centers, potentially leading to overly coarse or overly fine clustering, thus failing to generate a scientifically reasonable preliminary land spatial planning scheme. To address these problems, this invention proposes the following technical solution: using the K-means++ algorithm to select initial centers. When selecting initial cluster centers, the K-means++ algorithm prioritizes points that are far apart from each other, avoiding the randomness of initial center selection in the traditional K-means algorithm. This makes the initial cluster centers more representative, reduces the risk of the algorithm getting trapped in local optima, and lays a good foundation for subsequent clustering analysis; calculating the Euclidean distance from each spatial unit to each cluster center, and after calculating the Euclidean distance, reassigning each spatial unit to the cluster to which the nearest cluster center belongs. Euclidean distance is a commonly used indicator to measure the similarity between spatial units and cluster centers. By calculating the Euclidean distance and redistributing spatial units, spatial units within the same cluster can be made more similar in characteristics, making the clustering results more consistent with the actual spatial functional distribution. The silhouette coefficient is used to calculate the clustering effectiveness index, where a higher index value (closer to 1) indicates better clustering. The silhouette coefficient comprehensively considers intra-cluster cohesion and inter-cluster separation, and its calculation allows for an objective evaluation of the clustering effect. A value closer to 1 indicates high similarity among elements within clusters and significant differences between clusters, resulting in a better clustering effect. The clustering effectiveness index is then checked against a preset threshold. If the index is below the threshold, the number of cluster centers is increased and the iteration is repeated. If the index exceeds the threshold, a preliminary land spatial planning scheme is generated based on the corresponding clustering results. The preset threshold is a standard set based on actual needs and experience. When the clustering effectiveness index fails to reach this threshold, the clustering effect is poor, and the clustering results can be optimized by increasing the number of cluster centers and iterating again. When the index exceeds the threshold, it indicates a good clustering effect, which can be used to generate a preliminary land spatial planning scheme. In summary, this invention effectively solves the problems of traditional clustering methods easily getting trapped in local optima and the difficulty in evaluating and adjusting clustering effects by employing the K-means++ algorithm, allocating spatial units based on Euclidean distance, using silhouette coefficients to evaluate clustering effects, and adjusting the number of cluster centers based on the evaluation results. This allows for the generation of a scientifically sound preliminary land spatial planning scheme, providing a reliable basis for subsequent land spatial planning.

[0032] In one embodiment of the present invention, the step of generating an optimized land spatial planning scheme by solving for the optimal solution set using a multi-objective optimization algorithm based on the preset planning objective includes: S51, Establish a multi-objective optimization model, wherein the multi-objective optimization model includes setting the ecological protection objective as the objective function of minimizing the development intensity of ecologically sensitive areas, the economic development objective as the objective function of maximizing the regional economic growth rate, and the resource utilization objective as the objective function of minimizing energy consumption per unit of output. S52, Set the constraints of the multi-objective optimization model, including the lower limit of cultivated land protection area, the threshold of ecological protection area ratio and infrastructure coverage requirements; S53, The multi-objective optimization model is optimized using a multi-objective evolutionary algorithm, and a Pareto front solution set is generated through population iteration; S54. Based on the entropy weight method and the approximation ideal solution ranking method, the comprehensive optimal solution is selected from the Pareto front solution set as the optimized land spatial planning scheme.

[0033] As described in steps S51-S54 above, traditional planning methods face numerous difficulties in addressing multi-objective balance issues in the field of territorial spatial planning. On the one hand, objectives such as ecological protection, economic development, and resource utilization often conflict with each other. Traditional methods lack scientific quantitative means, making it difficult to find a reasonable balance among these objectives, often resulting in over-development that damages the ecosystem or inefficient resource utilization. On the other hand, the lack of effective integration of various constraints during the planning process leads to poor feasibility of planning schemes, such as the difficulty in simultaneously addressing requirements for farmland protection, ecological protection zone delineation, and infrastructure construction. To address these problems, this invention proposes the following technical solution: establishing a multi-objective optimization model, setting the ecological protection objective as the objective function of minimizing the development intensity of ecologically sensitive areas, the economic development objective as the objective function of maximizing the regional economic growth rate, and the resource utilization objective as the objective function of minimizing energy consumption per unit of output. By clearly quantifying each objective, the demands for ecological, economic, and resource utilization are transformed into specific and calculable functional forms, making the planning objectives clear and operable, providing a foundation for subsequent coordinated optimization of multiple objectives; setting constraints for the multi-objective optimization model, including a lower limit for farmland protection area, a threshold for the proportion of ecological protection zones, and infrastructure coverage requirements. These constraints are crucial for ensuring the rationality and feasibility of national land spatial planning. They clarify the bottom-line requirements for arable land, ecological protection zones, and infrastructure construction in the planning, preventing planning schemes from violating important principles and actual needs, and ensuring that the planning meets the needs of realistic development. A multi-objective evolutionary algorithm is used to optimize the multi-objective optimization model, generating a Pareto front solution set through population iteration. Based on the principles of biological evolution, the multi-objective evolutionary algorithm continuously searches and evolves in the multi-objective space by simulating population selection, crossover, and mutation. During the iteration process, the algorithm finds a series of non-dominated solutions, forming a Pareto front solution set. These solutions reach a balance among different objectives, providing a rich candidate set for subsequent selection of the optimal solution. The comprehensive optimal solution is selected from the Pareto front solution set as the optimized national land spatial planning scheme based on the entropy weight method and the approximation ideal solution ranking method. The entropy weight method objectively determines the weight of each objective in the decision-making process, reflecting the relative importance of different objectives; the approximation ideal solution ranking method ranks the schemes by calculating the distances between each scheme and the ideal solution and the negative ideal solution. Combining these two approaches allows for the selection of the optimal solution from the Pareto front solution set, demonstrating comprehensive performance across multiple objectives. This ensures that the final land spatial planning scheme balances multi-objective needs while possessing high scientific rigor and practicality. In summary, this invention effectively addresses the challenges of balancing multiple objectives and insufficient feasibility in traditional land spatial planning by establishing a multi-objective optimization model, setting reasonable constraints, employing a multi-objective evolutionary algorithm, and combining it with a scientific scheme selection method. It generates optimized land spatial planning schemes that balance ecological protection, economic development, and resource utilization while meeting practical constraints, thereby enhancing the scientific rigor and effectiveness of land spatial planning.

[0034] In one embodiment of the present invention, the step of collecting new data in real time and comparing it with the historical dataset, calculating the planning deviation rate, and updating the parameters of the land spatial planning scheme classification model based on the planning deviation rate includes: S61, acquire new data in real time through the remote sensing image data, the new data including information on newly added construction land, changes in vegetation cover and distribution of pollution sources; S62, compare the newly added data with the data in the historical dataset, and calculate the percentage of the sum of the absolute values ​​of the deviation rates of each indicator to the planned value; S63, determine whether the deviation rate exceeds a preset threshold. If the deviation rate exceeds the preset threshold, readjust the cluster center weights and update the parameters of the land spatial planning scheme classification model.

[0035] As described in steps S61-S63 above, traditional methods in land space planning and management often fail to respond promptly to the dynamic changes in land space. On the one hand, the actual situation of land space, such as the expansion of construction land, changes in vegetation cover, and the movement of pollution sources, is constantly changing, but traditional planning lacks real-time monitoring methods, making it difficult to obtain this information in a timely manner. On the other hand, even if changes are detected, there is no effective quantitative assessment method to judge the degree of deviation between the plan and the actual situation, let alone to adjust the planning model in a timely manner based on the deviation, leading to the gradual disconnect between the plan and reality and the loss of its guiding significance. To address the above problems, this invention proposes the following technical solution: real-time acquisition of new data through remote sensing image data, including information on newly added construction land, changes in vegetation cover, and the distribution of pollution sources. Remote sensing imagery has the characteristics of periodicity and large-scale monitoring, which can capture the dynamic changes in land space in a timely manner. Utilizing this characteristic to acquire relevant data in real time can ensure the timeliness and accuracy of information, provide the latest data for subsequent analysis, and solve the problem of data acquisition lag in traditional methods; the new data is compared with the data in the historical dataset, and the percentage of the sum of the absolute values ​​of the deviation rates of each indicator to the planned value is calculated. By comparing the actual situation with the plan, the discrepancy can be quantified. The planning deviation rate can intuitively reflect the degree of deviation on various indicators during the implementation of the plan, enabling planners to clearly understand the implementation status of the plan, providing data support for subsequent decision-making, and solving the problem of lack of quantitative assessment in traditional methods. The system determines whether the deviation rate exceeds a preset threshold. If the deviation rate exceeds the preset threshold, the cluster center weights are readjusted, and the parameters of the land spatial planning scheme classification model are updated. The preset threshold is a standard set according to the accuracy requirements of the plan and the actual situation. When the deviation rate exceeds this threshold, it indicates that there is a significant deviation between the plan and reality. Adjusting the cluster center weights and updating the model parameters at this time allows the model to adapt to changes in the actual situation, ensuring the scientific nature and effectiveness of the planning scheme, and solving the problem that traditional methods cannot adjust the planning model in a timely manner based on deviations. In summary, this invention establishes a dynamic monitoring and adjustment mechanism by collecting new data in real time, quantitatively calculating the deviation rate, and adjusting model parameters based on deviations. This effectively solves the problems of traditional land spatial planning management, such as the inability to respond to spatial dynamic changes in a timely manner, the lack of quantitative assessment, and the untimely adjustment of models, enabling land spatial planning to adapt to actual changes in real time and continuously play a guiding role.

[0036] like Figure 2 Furthermore, this invention also discloses a land spatial planning management system based on big data analysis, comprising: The data acquisition and preprocessing module 1 is used to acquire multi-source heterogeneous land space data, including remote sensing image data, geographic information system data, socio-economic statistical data and environmental monitoring data, and to establish a standardized land space dataset based on the multi-source heterogeneous land space data. Fusion analysis module 2 is used to assign weights to the multi-source data in the standardized territorial spatial dataset based on the entropy weight method, construct a fusion weight matrix, and generate a multi-dimensional territorial spatial feature matrix through weighted fusion. The functional zoning module is used to construct a hybrid analysis model, which includes a spatial feature extraction module and a classification prediction module. The multidimensional land space feature matrix is ​​input into the hybrid analysis model, and the spatial functional zoning prediction results are output. Classification module 4 is used to establish a classification model for land spatial planning schemes, input the spatial functional zoning prediction results into the land spatial planning scheme classification model for cluster analysis, calculate the cluster effectiveness index, and generate a preliminary land spatial planning scheme based on the cluster effectiveness index. Decision module 5 is used to obtain preset planning objectives and, based on the preset planning objectives, solve for the optimal solution set through a multi-objective optimization algorithm to generate an optimized land space planning scheme; The dynamic verification and optimization module 6 is used to establish a historical dataset based on the optimized territorial spatial planning scheme, collect new data in real time and compare it with the historical dataset, calculate the planning deviation rate, and update the parameters of the territorial spatial planning scheme classification model based on the planning deviation rate.

[0037] The present invention also discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of a land spatial planning and management method based on big data analysis.

[0038] The present invention also discloses a computer-readable storage medium storing a computer program thereon, characterized in that the computer program, when executed by a processor, implements the steps of a land spatial planning and management method based on big data analysis.

[0039] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0040] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A land spatial planning management method based on big data analysis, characterized in that, include: Collect multi-source heterogeneous land space data, including remote sensing image data, geographic information system data, socio-economic statistical data and environmental monitoring data, and establish a standardized land space dataset based on the multi-source heterogeneous land space data; Based on the entropy weight method, weights are assigned to the multi-source data in the standardized territorial spatial dataset to construct a fusion weight matrix, and a multi-dimensional territorial spatial feature matrix is ​​generated by weighted fusion. A hybrid analysis model is constructed, and the multidimensional land space feature matrix is ​​input into the hybrid analysis model to output the spatial functional zoning prediction results. A classification model for territorial spatial planning schemes is established. The spatial functional zoning prediction results are input into the territorial spatial planning scheme classification model for cluster analysis. A preliminary territorial spatial planning scheme is generated based on the clustering results. Obtain the preset planning objectives, and based on the preset planning objectives, use a multi-objective optimization algorithm to solve for the optimal solution set to generate an optimized land spatial planning scheme; Establish a historical dataset, collect new data in real time and compare it with the historical dataset, calculate the planning deviation rate, and update the parameters of the land spatial planning scheme classification model based on the planning deviation rate.

2. The land spatial planning management method based on big data analysis according to claim 1, characterized in that, The steps for establishing a standardized land space dataset based on the multi-source heterogeneous land space data include: An outlier detection algorithm is used to calculate the outlier threshold for pixel brightness in the remote sensing image data, and pixels whose brightness exceeds the outlier threshold in the remote sensing image data are removed. The process involves acquiring vector data, raster data, and coordinate system from the geographic information system data; converting the vector data into a standard geographic format; resampling the raster data to a uniform resolution; and unifying the coordinate system to a global geographic coordinate system. Based on the spatial distribution relationship of neighboring observation points, a weighted average method is used to complete the missing data of the environmental monitoring data, wherein the weights are inversely proportional to the distance between the observation point and the missing point. A random forest regression model is trained using population density, economic growth rate, and transportation network density as feature variables. The model is then used to predict and fill in missing fields in the socioeconomic statistics.

3. The land spatial planning management method based on big data analysis according to claim 1, characterized in that, The steps of assigning weights to the multi-source data in the standardized territorial spatial dataset based on the entropy weight method, constructing a fusion weight matrix, and generating a multidimensional territorial spatial feature matrix through weighted fusion include: Evaluation indicators are set for the remote sensing image data, the geographic information system data, the socio-economic statistical data, and the environmental monitoring data. The evaluation indicators include data completeness, timeliness, and spatial resolution. The entropy value of each evaluation indicator is calculated based on the information entropy theory, wherein the entropy value is negatively correlated with data uncertainty; Each evaluation indicator is assigned a weight based on its entropy value, wherein the lower the entropy value, the higher the weight. The multi-source heterogeneous land space data are superimposed and fused according to the weights to generate a multi-dimensional land space feature matrix containing the number of spatial units and the number of fused features.

4. The land spatial planning management method based on big data analysis according to claim 1, characterized in that, The steps for constructing the hybrid analysis model include: A deep convolutional network architecture is adopted. The multidimensional land space feature matrix is ​​input, and spatial texture features are extracted through multi-layer convolutional operations to output a high-dimensional feature map. The high-dimensional feature map is input into a random forest classifier, and several functional partition categories are classified based on the decision tree voting mechanism; An adaptive learning rate optimization algorithm is adopted, with classification accuracy as the training objective, and the parameters of the deep convolutional network architecture are adjusted through cross-validation. The classification accuracy is calculated using a confusion matrix, and the consistency coefficient is used to evaluate the consistency between the prediction results and the manually labeled results.

5. The land spatial planning management method based on big data analysis according to claim 1, characterized in that, The steps of inputting the spatial functional zoning prediction results into the land spatial planning scheme classification model for cluster analysis, and generating a preliminary land spatial planning scheme based on the clustering results include: The K-means++ algorithm is used to select the initial centers; Calculate the Euclidean distance from each spatial unit to each cluster center. After calculating the Euclidean distance, reassign each spatial unit to the cluster to which the nearest cluster center belongs. The silhouette coefficient is used to calculate the clustering effectiveness index, wherein the closer the clustering effectiveness index is to 1, the better the clustering effect. If the clustering effectiveness index is not greater than a preset threshold, the number of cluster centers is increased and the process is iterated again. If the clustering effectiveness index is greater than the preset threshold, a preliminary land spatial planning scheme is generated based on the corresponding clustering results.

6. The land spatial planning management method based on big data analysis according to claim 1, characterized in that, The step of generating an optimized land spatial planning scheme by solving for the optimal solution set using a multi-objective optimization algorithm based on the preset planning objectives includes: A multi-objective optimization model is established, which includes setting the ecological protection objective as the objective function of minimizing the development intensity of ecologically sensitive areas, the economic development objective as the objective function of maximizing the regional economic growth rate, and the resource utilization objective as the objective function of minimizing energy consumption per unit of output. The constraints of the multi-objective optimization model are set, including the lower limit of the cultivated land protection area, the threshold of the proportion of ecological protection areas, and the infrastructure coverage requirements. The multi-objective optimization model is optimized using a multi-objective evolutionary algorithm, and a Pareto front solution set is generated through population iteration. Based on the entropy weight method and the approximation ideal solution ranking method, the comprehensive optimal solution is selected from the Pareto front solution set as the optimized land spatial planning scheme.

7. The land spatial planning management method based on big data analysis according to claim 1, characterized in that, The steps of collecting new data in real time and comparing it with the historical dataset, calculating the planning deviation rate, and updating the parameters of the land spatial planning scheme classification model based on the planning deviation rate include: New data is acquired in real time using the remote sensing image data, including information on newly added construction land, changes in vegetation cover, and distribution of pollution sources. Compare the newly added data with the data in the historical dataset, and calculate the percentage of the sum of the absolute values ​​of the deviation rates of each indicator to the planned value; Determine whether the deviation rate exceeds a preset threshold. If the deviation rate exceeds the preset threshold, readjust the cluster center weights and update the parameters of the land spatial planning scheme classification model.

8. A land spatial planning management system based on big data analysis, characterized in that, include: The data acquisition and preprocessing module is used to acquire multi-source heterogeneous land space data, including remote sensing image data, geographic information system data, socio-economic statistical data and environmental monitoring data, and to establish a standardized land space dataset based on the multi-source heterogeneous land space data. The fusion analysis module is used to assign weights to the multi-source data in the standardized territorial spatial dataset based on the entropy weight method, construct a fusion weight matrix, and generate a multi-dimensional territorial spatial feature matrix through weighted fusion. The functional zoning module is used to construct a hybrid analysis model, which includes a spatial feature extraction module and a classification prediction module. The multidimensional land space feature matrix is ​​input into the hybrid analysis model, and the spatial functional zoning prediction results are output. The classification module is used to establish a classification model for land spatial planning schemes, input the spatial functional zoning prediction results into the land spatial planning scheme classification model for cluster analysis, calculate the cluster effectiveness index, and generate a preliminary land spatial planning scheme based on the cluster effectiveness index. The decision-making module is used to obtain preset planning objectives and, based on the preset planning objectives, solve for the optimal solution set through a multi-objective optimization algorithm to generate an optimized land spatial planning scheme. The dynamic verification and optimization module is used to establish a historical dataset based on the optimized land spatial planning scheme, collect new data in real time and compare it with the historical dataset, calculate the planning deviation rate, and update the parameters of the land spatial planning scheme classification model based on the planning deviation rate.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images