System and method for territorial space planning and design

Abstract

The invention discloses a territorial space planning and designing system, which relates to the technical field of geographic information science and comprises a data management module, a feature extraction and analysis module, a model construction and training module, an ecological environment evaluation module, a cultivated land protection evaluation module and a territorial space purpose control planning module. The data management module is used for collecting territorial space related data from various data sources, including geographic information, ecological environment, cultivated land protection and social economic data, and cleaning and preprocessing the data. The invention further discloses a territorial space planning and designing method. According to the system and method for territorial space planning and design provided by the invention, through mutual cooperation of all the modules, a full-chain solution from data collection, processing and analysis to planning making, implementation and monitoring is provided for territorial space planning and design, so that scientificity, accuracy and operability of planning are ensured; and green, low-carbon and high-quality development can be promoted.

Description

Technical Field

[0001] The present invention relates to the field of geographic information science and technology, and in particular to a system and method for national land space planning and design. Background Art

[0002] In the field of geographic information science and technology, national space planning and design is a complex and critical task, which involves the rational use of various natural resources in the national space, ecological environment protection, farmland protection and food security, as well as the sustainable development of social economy;

[0003] However, when dealing with land and space related data, traditional methods often find it difficult to efficiently collect and integrate data from multiple data sources. At the same time, there is a lack of effective means for data cleaning and preprocessing, resulting in low data quality and affecting the accuracy of subsequent analysis results. In terms of feature extraction, traditional methods often rely on manual judgment and experience summary, and it is difficult to comprehensively and accurately extract features closely related to ecological environment assessment, farmland protection assessment and urban development assessment. When constructing assessment models, traditional methods often lack systematic model construction and optimization processes, resulting in inaccurate model assessment results and difficulty in reflecting the actual situation. At the same time, in the formulation of land and space use control planning, traditional methods often lack scientificity and precision. It is difficult to formulate effective use control planning schemes based on the results of ecological environment assessment and farmland protection assessment, combined with various restrictive factors determined by feature extraction and analysis conclusions. In this regard, we propose a system and method for land and space planning design. Summary of the invention

[0004] In order to solve the above technical problems, a system and method for national land space planning and design are provided. This technical solution solves the above problems.

[0005] In order to achieve the above purpose, the technical solution adopted by the present invention is:

[0006] The system for national land space planning and design includes: data management module, feature extraction and analysis module, model building and training module, ecological environment assessment module, cultivated land protection assessment module and national land space use control planning module;

[0007] The data management module is used to collect land space-related data from various data sources, including geographic information, ecological environment and socio-economic data, and clean and pre-process the data;

[0008] The feature extraction and analysis module is used to extract the spatial features of geographic information data, analyze the time series laws of ecological environment, farmland protection and urban development data, and pay attention to the features related to the ecological environment in social and economic data;

[0009] The model building and training module is used to combine different modules into an overall model, optimize parameters using algorithms, and use preprocessed data for training, record changes in performance indicators during training, and obtain an accurate evaluation model;

[0010] The ecological environment assessment module is used to calculate the comprehensive evaluation indicators of ecological environment quality using the trained model, analyze the evaluation results, generate a visual report, and identify key areas and key issues for ecological environment protection;

[0011] The cultivated land protection assessment module is used to calculate and assess the comprehensive evaluation indicators of cultivated land and permanent basic farmland based on the trained model, analyze the assessment results, generate a visual report, and identify key areas and key issues for cultivated land protection;

[0012] The land use control planning module is used to formulate land use control plans based on the results of ecological environment assessments and farmland protection assessments, combined with various restrictive factors determined by feature extraction and analysis conclusions, including the formulation of ecological protection red lines, permanent basic farmland protection red lines and urban development boundaries, and formulate land use planning schemes based on the generated ecological restoration, farmland protection, urban development suggestions and combined with social and economic development needs, and can be adjusted and optimized.

[0013] A method for land space planning and design, and a system for implementing land space planning and design as described in any one of the items, comprising:

[0014] Collect data related to national land space, clean and pre-process the collected data, and remove outliers;

[0015] Extract features from the processed data, identify the spatial distribution characteristics of different land use types, capture the trends and laws therein, and extract features that are closely related to the land use control planning;

[0016] Based on the extracted feature data, a combination model is constructed and the model is optimized using an optimization algorithm;

[0017] Use the trained model to evaluate the ecological environment of the national land space and obtain comprehensive evaluation indicators of the ecological environment quality;

[0018] Use the trained model to evaluate the protection of cultivated land in the national territory and obtain a comprehensive evaluation index of the quality of cultivated land protection;

[0019] Based on the results of ecological environment assessment and farmland protection assessment, combined with various restrictive factors determined by feature extraction and analysis, a land use control plan is formulated. The land use control plan includes the demarcation of ecological protection red lines, the demarcation of permanent basic farmland protection red lines, the demarcation of urban development boundaries and the use control plan for rational resource utilization.

[0020] Preferably, data related to national land space is collected, and the collected data is cleaned and preprocessed to remove outliers, including:

[0021] Integrate data from different sources into a unified coordinate system and data format;

[0022] Based on the integrated data, missing values, duplicate values, and outliers are processed;

[0023] Among them, missing values ​​are checked by checking the data table and marking the cells containing null values. For raster data, check whether there are invalid values ​​in the pixels. If there are missing values, remove the entire column of data and obtain the data again;

[0024] Among them, duplicate values ​​are checked according to all key fields using data management software, and completely duplicate records are marked. Duplicate records are directly deleted and only one copy is retained;

[0025] Among them, outliers are judged based on the mean and standard deviation. If there are outliers, the entire column of data is removed and the data is obtained again; the calculation formula for the mean is:

[0026]

[0027] In the formula, μ represents the mean, s represents the total number of data points, and y represents the yth data point. Represents the value of the y-th data point

[0028] The calculation formula for standard deviation is:

[0029]

[0030] In the formula, ρ represents the standard deviation;

[0031] For raster data, a fixed-size window is used to calculate the mean of the pixel values ​​within the window, and the original value of the pixel in the center of the window is replaced by the mean value. The entire image is traversed to smooth the data.

[0032] Preferably, extracting features from the processed data specifically includes:

[0033] The spatial feature extraction method of geographic information data is:

[0034] Overlaying vector data of different land use types to analyze the spatial distribution relationship of different types, wherein the vector data includes but is not limited to: cultivated land, forest land, grassland, and urban and rural construction land;

[0035] The land use pattern index is calculated respectively, and the land use pattern index includes: patch area, patch number, patch density, shape index, and spread index. The land use pattern index reflects the degree of fragmentation, diversity, and connectivity of land use;

[0036] The terrain features were extracted using digital elevation model data, and the terrain analysis tools of ArcGIS software were used to process the digital elevation model data to obtain slope and aspect information;

[0037] The method for analyzing the time series regularity of ecological environment data and cultivated land protection data is as follows:

[0038] Collect long-term series of ecological environment and cultivated land data, and use linear regression analysis to analyze the changing trends and periodic characteristics of ecological environment and cultivated land protection indicators over time;

[0039] The method for extracting features related to ecological environment, farmland protection, and urban construction in social and economic data is as follows:

[0040] Associate population data with geographic spatial data, calculate population density in different regions, and analyze the relationship between population density and ecological environment, farmland protection, and urban construction indicators;

[0041] Collect economic data and analyze the relationship between economic development indicators and ecological environment, farmland protection and urban construction indicators through correlation coefficients. The economic data include: regional gross domestic product and industrial structure.

[0042] Preferably, based on the extracted feature number r, constructing the combination model specifically includes:

[0043] The pre-processed land space related data are used as the input of the model, wherein the land space related data include: geographic information data, ecological environment data, cultivated land protection data, urban construction data and social and economic data;

[0044] CNN layer: Use multiple convolution kernels to perform sliding convolution on the input data to extract spatial features. For two-dimensional input data, the convolution operation is expressed as:

[0045]

[0046] In the formula, C b，g represents the output value after convolution, L represents the input data, Q represents the convolution kernel, h and n represent the size of the convolution kernel;

[0047] The pooling operation uses maximum pooling to reduce the data dimension and extract the main features;

[0048] LSTM layer: The LSTM unit consists of a forget gate, an input gate, and an output gate;

[0049] Among them, the forget gate expression is:

[0050] f t =σ(W f ·[h t-1 , x t ]+b f )

[0051] In the formula, f t represents the output of the forget gate, σ represents the sigmoid function, W f represents the weight matrix of the forget gate, h t-1 represents the hidden state at the previous moment, x t represents the input at the current moment, b f Represents the bias term of the forget gate;

[0052] Among them, the input gate expression is:

[0053] i t =σ(W i ·[h t-1 , x t ]+b i )

[0054] In the formula, i t represents the output of the input gate, W i represents the weight matrix of the input gate, b i represents the bias term of the input gate;

[0055] Among them, the candidate unit state expression is:

[0056] C′ t =tanh(W C ·[h t-1 , x t ]+b c )

[0057] In the formula, C′ t represents the candidate unit state, W C The weight matrix representing the candidate unit state, b C represents the bias term;

[0058] Among them, the update unit state expression is:

[0059] C t =f t ⊙C t-1 +i t ⊙C′ t )

[0060] In the formula, C t represents the cell state at the current moment, ⊙ represents element-by-element multiplication;

[0061] Among them, the output gate expression is:

[0062] o t =σ(W o ·[h t-1 , x t ]+b o )

[0063] In the formula, o t represents the output of the output gate, W o represents the weight matrix of the output gate, b o represents the bias term of the output gate;

[0064] Among them, the hidden state expression is:

[0065] h t =o t ⊙tanh(C t )

[0066] The attention weight is calculated based on the input data and the query vector, where the attention calculation formula is:

[0067]

[0068] In the formula, θ represents the query vector, κ ι represents the key vector in the input data, d κ represents the dimension of the key vector, ∈ ι represents the attention score;

[0069] The attention weight is weighted and summed with the input data to get the attention output, where the calculation formula of the attention output is:

[0070]

[0071] In the formula, S ι Represents the value vector in the input data, softmax(∈ ι ) represents the normalized attention weight;

[0072] According to the task requirements, activation functions and output units are selected to construct a combined model architecture suitable for ecological environmental assessment, farmland protection assessment and land use control planning.

[0073] Preferably, optimizing the model using an optimization algorithm specifically includes:

[0074] A set of particles is randomly generated, each particle represents a set of parameters of the model, and each particle has a position vector and a time vector. The position vector represents the value of the model parameter, and the time vector determines the time direction and speed of the particle in the search space;

[0075] For each particle, its position vector is used as the parameter of the model, the model is trained using the training data, and the fitness value of the model is calculated;

[0076] Compare the current fitness value of each particle with the individual's historical optimal fitness value. If the current fitness value is better, update the individual's optimal position. Compare the current fitness value of each particle with the group's historical optimal fitness value. If the current fitness value is better, update the group's optimal position.

[0077] Among them, the calculation method for updating the time point and position of the particle is:

[0078]

[0079] In the formula, and denote the time point and position of particle i at the tth iteration, ω denotes the inertia weight, ε1 and ε2 denote the learning factors, ζ1 and ζ2 denote random numbers in the range of [0, 1], and pbest i represents the individual optimal position of particle i, and gbest represents the group optimal position;

[0080] Repeat the above steps until the stop condition is met;

[0081] The optimal position of the group is used as the optimal parameter of the model for the final model training and prediction, ensuring the accuracy and efficiency of the model in ecological environmental assessment, farmland protection assessment and land use control planning tasks.

[0082] Preferably, in the stage of cultivated land protection assessment, in view of the problems of cultivated land quality decline and area reduction, the protection strategy is determined based on the cultivated land quality evaluation model and the cultivated land change driving force model. The cultivated land quality evaluation model is constructed by a multi-factor comprehensive evaluation method, and the cultivated land change driving force model uses a Logistic regression model to identify key driving factors.

[0083] Among them, the expression of the cultivated land quality evaluation model constructed by the multi-factor comprehensive evaluation method is:

[0084]

[0085] In the formula, A represents the comprehensive score of cultivated land quality, L z Represents the weight of the zth evaluation factor, which is determined by the expert scoring method and satisfies U z represents the standardized value of the zth evaluation factor;

[0086] Among them, the expression of key driving factors identified by the Logistic regression model is:

[0087]

[0088] In the formula, V represents the probability of cultivated land change, e represents the natural constant, and s represents the linear combination of independent variables;

[0089] In areas where arable land quality has degraded, the obstacle factors are accurately located based on the evaluation results. If the soil fertility is insufficient, soil improvement projects will be implemented; if the irrigation facilities are imperfect, the irrigation system will be optimized. For areas where arable land is threatened by construction occupation, the use control will be strengthened based on driving force analysis, and construction land projects will be strictly approved. For those that meet the occupation conditions, the balance system of occupation and compensation will be implemented and the quality standards of supplementary arable land will be improved.

[0090] Preferably, in terms of rational resource utilization planning, for land resources, land suitability evaluation methods are used;

[0091] An evaluation system is constructed based on natural factors and socio-economic factors. The natural factors include soil type, topography and hydrological conditions, and the socio-economic factors include land use status and planned use:

[0092] The analytic hierarchy process was used to determine the weight of each evaluation factor, and the GIS spatial analysis technology was used to divide the land suitability level;

[0093] Provide a basis for land use layout for national land use control, give priority to improving the quality of land and the stability of land use within the protection red line of cultivated land and permanent basic farmland, guide the efficient and intensive use of construction land within the urban development boundary, strictly protect and restore ecological land within the ecological protection red line, and coordinate the allocation of land resources and control strategies for different uses.

[0094] Preferably, the expansion limit of the urban development boundary is determined by the following steps:

[0095] Resource and environmental carrying capacity assessment, using GIS technology to assess land, water resources and ecological environment capacity, determine the construction suitability level and water resource support scale, and ensure resource carrying capacity.

[0096] Prediction of population and industrial development trends, using models to predict population growth and industrial development trends, providing a basis for urban development scale and facility configuration.

[0097] Analyze the supporting capacity of facilities, evaluate the supporting capacity of transportation and public service facilities, and determine the facility configuration standards and construction scale requirements.

[0098] Comprehensive integration and boundary demarcation, build a multi-objective decision-making model, delineate urban development boundaries, optimize plans, and achieve multi-factor coordinated planning and comprehensive benefit balance.

[0099] Compared with the prior art, the present invention has the following beneficial effects:

[0100] The system and method of land space planning and design proposed in the present invention cooperate with each other through various modules, providing a full-chain solution from data collection, processing, analysis to planning, implementation and monitoring for land space planning and design, which helps to ensure the scientificity, accuracy and operability of the planning. By using algorithms and models, it can accurately and efficiently evaluate the ecological environment quality of the land space, identify key areas and key issues of ecological protection and farmland protection, and provide strong support for the formulation of targeted ecological environment protection plans. It can also formulate detailed land space use control plans based on the evaluation results. Flexible adjustment and optimization based on real-time monitoring data and ecological environment changes can help ensure the timeliness and adaptability of the plan. The use of resource management methods such as water resource balance analysis methods can help achieve optimal allocation and efficient utilization of resources. By considering the balance between carbon emissions and carbon sinks, it can help promote green and low-carbon development and promote the sustainable development of land space. BRIEF DESCRIPTION OF THE DRAWINGS

[0101] Figure 1 It is a framework diagram of the system composition of the present invention. DETAILED DESCRIPTION

[0102] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art may think of other obvious variations.

[0103] Reference Figure 1 As shown in the figure, the system of national land space planning and design. The system mainly covers several important modules, each of which cooperates and complements each other to jointly build a complete national land space planning and design system.

[0104] The data management module plays a fundamental and key role. It can collect land space-related data from a wide range of data sources, including detailed geographic information data provided by the geographic information system, such as topography, water system distribution, etc.; various ecological and environmental data recorded by the ecological and environmental monitoring system, including air quality, water quality, biodiversity indicators, etc.; and socio-economic data collected by the socio-economic statistics department, such as population density, industrial layout, and economic development level. After collecting these massive and complex data, the data management module will use professional data cleaning technology and pre-processing methods to carefully remove outliers, erroneous data, and duplicate data, so as to ensure that the data used for subsequent analysis and planning are highly accurate, complete, and reliable.

[0105] The feature extraction and analysis module focuses on the in-depth mining and analysis of data. For geographic information data, it is committed to extracting its unique spatial features, such as identifying the spatial form, distribution range and mutual relationship of geographical elements such as mountains and rivers through spatial analysis algorithms; for ecological environment, farmland protection and urban development data, this module will deeply analyze its time series laws, such as studying the changing trend of ecological environment indicators in the past few years, whether it has gradually improved or deteriorated, the increase and decrease of cultivated land area in different historical periods, and the speed and direction of urban scale expansion over time. At the same time, it will also focus on the characteristics of socio-economic data that are closely related to the ecological environment, such as the distribution of certain highly polluting industries and the relationship between ecologically sensitive areas, and the manifestation of population growth on ecological and environmental pressure, so as to provide comprehensive and in-depth information basis for subsequent planning decisions.

[0106] The model building and training module undertakes the important task of building an accurate assessment model. It cleverly integrates the data and analysis results provided by different modules into an overall model. In this process, advanced algorithms are used to optimize and adjust the model parameters, such as genetic algorithms and particle swarm optimization algorithms, to improve the accuracy and adaptability of the model. Subsequently, the model is repeatedly trained using the data preprocessed by the data management module, and the changes in performance indicators such as accuracy, recall rate, mean square error, etc. are closely recorded during the training process. Through continuous adjustment and optimization, an assessment model that can accurately reflect the status of national land space is finally obtained. This model will become the core tool for subsequent ecological environment assessment and cultivated land protection assessment.

[0107] The ecological environment assessment module works with the help of a trained model. It can accurately calculate the comprehensive evaluation indicators of ecological environment quality, which comprehensively consider factors such as air quality, water quality, soil quality, biodiversity, etc., and comprehensively and objectively reflect the ecological environment status in the national space. After obtaining the evaluation results, the module will use professional analysis methods to conduct in-depth analysis of the results, such as spatial interpolation analysis, cluster analysis, etc., to find out the spatial distribution differences and change patterns of ecological environment quality. In addition, it will also generate intuitive and clear visual reports, presenting the evaluation results in the form of charts, maps, etc., so that decision makers can understand the ecological environment status at a glance, and then determine the key areas for ecological environment protection, such as nature reserves, water conservation areas, etc., and clarify key issues, such as severely polluted river basins and ecologically fragile mountainous areas, to provide strong support for the targeted formulation of ecological protection strategies.

[0108] The cultivated land protection assessment module is also based on the trained model. It calculates and evaluates the comprehensive evaluation indicators of cultivated land and permanent basic farmland. These indicators involve the quality of cultivated land, area stability, production capacity level, rationality of permanent basic farmland demarcation, protection effectiveness and other dimensions. After completing the evaluation calculation, the evaluation results are carefully analyzed, such as comparing the differences in cultivated land quality in different regions and analyzing the threats faced by permanent basic farmland protection. Then a visual report is generated, showing the distribution of cultivated land, quality grade distribution and the protection status of permanent basic farmland through a map, so as to determine the key areas of cultivated land protection, such as high-quality cultivated land concentration areas, vulnerable cultivated land areas, etc., and key issues, such as areas with serious non-agriculturalization of cultivated land and risk points of permanent basic farmland being occupied, etc., to provide a clear direction for the formulation and implementation of cultivated land protection policies.

[0109] The land use control planning module is the core decision-making output part of the entire system. It formulates a land use control plan based on the results of ecological environment assessment and cultivated land protection assessment, and combines various restrictive factors determined by the conclusions of feature extraction and analysis, such as the requirements for restricting development in ecologically sensitive areas and the prohibition of non-agricultural construction within the cultivated land protection red line. This plan mainly includes the formulation of ecological protection red lines, clarifying which areas are ecological core areas that must be strictly protected, and prohibiting all development activities that damage the ecological environment; demarcating permanent basic farmland protection red lines to ensure that high-quality cultivated land and basic farmland are permanently protected and national food security; determining urban development boundaries, rationally planning the scale and direction of urban development, and preventing urban disorderly expansion from causing damage to the ecological environment and cultivated land. In addition, based on the generated ecological restoration suggestions, such as restoration plans for damaged ecosystems, cultivated land protection measures, such as land consolidation and farmland water conservancy construction, and urban development suggestions, such as industrial layout optimization and infrastructure construction planning, combined with social and economic development needs, a comprehensive and scientific land space planning plan is formulated. During the implementation of the plan, it can also be flexibly adjusted and optimized according to actual conditions to adapt to the changing social and economic development situation and ecological and environmental conditions.

[0110] The method for national land space planning and design, which is used to implement the system for national land space planning and design, has the following specific steps:

[0111] The first step is to collect data related to national land space. This requires staff to comprehensively and systematically obtain data from various relevant fields, including but not limited to obtaining accurate geographic information data from geographic surveying and mapping departments, collecting detailed ecological and environmental data from environmental monitoring agencies, and collating social and economic data from statistical departments, etc., to ensure the breadth and reliability of data sources. After collecting the data, the collected data is immediately cleaned and preprocessed. This process strictly follows the standard specifications for data processing, uses professional data processing software and algorithms, and carefully checks and removes outliers. For example, geographic coordinate data that obviously deviates from the normal range, illogical ecological and environmental indicator data, and social and economic data that are in serious conflict with other data are screened and corrected, laying a solid foundation for subsequent data analysis.

[0112] The second step is to extract features from the processed data. Using advanced geographic information system technology, data mining algorithms and other means, the spatial distribution characteristics of different land use types are identified, such as clearly dividing the spatial distribution range, shape, and mutual relationship of different types of land such as urban construction land, agricultural land, forest land, and water areas. At the same time, the trends and laws are captured through time series analysis, trend analysis and other methods, such as the trend of urban construction land gradually expanding to the surrounding areas over time, and the law that the area of ​​cultivated land decreases year by year in some areas. In addition, the features closely related to the land use control planning are accurately extracted, such as the boundary features of ecologically sensitive areas and the characteristics of cultivated land quality grades, etc., to provide key feature data for building accurate models.

[0113] The third step is to build a combined model based on the extracted feature data. According to the goals and needs of national land space planning, select a suitable model framework, such as a decision tree model based on machine learning, a neural network model, etc., to organically integrate feature data from different sources and types to build a combined model that can fully reflect the status of national land space. After the model is built, use optimization algorithms to optimize the model, such as using gradient descent algorithms, stochastic gradient descent algorithms, etc. to adjust the parameters of the model, and continuously improve the model's fit and prediction ability, so that it can more accurately evaluate the various conditions and trends of national land space.

[0114] The fourth step is to use the trained model to evaluate the ecological environment of the national land space. The processed and feature-extracted data is input into the trained model. The model will comprehensively consider various ecological and environmental factors such as air quality, water quality, biodiversity, etc. based on its internal algorithms and parameters, and calculate the comprehensive evaluation indicators of ecological and environmental quality. These indicators can quantitatively reflect the quality, change trend and existing problems of the ecological environment in the national land space, and provide a scientific basis for decision-making on ecological and environmental protection.

[0115] The fifth step is to use the trained model to evaluate the protection of cultivated land in the national territory. Similarly, the relevant data is input into the model, which will analyze and calculate from multiple dimensions such as cultivated land quality, area stability, production capacity level, and the demarcation and protection status of permanent basic farmland to obtain comprehensive evaluation indicators of cultivated land protection quality. Through these indicators, we can fully understand the current status, challenges and potential risks of cultivated land protection, and provide strong support for the formulation of effective cultivated land protection strategies.

[0116] The last step is to formulate a land use control plan based on the results of the ecological environment assessment and the results of the cultivated land protection assessment, combined with various restrictive factors determined by the feature extraction analysis conclusions. In this process, the ecological protection red line is delineated based on the key ecological protection areas and key issues determined by the ecological environment assessment, and the core areas of ecological protection and strict protection requirements are clarified; based on the results of the cultivated land protection assessment, the permanent basic farmland protection red line is delineated to ensure the permanent protection of cultivated land resources; at the same time, the urban development boundaries are determined by comprehensively considering the needs of urban development and the limitations of the ecological environment and cultivated land protection, and the layout and scale of the town are reasonably planned. In addition, a use control plan for the rational use of resources including ecological restoration, cultivated land protection, urban development, etc. is formulated to ensure that the development and protection of the land space can achieve a scientific, reasonable and sustainable balance, while promoting social and economic development, protecting the ecological environment and cultivated land resources to the greatest extent.

[0117] Collect data related to national land space, clean and pre-process the collected data, and remove outliers, including:

[0118] Integrate data from different sources into a unified coordinate system and data format. Since land and space related data come from multiple different channels, such as field surveys, satellite remote sensing, geographic information systems, etc., the coordinate systems and formats of these data are different. Therefore, they need to be integrated into a unified coordinate system and data format for subsequent processing and analysis.

[0119] Next, based on the integrated data, the data are processed for missing values, duplicate values, and outliers.

[0120] To handle missing values, you need to check the data table and carefully mark the cells containing null values. For raster data, you also need to check whether there are invalid values ​​in the pixels. If missing values ​​are found, remove the entire column of data and re-acquire the data. This is because missing values ​​will have a serious impact on subsequent analysis and planning, so it is necessary to ensure the integrity of the data as much as possible.

[0121] For the processing of duplicate values, use data management software to check for duplicates according to all key fields. In this way, completely duplicate records can be marked. Then, the duplicate records can be directly deleted and only one copy can be retained. The existence of duplicate values ​​will increase the redundancy of data, waste storage space, and also mislead the analysis results, so they need to be removed in time.

[0122] The judgment of outliers is based on the mean and standard deviation. If the deviation of a data point from the mean exceeds a certain standard deviation range, then the data point can be considered an outlier. If an outlier is found, the entire column of data is also removed and the data is re-acquired. Outliers are caused by measurement errors, data entry errors, etc., which will have a great impact on the analysis results, so they need to be handled with caution; the calculation formula for the mean is:

[0123]

[0124] In the formula, μ represents the mean, s represents the total number of data points, and y represents the yth data point. Represents the value of the y-th data point

[0125] The calculation formula for standard deviation is:

[0126]

[0127] In the formula, ρ represents the standard deviation;

[0128] For raster data, a fixed-size window is used to calculate the mean of the pixel values ​​within the window, and the original value of the pixel in the center of the window is replaced by the mean value. The entire image is traversed to smooth the data.

[0129] Feature extraction of processed data specifically includes:

[0130] The spatial feature extraction method of geographic information data is:

[0131] First of all, superimposing vector data of different land use types is a basic and important operation. For example, vector data such as cultivated land, forest land, grassland, and urban and rural construction land, each of which represents different land use methods and functions. By performing superposition analysis in the Geographic Information System (GIS) platform, the spatial distribution relationship of these different types can be clearly shown. For example, you can intuitively see the bordering area between cultivated land and forest land, and the expansion trend of urban and rural construction land around grassland. This analysis of distribution relationships helps to reveal the spatial pattern of land use and the mutual influence and action mechanism between different types of land.

[0132] Secondly, it is also an indispensable step to calculate the land use pattern index separately. Among them, the patch area reflects the size of each land use patch, which is of great significance for measuring the scale and concentration of land use. For example, a large area of ​​cultivated land patches may mean a large-scale agricultural production area, while a small area of ​​urban and rural construction land patches may represent scattered rural settlements or small town construction areas. The number of patches reflects the total number of land use patches in a specific area, which is closely related to the degree of land use fragmentation. The patch density is calculated by the ratio of the number of patches to the total area of ​​the region, which can more accurately reflect the degree of land use fragmentation, that is, the higher the patch density, the more fragmented the land use. The shape index is used to describe the complexity of the shape of the patch, which can help determine whether the land use is strongly disturbed by natural or human factors. For example, regular-shaped cultivated land patches may be the result of large-scale planning and arrangement, while complex-shaped forest patches may be affected by natural factors such as topography and landforms. The sprawl index focuses on measuring the spatial continuity and connectivity of land use types. For example, continuously distributed grassland patches have a higher sprawl, while grassland patches that are severely fragmented by other land use types have a lower sprawl. Through the comprehensive calculation and analysis of these land use pattern indices, key characteristics such as the degree of fragmentation, diversity, and connectivity of land use can be fully and accurately reflected, providing detailed spatial structure information for national land space planning.

[0133] Furthermore, using digital elevation model (DEM) data to extract terrain features is an important means to gain an in-depth understanding of the spatial characteristics of geographic information data. With the help of terrain analysis tools in GIS software, the digital elevation model data can be processed to obtain slope and aspect information. Slope information has important guiding significance for land use planning. For example, steep slope areas may be more suitable for developing forestry or maintaining natural ecology, while gentle slope areas are more conducive to farmland reclamation or urban construction. Slope aspect information should not be ignored either. Different slope aspects will affect natural conditions such as light, temperature, and moisture, thereby affecting vegetation growth and land use. For example, sunny slopes usually have sufficient light and are suitable for growing light-loving crops or building solar power generation facilities, while shady slopes may be more suitable for developing some shade-loving vegetation or serving as water conservation areas.

[0134] The time series analysis of ecological environment data and farmland protection data aims to reveal the dynamic change characteristics of these data over time, and its analysis method is systematic and scientific.

[0135] The primary task is to collect long-term series of ecological environment and cultivated land data. These data come from a wide range of sources, including multi-year meteorological data recorded by meteorological monitoring stations (such as temperature, precipitation, etc.), soil quality data provided by soil monitoring agencies (such as soil fertility, soil erosion degree, etc.), biodiversity data obtained by ecological monitoring networks (such as changes in the species and number of plants and animals, etc.), and cultivated land area, cultivated land quality and other data statistically collected by cultivated land resource management departments. The use of linear regression analysis to process these data can effectively analyze the changing trends and periodic characteristics of ecological environment and cultivated land protection indicators over time. For example, by performing linear regression analysis on multi-year temperature data, it can be determined whether the temperature in the region is rising, falling or relatively stable, and then analyze the impact of such temperature changes on the ecological environment (such as vegetation growth cycle, ecosystem balance, etc.) and cultivated land protection (such as crop growth suitability, irrigation demand, etc.). For cultivated land protection data, such as time series analysis of cultivated land area, it can be found whether the cultivated land area is decreasing, increasing or fluctuating year by year, and whether there is a periodic law in this change, such as whether it is related to the agricultural policy adjustment cycle or the economic development cycle. Through this time series law analysis, the development trend of ecological environment and farmland protection can be predicted in advance, providing a forward-looking basis for formulating corresponding policies and plans.

[0136] The feature extraction methods related to ecological environment, farmland protection and urban construction in socio-economic data focus on the association and relationship mining between data.

[0137] The key step is to associate population data with geospatial data. Through geocoding and other technical means, population data can be accurately located in geographic space to calculate the population density of different regions. For example, the population density is usually higher in the central area of ​​the city, while it is lower in remote mountainous areas or nature reserves. Analyzing the relationship between population density and ecological environment, farmland protection and urban construction indicators is of great significance in many aspects. Higher population density may cause greater pressure on the ecological environment, such as increasing energy consumption, generating more sewage and garbage, etc., thus affecting the quality of the ecological environment. In terms of farmland protection, areas with high population density may face greater pressure on farmland demand and are prone to non-agriculturalization of farmland. For urban construction, population density is an important basis for determining the scale of towns, the scale and layout of infrastructure construction.

[0138] In addition, it is also important to collect economic data and analyze the relationship between economic development indicators and ecological environment, farmland protection and urban construction indicators through correlation coefficients. The regional GDP reflects the overall economic scale and development level of a region, while the industrial structure shows the proportion and distribution of different industries in the regional economy. For example, regions with a high proportion of industry may produce more industrial pollution, which will have a negative impact on the ecological environment, but at the same time, they may also drive the rapid development of urban construction, such as building more industrial parks and transportation facilities. Through correlation coefficient analysis, the degree of correlation between these economic development indicators and ecological environment, farmland protection and urban construction indicators can be accurately quantified, such as calculating the correlation coefficient between regional GDP and air quality indicators to determine whether economic development is accompanied by the deterioration of air quality; analyzing the correlation between industrial structure adjustment and changes in cultivated land area to understand whether the decline in the proportion of agricultural industries leads to a reduction in cultivated land area. This analysis helps to achieve coordination and balance between economic development, ecological environment protection and cultivated land protection, and urban construction in national land space planning.

[0139] Based on the extracted feature data, building a combination model is one of the core links in the planning and design of national land space. The pre-processed national land space related data are used as the input of the model. These data cover multiple aspects such as geographic information data, ecological environment data, cultivated land protection data, urban construction data and socio-economic data. The spatial feature information provided by geographic information data, such as land use type distribution, topographic and geomorphic characteristics, etc., builds the spatial framework foundation for the model. The various indicators in the ecological environment data, such as air quality, water quality, biodiversity, etc., can reflect the ecological status of the national land space and are the key basis for evaluating the quality and change trend of the ecological environment. The cultivated land protection data focuses on the quality, area, and production capacity of cultivated land, and plays an irreplaceable role in ensuring national food security and sustainable agricultural development. The urban construction data records the scale, layout, infrastructure construction and other information of the town, which is helpful to plan the reasonable development direction and scale of the town. The characteristics of population density, economic development level and industrial structure reflected in the socio-economic data can reveal the socio-economic driving factors and constraints behind the development and utilization of national land space. By integrating these multi-source and multi-type data into the model, a comprehensive and integrated national land space planning model can be constructed, which can comprehensively consider the interrelationships and influences between various factors, thereby providing scientific and accurate decision-making support for national land space use control planning, ecological environmental protection planning, cultivated land protection planning, and urban construction planning. For example, in the model, according to the land use type and terrain characteristics in the geographic information data, combined with the ecological sensitivity indicators in the ecological environmental data, it can be determined which areas are suitable for development and construction, and which areas need to be protected; based on the population density and economic development needs in the socio-economic data, and the quality and area of ​​cultivated land in the cultivated land protection data, the expansion direction and scale of the town can be planned, and at the same time, a reasonable cultivated land protection strategy can be formulated to achieve sustainable use and coordinated development of national land space.

[0140] CNN layer: Use multiple convolution kernels to perform sliding convolution on the input data to extract spatial features. For two-dimensional input data, the convolution operation is expressed as:

[0141]

[0142] In the formula, C b，g represents the output value after convolution, L represents the input data, Q represents the convolution kernel, h and n represent the size of the convolution kernel;

[0143] The pooling operation uses maximum pooling to reduce the data dimension and extract the main features;

[0144] LSTM layer: The LSTM unit consists of a forget gate, an input gate, and an output gate;

[0145] Among them, the forget gate expression is:

[0146] f t =σ(W f ·[h t-1 , x t ]+b f )

[0147] In the formula, f t represents the output of the forget gate, σ represents the sigmoid function, W f represents the weight matrix of the forget gate, h t-1 represents the hidden state at the previous moment, x t represents the input at the current moment, b f Represents the bias term of the forget gate;

[0148] Among them, the input gate expression is:

[0149] i t =σ(W i ·[h t-1 , x t ]+b i )

[0150] In the formula, i t represents the output of the input gate, W i represents the weight matrix of the input gate, b i represents the bias term of the input gate;

[0151] Among them, the candidate unit state expression is:

[0152] C′ t =tanh(W C ·[h t-1 , x t ]+b C )

[0153] In the formula, C′ t represents the candidate unit state, W C The weight matrix representing the candidate unit state, b C represents the bias term;

[0154] Among them, the update unit state expression is:

[0155] C t =f t ⊙C t-1 +i t ⊙C′ t )

[0156] In the formula, C t represents the cell state at the current moment, ⊙ represents element-by-element multiplication;

[0157] Among them, the output gate expression is:

[0158] o t =σ(W o ·[h t-1 , x t ]+b o )

[0159] In the formula, o t represents the output of the output gate, W o represents the weight matrix of the output gate, b o represents the bias term of the output gate;

[0160] Among them, the hidden state expression is:

[0161] h t =ot⊙tanh(Ct)

[0162] The attention weight is calculated based on the input data and the query vector, where the attention calculation formula is:

[0163]

[0164] In the formula, θ represents the query vector, κ ι represents the key vector in the input data, d κ represents the dimension of the key vector, ∈ ι represents the attention score;

[0165] The attention weight is weighted and summed with the input data to get the attention output, where the calculation formula of the attention output is:

[0166]

[0167] In the formula, S ι Represents the value vector in the input data, softmax(∈ ι ) represents the normalized attention weight;

[0168] According to the task requirements, activation functions and output units are selected to construct a combined model architecture suitable for ecological environmental assessment, farmland protection assessment and land use control planning.

[0169] Optimizing the model using the optimization algorithm specifically includes:

[0170] In the complex process of building and optimizing models related to land space planning and design, an effective parameter optimization strategy involves the use of particle swarm optimization algorithm. First, a group of particles is randomly generated, which constitutes a group exploring in the model parameter search space. Each particle has unique properties, which are described by a position vector and a time vector. Among them, the position vector plays a very critical role because it directly represents the value of the model parameter. For example, in a model used to evaluate the ecological environment quality of land space, an element in the position vector may represent the setting of the weight of a certain pollutant, and another element may correspond to the quantitative parameters of a specific ecosystem service function. The time vector gives the particle a unique motion characteristic in the search space, which determines the time direction and speed of the particle in the search space. The time direction can be understood as the tendency of the particle to explore the parameter space in different iteration steps, whether to adjust the parameters in the direction of making the model more inclined to ecological protection, or to adjust in the direction of balancing economic development and ecological balance; the speed controls the magnitude of the change in the parameter value of the particle in each iteration. A faster speed means that the particle can quickly search for possible parameter combinations in a larger range, while a slower speed allows the particle to explore the parameter space of the local area more finely.

[0171] For each particle generated, the following operation is crucial. Its position vector is input into the model as a parameter of the model, and then the model is trained using carefully prepared training data. These training data come from observations and statistics of various aspects of national space, covering information such as geographic information data, ecological environment monitoring data, cultivated land protection data, and socio-economic data after preprocessing and feature extraction. After completing the model training, the fitness value of the model needs to be calculated. This fitness value is a comprehensive indicator to measure the quality of the model. It may comprehensively consider the degree of fit of the model to historical data, the accuracy of future trend prediction, and the ability of the model to balance between different national space planning goals (such as ecological environment protection, cultivated land protection, and reasonable urban development). For example, a model with a high fitness value can accurately reproduce the trend of past ecological environment indicators when simulating ecological environment data, and when predicting the future ecological environment based on current data, its prediction results are consistent with the judgment made by experts based on experience and other research methods. At the same time, the model can also provide reasonable planning suggestions for urban development and cultivated land protection under the premise of considering ecological protection, so that various goals can be coordinated rather than conflicting.

[0172] After obtaining the model fitness value corresponding to each particle, we enter the key stage of comparison and update. Compare the current fitness value of each particle with the individual historical optimal fitness value. The individual historical optimal fitness value records the best model performance achieved by the particle in the previous iteration. If the current fitness value is better, it means that the particle has found a better set of model parameter combinations in this iteration, which improves the performance of the model in all aspects. At this time, it is necessary to update the individual optimal position and record the current position vector as the individual optimal position of the particle, so as to continue to explore better parameter combinations with this as a reference point in subsequent iterations. At the same time, the current fitness value of each particle is also compared with the group's historical optimal fitness value. The group's historical optimal fitness value represents the best model performance found by the entire particle group in past iterations. When the current fitness value of a particle is better than the group's historical optimal fitness value, this is an extremely important discovery, which indicates that the particle has found a new and better model parameter configuration scheme, which may lead the entire group to develop in a better direction. Therefore, the optimal position of the group must be updated, and the position vector of the particle is determined as the optimal position of the group. In the subsequent iteration process, other particles in the group will refer to this optimal position of the group and adjust their own time vector and position vector, so as to explore more efficiently in the entire search space, hoping to find more and better model parameter combinations, so as to continuously improve the accuracy, reliability and practicality of the model in the planning and design of national land space, and provide solid model support for the scientific and reasonable planning of national land space.

[0173] Among them, the calculation method for updating the time point and position of the particle is:

[0174]

[0175] In the formula, and denote the time point and position of particle i at the tth iteration, ω denotes the inertia weight, ε1 and ε2 denote the learning factors, ζ1 and ζ2 denote random numbers in the range of [0, 1], and pbest i represents the individual optimal position of particle i, and gbest represents the group optimal position;

[0176] Repeat the above steps until the stop condition is met;

[0177] The optimal position of the group is used as the optimal parameter of the model for the final model training and prediction, ensuring the accuracy and efficiency of the model in ecological environmental assessment, farmland protection assessment and land use control planning tasks.

[0178] In the crucial link of cultivated land protection assessment, we are faced with two severe challenges: the decline of cultivated land quality and the reduction of cultivated land area. In order to effectively deal with these problems, we determine scientific and reasonable protection strategies based on specially constructed cultivated land quality evaluation models and cultivated land change driving force models. Among them, the cultivated land quality evaluation model is constructed using a multi-factor comprehensive evaluation method. By comprehensively considering multiple factors, it can comprehensively and accurately evaluate the quality of cultivated land. The cultivated land change driving force model uses the Logistic regression model to accurately identify the key driving factors that affect cultivated land changes.

[0179] In areas where the quality of cultivated land has degraded, we will accurately locate the obstacles that lead to quality degradation based on detailed evaluation results. For example, if the soil fertility is insufficient, we will vigorously implement soil improvement projects and use a series of scientific means to improve the fertility level of the soil, including reasonable fertilization and the use of soil conditioners. If the irrigation facilities are not perfect, we will focus on optimizing the irrigation system, such as repairing irrigation channels and updating irrigation equipment, to ensure that the cultivated land can be adequately and reasonably irrigated. For areas where the cultivated land area is threatened by construction occupation, we will strengthen the use control measures based on an in-depth analysis of the driving forces of cultivated land changes. For construction land projects, the approval process will be strictly followed, and only projects that meet the occupation conditions will be approved. At the same time, the occupation and compensation balance system will be strictly implemented, and the quality standards of supplementary cultivated land will be further improved to ensure that the total amount of cultivated land does not decrease and the quality does not decrease.

[0180] In terms of resource rational utilization planning, we use the land suitability evaluation method for the planning and utilization of land resources. A complete evaluation system is constructed based on natural factors and socio-economic factors, among which natural factors include soil types, topography and hydrological conditions. Different soil types have different physical and chemical properties, which will affect the land's suitability for planting crops and construction suitability; the ups and downs, slope and other characteristics of topography have a significant restriction or guiding effect on the development and utilization of land; hydrological conditions are related to the irrigation, drainage and convenience of water resources utilization of land. Socio-economic factors include the current land use status and planned use. The current land use status reflects the actual use of the current land and is the basic basis for planning adjustments, while the planned use reflects the future layout and functional positioning of the land. The analytic hierarchy process is used to determine the weights of each evaluation factor. The analytic hierarchy process can decompose complex decision-making problems into multiple levels, determine the relative importance of each factor by comparing them two by two, and then derive a scientific and reasonable weight distribution. Then, GIS spatial analysis technology is used to divide the land suitability level. GIS technology can combine geographic spatial data with attribute data, intuitively display the distribution of land at different suitability levels, and provide accurate land use layout basis for national land use control. In this process, priority will be given to ensuring the improvement of land quality and the stability of land use within the red line of cultivated land and permanent basic farmland protection, strictly restricting the occupation and destruction of non-agricultural construction, and ensuring the continuous development of its agricultural production function through a series of policy measures and technical means. At the same time, actively guide the efficient and intensive use of construction land within the urban development boundary, and encourage the realization of greater economic and social benefits on limited land resources through reasonable planning and layout, increasing building density and volume ratio. For ecological land within the ecological protection red line, strict protection and ecological restoration will be implemented to protect the integrity and stability of its ecosystem. Through ecological restoration projects such as afforestation and wetland restoration, the service function of the ecosystem and the supply capacity of ecological products will be improved, so as to coordinate the allocation and control strategies of land resources for different uses and realize the sustainable use of land resources and the orderly development of national land space.

[0181] There are a series of rigorous implementation steps to determine the expansion limit of urban development boundaries. The first is the assessment of resource and environmental carrying capacity, making full use of GIS technology to assess the capacity of land, water resources and ecological environment. When assessing the land resource capacity, comprehensively consider factors such as land topography, soil quality, and availability, determine the construction suitability level of different regions, and clarify which areas are suitable for high-intensity development and which areas are only suitable for limited development or should be protected. For the assessment of water resource capacity, analyze the total amount of water resources, available amount, water quality, and distribution changes of water resources in different seasons, determine the construction scale that water resources can support, ensure the sustainable supply of water resources during urban development, and ensure the carrying capacity of resources. The second is the prediction of population and industrial development trends, using professional models to predict population growth and industrial development trends. Through the analysis of historical population data, population migration patterns, birth rate and death rate and other factors, predict the quantitative changes of urban population in the future and the dynamic evolution of population distribution. For industrial development trends, study the development status, market demand, technological innovation trends of different industries, and predict the scale expansion or contraction of each industry, so as to provide a scientific basis for the scale of urban development and facility configuration, and ensure that urban construction can meet the needs of population growth and industrial development. Then comes the analysis of facility support capacity, which comprehensively evaluates the support capacity of transportation and public service facilities. In terms of transportation facilities, the density, traffic capacity, and connectivity with external transportation hubs of the road network are considered to determine the configuration standards of transportation facilities in different areas, such as the width and grade of roads, as well as the construction scale requirements, including the mileage of new roads and the scale of parking lots. For public service facilities, the existing service level, coverage, and number of people served of education, medical care, culture and other facilities are analyzed to determine their configuration standards and construction scale requirements in the process of urban expansion, so as to ensure that residents can enjoy convenient and high-quality public services. Finally, comprehensive integration and boundary delineation are carried out to build a multi-objective decision-making model, and factors such as resource and environmental carrying capacity, population and industrial development trends, and facility support capacity are incorporated into the model. Through the comprehensive analysis and calculation of the model, the urban development boundary is delineated. In the delineation process, the plan is continuously optimized, and the balance between the demands of different stakeholders and different planning objectives is fully considered to achieve a balance between the coordinated planning of multiple factors and the comprehensive benefits, so as to ensure that the delineation of urban development boundaries not only meets the long-term needs of urban development, but also effectively protects resources, the environment and the ecosystem.

[0182] The use process of the present invention is: collecting and cleaning national land space data; removing outliers; extracting land use and ecological characteristics; building an optimization combination model; evaluating the quality of the ecological environment; and formulating an ecological protection plan.

[0183] To sum up, the advantages of the present invention are: providing a full-chain solution, using algorithms and models to accurately evaluate ecological environmental quality, identify key areas and key issues, support the formulation of land use control plans and adjust and optimize them according to monitoring data, use resource management methods to achieve optimal resource allocation and efficient utilization, consider the balance of carbon emissions and carbon sinks to promote green and low-carbon development and promote high-quality development.

[0184] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention. These changes and improvements fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the attached claims and their equivalents.

Claims

1. The system of national land space planning and design is characterized by: include: Data management module, feature extraction and analysis module, model building and training module, ecological environment assessment module, cultivated land protection assessment module and land space use control planning module; The data management module is used to collect land space related data from various data sources, including geographic information, ecological environment and socio-economic data, and clean and pre-process the data; The feature extraction and analysis module is used to extract the spatial features of geographic information data, analyze the time series laws of ecological environment, farmland protection and urban development data, and pay attention to the features related to the ecological environment in social and economic data; The model building and training module is used to combine different modules into an overall model, optimize parameters using algorithms, and use preprocessed data for training, record changes in performance indicators during training, and obtain an accurate evaluation model; The ecological environment assessment module is used to calculate the comprehensive evaluation indicators of ecological environment quality using the trained model, analyze the evaluation results, generate a visual report, and identify key areas and key issues for ecological environment protection; The cultivated land protection assessment module is used to calculate and assess the comprehensive evaluation indicators of cultivated land and permanent basic farmland based on the trained model, analyze the assessment results, generate a visual report, and identify key areas and key issues for cultivated land protection; The land use control planning module is used to formulate land use control plans based on the results of ecological environment assessments and farmland protection assessments, combined with various restrictive factors determined by feature extraction and analysis conclusions, including the formulation of ecological protection red lines, permanent basic farmland protection red lines and urban development boundaries, and formulate land use planning schemes based on the generated ecological restoration, farmland protection, urban development suggestions and combined with social and economic development needs, and can be adjusted and optimized.

2. The method of national land space planning and design is characterized by: A system for implementing the national land space planning design as claimed in any one of claim 1, comprising: Collect data related to national land space, clean and pre-process the collected data, and remove outliers; Extract features from the processed data, identify the spatial distribution characteristics of different land use types, capture the trends and laws therein, and extract features that are closely related to the land use control planning; Based on the extracted feature data, a combination model is constructed and the model is optimized using an optimization algorithm; The trained model is used to evaluate the ecological environment of the national land space and obtain the comprehensive evaluation indicators of the ecological environment quality: The trained model is used to evaluate the protection of cultivated land in the national land space, and the comprehensive evaluation index of cultivated land protection quality is obtained: Based on the results of ecological environment assessment and farmland protection assessment, combined with various restrictive factors determined by feature extraction and analysis, a land use control plan is formulated. The land use control plan includes the demarcation of ecological protection red lines, the demarcation of permanent basic farmland protection red lines, the demarcation of urban development boundaries and the use control plan for rational resource utilization.

3. The method for national land space planning and design according to claim 2, characterized in that: Collect data related to national land space, clean and pre-process the collected data, and remove outliers, including: Integrate data from different sources into a unified coordinate system and data format; Based on the integrated data, missing values, duplicate values, and outliers are processed; Among them, missing values ​​are checked by checking the data table and marking the cells containing null values. For raster data, check whether there are invalid values ​​in the pixels. If there are missing values, remove the entire column of data and obtain the data again; Among them, duplicate values ​​are checked according to all key fields using data management software, and completely duplicate records are marked. Duplicate records are directly deleted and only one copy is retained; Among them, outliers are judged based on the mean and standard deviation. If there are outliers, the entire column of data is removed and the data is obtained again; the calculation formula for the mean is: In the formula, μ represents the mean, s represents the total number of data points, and y represents the yth data point. Represents the value of the y-th data point The calculation formula for standard deviation is: In the formula, ρ represents the standard deviation; For raster data, a fixed-size window is used to calculate the mean of the pixel values ​​within the window, and the original value of the pixel in the center of the window is replaced by the mean value. The entire image is traversed to smooth the data.

4. The method for national land space planning and design according to claim 2, characterized in that: Feature extraction of processed data specifically includes: The spatial feature extraction method of geographic information data is: Overlaying vector data of different land use types to analyze the spatial distribution relationship of different types, wherein the vector data includes but is not limited to: cultivated land, forest land, grassland, and urban and rural construction land; The land use pattern index is calculated respectively, and the land use pattern index includes: patch area, patch number, patch density, shape index, and spread index. The land use pattern index reflects the degree of fragmentation, diversity, and connectivity of land use; The terrain features were extracted using digital elevation model data, and the terrain analysis tools of ArcGIS software were used to process the digital elevation model data to obtain slope and aspect information; The method for analyzing the time series regularity of ecological environment data and cultivated land protection number r is as follows: Collect long-term series of ecological environment and cultivated land data, and use linear regression analysis to analyze the changing trends and periodic characteristics of ecological environment and cultivated land protection indicators over time; The method for extracting features related to ecological environment, farmland protection, and urban construction in social and economic data is as follows: Associate population data with geographic spatial data, calculate population density in different regions, and analyze the relationship between population density and ecological environment, farmland protection, and urban construction indicators; Collect economic data and analyze the relationship between economic development indicators and ecological environment, farmland protection and urban construction indicators through correlation coefficients. The economic data include: regional gross domestic product and industrial structure.

5. The method for national land space planning and design according to claim 2, characterized in that: Based on the extracted feature data, building a combination model specifically includes: The pre-processed land space related data are used as the input of the model, wherein the land space related data include: geographic information data, ecological environment data, cultivated land protection data, urban construction data and social and economic data; CNN layer: Use multiple convolution kernels to perform sliding convolution on the input data to extract spatial features. For two-dimensional input data, the convolution operation is expressed as: In the formula, C b，g represents the output value after convolution, L represents the input data, Q represents the convolution kernel, h and n represent the size of the convolution kernel; The pooling operation uses maximum pooling to reduce the data dimension and extract the main features; LSTM layer: The LSTM unit consists of a forget gate, an input gate, and an output gate; Among them, the forget gate expression is: f t =σ(W f ·[h t-1 ,x t ]+b f ) In the formula, f t represents the output of the forget gate, σ represents the sigmoid function, W f represents the weight matrix of the forget gate, h t-1 represents the hidden state at the previous moment, x t represents the input at the current moment, b f Represents the bias term of the forget gate; Among them, the input gate expression is: i t =σ(W i ·[h t-1 ,x t ]+b i ) In the formula, i t represents the output of the input gate, W i represents the weight matrix of the input gate, b i represents the bias term of the input gate; Among them, the candidate unit state expression is: C' t =tanh(W C ·[h t-1 ,x t ]+b C ) In the formula, C′ t represents the candidate unit state, W C The weight matrix representing the candidate unit state, b C represents the bias term; Among them, the update unit state expression is: C t =f t ⊙c t-1 +i t ⊙C′ t ) In the formula, C t represents the cell state at the current moment, ⊙ represents element-by-element multiplication; Among them, the output gate expression is: the t =σ(W o ·[h t-1 ,x t ]+b o ) In the formula, o t represents the output of the output gate, W o represents the weight matrix of the output gate, b o Represents the bias term of the output gate; Among them, the hidden state expression is: ht=ot⊙tanh(Ct) The attention weight is calculated based on the input data and the query vector, where the attention calculation formula is: In the formula, θ represents the query vector, κ ι represents the key vector in the input data, d κ represents the dimension of the key vector, ∈ ι represents the attention score; The attention weight is weighted and summed with the input data to get the attention output, where the calculation formula of the attention output is: In the formula, S ι Represents the value vector in the input data, softmax(∈ ι ) represents the normalized attention weight; According to the task requirements, activation functions and output units are selected to construct a combined model architecture suitable for ecological environmental assessment, farmland protection assessment and land use control planning.

6. The method for national land space planning and design according to claim 2, characterized in that: Optimizing the model using the optimization algorithm specifically includes: A set of particles is randomly generated, each particle represents a set of parameters of the model, and each particle has a position vector and a time vector. The position vector represents the value of the model parameter, and the time vector determines the time direction and speed of the particle in the search space; For each particle, its position vector is used as the parameter of the model, the model is trained using the training data, and the fitness value of the model is calculated; Compare the current fitness value of each particle with the individual's historical optimal fitness value. If the current fitness value is better, update the individual's optimal position. Compare the current fitness value of each particle with the group's historical optimal fitness value. If the current fitness value is better, update the group's optimal position: Among them, the calculation method for updating the time point and position of the particle is: In the formula, and denote the time point and position of particle i at the tth iteration, ω denotes the inertia weight, ε1 and ε2 denote the learning factors, ζ1 and ζ2 denote random numbers in the range of [0, 1], and pbest i represents the individual optimal position of particle i, and gbest represents the group optimal position; Repeat the above steps until the stop condition is met; The optimal position of the group is used as the optimal parameter of the model for the final model training and prediction, ensuring the accuracy and efficiency of the model in ecological environmental assessment, farmland protection assessment and land use control planning tasks.

7. The method for national land space planning and design according to claim 2, characterized in that: In the assessment of cultivated land protection, in response to the decline in cultivated land quality and reduction in area, the protection strategy is determined based on the cultivated land quality evaluation model and the cultivated land change driving force model. The cultivated land quality evaluation model is constructed using a multi-factor comprehensive evaluation method, and the cultivated land change driving force model uses a Logistic regression model to identify key driving factors. Among them, the expression of the cultivated land quality evaluation model constructed by the multi-factor comprehensive evaluation method is: In the formula, A represents the comprehensive score of cultivated land quality, L z Represents the weight of the zth evaluation factor, which is determined by the expert scoring method and satisfies U z represents the standardized value of the zth evaluation factor; Among them, the expression of key driving factors identified by the Logistic regression model is: In the formula, V represents the probability of cultivated land change, e represents the natural constant, and s represents the linear combination of independent variables; In areas where arable land quality has degraded, the obstacle factors are accurately located based on the evaluation results. If the soil fertility is insufficient, soil improvement projects will be implemented; if the irrigation facilities are imperfect, the irrigation system will be optimized. For areas where arable land is threatened by construction occupation, the use control will be strengthened based on driving force analysis, and construction land projects will be strictly approved. For those that meet the occupation conditions, the balance system of occupation and compensation will be implemented and the quality standards of supplementary arable land will be improved.

8. The method for national land space planning and design according to claim 2, characterized in that: In terms of resource rational utilization planning, for land resources, land suitability evaluation methods are used; An evaluation system is constructed based on natural factors and socio-economic factors. The natural factors include soil type, topography and hydrological conditions, and the socio-economic factors include land use status and planned use: The analytic hierarchy process was used to determine the weight of each evaluation factor, and the GIS spatial analysis technology was used to divide the land suitability level; Provide a basis for land use layout for national land use control, give priority to improving the quality of land and the stability of land use within the protection red line of cultivated land and permanent basic farmland, guide the efficient and intensive use of construction land within the urban development boundary, strictly protect and restore ecological land within the ecological protection red line, and coordinate the allocation of land resources and control strategies for different uses.

9. The method for national land space planning and design according to claim 2, characterized in that: Determine the extension limit of the urban development boundary, and the implementation steps are as follows: Resource and environmental carrying capacity assessment: using GIS technology to assess the land, water resources and ecological environment capacity, determine the construction suitability level and water resources support scale, and ensure resource carrying capacity; Prediction of population and industrial development trends: using models to predict population growth and industrial development trends, providing a basis for urban development scale and facility configuration; Facility support capacity analysis, assessing the support capacity of transportation and public service facilities, and determining facility configuration standards and construction scale requirements; Comprehensive integration and boundary demarcation, build a multi-objective decision-making model, delineate urban development boundaries, optimize plans, and achieve multi-factor coordinated planning and comprehensive benefit balance.

Images