A method and system for constructing an early warning model for land characteristics affecting supply and demand balance

Abstract

The present invention relates to the technical field of analysis and early warning models, and in particular to a method and system for constructing an early warning model in which land characteristics affect the balance between supply and demand. The method comprises the following steps: collecting all-weather environmental data of the land area to be analyzed and performing terrain structure analysis to obtain regional land terrain structure data; performing slope soil attribute analysis based on the regional land terrain structure data, and then constructing a three-dimensional land structure attribute model to obtain a three-dimensional land structure attribute model; performing dynamic soil fertility accumulation path deduction based on the three-dimensional land structure attribute model, and performing fertility trait stratification analysis to obtain regional fertility trait stratification data; using a land trait analysis and early warning model to obtain land characteristics, and then regulating land supply and demand according to the land characteristics. The present invention makes the analysis and early warning model technology more perfect by optimizing the analysis and early warning model technology.

Description

Technical Field

[0001] The present invention relates to the technical field of analytical early warning models, and in particular to a method and system for constructing an early warning model of land characteristics affecting supply and demand balance. Background Art

[0002] In the process of land property analysis, multiple influencing factors must be considered, such as climate change, soil management practices, topography, etc. The interaction of these factors often leads to complex changes in land properties, making it impossible for the analysis of a single variable to fully reflect the true condition of the land. Therefore, the construction of a comprehensive early warning model can integrate multiple data sources and use methods such as multivariate regression analysis and machine learning to conduct multi-dimensional evaluation and prediction of land properties. The results of this comprehensive analysis can provide scientific guidance for the optimal allocation of land resources and help achieve the sustainable development of land resources. In addition, the construction of an early warning model for the impact of land characteristics on the balance between supply and demand needs to be combined with local realities and consider the balance between regional economic and social development and the ecological environment. For example, in farmland soil management, targeted soil improvement measures should be designed in combination with the growth requirements of different crops to ensure the sustainability of agricultural production. At the same time, local governments should introduce model-based decision support systems in land management policies, adjust policies in a timely manner through regular monitoring and analysis, optimize land use structure, and reduce waste and degradation of land resources. However, the traditional method of constructing an early warning model for the impact of land characteristics on the balance between supply and demand has the problem of inaccurate analysis of land properties, resulting in low accuracy of the early warning model, and thus the early warning model cannot be better utilized to obtain land characteristics to adjust the balance between supply and demand. Summary of the invention

[0003] Based on this, it is necessary to provide a method and system for constructing an early warning model of the impact of land characteristics on supply and demand balance to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a method for constructing an early warning model of land characteristics affecting supply and demand balance is provided, the method comprising the following steps:

[0005] Step S1: Acquire the land area to be analyzed; collect all-weather environmental data of the land area to be analyzed by ground sensors to obtain all-weather environmental data of the land analysis area; perform terrain structure analysis on the land area to be analyzed to obtain regional land terrain structure data; perform slope soil attribute analysis based on the regional land terrain structure data to obtain slope soil attribute data; construct a three-dimensional land structure attribute model based on the slope soil attribute data to obtain a three-dimensional land structure attribute model;

[0006] Step S2: Performing an initial property analysis on the land area to be analyzed to obtain initial property data of the land area; performing dynamic soil fertility accumulation path deduction based on the all-weather environmental data of the land analysis area and the three-dimensional land structure attribute model to obtain soil fertility accumulation path deduction data; performing fertility stratification analysis on the initial property data of the land area according to the soil fertility accumulation path deduction data to obtain regional fertility stratification data;

[0007] Step S3: Based on the Q-learning algorithm, a land property analysis and early warning model is constructed for the property boundary feature data to obtain the land property analysis and early warning model, and the land characteristics are obtained by using the land property analysis and early warning model, and the land supply and demand are further adjusted according to the land characteristics.

[0008] The present invention collects all-weather environmental data of the land area to be analyzed through ground sensors, and can systematically collect environmental variables such as climate, humidity, temperature, etc. in the area, providing a solid data foundation for subsequent analysis. Next, terrain structure analysis is performed to gain a deeper understanding of the terrain characteristics in the area, so that key elements such as slope and slope direction can be identified. By analyzing the soil properties of the slope, the physical and chemical properties of the soil can be obtained, laying the foundation for building a three-dimensional land structure attribute model, ensuring that the model can accurately reflect the actual situation of the land. The comprehensive effect of this step significantly improves the understanding of the land environment and its soil characteristics, and provides important support for subsequent fertility analysis. First, the initial characteristics of the land area are analyzed for the land area to be analyzed, and the basic characteristics of the soil in the area, such as pH value, organic matter content, etc., are obtained. These data are crucial to understanding the fertility status of the soil. Then, based on the all-weather environmental data and the three-dimensional land structure attribute model, the dynamic soil fertility accumulation path is deduced, which helps to reveal the changing laws of soil fertility in time and space. This process can not only effectively predict the changing trend of fertility, but also analyze the regional fertility characteristics through the analysis of the deduction data, so as to more clearly understand the fertility characteristics of different soil layers and ensure the pertinence and effectiveness of agricultural management measures. The use of Q-learning algorithm to construct an early warning model for the influence of land characteristics on supply and demand balance is an advanced method of using reinforcement learning, which can predict future changes in land characteristics by learning historical data. The model can not only monitor the properties and changes of soil in real time, but also identify potential risks and problems through intelligent algorithms, and provide early warning information for land managers. By establishing such an early warning model, the scientificity and accuracy of land management can be effectively improved, the agricultural production risks caused by changes in soil characteristics can be reduced, and sustainable land use and management can be promoted. Therefore, the present invention is an optimization treatment of a traditional method for constructing an early warning model for the influence of land characteristics on supply and demand balance, which solves the problem that the traditional method for constructing an early warning model for the influence of land characteristics on supply and demand balance has inaccurate analysis of land characteristics, thereby causing low accuracy of the early warning model, improves the accuracy of land characteristic analysis, and enhances the early warning ability of the early warning model.

[0009] Preferably, step S1 comprises the following steps:

[0010] Step S11: obtaining the land area to be analyzed;

[0011] Step S12: collecting all-weather environmental data of the land area to be analyzed through ground sensors to obtain all-weather environmental data of the land analysis area;

[0012] Step S13: Performing terrain structure analysis on the land area to be analyzed to obtain regional land terrain structure data;

[0013] Step S14: performing slope soil property analysis on the land area to be analyzed based on the regional land topography data to obtain slope soil property data;

[0014] Step S15: constructing a three-dimensional land structure attribute model according to the regional land topography data and the slope soil attribute data to obtain a three-dimensional land structure attribute model.

[0015] The present invention obtains the land area to be analyzed, which is the starting point of the entire analysis process. By clarifying the target area, it is ensured that the subsequent data collection and analysis can be concentrated in a specific geographical location, which is crucial to the characteristics of the study area. This step involves defining the boundaries, attributes and environmental factors that affect soil and plant growth in the area, laying the foundation for subsequent research. This clear definition helps to organize resources efficiently and ensure the accuracy and effectiveness of data collection. The ground sensor is used to collect all-weather environmental data for the area to be analyzed. The beneficial effect of this step is that all-weather data collection can comprehensively record the climatic conditions of the area, including temperature, humidity, precipitation and other factors. These environmental variables have a direct impact on soil properties and plant growth. Therefore, through systematic data collection, an important information basis is provided for understanding the ecological dynamics in the region. These data can not only reflect the current environmental conditions, but also provide support for long-term trend analysis. The terrain structure analysis is performed on the land area to be analyzed to obtain the terrain feature data of the area, including slope, aspect and elevation changes. The importance of this step is that terrain is a key factor affecting soil moisture, nutrient distribution and plant growth. Through in-depth analysis of the terrain structure, the potential impact of different terrains on soil properties and plant growth can be identified. These data provide the necessary geographical background for subsequent soil property analysis and model construction, making the analysis more comprehensive and scientific. Slope soil property analysis is performed based on regional land topography data to obtain slope soil property data. The beneficial effect of this step is that by analyzing the physical, chemical and biological properties of slope soil, the soil fertility and its relationship with the terrain can be revealed. Slope soil is affected by factors such as water flow and wind, and often shows different properties and characteristics. Therefore, through this analysis, we can have a deeper understanding of the law of soil change and provide a scientific basis for the rational use of land. A three-dimensional land structure attribute model is constructed based on regional land topography data and slope soil attribute data to obtain a three-dimensional land structure attribute model. The construction of this model can integrate the previously collected environmental, topographic and soil data to form a comprehensive land property visualization tool. Through the three-dimensional model, researchers can more intuitively analyze the interaction between different factors and predict soil fertility changes and their impact on plant growth. This comprehensive model not only helps scientific research, but also provides decision support for agricultural management and land planning.

[0016] Preferably, step S2 comprises the following steps:

[0017] Step S21: performing an initial land property analysis on the land area to be analyzed based on the regional land topography data and the slope soil property data to obtain initial land property data;

[0018] Step S22: based on the all-weather environmental data of the land analysis area and the three-dimensional land structure attribute model, a dynamic soil fertility accumulation path is deduced for the regional land terrain structure data and the slope soil attribute data to obtain soil fertility accumulation path deduction data;

[0019] Step S23: performing a stacking intensity spatial heterogeneity analysis on the soil fertility stacking path deduction data to obtain stacking path intensity spatial heterogeneity data;

[0020] Step S24: performing fertility trait hierarchical analysis on the initial trait data of the land area according to the spatial heterogeneous data of the accumulation path intensity, and obtaining regional fertility trait hierarchical data.

[0021] The present invention generates initial land area property data by analyzing the initial land area property of regional land terrain structure data and slope soil attribute data. The beneficial effect of this step is that it lays a solid foundation for subsequent analysis and can fully understand the physical and chemical properties of the soil. These initial property data not only provide important parameters for the construction of the model, but also reveal the basic conditions of the land, which helps to identify potential risks and opportunities, so as to better manage and plan land resources. Using all-weather environmental data and a three-dimensional land structure attribute model, dynamic soil fertility accumulation path deduction is performed on regional land terrain structure and slope soil attribute data. The importance of this step is that it can simulate the dynamic evolution process of soil properties with environmental changes. This dynamic deduction provides time-effective data for establishing an early warning model, helps identify changes in land properties under different climatic conditions, and then predicts potential environmental risks, ensuring the foresight and adaptability of land management strategies. The spatial heterogeneity of accumulation intensity is analyzed for the soil fertility accumulation path deduction data to generate spatial heterogeneous data of accumulation path intensity. The beneficial effect of this step is that it reveals the spatial distribution characteristics of soil properties and can identify differences in soil properties and potential risk areas in different regions. By analyzing these spatial heterogeneities, the early warning model can more effectively monitor and evaluate the impact of changes in land properties, provide a scientific basis for relevant decision-making, and improve the efficiency and flexibility of land management. Based on the spatial heterogeneous data of accumulation path intensity, the initial property data of the land area is analyzed by stratification of fertility traits to obtain regional fertility stratification data. The beneficial effect of this step is that it reveals the complexity of soil properties through stratified analysis, allowing the model to more accurately identify and predict soil changes at different levels. This detailed stratified analysis provides more in-depth information for the early warning model, which can help land managers formulate targeted response measures and improve the sustainable management capabilities of land resources.

[0022] Preferably, step S22 includes the following steps:

[0023] Step S221: extracting rainfall and temperature changes from the all-weather environmental data of the land analysis area to obtain regional rainfall data and regional temperature change data respectively;

[0024] Step S222: evaluating the water penetration efficiency of the slope soil property data according to the regional rainfall data and the regional temperature change data, and obtaining the slope soil water penetration efficiency data;

[0025] Step S223: performing soil component loss path analysis on the regional land topography data based on the slope soil water penetration efficiency data to obtain soil component loss path data;

[0026] Step S224: simulating the movement of nutrient soil layers on the slope soil water penetration efficiency data and the soil component loss path data according to the three-dimensional land structure attribute model to obtain nutrient soil layer movement simulation data;

[0027] Step S225: performing nutrient accumulation and convergence regional distribution analysis on the nutrient soil layer movement simulation data to obtain nutrient accumulation and convergence regional data;

[0028] Step S226: Dynamic soil fertility accumulation path deduction is performed based on the nutrient accumulation convergence area data to obtain soil fertility accumulation path deduction data.

[0029] The present invention extracts rainfall and temperature changes from the all-weather environmental data of the land analysis area to obtain regional rainfall data and regional temperature change data respectively. This process provides important basic climate data for subsequent analysis, which helps to understand how the climate conditions in the region affect soil properties. Accurate rainfall and temperature data are the core elements of building an early warning model, which helps to analyze the impact of potential environmental changes on soil dynamics, thereby laying the foundation for formulating effective land management strategies. Combined with regional rainfall data and temperature change data, the slope soil attribute data is evaluated for water infiltration efficiency to obtain slope soil water infiltration efficiency data. The importance of this step is that it reveals the soil's ability to absorb and retain water, and provides a key indicator for understanding the soil's response under different climatic conditions. By effectively evaluating the water infiltration efficiency, the early warning model can predict the water stress faced by the soil under extreme weather, thereby guiding the formulation of response measures. Based on the slope soil water infiltration efficiency data, the soil component loss path analysis of the regional land terrain structure data is carried out to obtain soil component loss path data. This analysis reveals the potential risk of soil component loss under different terrain conditions and helps identify areas prone to soil erosion. By determining the loss path, the early warning model can implement risk monitoring for specific areas, take protective measures in advance, and reduce the impact of soil component loss on land resources. According to the three-dimensional land structure attribute model, the nutrient soil layer movement simulation is performed on the slope soil water infiltration efficiency data and the soil component loss path data, so as to obtain the nutrient soil layer movement simulation data. The beneficial effect of this step is that it can dynamically simulate the movement of nutrients in the soil layer and its distribution law, providing timely and accurate data support for soil property analysis. This simulation provides important information for the early warning model, helping to identify the impact of nutrient changes on soil health, thereby providing a basis for land management. By analyzing the nutrient accumulation and convergence area distribution of the nutrient soil layer movement simulation data, nutrient accumulation and convergence area data are obtained. This analysis can reveal the aggregation and distribution characteristics of nutrients in different regions and help understand the performance of soil under different environmental conditions. By analyzing the nutrient accumulation area, the early warning model can identify potential environmental risks and change trends, provide a basis for land management decisions, and help formulate more scientific monitoring and response strategies. Based on the nutrient accumulation and convergence area data, dynamic soil fertility accumulation path deduction is performed to obtain soil fertility accumulation path deduction data. The importance of this step is that it can track the changes in soil fertility under different time and space conditions, providing support for the dynamic nature of the model. By deducing the fertility accumulation path, the early warning model can identify soil degradation trends and regional risks in advance, thereby providing a scientific basis for effective land management and decision-making, and ensuring the sustainable use of land resources.

[0030] Preferably, step S225 includes the following steps:

[0031] Conduct movement dynamics attenuation analysis on the simulated data of nutrient soil layer movement to obtain nutrient movement dynamics attenuation data;

[0032] According to the nutrient movement dynamic attenuation data, the regional agglomeration influencing factors of the nutrient soil layer movement simulation data are identified to obtain the regional agglomeration influencing factors;

[0033] According to the nutrient movement dynamic attenuation data and regional agglomeration influencing factors, the nutrient soil layer movement simulation data is processed by cumulative continuous interpolation to obtain the nutrient soil layer cumulative continuous interpolation data;

[0034] The nutrient accumulation and convergence regional distribution analysis was performed on the nutrient soil layer accumulation continuous interpolation data to obtain the nutrient accumulation and convergence regional data.

[0035] The present invention performs a motion dynamic decay analysis on the nutrient soil layer movement simulation data, thereby obtaining the nutrient movement dynamic decay data. This analysis can reveal the movement speed and dynamic change trend of nutrients in the soil layer, and help identify the stability and mobility of nutrients under different environmental conditions. By understanding the dynamic decay of nutrients, the early warning model can more accurately assess potential soil health problems and identify factors that lead to the decline of soil layer functions, thereby providing a scientific basis for subsequent land management strategies. According to the obtained nutrient movement dynamic decay data, the regional agglomeration influencing factors of the nutrient soil layer movement simulation data are identified. The importance of this process lies in that the key factors that have a significant impact on nutrient aggregation and distribution can be identified from multiple influencing factors. The identification of these influencing factors enables the early warning model to focus on the main driving factors affecting soil layer health, thereby improving the sensitivity and responsiveness to land resource changes, and providing guidance for further analysis and decision-making. Combined with the nutrient movement dynamic decay data and the regional agglomeration influencing factors, the nutrient soil layer movement simulation data is subjected to cumulative continuous interpolation processing to obtain nutrient soil layer cumulative continuous interpolation data. This technology can fill the gaps in spatial data and generate more accurate nutrient distribution maps, thereby better reflecting the spatial variability of nutrients in the soil layer. Through continuous interpolation, the early warning model can capture subtle changes in nutrient distribution and provide more detailed and reliable basic data for land management. By analyzing the nutrient accumulation and convergence area distribution of the nutrient soil layer continuous interpolation data, the nutrient accumulation and convergence area data is obtained. The key to this step is that by analyzing the convergence of nutrients in the soil layer, the existing environmental risk areas can be identified. Mastering the distribution characteristics of nutrient convergence enables the early warning model to detect potential soil layer problems in a timely manner, providing a basis for formulating targeted management measures to ensure the sustainable use and healthy status of land resources.

[0036] Preferably, step S24 comprises the following steps:

[0037] Step S241: performing spatial intensity numerical simulation on the stacking path intensity spatial heterogeneous data to obtain stacking path intensity numerical data;

[0038] Step S242: performing spatial autocorrelation feature analysis on the stacking path strength numerical data to obtain stacking strength spatial autocorrelation data;

[0039] Step S243: performing multivariate regression analysis on the stacking path strength numerical data according to the stacking strength spatial autocorrelation data to obtain stacking strength multivariate regression data;

[0040] Step S244: Based on the accumulation intensity multivariate regression data and the accumulation intensity spatial autocorrelation data, the initial property data of the land area are subjected to stratified analysis of fertility traits to obtain regional fertility stratified data.

[0041] The present invention performs spatial intensity numerical simulation on the spatial heterogeneous data of the accumulation path intensity, thereby obtaining the accumulation path intensity numerical data. This simulation can effectively reveal the intensity changes of the accumulation paths in different plots, and provide important quantitative data support for subsequent analysis. By modeling spatial heterogeneity, the early warning model can identify the distribution characteristics of the accumulation intensity, lay the foundation for the in-depth analysis of land properties, and ensure the accuracy of land management and monitoring. The obtained accumulation path intensity numerical data is subjected to spatial autocorrelation feature analysis, and then the accumulation intensity spatial autocorrelation data is obtained. This analysis process reveals the spatial correlation of accumulation intensity and helps understand the mutual influence relationship between accumulation intensity in different regions. By identifying spatial autocorrelation features, the early warning model can better evaluate the regional changes in land properties, thereby providing a scientific basis for formulating targeted management strategies and effectively reducing the potential risks of land resources. Based on the accumulation intensity spatial autocorrelation data, the accumulation path intensity numerical data is subjected to multivariate regression analysis, and finally the accumulation intensity multivariate regression data is obtained. This analysis can reveal the relationship between multiple factors affecting accumulation intensity and identify key driving variables. Through multivariate regression, the early warning model can more accurately predict the changing trend of accumulation intensity, enhance the sensitivity to the dynamics of land properties, provide more reliable data support for decision-making, and thus optimize land management strategies. Combining the multivariate regression data of accumulation intensity and the spatial autocorrelation data of accumulation intensity, the initial property data of the land area is analyzed by stratification of fertility traits to obtain regional fertility stratification data. This stratification analysis can reveal the characteristics and fertility status of different soil layers in the region and help identify potential risk areas. By deeply analyzing the property characteristics of the land area, the early warning model can develop more accurate monitoring and management plans to ensure the effective use and sustainable development of land resources.

[0042] Preferably, step S3 comprises the following steps:

[0043] Step S31: extracting trait boundary features from the regional fertility trait stratification data to obtain trait boundary feature data;

[0044] Step S32: performing fertility loss area association on the soil fertility accumulation path deduction data according to the trait boundary characteristic data to obtain fertility loss area association data;

[0045] Step S33: Based on the Q-learning algorithm, a land property analysis and early warning model is constructed for the fertility loss area correlation data and the property boundary feature data to obtain the land property analysis and early warning model, and the land characteristics are obtained using the land property analysis and early warning model, and the land supply and demand are further adjusted according to the land characteristics.

[0046] The present invention extracts the boundary characteristics of the traits for the stratified data of regional fertility traits, thereby obtaining the boundary characteristic data of the traits. This process helps to identify the significant trait changes between different plots and reveals the boundary conditions of soil characteristics at each level. By clarifying the boundary characteristics of the traits, the early warning model can better understand the spatial distribution and heterogeneity of the soil, provide a clear basis for subsequent analysis, and ensure the scientificity and accuracy of the subsequent steps. Based on the extracted boundary characteristic data of the traits, the fertility loss area is analyzed for the association of the soil fertility accumulation path deduction data, thereby obtaining the fertility loss area association data. This analysis not only helps to identify potential fertility loss areas, but also reveals its relationship with specific trait boundaries. Through this association analysis, the early warning model can identify key areas, discover the decline trend of soil functions in a timely manner, and provide data support for the formulation of effective management strategies to reduce future environmental risks. Based on the Q-learning algorithm, the fertility loss area association data and the trait boundary characteristic data are combined to construct a land trait analysis early warning model. This process takes advantage of reinforcement learning, so that the model can self-optimize and improve its responsiveness to changes in land traits. Through dynamic learning and adjustment, the early warning model can effectively predict potential land problems, enhance sensitivity to land changes under different environmental conditions, provide a scientific basis for decision-making, and promote the sustainable management and utilization of land resources.

[0047] Preferably, step S33 includes the following steps:

[0048] Step S331: performing feature importance evaluation on fertility loss area correlation data to obtain loss feature importance evaluation data;

[0049] Step S332: performing boundary characteristic loss coupling according to the loss characteristic importance evaluation data and the characteristic boundary characteristic data to obtain boundary characteristic loss coupling data;

[0050] Step S333: dividing the boundary characteristic loss coupling data into a test set and a training set to obtain a boundary characteristic loss test set and a boundary characteristic loss training set respectively;

[0051] Step S334: performing sampling balancing processing on the boundary characteristic loss training set to obtain a boundary characteristic loss sampling balanced training set;

[0052] Step S335: constructing an initial land property analysis and early warning model for the boundary property loss sampling balance training set based on the Q-learning algorithm to obtain an initial land property analysis and early warning model;

[0053] Step S336: Test and verify the initial land property analysis and early warning model according to the boundary property loss test set to obtain the land property analysis and early warning model, use the land property analysis and early warning model to obtain land characteristics, and then regulate land supply and demand according to the land characteristics.

[0054] The present invention evaluates the importance of features in fertility loss area association data, aiming to determine the role of each feature in fertility loss. By analyzing the contribution of different features, the factors that have the greatest impact on land properties can be identified. This evaluation provides a basis for subsequent coupling analysis, allowing the model to focus on the most critical features, improve the efficiency and accuracy of the analysis, and thus lay a solid foundation for building a more accurate early warning model. Boundary trait loss is coupled based on the loss feature importance assessment data and the trait boundary feature data. The goal of this process is to combine key features with boundary features to identify trait loss patterns. Through this coupling, the model can better understand the relationship between different features and then reveal potential risks under boundary conditions. Such coupling analysis provides richer and more accurate data support for subsequent model training and improves the predictive ability of the early warning system. The boundary trait loss coupling data is divided into a test set and a training set, thereby obtaining a boundary trait loss test set and a boundary trait loss training set respectively. This division is a key step in establishing an effective model. By dividing the data into two parts, training and testing, it is ensured that the model can be verified on unseen data. Such a structure not only improves the generalization ability of the model, but also ensures the scientific nature of the evaluation process, providing a reliable guarantee for the accuracy of the final model. The sampling balance processing of the boundary trait loss training set aims to solve the problem of class imbalance and ensure that the model can fairly learn the characteristics of each category during training. This processing adjusts the sample ratio so that the model can get equal attention when facing different categories. By balancing the training set, the early warning model can better capture the characteristics of rare events, improve its sensitivity and accuracy in practical applications, and lay a solid foundation for subsequent model training. Based on the Q-learning algorithm, the initial land characteristics affect the supply and demand balance of the early warning model constructed for the boundary trait loss sampling balance training set. This process uses the framework of reinforcement learning to enable the model to optimize its decision-making process through continuous trial and error learning. Through the Q-learning algorithm, the model can adaptively adjust its strategy to cope with different changes in land traits. This intelligent learning mechanism enables the early warning model to more accurately predict potential risks and improve the efficiency of land management. The initial land trait analysis early warning model is tested and verified using the boundary trait loss test set to evaluate the accuracy and effectiveness of the model. By verifying the test set, the performance of the model can be fully evaluated and its deficiencies in practical applications can be identified. This process not only helps to confirm the practical applicability of the model, but also provides feedback for further optimization of the model, thus ensuring that the final early warning model has high predictive ability and reliability.

[0055] Preferably, step S334 includes the following steps:

[0056] Perform feature clustering on the boundary characteristic loss training set to obtain the boundary characteristic loss clustering training set;

[0057] Under-sampling the boundary trait loss training set based on the boundary trait loss clustering training set to obtain an under-sampled boundary trait loss training set;

[0058] Perform stratified sampling on the under-sampled boundary trait loss training set to obtain a boundary trait stratified sampling training set;

[0059] The boundary trait loss training set is sampled and balanced according to the boundary trait stratified sampling training set and the under-sampling boundary trait loss training set to obtain the boundary trait loss sampling balanced training set.

[0060] When the present invention performs feature clustering on the boundary trait loss training set, the purpose is to classify similar features to identify data points with similar loss patterns. Through feature clustering, the complexity of the data can be reduced and representative feature combinations can be extracted. This method not only helps the model to better understand the intrinsic relationship between the data, but also provides a more accurate basis for subsequent undersampling and sampling processing, and improves the effect of subsequent analysis. Undersampling processing based on the boundary trait loss clustering training set is intended to reduce certain overly dominant categories in the data set in order to balance the category distribution of the training set. By randomly extracting samples from the dominant category, the bias of the model against these categories during training can be reduced, so that the model can learn the characteristics of each category fairly. This processing step helps to improve the model's prediction ability for minority classes and ensure the comprehensiveness and effectiveness of the early warning system. Stratified sampling processing is performed on the undersampled boundary trait loss training set to ensure that the proportion of each category in the training set is reasonable. By independently sampling in each category, model failure caused by sampling bias can be effectively avoided. This method ensures that the model can represent the characteristics of each category during the learning process, so that the final model can more comprehensively consider various situations when predicting, and improve the generalization ability of the model. The boundary trait loss training set is sampled and balanced according to the boundary trait stratified sampling training set and the under-sampled boundary trait loss training set, and finally the boundary trait loss sampling balanced training set is obtained. This process combines the two sampling methods to ensure that the model can obtain a balanced sample distribution during training. This not only improves the model's ability to recognize various features, but also enhances the accuracy of predicting potential risks, making the early warning model more reliable and effective in practical applications.

[0061] Preferably, the present invention further provides a system for constructing an early warning model of land characteristics affecting supply and demand balance, which is used to execute the method for constructing an early warning model of land characteristics affecting supply and demand balance as described above. The system for constructing an early warning model of land characteristics affecting supply and demand balance comprises:

[0062] The land structure attribute analysis module is used to obtain the land area to be analyzed; collect all-weather environmental data of the land area to be analyzed through ground sensors to obtain all-weather environmental data of the land analysis area; perform terrain structure analysis on the land area to be analyzed to obtain regional land terrain structure data; perform slope soil attribute analysis based on the regional land terrain structure data to obtain slope soil attribute data; construct a three-dimensional land structure attribute model based on the slope soil attribute data to obtain a three-dimensional land structure attribute model;

[0063] The fertility trait hierarchical analysis module is used to perform initial trait analysis on the land area to be analyzed to obtain initial trait data of the land area; to perform dynamic soil fertility accumulation path deduction based on the all-weather environmental data of the land analysis area and the three-dimensional land structure attribute model to obtain soil fertility accumulation path deduction data; to perform fertility trait hierarchical analysis on the initial trait data of the land area according to the soil fertility accumulation path deduction data to obtain regional fertility trait hierarchical data;

[0064] The analysis and early warning model construction module is used to construct a land property analysis and early warning model for property boundary feature data based on the Q-learning algorithm, obtain a land property analysis and early warning model, use the land property analysis and early warning model to obtain land characteristics, and then regulate land supply and demand based on the land characteristics.

[0065] The beneficial effect of the present invention is that the ground sensor collects all-weather environmental data of the land area to be analyzed, and can systematically collect environmental variables such as climate, humidity, temperature, etc. in the area, providing a solid data foundation for subsequent analysis. Next, the terrain structure analysis is carried out, so that the terrain characteristics in the area can be more deeply understood, so that key elements such as slope and slope direction can be identified. By analyzing the soil properties of the slope, the physical and chemical properties of the soil can be obtained, which lays the foundation for building a three-dimensional land structure attribute model and ensures that the model can accurately reflect the actual situation of the land. The comprehensive effect of this step significantly improves the understanding of the land environment and its soil characteristics, and provides important support for subsequent fertility analysis. First, the initial characteristics of the land area are analyzed for the land area to be analyzed, and the basic characteristics of the soil in the area, such as pH value, organic matter content, etc., are obtained. These data are crucial to understanding the fertility of the soil. Then, based on the all-weather environmental data and the three-dimensional land structure attribute model, the dynamic soil fertility accumulation path is deduced, which helps to reveal the changing laws of soil fertility in time and space. This process can not only effectively predict the changing trend of fertility, but also analyze the regional fertility characteristics through the analysis of the deduction data, so as to more clearly understand the fertility characteristics of different soil layers and ensure the pertinence and effectiveness of agricultural management measures. The use of Q-learning algorithm to construct an early warning model for the influence of land characteristics on supply and demand balance is an advanced method of using reinforcement learning, which can predict future changes in land characteristics by learning historical data. The model can not only monitor the properties and changes of soil in real time, but also identify potential risks and problems through intelligent algorithms, and provide early warning information for land managers. By establishing such an early warning model, the scientificity and accuracy of land management can be effectively improved, the agricultural production risks caused by changes in soil characteristics can be reduced, and sustainable land use and management can be promoted. Therefore, the present invention is an optimization treatment of a traditional method for constructing an early warning model for the influence of land characteristics on supply and demand balance, which solves the problem that the traditional method for constructing an early warning model for the influence of land characteristics on supply and demand balance has inaccurate analysis of land characteristics, thereby causing low accuracy of the early warning model, improves the accuracy of land characteristic analysis, and enhances the early warning ability of the early warning model. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 A schematic diagram of the steps of a method for constructing an early warning model for the impact of land characteristics on the balance between supply and demand;

[0067] Figure 2 for Figure 1 Detailed implementation steps of step S2 in FIG.

[0068] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0069] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0070] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0071] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0072] To achieve this, please refer to Figure 1 to Figure 2 , a method for constructing an early warning model of land characteristics affecting supply and demand balance, the method comprising the following steps:

[0073] Step S1: Acquire the land area to be analyzed; collect all-weather environmental data of the land area to be analyzed by ground sensors to obtain all-weather environmental data of the land analysis area; perform terrain structure analysis on the land area to be analyzed to obtain regional land terrain structure data; perform slope soil attribute analysis based on the regional land terrain structure data to obtain slope soil attribute data; construct a three-dimensional land structure attribute model based on the slope soil attribute data to obtain a three-dimensional land structure attribute model;

[0074] Step S2: Performing an initial property analysis on the land area to be analyzed to obtain initial property data of the land area; performing dynamic soil fertility accumulation path deduction based on the all-weather environmental data of the land analysis area and the three-dimensional land structure attribute model to obtain soil fertility accumulation path deduction data; performing fertility stratification analysis on the initial property data of the land area according to the soil fertility accumulation path deduction data to obtain regional fertility stratification data;

[0075] Step S3: Based on the Q-learning algorithm, a land property analysis and early warning model is constructed for the property boundary feature data to obtain the land property analysis and early warning model, and the land characteristics are obtained by using the land property analysis and early warning model, and the land supply and demand are further adjusted according to the land characteristics.

[0076] In the embodiment of the present invention, reference Figure 1 The above is a schematic flow chart of the steps of a method for constructing an early warning model of land characteristics affecting supply and demand balance according to the present invention. In this example, the method for constructing an early warning model of land characteristics affecting supply and demand balance includes the following steps:

[0077] Step S1: Acquire the land area to be analyzed; collect all-weather environmental data of the land area to be analyzed by ground sensors to obtain all-weather environmental data of the land analysis area; perform terrain structure analysis on the land area to be analyzed to obtain regional land terrain structure data; perform slope soil attribute analysis based on the regional land terrain structure data to obtain slope soil attribute data; construct a three-dimensional land structure attribute model based on the slope soil attribute data to obtain a three-dimensional land structure attribute model;

[0078] In the embodiment of the present invention, the first step of obtaining the land area to be analyzed is to use ground sensors to collect all-weather environmental data of the selected area. The sensors used have the real-time monitoring capability of multiple parameters such as temperature, humidity, soil moisture and light intensity, and upload the data to the data processing center in real time through wireless transmission technology. After the data collection is completed, the terrain of the land area to be analyzed is analyzed using GIS (Geographic Information System) technology to extract terrain features such as slope, slope direction and altitude changes to form regional land terrain structure data. Next, based on the extracted terrain structure data, the gravity model and hydrological model are used to analyze the slope soil properties, focusing on the texture, density and water retention capacity of the soil, so as to obtain the slope soil property data. Based on these data, a three-dimensional land structure attribute model is constructed, and three-dimensional modeling software is used to combine various soil properties and terrain features to achieve three-dimensional visualization of the land structure. Finally, the three-dimensional land structure attribute model formed provides a basis for subsequent soil fertility and property analysis.

[0079] Step S2: Performing an initial property analysis on the land area to be analyzed to obtain initial property data of the land area; performing dynamic soil fertility accumulation path deduction based on the all-weather environmental data of the land analysis area and the three-dimensional land structure attribute model to obtain soil fertility accumulation path deduction data; performing fertility stratification analysis on the initial property data of the land area according to the soil fertility accumulation path deduction data to obtain regional fertility stratification data;

[0080] In an embodiment of the present invention, the implementation steps of performing an initial property analysis of the land area to be analyzed in the land area to be analyzed include a comprehensive investigation of the physical and chemical properties of the soil in the area. By collecting samples and conducting laboratory analysis, the pH value, organic matter content and nutrient content of the soil in the area are recorded, and finally the initial property data of the land area are obtained. Then, the dynamic soil fertility accumulation path is deduced by combining the all-weather environmental data obtained in step S1 with the three-dimensional land structure attribute model. In this process, a dynamic simulation method is adopted to analyze the change path of soil fertility under different climatic conditions by constructing a time series model, predict the accumulation trend of soil fertility, and obtain the soil fertility accumulation path deduction data. Finally, the fertility traits are layered and analyzed based on the deduced data and the initial property data of the land area, and the fertility distribution of the soil at different depth levels is analyzed by using geostatistical methods, and finally the regional fertility trait layered data is obtained. This process ensures a comprehensive understanding of the regional soil fertility and provides support for subsequent decision-making.

[0081] Step S3: Based on the Q-learning algorithm, a land property analysis and early warning model is constructed for the property boundary feature data to obtain the land property analysis and early warning model, and the land characteristics are obtained by using the land property analysis and early warning model, and the land supply and demand are further adjusted according to the land characteristics.

[0082] In the embodiment of the present invention, the implementation process of constructing a land property analysis and early warning model based on the Q-learning algorithm first needs to collect and organize the land property boundary feature data. These data include soil moisture, nutrient content and physical properties, etc., which constitute the feature input. Subsequently, a Q-learning model is constructed, the state space of which is composed of different soil properties, and the action space is defined as different management measures, such as fertilization, irrigation, etc. When training the model, the land management strategy is optimized by continuously adjusting the Q value. At each update, the reward mechanism is used to evaluate the effect of each management measure, compare it with the preset target, and gradually learn the best land management strategy. After the training is completed, the model can perform real-time analysis on the newly input soil property data and warn of potential soil property change risks. Finally, through the output of the model, a land property analysis and early warning model is formed to provide a scientific decision-making basis for land managers. This process effectively utilizes the principle of reinforcement learning to ensure accurate analysis of complex soil properties.

[0083] Preferably, step S1 comprises the following steps:

[0084] Step S11: obtaining the land area to be analyzed;

[0085] Step S12: collecting all-weather environmental data of the land area to be analyzed through ground sensors to obtain all-weather environmental data of the land analysis area;

[0086] Step S13: Performing terrain structure analysis on the land area to be analyzed to obtain regional land terrain structure data;

[0087] Step S14: performing slope soil property analysis on the land area to be analyzed based on the regional land topography data to obtain slope soil property data;

[0088] Step S15: constructing a three-dimensional land structure attribute model according to the regional land topography data and the slope soil attribute data to obtain a three-dimensional land structure attribute model.

[0089] In the embodiment of the present invention, the process of obtaining the land area to be analyzed first requires clarifying the geographical location and boundaries of the study area. By consulting cadastral data, aerial images and relevant literature, the specific coordinates and characteristics of the area to be analyzed are determined. Then, the global positioning system (GPS) equipment is used to confirm the boundaries of the selected area on the spot. In this process, it is necessary to pay attention to selecting representative plots to ensure that the soil and environmental characteristics in the area can reflect the overall situation. At the same time, the current land use status of the area, including information such as crop types and land use history, is recorded to provide background information for subsequent data analysis. The step of collecting all-weather environmental data of the land area to be analyzed by ground sensors first deploys multiple sensors, covering parameters such as temperature, humidity, soil moisture, light intensity and wind speed. The sensor should have high-precision measurement capabilities and be able to withstand the influence of different climatic conditions to ensure the continuity and accuracy of data collection. During the data collection process, the sensor records the environmental parameters once every certain time (such as 5 minutes), and the data is transmitted to the central database in real time via a wireless network. The data storage format adopts a standardized format to ensure the convenience of subsequent analysis. After the data collection is completed, the data is preliminarily screened and corrected to remove outliers and obtain effective all-weather environmental data for the land analysis area. When analyzing the terrain structure of the land area to be analyzed, the digital elevation model (DEM) is first used to obtain the elevation information in the area. The acquired DEM data is analyzed through the geographic information system (GIS) tool to extract the slope, slope direction and other terrain features. The terrain in the area is classified into types such as flat, hilly and mountainous, so as to classify and process different terrain features. Next, combined with remote sensing images, the land cover type is extracted, and the impact of different cover types on the terrain is analyzed to form regional land terrain structure data. The entire process uses terrain analysis tools and spatial analysis algorithms to ensure the accuracy and comprehensiveness of terrain feature extraction, providing basic data for subsequent soil property analysis. The steps for slope soil property analysis based on regional land terrain structure data first require the selection of appropriate soil sampling points, which should be evenly distributed according to the terrain characteristics and cover areas with different slopes and slope directions. Then, field sampling is carried out and the soil samples are taken back to the laboratory for analysis to test their physical and chemical properties, including soil particle size, porosity, organic matter content and nutrient status. Standard laboratory analysis methods, such as gas chromatography or mass spectrometry, are used to ensure the accuracy of the analysis results. Finally, the slope soil attribute data obtained should be systematically organized to form a data set to provide a detailed soil attribute basis for the subsequent three-dimensional model construction. The step of constructing a three-dimensional land structure attribute model based on regional land terrain structure data and slope soil attribute data first requires the integration of the terrain and soil data obtained in the previous step. Through the three-dimensional modeling software, a three-dimensional model based on terrain data is constructed, combined with soil attribute data, and the model is given corresponding soil characteristics.Finite element analysis was used to divide the model in detail to ensure that soil characteristics in different areas could be accurately reflected. Then, the terrain and soil properties were coupled using meshing and interpolation algorithms to generate the final three-dimensional land structure attribute model. This model can intuitively display the land structure characteristics in the region and lay the foundation for subsequent soil property analysis.

[0090] Preferably, step S2 comprises the following steps:

[0091] Step S21: performing an initial land property analysis on the land area to be analyzed based on the regional land topography data and the slope soil property data to obtain initial land property data;

[0092] Step S22: based on the all-weather environmental data of the land analysis area and the three-dimensional land structure attribute model, a dynamic soil fertility accumulation path is deduced for the regional land terrain structure data and the slope soil attribute data to obtain soil fertility accumulation path deduction data;

[0093] Step S23: performing a stacking intensity spatial heterogeneity analysis on the soil fertility stacking path deduction data to obtain stacking path intensity spatial heterogeneity data;

[0094] Step S24: performing fertility trait hierarchical analysis on the initial trait data of the land area according to the spatial heterogeneous data of the accumulation path intensity, and obtaining regional fertility trait hierarchical data.

[0095] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes:

[0096] Step S21: performing an initial land property analysis on the land area to be analyzed based on the regional land topography data and the slope soil property data to obtain initial land property data;

[0097] In an embodiment of the present invention, the step of performing initial property analysis of a land region based on regional land terrain structure data and slope soil attribute data first requires integrating the terrain and soil data obtained previously. Using GIS technology, the terrain structure data and soil attribute data are superimposed and analyzed to identify soil properties under different terrain units. During the analysis process, key soil indicators such as pH value, soil organic matter content, and concentrations of elements such as nitrogen, phosphorus, and potassium are selected, and a spatial distribution map of regional soil properties is generated through spatial interpolation methods. These data can be implemented through the Kriging interpolation method to ensure the continuity and rationality of the generated data in space. Finally, all analysis results are sorted out to form initial property data of the land region, including the relationship between soil characteristics and terrain characteristics of different plots in the region, laying the foundation for subsequent fertility analysis.

[0098] Step S22: based on the all-weather environmental data of the land analysis area and the three-dimensional land structure attribute model, a dynamic soil fertility accumulation path is deduced for the regional land terrain structure data and the slope soil attribute data to obtain soil fertility accumulation path deduction data;

[0099] In an embodiment of the present invention, the step of dynamically deducing the soil fertility accumulation path of regional land terrain structure data and slope soil attribute data based on the all-weather environmental data of the land analysis area and the three-dimensional land structure attribute model first integrates the all-weather environmental data, terrain structure data and slope soil attribute data. By establishing a dynamic model, the change process of soil fertility under different climatic conditions is simulated. The material balance method is adopted to combine the environmental data with the soil data, establish a continuous time series model, and deduce the accumulation path of soil fertility. In the specific operation, it is necessary to set the initial conditions, such as the initial soil fertility and climate variables, and deduce step by step using the finite difference method. At each time step, the model will update the soil fertility status, record the soil fertility change data at different time points, and finally form the soil fertility accumulation path deduction data, reflecting the dynamic change trend.

[0100] Step S23: performing a stacking intensity spatial heterogeneity analysis on the soil fertility stacking path deduction data to obtain stacking path intensity spatial heterogeneity data;

[0101] In an embodiment of the present invention, the step of performing spatial heterogeneity analysis on the soil fertility accumulation path deduction data of the accumulation intensity first requires performing spatial statistical analysis on the deduction results. Through spatial autocorrelation analysis, the spatial distribution characteristics of fertility accumulation are evaluated to determine whether there is an aggregation phenomenon. Next, the semivariogram method is used to calculate the difference in fertility accumulation intensity between different locations to evaluate the spatial variability of accumulation intensity. The fertility accumulation intensity data is processed by a geographic information system to generate a spatial distribution map of accumulation intensity, showing the heterogeneity characteristics of accumulation intensity in different regions. The analysis results can provide spatial heterogeneous data on accumulation intensity, reveal the distribution law of soil fertility, and provide a basis for subsequent hierarchical analysis of fertility traits.

[0102] Step S24: performing fertility trait hierarchical analysis on the initial trait data of the land area according to the spatial heterogeneous data of the accumulation path intensity, and obtaining regional fertility trait hierarchical data.

[0103] In an embodiment of the present invention, the step of performing stratified analysis of fertility traits on initial property data of land regions according to spatial heterogeneous data of accumulation path intensity first integrates spatial heterogeneous data of accumulation intensity with initial property data of land regions. A stratified sampling method is adopted to divide the region into different fertility levels based on the spatial distribution of accumulation intensity. In a specific implementation, high, medium and low fertility areas are identified by setting thresholds. Then, the variance analysis method is applied to compare soil properties at different fertility levels, and the mean value and coefficient of variation of each level are calculated to evaluate its fertility characteristics. Finally, regional fertility trait stratified data is formed to clearly show the fertility differences of soils at different levels, providing basic data support for further land management and utilization.

[0104] Preferably, step S22 includes the following steps:

[0105] Step S221: extracting rainfall and temperature changes from the all-weather environmental data of the land analysis area to obtain regional rainfall data and regional temperature change data respectively;

[0106] Step S222: evaluating the water penetration efficiency of the slope soil property data according to the regional rainfall data and the regional temperature change data, and obtaining the slope soil water penetration efficiency data;

[0107] Step S223: performing soil component loss path analysis on the regional land topography data based on the slope soil water penetration efficiency data to obtain soil component loss path data;

[0108] Step S224: simulating the movement of nutrient soil layers on the slope soil water penetration efficiency data and the soil component loss path data according to the three-dimensional land structure attribute model to obtain nutrient soil layer movement simulation data;

[0109] Step S225: performing nutrient accumulation and convergence regional distribution analysis on the nutrient soil layer movement simulation data to obtain nutrient accumulation and convergence regional data;

[0110] Step S226: Dynamic soil fertility accumulation path deduction is performed based on the nutrient accumulation convergence area data to obtain soil fertility accumulation path deduction data.

[0111] In the embodiment of the present invention, the step of extracting rainfall and temperature changes from the all-weather environmental data of the land analysis area first requires separating the time series data related to rainfall and temperature from the data set. Using data analysis tools, statistical processing is performed on the continuous observation records to calculate the rainfall changes and temperature fluctuations within a specific time period. The rainfall data can be obtained by directly reading the precipitation records from the sensor, and the simple moving average method is used to smooth the rainfall changes to reduce the impact of extreme values ​​on the analysis results. The temperature change data can be fitted with the historical temperature records by the least squares method to extract the daily change trend and seasonal fluctuation of the temperature. Finally, clear regional rainfall data and regional temperature change data are obtained, which provide a basis for the subsequent water infiltration efficiency evaluation. The step of evaluating the water infiltration efficiency of the slope soil attribute data based on the regional rainfall data and the regional temperature change data first requires determining the physical properties of the soil, including soil type, particle composition and structural characteristics. This information is obtained through preliminary soil sampling and laboratory analysis. Next, the permeability experimental data is used to combine the rainfall and temperature change data to establish a soil moisture infiltration model. The double exponential model is used to simulate the infiltration process and calculate the water infiltration efficiency under different conditions. In the specific operation, the rainfall and temperature changes are combined with soil characteristics, and the water infiltration efficiency in different time periods is evaluated through numerical calculation methods. Finally, the slope soil water infiltration efficiency data is obtained, showing the dynamic movement characteristics of water in the soil. The steps of analyzing the soil component loss path based on the slope soil water infiltration efficiency data and the regional land terrain structure data first need to determine the water flow path in different slope areas. Using the terrain analysis tool, the slope flow direction is calculated through DEM data, and the potential water flow concentration area is identified by combining the slope and aspect information. Next, combined with the water infiltration efficiency data, the soil and water conservation model is used to deduce the soil component loss. The model takes into account the rainfall intensity, slope characteristics and soil moisture status, and the mass balance method is used to calculate the soil component loss. By setting parameters, the soil component loss path under different rainfall and temperature conditions is simulated to form soil component loss path data, revealing the loss dynamics of soil components under specific conditions. The steps of simulating the movement of nutrient soil layers based on the slope soil water infiltration efficiency data and soil component loss path data according to the three-dimensional land structure attribute model first need to integrate the slope water infiltration efficiency and soil component loss path information. This process can be done by using the finite element analysis method to divide the three-dimensional land structure model into multiple units. Then, different nutrient flow conditions are set, and the movement of nutrients in the soil layer and their interaction with water are simulated through the principle of conservation of matter. In each calculation unit, the solubility of nutrients and soil characteristics are considered, and the movement trajectory of nutrients is deduced by combining the molecular diffusion model and the convection model. Finally, the simulation data of nutrient soil layer movement is obtained, showing the distribution characteristics and change laws of nutrients in different soil layers.The steps of analyzing the regional distribution of nutrient accumulation and convergence of nutrient soil layer movement simulation data first require spatial statistical analysis of the nutrient movement simulation results. Use the Kriging interpolation method to spatially interpolate nutrient concentrations and generate a spatial distribution map of nutrient concentrations. Then, combined with specific thresholds, identify high-concentration areas of nutrient accumulation. Use spatial cluster analysis methods to evaluate the cumulative effect of nutrients in different regions and determine the uniformity or aggregation of their distribution. Finally, nutrient accumulation and convergence area data are obtained, which shows the distribution characteristics of nutrients on the land in detail, providing data support for subsequent dynamic soil fertility deduction. The steps of deducing the dynamic soil fertility accumulation path based on nutrient accumulation and convergence area data first require combining the nutrient accumulation area data with environmental data to determine the dynamic changes in fertility status. Establish a time series model to simulate the changes in nutrients under different seasons and rainfall conditions. Use the piecewise linear regression method to fit the fertility status of the nutrient accumulation area and obtain the nutrient accumulation amount in different time periods. Next, based on the water penetration efficiency and component loss path data, the dynamic changes of soil fertility were deduced, the accumulation and loss paths of fertility were identified, and finally the soil fertility accumulation path deduction data was obtained, showing the dynamic accumulation process of nutrients in the soil.

[0112] Preferably, step S225 includes the following steps:

[0113] Conduct movement dynamics attenuation analysis on the simulated data of nutrient soil layer movement to obtain nutrient movement dynamics attenuation data;

[0114] According to the nutrient movement dynamic attenuation data, the regional agglomeration influencing factors of the nutrient soil layer movement simulation data are identified to obtain the regional agglomeration influencing factors;

[0115] According to the nutrient movement dynamic attenuation data and regional agglomeration influencing factors, the nutrient soil layer movement simulation data is processed by cumulative continuous interpolation to obtain the nutrient soil layer cumulative continuous interpolation data;

[0116] The nutrient accumulation and convergence regional distribution analysis was performed on the nutrient soil layer accumulation continuous interpolation data to obtain the nutrient accumulation and convergence regional data.

[0117] In the embodiment of the present invention, the step of analyzing the dynamic decay of nutrient soil layer movement simulation data first requires extracting the motion data of each time node during the nutrient movement process, and calculating the dynamic decay value of nutrient movement by comparing the changes in nutrient concentrations in different time periods. The exponential decay model is used to analyze the migration and dissipation of nutrients in the soil layer. In this process, by setting different initial nutrient concentrations, the decay curve of nutrient concentration is fitted using the least squares method to obtain the decay coefficient, thereby forming nutrient movement dynamic decay data, reflecting the dynamic change trend of nutrients in the soil. According to the nutrient movement dynamic decay data, the step of identifying regional agglomeration influencing factors of nutrient soil layer movement simulation data first requires determining the key factors affecting nutrient aggregation, including soil type, precipitation, temperature, etc. Through multiple linear regression analysis, the decay data is associated with environmental factors, and the contribution of each factor to nutrient movement is calculated. Using correlation coefficient analysis, the factors that significantly affect nutrient aggregation are identified, and finally the regional agglomeration influencing factors are obtained, which provide important references for subsequent nutrient analysis. The steps of cumulative continuous interpolation processing of nutrient soil layer movement simulation data based on nutrient movement dynamic decay data and regional aggregation influencing factors are as follows: first, the movement simulation data needs to be gridded and mapped to a regular grid. The spatial interpolation is performed using the Kriging interpolation method, and the nutrient accumulation in each grid is calculated in combination with the movement dynamic decay data. By setting the interpolation parameters, the continuity and spatial consistency of the interpolation results are ensured, and finally the nutrient soil layer cumulative continuous interpolation data are obtained, which reflects the spatial distribution and change of nutrients in the soil layer. The steps of nutrient accumulation and aggregation regional distribution analysis of nutrient soil layer cumulative continuous interpolation data are as follows: first, spatial analysis technology is used to divide the interpolation results into regions. The cluster analysis method is used to identify areas with high nutrient concentrations and analyze their relationship with the surrounding environment. The geographic information system (GIS) tool is used to visualize each aggregation area to form a distribution map of the nutrient aggregation area. The spatial statistical method is used to further evaluate the characteristics and distribution laws of each aggregation area, and finally the nutrient accumulation and aggregation area data are obtained, which provides a basis for subsequent soil management and optimization.

[0118] Preferably, step S24 comprises the following steps:

[0119] Step S241: performing spatial intensity numerical simulation on the stacking path intensity spatial heterogeneous data to obtain stacking path intensity numerical data;

[0120] Step S242: performing spatial autocorrelation feature analysis on the stacking path strength numerical data to obtain stacking strength spatial autocorrelation data;

[0121] Step S243: performing multivariate regression analysis on the stacking path strength numerical data according to the stacking strength spatial autocorrelation data to obtain stacking strength multivariate regression data;

[0122] Step S244: Based on the accumulation intensity multivariate regression data and the accumulation intensity spatial autocorrelation data, the initial property data of the land area are subjected to stratified analysis of fertility traits to obtain regional fertility stratified data.

[0123] In an embodiment of the present invention, the step of performing spatial intensity numerical simulation on the spatial heterogeneous data of the stacking path intensity is to first convert the stacking path intensity data into a spatial data format and perform point calibration according to the actual geographic coordinates. By using the finite element analysis method, a spatial model of the plot is established, and appropriate boundary conditions and initial conditions are set in the model to accurately reflect the distribution of the stacking intensity at different spatial positions. During the simulation process, the model parameters need to be adjusted, including material properties and external environmental influences, and finally the stacking path intensity numerical data are obtained, which show the spatial distribution of nutrients in the soil and their changes. The step of performing spatial autocorrelation feature analysis on the stacking path intensity numerical data is to first calculate the local and global autocorrelation of the stacking path intensity. Statistical indicators such as Moran's I are used to evaluate the spatial aggregation and dispersion of data. By combining numerical data with spatial position, the correlation of stacking intensity in different regions can be revealed. The results of the analysis form the spatial autocorrelation data of stacking intensity, which characterizes the distribution law of stacking intensity in space and provides a basis for subsequent analysis. The steps of multivariate regression analysis of the numerical data of accumulation path intensity based on the spatial autocorrelation data of accumulation intensity require the construction of a regression model, with accumulation intensity as the dependent variable and spatial autocorrelation data and other influencing factors (such as climate and topography) as independent variables. The regression coefficient is calculated by the least squares method to evaluate the influence of each independent variable on accumulation intensity. In this process, the goodness of fit test of the model is required to ensure the reliability and predictive ability of the model, and finally the multivariate regression data of accumulation intensity is obtained, which reflects the comprehensive influence of different factors on soil fertility. The steps of stratifying and analyzing the fertility traits of the initial trait data of the land area based on the multivariate regression data of accumulation intensity and the spatial autocorrelation data of accumulation intensity are as follows: first, the initial trait data of the land area needs to be connected with the accumulation intensity data to determine the fertility trait standards at different levels. Using the hierarchical analysis method, the region is divided according to the regression analysis results, the soil is divided into multiple levels, and the fertility characteristics of each level are identified. Finally, the regional fertility trait stratification data is formed, which reflects the characteristics of soil fertility at different levels and provides data support for further land management.

[0124] Preferably, step S3 comprises the following steps:

[0125] Step S31: extracting trait boundary features from the regional fertility trait stratification data to obtain trait boundary feature data;

[0126] Step S32: performing fertility loss area association on the soil fertility accumulation path deduction data according to the trait boundary characteristic data to obtain fertility loss area association data;

[0127] Step S33: Based on the Q-learning algorithm, a land property analysis and early warning model is constructed for the fertility loss area correlation data and the property boundary feature data to obtain the land property analysis and early warning model, and the land characteristics are obtained using the land property analysis and early warning model, and the land supply and demand are further adjusted according to the land characteristics.

[0128] In an embodiment of the present invention, the step of extracting the boundary characteristics of the regional fertility trait stratified data first requires identifying and analyzing the key trait parameters in the stratified data. This process extracts the boundaries where the fertility traits change significantly by setting thresholds and conditions. For example, the gradient boosting method is used to evaluate the contribution of each trait parameter, and the trait boundary characteristic data is obtained by comparing the changes in fertility indicators between different levels. This data can clearly reflect the distribution characteristics of soil fertility and provide a basis for subsequent analysis. According to the trait boundary characteristic data, the step of associating the fertility loss area with the soil fertility accumulation path deduction data needs to match the fertility loss area with the boundary characteristics. Through spatial data analysis technology, the fertility loss points in the accumulation path deduction data are combined with the trait boundary characteristics, and the spatial interpolation method (such as Kriging interpolation) is used for analysis to identify the key areas of fertility loss. In this process, the influence of environmental factors needs to be considered and appropriate adjustments need to be made to obtain fertility loss area association data, which clearly shows the relationship between fertility loss and regional traits. The steps of constructing a land trait analysis and early warning model based on the Q-learning algorithm for the fertility loss area correlation data and trait boundary feature data require the establishment of state space and action space, with the fertility loss area as the state and the trait boundary feature as the environmental feedback. Through repeated experiments, the appropriate strategy is optimized and selected, and the value function is gradually updated. Using the iterative update formula of Q-learning, the benefits of the current state are combined with future benefits to obtain the optimal strategy. Finally, a land trait analysis and early warning model is generated, which can predict soil fertility changes, identify potential risk areas, and provide data support for subsequent land management and utilization.

[0129] Preferably, step S33 includes the following steps:

[0130] Step S331: performing feature importance evaluation on fertility loss area correlation data to obtain loss feature importance evaluation data;

[0131] Step S332: performing boundary characteristic loss coupling according to the loss characteristic importance evaluation data and the characteristic boundary characteristic data to obtain boundary characteristic loss coupling data;

[0132] Step S333: dividing the boundary characteristic loss coupling data into a test set and a training set to obtain a boundary characteristic loss test set and a boundary characteristic loss training set respectively;

[0133] Step S334: performing sampling balancing processing on the boundary characteristic loss training set to obtain a boundary characteristic loss sampling balanced training set;

[0134] Step S335: constructing an initial land property analysis and early warning model for the boundary property loss sampling balance training set based on the Q-learning algorithm to obtain an initial land property analysis and early warning model;

[0135] Step S336: Test and verify the initial land property analysis and early warning model according to the boundary property loss test set to obtain the land property analysis and early warning model, use the land property analysis and early warning model to obtain land characteristics, and then regulate land supply and demand according to the land characteristics.

[0136] In an embodiment of the present invention, the step of evaluating the importance of features for the fertility loss area association data requires the use of feature selection technology to analyze different features in the data set. A feature importance evaluation method based on a tree model is used, and the specific method is Random Forest. In this process, the importance score of each feature to the prediction result is calculated by training the random forest model. This score is based on the information gain contributed by the feature when the model splits the node, identifies the feature with the highest correlation with fertility loss, and finally generates loss feature importance evaluation data to provide a basis for subsequent analysis. In the step of coupling boundary trait loss according to the loss feature importance evaluation data and the trait boundary feature data, the two sets of data need to be integrated and multivariate linear regression analysis needs to be performed. In this process, by defining the linear relationship between the loss feature and the boundary feature, a regression model is established to calculate the degree of influence of each feature on the loss. In the coupling process, the least squares method is used to solve the regression equation to obtain the boundary trait loss coupling data, and the influence mechanism of each boundary feature on fertility loss is clarified. The step of dividing the boundary trait loss coupling data into a test set and a training set needs to be divided according to the proportion of the data set. The specific operation is to use an 80 / 20 ratio to randomly divide the data into a training set and a test set. During the division process, ensure that the two parts of the data are consistent in the distribution of fertility loss characteristics to avoid introducing bias. The training set is used for model construction, and the test set is used for subsequent model verification, and the boundary trait loss test set and the boundary trait loss training set are obtained respectively. The step of sampling and balancing the boundary trait loss training set requires the use of oversampling or undersampling technology to ensure that the number of samples of each type is balanced in the training set. The specific operation is to randomly copy or synthesize minority class samples (such as using the SMOTE method) for categories with a small number of samples to increase the sample size, or randomly delete samples from the majority class to achieve a balanced number of categories. This processing process ensures the diversity of the training set, improves the training effect of the model, and finally obtains a boundary trait loss sampling and balanced training set. The step of constructing the initial land trait analysis and early warning model for the boundary trait loss sampling and balanced training set based on the Q-learning algorithm requires that fertility loss be regarded as a state, and corresponding actions and rewards be defined. By constructing the state-action value function (Q function), the Q value is optimized using the iterative update method to achieve the learning of the optimal decision-making strategy. The Bellman equation is used for value update. After multiple iterations, it gradually converges and finally generates an initial land property analysis and early warning model, which can effectively warn of fertility loss. The step of testing and verifying the initial land property analysis and early warning model based on the boundary property loss test set is to evaluate the model's prediction performance by inputting the test set into the model. The specific operation is to calculate the model's accuracy, precision, and recall on the test set to evaluate its ability to predict fertility loss.At the same time, the prediction results of the model are visualized using the confusion matrix, and the model parameters are adjusted to further optimize it, ultimately obtaining an effective land property analysis and early warning model.

[0137] Preferably, step S334 includes the following steps:

[0138] Perform feature clustering on the boundary characteristic loss training set to obtain the boundary characteristic loss clustering training set;

[0139] Under-sampling the boundary trait loss training set based on the boundary trait loss clustering training set to obtain an under-sampled boundary trait loss training set;

[0140] Perform stratified sampling on the under-sampled boundary trait loss training set to obtain a boundary trait stratified sampling training set;

[0141] The boundary trait loss training set is sampled and balanced according to the boundary trait stratified sampling training set and the under-sampling boundary trait loss training set to obtain the boundary trait loss sampling balanced training set.

[0142] In an embodiment of the present invention, the step of performing feature clustering on the boundary trait loss training set adopts the K-means clustering algorithm. First, the features in the training set need to be standardized to eliminate the dimensional differences between different features. Then, a suitable K value is selected, and the number of clusters is determined by the Elbow Method. Next, the K-means algorithm is applied to cluster the data, and the clustering process is continuously iterated to minimize the distance from the sample to the cluster center. Finally, a boundary trait loss cluster training set is obtained, and each sample is assigned to a corresponding cluster according to its characteristics, which is convenient for subsequent processing. Based on the boundary trait loss cluster training set, the step of performing undersampling processing on the boundary trait loss training set is to randomly select some samples for deletion for cluster categories with a large number of samples to balance the number of samples in each category. During the processing, the number of samples in each cluster is first counted, and the cluster with the largest number of samples is determined as a benchmark. Then, the same number of samples are randomly selected from each cluster according to the set ratio, and the excess samples are deleted, and finally an undersampled boundary trait loss training set is generated. This processing ensures that the number of samples of each category in the training set is balanced, which helps to improve the stability of model training. The steps of stratified sampling processing for the undersampled boundary trait loss training set first require stratifying the data according to the category label to ensure that each category is represented in the training set. During the implementation process, define the required sample ratio, usually 70% for training and 30% for validation. Randomly extract the corresponding proportion of samples in each category to ensure that the number of samples in each category is consistent between the training set and the validation set. Finally, the boundary trait stratified sampling training set is generated to facilitate subsequent model validation and testing and improve the generalization ability of the model. The steps of sampling and balancing the boundary trait loss training set based on the boundary trait stratified sampling training set and the undersampled boundary trait loss training set need to combine the characteristics of the two and adjust the distribution of the training set samples through a comprehensive strategy. The specific operation is to merge the samples obtained by stratified sampling with the samples obtained by undersampling to ensure that the number of samples in each category is consistent. Then, use random sampling or weighted sampling methods to adjust the sample ratio in the final training set to achieve the balance of the overall samples, and finally obtain the boundary trait loss sampling balanced training set. This step ensures that the training set is representative and can effectively support subsequent model training and validation.

[0143] Preferably, the present invention further provides a system for constructing an early warning model of land characteristics affecting supply and demand balance, which is used to execute the method for constructing an early warning model of land characteristics affecting supply and demand balance as described above. The system for constructing an early warning model of land characteristics affecting supply and demand balance comprises:

[0144] The land structure attribute analysis module is used to obtain the land area to be analyzed; collect all-weather environmental data of the land area to be analyzed through ground sensors to obtain all-weather environmental data of the land analysis area; perform terrain structure analysis on the land area to be analyzed to obtain regional land terrain structure data; perform slope soil attribute analysis based on the regional land terrain structure data to obtain slope soil attribute data; construct a three-dimensional land structure attribute model based on the slope soil attribute data to obtain a three-dimensional land structure attribute model;

[0145] The fertility trait hierarchical analysis module is used to perform initial trait analysis on the land area to be analyzed to obtain initial trait data of the land area; to perform dynamic soil fertility accumulation path deduction based on the all-weather environmental data of the land analysis area and the three-dimensional land structure attribute model to obtain soil fertility accumulation path deduction data; to perform fertility trait hierarchical analysis on the initial trait data of the land area according to the soil fertility accumulation path deduction data to obtain regional fertility trait hierarchical data;

[0146] The analysis and early warning model construction module is used to construct a land property analysis and early warning model for property boundary feature data based on the Q-learning algorithm, obtain a land property analysis and early warning model, use the land property analysis and early warning model to obtain land characteristics, and then regulate land supply and demand based on the land characteristics.

[0147] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0148] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for constructing an early warning model of land characteristics affecting supply and demand balance, characterized in that: The following steps are involved: Step S1: Acquire the land area to be analyzed; collect all-weather environmental data of the land area to be analyzed through ground sensors to obtain all-weather environmental data of the land analysis area; Conduct terrain structure analysis on the land area to be analyzed to obtain regional land terrain structure data; Perform slope soil attribute analysis based on regional land topography data to obtain slope soil attribute data; construct a three-dimensional land structure attribute model based on the slope soil attribute data to obtain a three-dimensional land structure attribute model; Step S2: performing an initial land property analysis on the land area to be analyzed to obtain initial land property data; Based on the all-weather environmental data of the land analysis area and the three-dimensional land structure attribute model, the dynamic soil fertility accumulation path is deduced to obtain the soil fertility accumulation path deduction data; based on the soil fertility accumulation path deduction data, the initial property data of the land area is analyzed by fertility stratification to obtain the regional fertility stratification data; Step S3: constructing a land property analysis and early warning model for the property boundary feature data based on the Q-learning algorithm to obtain a land property analysis and early warning model, using the land property analysis and early warning model to obtain land characteristics, and then regulating land supply and demand according to the land characteristics; Wherein step S3 comprises: Step S31: extracting trait boundary features from the regional fertility trait stratification data to obtain trait boundary feature data; Step S32: performing fertility loss area association on the soil fertility accumulation path deduction data according to the trait boundary characteristic data to obtain fertility loss area association data; Step S33: constructing a land property analysis and early warning model based on the fertility loss area correlation data and the property boundary feature data based on the Q-learning algorithm to obtain the land property analysis and early warning model, using the land property analysis and early warning model to obtain land characteristics, and then regulating land supply and demand according to the land characteristics; wherein step S33 includes: Step S331: performing feature importance evaluation on fertility loss area correlation data to obtain loss feature importance evaluation data; Step S332: performing boundary characteristic loss coupling according to the loss characteristic importance evaluation data and the characteristic boundary characteristic data to obtain boundary characteristic loss coupling data; Step S333: dividing the boundary characteristic loss coupling data into a test set and a training set to obtain a boundary characteristic loss test set and a boundary characteristic loss training set respectively; Step S334: performing sampling balance processing on the boundary property loss training set to obtain a boundary property loss sampling balanced training set; wherein step S334 includes: Perform feature clustering on the boundary characteristic loss training set to obtain the boundary characteristic loss clustering training set; Under-sampling the boundary trait loss training set based on the boundary trait loss clustering training set to obtain an under-sampled boundary trait loss training set; Perform stratified sampling on the under-sampled boundary trait loss training set to obtain a boundary trait stratified sampling training set; According to the boundary trait stratified sampling training set and the under-sampling boundary trait loss training set, the boundary trait loss training set is subjected to sampling balance processing to obtain the boundary trait loss sampling balanced training set; Step S335: constructing an initial land property analysis and early warning model for the boundary property loss sampling balance training set based on the Q-learning algorithm to obtain an initial land property analysis and early warning model; Step S336: Test and verify the initial land property analysis and early warning model according to the boundary property loss test set to obtain the land property analysis and early warning model, use the land property analysis and early warning model to obtain land characteristics, and then regulate land supply and demand according to the land characteristics.

2. The method for constructing an early warning model of land characteristics affecting supply and demand balance according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: obtaining the land area to be analyzed; Step S12: collecting all-weather environmental data of the land area to be analyzed through ground sensors to obtain all-weather environmental data of the land analysis area; Step S13: Performing terrain structure analysis on the land area to be analyzed to obtain regional land terrain structure data; Step S14: performing slope soil property analysis on the land area to be analyzed based on the regional land topography data to obtain slope soil property data; Step S15: constructing a three-dimensional land structure attribute model according to the regional land topography data and the slope soil attribute data to obtain a three-dimensional land structure attribute model.

3. The method for constructing an early warning model of land characteristics affecting supply and demand balance according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: performing an initial land property analysis on the land area to be analyzed based on the regional land topography data and the slope soil property data to obtain initial land property data; Step S22: based on the all-weather environmental data of the land analysis area and the three-dimensional land structure attribute model, a dynamic soil fertility accumulation path is deduced for the regional land terrain structure data and the slope soil attribute data to obtain soil fertility accumulation path deduction data; Step S23: performing a stacking intensity spatial heterogeneity analysis on the soil fertility stacking path deduction data to obtain stacking path intensity spatial heterogeneity data; Step S24: performing fertility trait hierarchical analysis on the initial trait data of the land area according to the spatial heterogeneous data of the accumulation path intensity, and obtaining regional fertility trait hierarchical data.

4. The method for constructing an early warning model of land characteristics affecting supply and demand balance according to claim 3, characterized in that: Step S22 includes the following steps: Step S221: extracting rainfall and temperature changes from the all-weather environmental data of the land analysis area to obtain regional rainfall data and regional temperature change data respectively; Step S222: evaluating the water penetration efficiency of the slope soil property data according to the regional rainfall data and the regional temperature change data, and obtaining the slope soil water penetration efficiency data; Step S223: performing soil component loss path analysis on the regional land topography data based on the slope soil water penetration efficiency data to obtain soil component loss path data; Step S224: simulating the movement of nutrient soil layers on the slope soil water penetration efficiency data and the soil component loss path data according to the three-dimensional land structure attribute model to obtain nutrient soil layer movement simulation data; Step S225: performing nutrient accumulation and convergence regional distribution analysis on the nutrient soil layer movement simulation data to obtain nutrient accumulation and convergence regional data; Step S226: Dynamic soil fertility accumulation path deduction is performed based on the nutrient accumulation convergence area data to obtain soil fertility accumulation path deduction data.

5. The method for constructing an early warning model of land characteristics affecting supply and demand balance according to claim 4, characterized in that: Step S225 includes the following steps: Conduct movement dynamics attenuation analysis on the simulated data of nutrient soil layer movement to obtain nutrient movement dynamics attenuation data; According to the nutrient movement dynamic attenuation data, the regional agglomeration influencing factors of the nutrient soil layer movement simulation data are identified to obtain the regional agglomeration influencing factors; According to the nutrient movement dynamic attenuation data and regional agglomeration influencing factors, the nutrient soil layer movement simulation data is processed by cumulative continuous interpolation to obtain the nutrient soil layer cumulative continuous interpolation data; The nutrient accumulation and convergence regional distribution analysis was performed on the nutrient soil layer accumulation continuous interpolation data to obtain the nutrient accumulation and convergence regional data.

6. The method for constructing an early warning model of land characteristics affecting supply and demand balance according to claim 3, characterized in that: Step S24 includes the following steps: Step S241: performing spatial intensity numerical simulation on the stacking path intensity spatial heterogeneous data to obtain stacking path intensity numerical data; Step S242: performing spatial autocorrelation feature analysis on the stacking path strength numerical data to obtain stacking strength spatial autocorrelation data; Step S243: performing multivariate regression analysis on the stacking path strength numerical data according to the stacking strength spatial autocorrelation data to obtain stacking strength multivariate regression data; Step S244: Based on the accumulation intensity multivariate regression data and the accumulation intensity spatial autocorrelation data, the initial property data of the land area are subjected to stratified analysis of fertility traits to obtain regional fertility stratified data.

7. A system for constructing an early warning model of land characteristics affecting supply and demand balance, characterized in that: The method for constructing an early warning model of land characteristics affecting supply and demand balance according to claim 1 is used to construct an early warning model of land characteristics affecting supply and demand balance, and the system comprises: The land structure attribute analysis module is used to obtain the land area to be analyzed; collect all-weather environmental data of the land area to be analyzed through ground sensors to obtain all-weather environmental data of the land analysis area; perform terrain structure analysis on the land area to be analyzed to obtain regional land terrain structure data; perform slope soil attribute analysis based on the regional land terrain structure data to obtain slope soil attribute data; construct a three-dimensional land structure attribute model based on the slope soil attribute data to obtain a three-dimensional land structure attribute model; The fertility trait hierarchical analysis module is used to perform initial trait analysis on the land area to be analyzed to obtain initial trait data of the land area; to perform dynamic soil fertility accumulation path deduction based on the all-weather environmental data of the land analysis area and the three-dimensional land structure attribute model to obtain soil fertility accumulation path deduction data; to perform fertility trait hierarchical analysis on the initial trait data of the land area according to the soil fertility accumulation path deduction data to obtain regional fertility trait hierarchical data; The analysis and early warning model construction module is used to construct a land property analysis and early warning model for property boundary feature data based on the Q-learning algorithm, obtain a land property analysis and early warning model, use the land property analysis and early warning model to obtain land characteristics, and then regulate land supply and demand based on the land characteristics.

Images