Land utilization change prediction and decision support system based on machine learning

Abstract

The invention belongs to the technical field of land resource management and planning, and provides a land utilization change prediction and decision support system based on machine learning, which comprises a data acquisition and preprocessing module used for collecting multi-source land utilization data; the feature extraction and modeling module comprises a time sequence analysis sub-module and a spatial feature extraction sub-module; the prediction model module comprises a machine learning model training sub-module and a model evaluation sub-module; according to the method, a comprehensive land utilization data set is formed by integrating multi-source information such as satellite remote sensing data, geographic information system data, meteorological data and economic data; data cleaning and standardization processing are performed by using a deep learning algorithm, so that the quality and consistency of the data are ensured; through the time sequence analysis and spatial feature extraction technology, the spatial and temporal features of the land utilization change are comprehensively captured, then a high-precision prediction model is constructed, and the prediction precision of the land utilization change is remarkably improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of land resource management and planning, and specifically relates to a land use change prediction and decision support system based on machine learning. Background Art

[0002] In the process of land resource management and planning, accurately predicting land use changes is of great significance for achieving sustainable development and optimizing resource allocation. However, traditional land use change prediction methods often rely on manual experience judgment or simple statistical models, and these methods have limitations in dealing with complex and variable land use data, making it difficult to accurately capture the spatio-temporal characteristics and potential laws of land use changes. In addition, existing decision support systems usually lack intelligent prediction capabilities and comprehensive decision analysis functions, and cannot meet the current requirements of land resource management for high-precision prediction and diversified decision support.

[0003] With the rapid development of machine learning technology, it has shown strong application potential in fields such as data mining and pattern recognition. Applying machine learning technology to land use change prediction and decision support is expected to achieve intelligent prediction and scientific management of land use changes. However, how to integrate multi-source land use data, effectively extract spatio-temporal characteristics, build a high-precision prediction model, and provide comprehensive decision support remains a technical problem faced by the current field.

[0004] Therefore, those skilled in the art have proposed a land use change prediction and decision support system based on machine learning to solve the problems raised in the background art. Summary of the Invention

[0005] Based on this, it is necessary to provide a system that can realize intelligent digital management of land use change prediction and decision support for the above technical problems.

[0006] In a first aspect, the present application provides a land use change prediction and decision support system based on machine learning, including:

[0007] A data collection and preprocessing module, configured to collect multi-source land use data and perform cleaning, fusion, and standardization processing on the data;

[0008] A feature extraction and modeling module, including a time series analysis sub-module and a spatial feature extraction sub-module; the time series analysis sub-module is configured to extract the time features of land use changes from historical data; the spatial feature extraction sub-module is configured to extract the spatial features of land use changes;

[0009] A prediction model module, including a machine learning model training sub-module and a model evaluation sub-module; the machine learning model training sub-module is used to train a prediction model based on the extracted spatio-temporal features; the model evaluation sub-module is used to evaluate the accuracy and stability of the prediction model;

[0010] A decision support module, including a planning recommendation sub-module and a risk assessment sub-module; the planning recommendation sub-module is used to generate land use planning recommendations based on the prediction results; the risk assessment sub-module is used to evaluate the risks and feasibility of different planning schemes.

[0011] In one embodiment, the data collection and preprocessing module is further used for:

[0012] Through multi-source data fusion technology, integrate satellite remote sensing data, geographic information system (GIS) data, meteorological data and economic data to form a comprehensive land use dataset;

[0013] Use a deep learning-based data cleaning algorithm to automatically detect and repair outliers and missing values in the data;

[0014] Adopt a standardization processing method to unify data from different sources to the same scale and format to ensure data consistency and comparability.

[0015] In one embodiment, the feature extraction and modeling module is further used for:

[0016] Based on time series analysis methods, extract the periodic, trend and seasonal characteristics of land use changes;

[0017] Use spatial clustering algorithms to identify the spatial patterns and hotspots of land use changes;

[0018] Combine spatio-temporal features to construct a multi-dimensional land use change feature vector.

[0019] In one embodiment, the prediction model module is further used for:

[0020] Use a variety of machine learning algorithms, including random forest, support vector machine (SVM) and deep neural network (DNN), to train a prediction model for land use changes;

[0021] Through cross-validation methods, evaluate the prediction performance of different models and select the optimal model;

[0022] Based on the optimal model, generate prediction results of land use changes at different future time points.

[0023] In one embodiment, in the prediction model module, the model training sub-module includes:

[0024] Construct a feature selector according to the spatio-temporal characteristics of land use change to select the most predictive feature combination;

[0025] Train multiple machine learning models based on the feature vectors output by the feature selector;

[0026] Integrate the prediction results of multiple models through model fusion technology to improve the accuracy and stability of prediction.

[0027] In one embodiment, in the prediction model module, the model evaluation sub-module includes:

[0028] Use metrics such as mean squared error (MSE), mean absolute error (MAE), and coefficient of determination (R 2 ) to evaluate the prediction performance of the model;

[0029] Analyze the classification performance of the model through a confusion matrix to identify the main reasons for prediction errors;

[0030] Based on the evaluation results, adjust the model parameters to optimize the model performance.

[0031] In one embodiment, the decision support module is also used for:

[0032] Generate land use planning suggestions under different scenarios based on the prediction results, including various scenarios such as current situation optimization, ecological priority, and economic development;

[0033] Use a risk assessment algorithm to calculate the environmental risk, economic risk, and social risk of each planning scheme;

[0034] Generate a comprehensive evaluation report in combination with the risk assessment results to provide a scientific basis for decision-makers.

[0035] In one embodiment, in the decision support module, the planning suggestion sub-module includes:

[0036] Use a multi-objective optimization algorithm to comprehensively consider economic benefits, environmental impacts, and social needs to generate an optimal planning scheme;

[0037] Dynamically adjust the planning scheme based on the prediction results to ensure the feasibility and sustainability of the plan;

[0038] Provide a visualization tool to display the expected effects of different planning schemes to help decision-makers understand and select the optimal scheme.

[0039] In one embodiment, in the decision support module, the risk assessment sub-module includes:

[0040] Use the analytic hierarchy process (AHP) and fuzzy comprehensive evaluation method to evaluate the multi-dimensional risks of different planning schemes;

[0041] Based on historical data and expert knowledge, establish a risk assessment model to calculate the probability and impact degree of each risk;

[0042] Generate a risk assessment report to provide risk warnings and suggestions for countermeasures to decision-makers.

[0043] In one embodiment, the land use change prediction and decision support system based on machine learning further includes:

[0044] A user interaction module, which includes a data input sub-module and a result display sub-module; the data input sub-module is used to receive the land use data and parameter settings input by the user; the result display sub-module is used to display the prediction results and planning suggestions to the user in a graphical manner, improving the user-friendliness and operability of the system.

[0045] In one embodiment, the land use change prediction and decision support system based on machine learning further includes:

[0046] A data update and maintenance module, which includes a data update sub-module and a system maintenance sub-module; the data update sub-module is used to regularly update the land use data to ensure the timeliness of the data; the system maintenance sub-module is used to monitor the running state of the system, timely detect and repair system failures, and ensure the stable operation of the system.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] 1. By integrating multi-source information such as satellite remote sensing data, geographic information system (GIS) data, meteorological data, and economic data, the present invention forms a comprehensive land use data set. Using deep learning algorithms for data cleaning and standardization processing ensures the quality and consistency of the data. Through time series analysis and spatial feature extraction techniques, the spatio-temporal characteristics of land use changes are comprehensively captured, and then a high-precision prediction model is constructed, significantly improving the prediction accuracy of land use changes.

[0049] 2. The present invention not only provides the prediction results of land use changes, but also generates land use planning suggestions under different scenarios based on the prediction results. Through risk assessment algorithms, the environmental risks, economic risks, and social risks of each planning scheme are calculated, providing a comprehensive risk assessment report for decision-makers. Combining visualization tools, the prediction results and planning suggestions are displayed in a graphical manner to help decision-makers intuitively understand and make scientific decisions.

[0050] 3. The system of the present invention has an intelligent data update and maintenance function, which can regularly update land use data to ensure the timeliness of the data. At the same time, the system monitors the operating status, discovers and repairs system failures in a timely manner to ensure the stable operation of the system. This not only reduces the cost of manual management but also improves the intelligent level of land use management. Brief Description of the Drawings

[0051] Figure 1 It is a system structure block diagram of a land use change prediction and decision support system based on machine learning provided by an embodiment of the present invention; the figure includes a data collection and preprocessing module 101, a feature extraction and modeling module 102, a prediction model module 103, a decision support module 104, a user interaction module 105, and a data update and maintenance module 106. Detailed Embodiment

[0052] The following further describes in detail the embodiments of the present invention with reference to the drawings and embodiments. The following embodiments are used to illustrate the present invention but cannot be used to limit the scope of the present invention.

[0053] The present invention provides a land use change prediction and decision support system based on machine learning. The system realizes the intelligent prediction and management of land use changes by integrating multi-source data, extracting spatio-temporal features, training prediction models, and providing decision support. The following Figure 1 details the specific embodiments of the present invention.

[0054] Figure 1This is a system structure block diagram of a land use change prediction and decision support system based on machine learning provided by an embodiment of the present invention. The system mainly includes a data acquisition and preprocessing module 101, a feature extraction and modeling module 102, a prediction model module 103, a decision support module 104, a user interaction module 105, and a data update and maintenance module 106. First, the data acquisition and preprocessing module 101 is responsible for collecting and processing multi-source land use data, including satellite remote sensing data, geographic information system (GIS) data, meteorological data, and economic data. This module integrates these data into a comprehensive land use dataset through multi-source data fusion technology. Specifically, satellite remote sensing data can be obtained through multi-spectral and high-resolution images, GIS data can include land use classification maps, topographic maps, transportation network maps, etc., meteorological data can include temperature, precipitation, wind speed, etc., and economic data can include population, employment, economic activities, etc. Data preprocessing mainly includes two parts: data cleaning and standardization processing. Data cleaning uses deep learning-based algorithms, such as DeepAutoencoder, to automatically detect and repair outliers and missing values to ensure the quality of the data. Standardization processing unifies data from different sources to the same scale and format. For example, the resolution of remote sensing images is unified to 10 meters, and the time interval of time series data is unified to monthly or annual to ensure the consistency and comparability of the data.

[0055] The feature extraction and modeling module 102 includes a time series analysis sub-module and a spatial feature extraction sub-module. The time series analysis sub-module extracts periodic, trend, and seasonal features by analyzing historical land use data. Specific time series analysis methods can use algorithms such as ARIMA (Autoregressive Integrated Moving Average Model) and LSTM (Long Short-Term Memory Network). The ARIMA model fits time series data through three parts: autoregression, differencing, and moving average, and the LSTM model uses a gating mechanism to handle the problem of information retention in long sequence data. The spatial feature extraction sub-module uses spatial clustering algorithms to identify spatial patterns and hotspots of land use change. Specific clustering algorithms can use K-means, DBSCAN (Density-Based Spatial Clustering of Applications with Noise), and SOM (Self-Organizing Map), etc. Taking the K-means algorithm as an example, its principle is to iteratively divide data points into K clusters, each cluster is represented by its centroid, and finally the distance between data points within each cluster is minimized. Spatial feature extraction also includes combining the extracted features with time features to construct a multi-dimensional land use change feature vector. Specific feature vector construction methods can use dimensionality reduction techniques such as PCA (Principal Component Analysis) or t-SNE (t-Distributed Stochastic Neighbor Embedding) to reduce high-dimensional features to low-dimensional features for easy model training and processing.

[0056] The prediction model module 103 is the core part of the system, including a machine learning model training sub-module and a model evaluation sub-module. The machine learning model training sub-module trains a prediction model based on the extracted spatio-temporal features. Specific machine learning algorithms include Random Forest, Support Vector Machine (SVM), and Deep Neural Network (DNN), etc. The Random Forest algorithm improves the accuracy and stability of the model by constructing multiple decision trees and integrating their prediction results. The Support Vector Machine algorithm classifies data points by finding the optimal hyperplane and is suitable for classification problems of high-dimensional data. The Deep Neural Network algorithm extracts deep features of data through the connection of multiple layers of neurons and is suitable for modeling complex non-linear relationships. The model evaluation sub-module evaluates the prediction performance of different models through cross-validation methods and selects the optimal model. Specific evaluation metrics include Mean Squared Error (MSE), Mean Absolute Error (MAE), and Coefficient of Determination (R 2 ) etc.

[0057] The formula for Mean Squared Error MSE is:

[0058]

[0059] where, y i is the true value, is the predicted value, and n is the number of samples. The formula for Mean Absolute Error MAE is:

[0060]

[0061] The formula for Coefficient of Determination R 2 is:

[0062]

[0063] where, is the average value of the true values. The classification performance of the model can be analyzed through a confusion matrix to identify the main reasons for prediction errors. Based on the evaluation results, the model parameters can be adjusted to optimize the model performance.

[0064] The model training sub-module further includes the construction of a feature selector and the training of multiple machine learning models. The feature selector improves the prediction accuracy of the model by screening the most predictive feature combinations. Common feature selection methods include Recursive Feature Elimination (RFE), LASSO (Least Absolute Shrinkage and Selection Operator), and Random Forest feature importance evaluation, etc. Taking LASSO as an example, its objective function is:

[0065]

[0066] Among them, y is the response variable, X is the feature matrix, β is the feature weight, and α is the regularization parameter. The output of the feature selector is used to train multiple machine learning models, and the prediction results of multiple models are integrated through model fusion technology to improve the accuracy and stability of prediction. Specific model fusion methods can include weighted average, voting mechanism, and stacking, etc.

[0067] The decision support module 104 includes a planning recommendation sub-module and a risk assessment sub-module. The planning recommendation sub-module generates land use planning recommendations under different scenarios based on the prediction results, including scenarios such as current situation optimization, ecological priority, and economic development. Specific generation methods can use multi-objective optimization algorithms, such as NSGA-II (Non-dominated Sorting Genetic Algorithm) and MOEA / D (Multi-Objective Optimization Algorithm / Decomposition), etc. Taking NSGA-II as an example, it screens the optimal solutions through non-dominated sorting and crowding distance calculation to ensure that the generated planning scheme achieves a balance among multiple objectives. Based on the prediction results, the planning recommendation sub-module can dynamically adjust the planning scheme to ensure the feasibility and sustainability of the plan. In addition, this sub-module also provides a visualization tool to display the expected effects of different planning schemes in a graphical form to users, helping decision-makers understand and select the optimal scheme.

[0068] The risk assessment sub-module calculates the environmental risk, economic risk, and social risk of each planning scheme using risk assessment algorithms. Common risk assessment methods include the Analytic Hierarchy Process (AHP) and the Fuzzy Comprehensive Evaluation (FCE), etc. Taking the Analytic Hierarchy Process as an example, its steps include constructing a hierarchical structure model, establishing a judgment matrix, calculating the weight vector, and conducting a consistency test. Suppose the hierarchical structure model includes an objective layer, a criterion layer, and a scheme layer. The criterion layer can include environmental impact, economic benefits, and social needs, etc. The relative importance of each pair of criteria is determined through expert scoring or historical data analysis to form a judgment matrix. Suppose the judgment matrix A is:

[0069]

[0070] Among them, a ijIt represents the importance ratio of criterion i relative to criterion j. The weight vector can be calculated using the eigenvector method or the geometric mean method. For example, the eigenvector method determines the weight vector by solving the eigenvector corresponding to the largest eigenvalue of matrix A. The fuzzy comprehensive evaluation method uses fuzzy mathematics to handle uncertain and imprecise data and evaluate the risks of each planning scheme. The specific steps include determining the evaluation index set, establishing a fuzzy relation matrix, and conducting a comprehensive evaluation, etc. Through risk assessment, a comprehensive evaluation report is generated to provide a scientific basis and risk warning for decision-makers.

[0071] The user interaction module 105 includes a data input sub-module and a result display sub-module. The data input sub-module receives the land use data and parameter settings input by the user. For example, the user can upload remote sensing images, GIS data, meteorological data, etc. through the graphical interface and set the parameters for model training, such as the regularization parameter α, the learning rate η, and the number of training epochs T, etc. The result display sub-module presents the prediction results and planning suggestions to the user in a graphical way, improving the user-friendliness and operability of the system. The specific display methods can include map visualization, time series charts, risk assessment reports, etc. Map visualization uses GIS technology to display the land use changes of different planning schemes on the map. The time series chart shows the changing trend of the prediction results over time, and the risk assessment report elaborates on the risks and countermeasures of each planning scheme in the form of text and charts.

[0072] The data update and maintenance module 106 includes a data update sub-module and a system maintenance sub-module. The data update sub-module updates the land use data regularly to ensure the timeliness of the data. The specific update cycle can be determined according to the type of data and the application scenario. For example, meteorological data can be updated hourly, and economic data can be updated quarterly. The system maintenance sub-module monitors the running status of the system, discovers and repairs system failures in a timely manner to ensure the stable operation of the system. The specific monitoring methods can include log recording, performance monitoring, and fault troubleshooting, etc. For example, the log recording module records the operation logs and error logs of the system. The performance monitoring module monitors indicators such as the running time and memory occupancy of the system in real time. The fault troubleshooting module will automatically trigger an alarm when an anomaly is detected and provide fault troubleshooting tools and methods.

[0073] The following further illustrates the implementation manner of the present invention through a specific application scenario. Suppose a certain urban planning department needs to predict the land use changes in the next 5 years and generate corresponding planning suggestions. First, through the data collection and preprocessing module 101, satellite remote sensing images, GIS data, meteorological data, and economic data of the city are collected. The remote sensing images can be obtained from the National Satellite Meteorological Center, the GIS data can be obtained from the Urban Planning Bureau, the meteorological data can be obtained from the Meteorological Bureau, and the economic data can be obtained from the statistical department. After data cleaning and standardization processing, a comprehensive land use data set is formed. Then, the feature extraction and modeling module 102 extracts the temporal features and spatial features of land use changes from the data set. The time series analysis sub-module uses the ARIMA model to extract the periodic and trend features of land use changes, and uses the LSTM model to extract seasonal features. The spatial feature extraction sub-module identifies the hot spots of land use changes through the DBSCAN algorithm, and uses the PCA technique to reduce the dimension of the extracted features to form a multi-dimensional feature vector.

[0074] The prediction model module 103 trains a prediction model based on the extracted feature vector. Specifically, the machine learning model training sub-module first selects the most predictive feature combination through the LASSO feature selection method, and then trains the random forest, support vector machine, and deep neural network models respectively based on the output of the feature selector. The prediction performance of different models is evaluated through the 10-fold cross-validation method, and the optimal model is selected. The selection criterion for the optimal model is to comprehensively consider the MSE, MAE, and R 2 indexes of the model, as well as the complexity and calculation time of the model. Suppose the MSE of the random forest model in the 10-fold cross-validation is 0.05, the MAE is 0.02, and the R 2 is 0.85, the MSE of the support vector machine model is 0.06, the MAE is 0.03, and the R 2 is 0.83, and the MSE of the deep neural network model is 0.04, the MAE is 0.01, and the R 2 is 0.87, then the deep neural network model is selected as the optimal model. Based on the optimal model, the prediction results of land use changes at different future time points are generated, such as predicting the land use changes in the next 1 year, 2 years, 3 years, 4 years, and 5 years.

[0075] The decision support module 104 generates land use planning suggestions under different scenarios based on the prediction results. Specifically, the planning suggestion sub-module uses the NSGA-II multi-objective optimization algorithm to comprehensively consider economic benefits, environmental impacts, and social needs, and generates the optimal planning scheme. Suppose the planning scheme includes three scenarios: current situation optimization, ecological priority, and economic development. The goal of the current situation optimization scheme is to maintain the current land use pattern, the goal of the ecological priority scheme is to protect the ecological environment, and the goal of the economic development scheme is to promote economic development. Through the analytic hierarchy process, a judgment matrix is established to calculate the weight vector of each criterion. For example, the weight of economic impact is 0.4, the weight of environmental impact is 0.3, and the weight of social needs is 0.3. Based on the weight vector, the comprehensive scores of different planning schemes are calculated to generate the optimal planning suggestions. At the same time, the risk assessment sub-module uses the AHP method and the FCE method to evaluate the environmental risk, economic risk, and social risk of each planning scheme. For example, environmental risk assessment can be carried out by calculating the impact degree of the scheme on water quality, air quality, and soil quality, economic risk assessment can be carried out by calculating the impact degree of the scheme on GDP, employment, and investment, and social risk assessment can be carried out by calculating the impact degree of the scheme on the quality of life of residents, social security, etc. Finally, a risk assessment report is generated to provide risk warnings and suggestions for countermeasures to decision-makers.

[0076] The user interaction module 105 receives the land use data and parameter settings input by the user through the graphical interface. For example, the user can upload the latest remote sensing images and meteorological data, and set the learning rate of the deep neural network model to 0.001, the regularization parameter to 0.1, and the number of training rounds to 1000. The result display sub-module displays the prediction results and planning suggestions to the user in a graphical form. For example, the map visualizes the land use changes in the next 5 years, the time series chart shows the prediction effects of different planning schemes, and the risk assessment report elaborates on the risks and countermeasures of each planning scheme in detail in the form of text and charts.

[0077] The data update and maintenance module 106 ensures the stability and timeliness of the system's data and running status. The data update sub-module regularly obtains the latest data from various data sources. For example, it obtains the latest meteorological data from the meteorological bureau every month and the latest economic data from the statistical department every quarter to ensure the timeliness of the data. The system maintenance sub-module monitors the running status of the system through methods such as log recording, performance monitoring, and fault troubleshooting. For example, the log recording module records the operation logs and error logs of the system, the performance monitoring module monitors the running time and memory occupancy of the system in real time, and the fault troubleshooting module triggers an alarm quickly when an abnormality is found and provides professional fault troubleshooting tools and methods, thus ensuring the stable operation and timely repair of the system.

[0078] Working principle: Through modules such as data acquisition and preprocessing, feature extraction and modeling, prediction model construction and evaluation, and decision support, multi-source data such as satellite remote sensing, GIS, meteorology, and economy are integrated to extract the spatio-temporal characteristics of land use changes, train prediction models and evaluate their performance, and ultimately provide a scientific basis for land use planning and decision-making.

[0079] An embodiment of the present application provides an electronic device, which is applicable to the above-mentioned land use change prediction and decision support system based on machine learning, and includes:

[0080] A memory for storing computer programs and data;

[0081] A processor for running system programs.

[0082] An embodiment of the present application provides a computer storage medium, which is applicable to the above-mentioned land use change prediction and decision support system based on machine learning, and performs hierarchical confidentiality management on the above system and data according to the requirements of confidentiality management.

[0083] Those skilled in the art should understand that the embodiments of the present application can be provided as a system or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0084] The present application is described with reference to the flowcharts and / or block diagrams of devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0085] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means realizes the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0086] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one Figure 1 process or multiple processes and / or blocks Figure 1 or multiple blocks.

[0087] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and memory.

[0088] The memory may include non-permanent memory in the form of computer-readable media, random access memory (RAM) and / or non-volatile memory such as read-only memory (ROM) or flash memory (flash RAM). The memory is an example of computer-readable media.

[0089] Computer-readable media includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette tapes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transitory media such as modulated data signals and carrier waves.

[0090] It should also be noted that the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or elements inherent to such process, article or device. Without further limitation, an element defined by the statement "comprising a..." does not exclude the presence of additional identical elements in the process, article or device comprising the element.

[0091] Embodiments of the present invention are provided for purposes of illustration and description. Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. A land use change prediction and decision support system based on machine learning, characterized in that: include: A data collection and preprocessing module (101) is used to collect multi-source land use data and clean, fuse and standardize the data; A feature extraction and modeling module (102), including a time series analysis submodule and a spatial feature extraction submodule; The time series analysis submodule is used to extract the temporal characteristics of land use changes from historical data; the spatial feature extraction submodule is used to extract the spatial characteristics of land use changes; A prediction model module (103), comprising a machine learning model training submodule and a model evaluation submodule; the machine learning model training submodule is used to train a prediction model based on the extracted spatiotemporal features; The model evaluation submodule is used to evaluate the accuracy and stability of the prediction model; A decision support module (104), including a planning suggestion submodule and a risk assessment submodule; The planning suggestion submodule is used to generate land use planning suggestions based on the prediction results; The risk assessment submodule is used to evaluate the risks and feasibility of different planning options.

2. The land use change prediction and decision support system based on machine learning according to claim 1 is characterized in that: The data acquisition and preprocessing module (101) is also used for: Through multi-source data fusion technology, satellite remote sensing data, geographic information system (GIS) data, meteorological data and economic data are integrated to form a comprehensive land use data set; Use deep learning-based data cleaning algorithms to automatically detect and repair outliers and missing values ​​in the data; Standardized processing methods are used to unify data from different sources into the same scale and format to ensure data consistency and comparability.

3. The land use change prediction and decision support system based on machine learning according to claim 1 is characterized in that: The feature extraction and modeling module (102) is also used for: Based on time series analysis methods, the periodicity, trend and seasonal characteristics of land use change are extracted; Using spatial clustering algorithms, identify spatial patterns and hotspots of land use change; Combining spatiotemporal characteristics, a multi-dimensional land use change feature vector is constructed.

4. The land use change prediction and decision support system based on machine learning according to claim 1 is characterized in that: The prediction model module (103) is also used for: Using a variety of machine learning algorithms, including random forests, support vector machines, and deep neural networks, to train predictive models for land use change; Through the cross-validation method, the prediction performance of different models is evaluated and the optimal model is selected; Based on the optimal model, the prediction results of land use change at different time points in the future are generated.

5. The land use change prediction and decision support system based on machine learning according to claim 4 is characterized in that: In the prediction model module (103), the model training submodule includes: According to the spatiotemporal characteristics of land use change, a feature selector is constructed to select the most predictive feature combination; Train multiple machine learning models based on the feature vectors output by the feature selector; Through model fusion technology, the prediction results of multiple models are integrated to improve the accuracy and stability of the prediction.

6. The land use change prediction and decision support system based on machine learning according to claim 4 is characterized in that: In the prediction model module (103), the model evaluation submodule includes: The prediction performance of the model was evaluated using mean square error, mean absolute error, and coefficient of determination indicators; Analyze the classification performance of the model through the confusion matrix and identify the main causes of prediction errors; Based on the evaluation results, adjust the model parameters to optimize the model performance.

7. The land use change prediction and decision support system based on machine learning according to claim 1 is characterized in that: The decision support module (104) is also used for: Based on the prediction results, generate land use planning recommendations under different scenarios, including current situation optimization, ecological priority and economic development scenarios; Using risk assessment algorithms, calculate the environmental, economic and social risks of each planning option; Combined with the risk assessment results, a comprehensive assessment report is generated to provide a scientific basis for decision makers.

8. The land use change prediction and decision support system based on machine learning according to claim 7 is characterized in that: In the decision support module (104), the planning suggestion submodule includes: Use multi-objective optimization algorithms to comprehensively consider economic benefits, environmental impacts and social needs to generate the best planning solution; Based on the forecast results, dynamically adjust the planning scheme to ensure the feasibility and sustainability of the plan; Provide visualization tools to show the expected effects of different planning schemes, helping decision makers understand and choose the best option.

9. The land use change prediction and decision support system based on machine learning according to claim 7, characterized in that: In the decision support module (104), the risk assessment submodule includes: Use the analytic hierarchy process and fuzzy comprehensive evaluation method to assess the multi-dimensional risks of different planning schemes; Based on historical data and expert knowledge, a risk assessment model is established to calculate the probability and impact of each risk; Generate risk assessment reports to provide decision makers with risk warnings and response measures.

10. The land use change prediction and decision support system based on machine learning according to claim 1, characterized in that: Also includes: A user interaction module (105), the user interaction module includes a data input submodule and a result display submodule; The data input submodule is used to receive land use data and parameter settings input by users; the result display submodule is used to display the prediction results and planning suggestions to users in a graphical way to improve the user-friendliness and operability of the system; The data updating and maintenance module (106) includes a data updating submodule and a system maintenance submodule; the data updating submodule is used to regularly update the land use data to ensure the timeliness of the data; the system maintenance submodule is used to monitor the operating status of the system, timely discover and repair system faults, and ensure the stable operation of the system.

Images