Saline-alkali soil improvement control analysis method and system based on digital twinning

Abstract

The invention discloses a saline-alkali soil improvement control analysis method and system based on digital twinning, and belongs to the technical field of saline-alkali soil improvement control analysis, and the method comprises the steps: firstly obtaining and preprocessing the terrain, soil permeability and precipitation data of saline-alkali soil, predicting a retention water accumulation region through employing a support vector machine algorithm, and estimating the amount of retained water by using a Kriging interpolation method. Then, the dynamic change process of the retained water is simulated through an artificial neural network algorithm, the information is integrated into a digital twinborn model, and visualization of spatial distribution and dynamic change of the retained water is achieved. And finally, on the basis of retained water information in the digital twin model, optimizing the layout and depth of the drainage ditch, selecting proper drainage materials and modes, and forming a complete drainage design scheme. According to the method, accurate prediction, dynamic simulation and visual analysis of the retention water of the saline-alkali soil are realized through a method of combining data driving and an intelligent algorithm, and a scientific basis and decision support are provided for improvement of the saline-alkali soil and agricultural production.

Description

Technical Field

[0001] The present invention belongs to the technical field of saline-alkali land improvement control analysis, and in particular, to a saline-alkali land improvement control analysis method and system based on digital twins. Background Art

[0002] In the process of using digital twin technology to dynamically simulate the retained water in saline-alkali land, many technical difficulties and contradictions are faced. First, the saline-alkali land has complex and changeable terrain and undulating micro-topography, which makes the collection and processing of terrain data extremely difficult. Secondly, the soil properties of saline-alkali land are unique, the permeability varies greatly, and the infiltration capacity and rate in different regions vary significantly, which brings challenges to the acquisition and analysis of soil permeability data. Furthermore, the precipitation in saline-alkali land is unevenly distributed in time and space, and the precipitation varies greatly. The accurate collection and dynamic update of precipitation data face many technical bottlenecks. At the same time, the dynamic change process of retained water involves the complex interaction of multiple factors such as terrain, soil, and precipitation. To build a digital twin model that can accurately reflect this process, it is necessary to have a deep understanding of the internal connection and action mechanism between the various factors, and to establish a reasonable mathematical model and algorithm. In addition, the computational efficiency, stability, and visualization of the model also put forward higher requirements for digital twin technology. Summary of the invention

[0003] In order to solve the above technical problems, the present invention proposes a saline-alkali land improvement control and analysis method and system based on digital twin to solve the problems existing in the above-mentioned prior art.

[0004] To achieve the above object, the present invention provides a saline-alkali land improvement control and analysis method based on digital twin, comprising:

[0005] Obtain topographic data, soil permeability data and precipitation data of saline-alkali land;

[0006] The acquired data were preprocessed to obtain normalized topography, soil permeability and precipitation datasets;

[0007] Constructing a retained water accumulation area prediction model through a support vector machine model, and calculating the normalized terrain, soil permeability and precipitation data set through the retained water accumulation area prediction model to obtain the retained water accumulation area;

[0008] The amount of retained water in the accumulated area of ​​the retained water is estimated based on the Kriging interpolation method to obtain a spatial distribution map of the amount of retained water;

[0009] The accumulation area of ​​the retained water and the spatial distribution map of the retained water volume are simulated by an artificial neural network algorithm to obtain a simulated and predicted dynamic change process of the retained water;

[0010] Integrate the accumulation area of ​​the retained water, the spatial distribution map of the retained water volume, and the simulated predicted dynamic change process of the retained water into the digital twin model to generate a digital twin model of retained water in saline-alkali land;

[0011] A complete drainage design plan is generated based on the digital twin model of the retained water in the saline-alkali land.

[0012] Preferably, the process of obtaining the topographic data, soil permeability data and precipitation data of the saline-alkali land includes:

[0013] Obtain elevation and slope terrain parameter data of saline-alkali land areas, obtain soil texture, soil bulk density, soil porosity soil parameter data of saline-alkali land areas, and obtain daily and monthly precipitation meteorological data of saline-alkali land areas;

[0014] Processing the terrain parameter data using a digital elevation model to extract terrain features and obtain the terrain data;

[0015] Based on the soil moisture characteristic curve model, calculating the soil permeability coefficient corresponding to the soil parameter data to obtain the soil permeability data;

[0016] The multi-year average precipitation of the meteorological data is calculated to obtain precipitation data.

[0017] Preferably, the process of preprocessing the acquired data to obtain a normalized terrain, soil permeability and precipitation data set includes: preprocessing the acquired data, removing outliers and noise data, and normalizing the data to obtain a normalized terrain, soil permeability and precipitation data set.

[0018] Preferably, the process of constructing a prediction model for the accumulation area of ​​retained water by using a support vector machine model comprises:

[0019] Build a support vector machine model and construct an input data set for the support vector machine model;

[0020] Optimizing the parameters of the support vector machine model to obtain an optimized model;

[0021] The optimization model is trained using the input data set to generate the retained water accumulation area prediction model.

[0022] Preferably, the parameters of the support vector machine model are optimized to obtain an optimized model, which includes:

[0023] A grid search method is used to optimize the hyperparameters of the support vector machine model, the model performance is evaluated through a cross-validation method, the model parameters with the highest prediction accuracy are selected, and the support vector machine model is optimized based on the model parameters with the highest accuracy to obtain an optimized model.

[0024] Preferably, the process of estimating the amount of retained water in the accumulation area of ​​the retained water based on the Kriging interpolation method to obtain a spatial distribution map of the amount of retained water includes:

[0025] Determining the spatial distribution of sampling points based on the accumulation area of ​​the retained water, wherein the sampling points cover the entire accumulation area;

[0026] Obtain the geographic coordinate information and measured retained water volume data of each sampling point to establish a sampling point data set;

[0027] Analyze the sampling point data set, determine the distribution characteristics, trends and outliers of the data, and generate analysis results;

[0028] Preprocessing the sampling point data set based on the analysis result to generate a preprocessed sampling point data set;

[0029] Constructing a Kriging interpolation model based on the preprocessed sampling point data set;

[0030] Optimizing the variogram and interpolation parameters of the Kriging interpolation model through cross-validation to generate an optimized Kriging interpolation model;

[0031] Based on the optimized Kriging interpolation model, the spatial interpolation estimation of the entire retained water accumulation area is carried out to obtain the predicted value of retained water volume at each location in the area;

[0032] A spatial distribution map of the retained water volume is obtained based on the predicted value of the retained water volume at each location in the area.

[0033] Preferably, the process of obtaining the simulated predicted dynamic change process of retained water comprises:

[0034] Obtain historical retained water detection datasets and build a recurrent neural network model;

[0035] Preprocessing the historical retained water detection data set, extracting key features, and generating a training set and a test set;

[0036] Inputting the training set into the recurrent neural network model, training the recurrent neural network model through forward propagation and back propagation algorithms, and obtaining a trained recurrent neural network model;

[0037] Inputting the test set into the trained recurrent neural network model to obtain a test result; and calculating the prediction accuracy of the trained recurrent neural network model according to the test result;

[0038] Adjusting the structure and parameters of the trained recurrent neural network model based on the test accuracy to generate a retained water prediction model;

[0039] The accumulated area of ​​the retained water and the spatial distribution diagram of the retained water volume are simulated based on the retained water prediction model to generate the simulated predicted dynamic change process of the retained water.

[0040] Preferably, the process of generating a digital twin model of saline-alkali land retention water includes:

[0041] Obtain remote sensing image data and geographic information data of saline-alkali land retention water;

[0042] Performing image segmentation processing on the remote sensing image data to obtain vector boundaries of the retained water accumulation area;

[0043] A digital twin model of the retained water in the saline-alkali land is constructed based on the vector boundary of the retained water accumulation area and the simulated and predicted dynamic change process of the retained water.

[0044] Preferably, the process of generating a complete drainage design scheme based on the digital twin model of saline-alkali land retention water includes:

[0045] Based on the digital twin model of the retained water in the saline-alkali land, information on the accumulated area and amount of retained water is obtained;

[0046] Optimizing drainage ditch layout and depth parameters according to the topographical features and water distribution of the accumulation area;

[0047] Adopt hydraulic simulation algorithm and use hydraulic model to simulate and analyze various drainage methods to obtain drainage effect and efficiency;

[0048] The complete drainage design scheme is constructed based on the drainage ditch layout and depth parameters, as well as the drainage effect and efficiency.

[0049] To achieve the above purpose, the present invention also provides a saline-alkali land improvement control and analysis system based on digital twin, comprising:

[0050] The data acquisition and preprocessing module is used to acquire the terrain data, soil permeability data and precipitation data of saline-alkali land, and preprocess the acquired data to obtain standardized terrain, soil permeability and precipitation data sets;

[0051] A retained water accumulation area prediction module is used to construct a retained water accumulation area prediction model through a support vector machine model, and calculate the normalized terrain, soil permeability and precipitation data set through the retained water accumulation area prediction model to obtain the retained water accumulation area;

[0052] A retained water volume estimation module is used to estimate the volume of retained water in the accumulated area of ​​the retained water based on the Kriging interpolation method to obtain a spatial distribution map of the retained water volume;

[0053] A retained water dynamic change simulation module is used to simulate the accumulation area of ​​the retained water and the spatial distribution map of the retained water volume through an artificial neural network algorithm to obtain a simulated and predicted retained water dynamic change process;

[0054] A digital twin model construction module is used to integrate the accumulation area of ​​the retained water, the spatial distribution map of the retained water volume, and the simulated predicted dynamic change process of the retained water into the digital twin model to generate a digital twin model of the retained water in the saline-alkali land;

[0055] The drainage design scheme formulation module is used to generate a complete drainage design scheme based on the digital twin model of the saline-alkali land retained water.

[0056] Compared with the prior art, the present invention has the following advantages and technical effects:

[0057] The present invention discloses a method for constructing a digital twin model of retained water in saline-alkali land and a drainage design method. The method first obtains and preprocesses the terrain, soil permeability and precipitation data of the saline-alkali land, uses a support vector machine algorithm to predict the retained water accumulation area, and uses the Kriging interpolation method to estimate the amount of retained water. Subsequently, the dynamic change process of the retained water is simulated by an artificial neural network algorithm, and this information is integrated into the digital twin model to realize the visualization of the spatial distribution and dynamic changes of the retained water. Finally, based on the retained water information in the digital twin model, the layout and depth of the drainage ditch are optimized, and appropriate drainage materials and methods are selected to form a complete drainage design scheme. The present invention realizes the accurate prediction, dynamic simulation and visual analysis of retained water in saline-alkali land through a method combining data-driven and intelligent algorithms, providing a scientific basis and decision-making support for saline-alkali land improvement and agricultural production. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0059] Figure 1 This is a flow chart of a saline-alkali land improvement control and analysis method based on digital twins according to an embodiment of the present invention;

[0060] Figure 2 A schematic diagram of a model training test according to an embodiment of the present invention;

[0061] Figure 3 A schematic diagram of a visualization process of saline-alkali land retention water according to an embodiment of the present invention;

[0062] Figure 4 This is a schematic diagram of the structure of a saline-alkali land improvement control and analysis system based on digital twins according to an embodiment of the present invention. DETAILED DESCRIPTION

[0063] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0064] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0065] Embodiment 1

[0066] like Figure 1-3 As shown, this embodiment provides a saline-alkali land improvement control analysis method based on digital twins, including:

[0067] Step S101, obtaining terrain data, soil permeability data and precipitation data of saline-alkali land, preprocessing the obtained data, removing outliers and noise data, and standardizing the data to obtain standardized terrain, soil permeability and precipitation data sets.

[0068] Obtain elevation and slope terrain parameter data of saline-alkali land areas, obtain soil texture, soil bulk density, soil porosity soil parameter data of saline-alkali land areas, and obtain daily and monthly precipitation meteorological data of saline-alkali land areas;

[0069] The terrain parameter data is processed by using a digital elevation model to extract terrain features and obtain terrain feature data; based on a soil moisture characteristic curve model, the soil permeability coefficient corresponding to the soil parameter data is calculated to obtain soil permeability data; the multi-year average precipitation of the meteorological data is counted to obtain precipitation data;

[0070] Performing exploratory data analysis on the terrain feature data, the soil permeability data and the precipitation data, identifying outliers using a box plot, and removing noise data using a smoothing filter to obtain optimized terrain feature data, optimized soil permeability data and optimized precipitation data;

[0071] The optimized terrain feature data, the optimized soil permeability data and the optimized precipitation data are standardized by using a minimum-maximum standardization method, the data are scaled to within an interval, the dimension effect is eliminated, and standardized terrain feature data, standardized soil permeability data and standardized precipitation data are obtained;

[0072] Performing attribute selection on the standardized terrain feature data, the standardized soil permeability data, and the standardized precipitation data, removing redundant attributes, and reducing data dimensions to obtain reduced-dimensional terrain feature data, reduced-dimensional soil permeability data, and reduced-dimensional precipitation data;

[0073] Randomly dividing the terrain feature data after dimensionality reduction, the soil permeability data after dimensionality reduction, and the precipitation data after dimensionality reduction into a training set and a test set;

[0074] The training set is trained using a support vector machine algorithm to obtain a saline-alkali land classification prediction model;

[0075] The test set is input into the saline-alkali land classification prediction model to predict the type of saline-alkali land in an unknown area.

[0076] To further optimize the scheme, in order to obtain the elevation and slope terrain parameter data of the saline-alkali land area, a high-precision GPS measuring instrument can be used for field measurement to obtain the elevation and slope point data in the area. At the same time, soil samples in the saline-alkali land area are collected using a soil sampler, and soil parameter data such as soil texture, soil bulk density, and soil porosity are obtained through indoor soil physical property measurement experiments. In addition, meteorological data such as daily precipitation and monthly precipitation in the saline-alkali land area are obtained through long-term observation data of meteorological stations. The acquired terrain parameter data are processed using a digital elevation model, and terrain feature extraction algorithms such as terrain relief index and terrain moisture index are used to extract terrain features to obtain terrain feature data. Based on soil moisture characteristic curve models such as the van Genuchten model, the soil permeability coefficient corresponding to the soil parameter data is calculated to obtain soil permeability data. Statistics of meteorological data for many years are used to calculate the average precipitation for many years to obtain precipitation data. Exploratory data analysis is performed on terrain feature data, soil permeability data, and precipitation data, and box plots are used to identify outliers, such as data that exceeds 5 times the upper and lower quartiles.

[0077] Step S102, based on the pre-processed terrain, soil permeability and precipitation data, the support vector machine algorithm is used to predict the accumulation area of ​​retained water. By training the existing retained water accumulation area data, a retained water accumulation area prediction model is established. If the prediction accuracy reaches a preset threshold, the model is applied to new saline-alkali land data to predict the accumulation area of ​​retained water.

[0078] Obtain the terrain, soil permeability and precipitation data of the target saline-alkali area, pre-process the data, and unify the format and unit of the data; extract feature vectors based on the pre-processed terrain, soil permeability and precipitation data, and construct an input data set for a support vector machine model; use a grid search method to optimize the hyperparameters of the support vector machine model, evaluate the model performance through a cross-validation method, and select the model parameters with the highest prediction accuracy; use the pre-acquired retained water accumulation area data to train the support vector machine model to obtain a retained water accumulation area prediction model; use the trained support vector machine model to predict the terrain, soil permeability and precipitation data of the target saline-alkali area to obtain a retained water accumulation area distribution map;

[0079] To further optimize the scheme, in order to obtain the topography, soil permeability and precipitation data of the target saline-alkali area, the target area can be scanned by a drone equipped with a high-precision laser radar to obtain high-precision digital elevation model (DEM) data with a spatial resolution of up to 5 meters. At the same time, soil samples in the target area are collected using a soil sampler, and parameters such as soil permeability coefficient are obtained through indoor soil permeability test experiments. The spatial distribution density of sampling points is 1 per square kilometer. In addition, daily precipitation data are obtained by collecting precipitation observation data from meteorological stations near the target area, with a time span of 10 years. The obtained DEM data, soil permeability data and precipitation data are preprocessed, and the coordinate system is unified to WGS84, the unit of soil permeability coefficient is unified to millimeter / hour, and the unit of precipitation is unified to millimeter. In ArcGIS software, the terrain features of DEM data, including elevation, slope, slope aspect, convexity, etc., are extracted using hydrological analysis tools to generate terrain feature raster data. The soil permeability sample data is interpolated into raster data using the Kriging interpolation method, and the spatial resolution is consistent with the DEM. The daily precipitation data was statistically analyzed, and the annual precipitation was calculated to obtain the annual precipitation raster data. The terrain feature raster data, soil permeability raster data and annual precipitation raster data were raster superimposed, and the feature vector of each raster unit was extracted to construct the input data set of the support vector machine model. The grid search method was used to optimize the hyperparameters of the support vector machine model, including the penalty coefficient C, the kernel function type, etc. The model performance was evaluated by the 5-fold cross-validation method, and the model parameters with the highest prediction accuracy were selected. The sample data of the retained water accumulation area interpreted by remote sensing were used to extract the corresponding feature vectors, and the support vector machine model was trained to obtain the retained water accumulation area prediction model. The feature vector of the target saline-alkali area was input into the prediction model to obtain the retained water accumulation area distribution map, and the spatial resolution of the distribution map was 1 meter. The prediction accuracy of the model was evaluated by comparing the prediction results with the actual survey data. If the accuracy rate reached more than 90%, the model was applied to the prediction of retained water accumulation areas in other saline-alkali areas. According to the predicted distribution map of the area of ​​retained water accumulation, artificial drainage ditches are planned to be built in areas with serious water accumulation and poor drainage. The depth of the ditches should be no less than 1 meter and the spacing should be no more than 100 meters. Around the area of ​​retained water accumulation, salt-tolerant plants such as Suaeda salsa and Paspalum marissima are planted with a planting density of no less than 10 plants per square meter to reduce the accumulation of retained water and promote the leaching of soil salt.

[0080] Step S103, in the predicted retained water accumulation area, the Kriging interpolation method is used to estimate the amount of retained water, by selecting several sampling points in the accumulation area, obtaining the retained water volume data of these sampling points, and then using the Kriging interpolation method to perform spatial interpolation estimation on the retained water volume of the entire accumulation area to obtain a spatial distribution map of the retained water volume.

[0081] The predicted retained water accumulation area is obtained, and the spatial distribution of sampling points is determined according to the accumulation area, wherein the sampling points cover the entire accumulation area and have a certain degree of representativeness; the geographical coordinate information and the measured retained water volume data of each sampling point are obtained to establish a sampling point data set; an exploratory analysis is performed on the sampling point data set to determine the distribution characteristics, trends and outliers of the data, and the sampling point data set is preprocessed according to the analysis results, wherein the preprocessing includes removing outliers and normalizing the data; a Kriging interpolation model is established according to the preprocessed sampling point data set, and the variogram and interpolation parameters of the Kriging interpolation model are optimized through cross-validation; the optimized Kriging interpolation model is used to perform spatial interpolation estimation on the entire retained water accumulation area to obtain the predicted value of the retained water volume at each location in the area; a spatial distribution map of the retained water volume is generated according to the interpolation estimation results, wherein the spatial distribution map includes a contour map and a three-dimensional surface map; error analysis and uncertainty assessment are performed on the interpolation estimation results, and the root mean square error and Kriging variance of the interpolation estimation results are calculated to determine the reliability and applicability of the interpolation estimation results.

[0082] To further optimize the scheme, a systematic sampling method was used to determine the spatial distribution of sampling points in the predicted retained water accumulation area. The sampling point spacing was 50 meters, and a total of 120 sampling points were set up. The geographical coordinates of each sampling point were measured by GPS, and the measured retained water volume at each point was measured by a portable water level meter to obtain the sampling point data set. The box plot analysis of the sampling point data showed that the data was right-skewed and there were 3 outliers. After removing the outliers, the data was normalized to the maximum and minimum values. Based on the preprocessed sampling point data, a Kriging interpolation model was established. The optimal variogram was determined to be a Gaussian model through 5-fold cross validation, and the interpolation search radius was 100 meters. Using the optimized Kriging interpolation model, spatial interpolation was performed on the retained water accumulation area to estimate the retained water volume of each grid unit, with a grid size of 10 meters. According to the interpolation results, the retained water volume contour map and three-dimensional surface map were drawn to intuitively display the spatial distribution characteristics of the retained water volume. Through leave-one-out cross-validation, the root mean square error of the interpolation result is 15, and the Kriging variance is 12, indicating that the interpolation result has high reliability and applicability. In the interpolation process, the octree algorithm is used to optimize the storage and calculation of the Kriging matrix to improve the interpolation efficiency. The three-dimensional surface map is illuminated and the transparency is set to enhance the three-dimensional sense and realism of the image. Comprehensive analysis of the interpolation results and spatial distribution map shows that the area with the largest amount of retained water is located in the northwest of the study area. The area is low-lying and the soil permeability is poor. It is a high-incidence area for retained water accumulation and needs to be focused on.

[0083] Step S104, based on the predicted retained water accumulation area and the estimated retained water volume, an artificial neural network algorithm is used to simulate the dynamic change process of retained water. By constructing a neural network model that includes factors such as topography, soil permeability, and precipitation, and using historical retained water dynamic change data to train the model, a simulation prediction of the dynamic change process of retained water in the future can be achieved.

[0084] Obtain data related to the topography, soil permeability, and precipitation of the target area as input features of the neural network model;

[0085] Inputting the relevant data into a pre-trained neural network model to obtain a prediction result of the dynamic change of retained water in the target area within a period of time in the future;

[0086] If the prediction results show that the retained water is accumulating, it is determined that drainage and diversion measures must be taken.

[0087] To further optimize the scheme, in order to predict the dynamic changes of future retained water in the target area, the elevation, slope, soil texture, precipitation and other data of the area were first collected and sorted into a raster data set with a resolution of 50 meters. The data set was randomly divided into training set, validation set and test set with a ratio of 7:2:1. A three-layer BP neural network was used to construct the prediction model. The input layer included 12 features such as elevation, slope, soil texture, and precipitation in the previous three days. The hidden layer contained 24 neurons, and the output layer was the amount of retained water in the next seven days. The model used the Adam optimizer, the initial learning rate was 01, and the loss function was the mean square error. After training 500 epochs, the mean absolute error on the test set was 15 mm. The precipitation forecast data for the next seven days in the area was input into the trained neural network model to predict the trend of retained water changes in the next week. The results showed that in the event of continuous heavy rainfall, the amount of retained water in the low-lying areas in the southwest of the region would reach a peak on the third day, and the maximum water depth might exceed 200 mm, so drainage measures needed to be taken in advance. Using GIS spatial analysis functions, waterlogging risk areas can be automatically identified and risk level maps can be generated. Differentiated drainage plans can be developed based on the prediction results and risk distribution, with infiltration wells and drainage ditches being deployed in high-risk areas first, and waterlogging dynamics monitored in real time. By combining neural network prediction with GIS spatial analysis, the risk of retained water can be accurately predicted and precise governance can be guided.

[0088] Artificial neural network algorithm is used to simulate the dynamic change process of retained water.

[0089] The method comprises the following steps: obtaining a historical monitoring data set of retained water, wherein the historical monitoring data set includes water level data and water volume data; preprocessing the historical monitoring data set, removing outliers and noise, extracting key features, and obtaining a training sample set and a test sample set; constructing a recurrent neural network model according to the dynamic change characteristics of retained water, wherein the recurrent neural network model includes an input layer, at least one hidden layer and an output layer; determining the number of neurons in the input layer, the hidden layer and the output layer, and initializing the parameters of the recurrent neural network model; inputting the training sample set into the recurrent neural network model, training the recurrent neural network model through forward propagation and back propagation algorithms, and obtaining a trained recurrent neural network model; in the training process, adopting an early stopping method to avoid overfitting of the recurrent neural network model; inputting the test sample set into the trained recurrent neural network model, and obtaining a test result; according to the test result, calculating the prediction accuracy of the trained recurrent neural network model, and if the prediction accuracy does not reach a preset threshold, adjusting the structure and parameters of the recurrent neural network model, and re-executing the training step until the prediction accuracy reaches the preset threshold, and obtaining a retained water prediction model.

[0090] To further optimize the scheme, in order to predict the dynamic changes of retained water in a certain area in the next week, the water level and water volume monitoring data of the area in the past three months were first collected, totaling 1000 records. The original data was cleaned, 5% of the outliers were removed, and 10 key features such as the water level change rate and the water volume change rate were extracted. The processed data was randomly divided into a training set and a test set in a ratio of 8:2. Considering that the dynamic changes of retained water have time series dependence, a long short-term memory (LSTM) recurrent neural network model was chosen to be constructed. The input layer contains 10 neurons, corresponding to the 10 extracted features; the hidden layer contains two LSTM layers, each containing 64 neurons; the output layer contains 7 neurons, which represent the predicted water volume for each day in the next week. The model parameters are initialized using the Xavier method, the optimizer uses Adam, and the learning rate is set to 001. After each epoch, the model performance is evaluated on the validation set. If the validation loss does not decrease for 5 consecutive epochs, the early stopping mechanism is triggered. After the model was trained for 100 epochs, the average absolute error on the test set was 1 mm, meeting the preset threshold of 8 mm. Finally, the trained LSTM model was used to predict the trend of retained water volume in the next 7 days in combination with the precipitation forecast data for the next week. To further improve the prediction accuracy, more factors affecting retained water, such as soil infiltration capacity and evaporation intensity, can be considered to be introduced, the model structure can be optimized, and the amount of training data can be increased. At the same time, the model is regularly updated online using newly collected monitoring data to adapt to potential changes in the dynamic changes of retained water.

[0091] Step S105, integrate the predicted retained water accumulation area, the estimated retained water volume and the simulated retained water dynamic change process into the digital twin model, and generate a digital twin model of retained water in saline-alkali land through three-dimensional visualization technology to achieve intuitive display and analysis of the spatial distribution and dynamic changes of retained water.

[0092] Obtain remote sensing image data and geographic information data of saline-alkali land retention water;

[0093] Performing image segmentation processing on the remote sensing image data to obtain vector boundaries of the retained water accumulation area;

[0094] Calculating the storage capacity of the retained water by a water volume estimation model according to the area and water depth information of the retained water accumulation area;

[0095] The vector boundary of the retained water accumulation area and the water volume estimation result are integrated into the digital twin platform to construct a three-dimensional visualization scene of retained water in saline-alkali land.

[0096] To further optimize the scheme, a multispectral camera equipped with a drone was used to take aerial photos of the target saline-alkali land, obtain high-definition images with a resolution of 1 meter, and record the geographic coordinates of the images through the RTK differential positioning system. At the same time, the airborne laser radar was used to scan the ground to obtain high-precision three-dimensional point cloud data. The multispectral image and point cloud data were aligned to build a true three-dimensional model. The image was processed based on the U-Net semantic segmentation algorithm, the pixel boundaries of the retained water area were extracted, and vectorized into polygonal features. The vector boundaries and DEM elevation data were superimposed on the GIS platform, and the area and average water depth of the retained water were obtained through raster operations. Assuming that the retained water area is 15 hectares and the average water depth is 3 meters, the retained water storage capacity is about 37,500 cubic meters. The retained water vector boundary was stretched into a solid using three-dimensional modeling software, and a translucent blue material was given, which was then superimposed on the three-dimensional surface model to generate a realistic retained water landscape. The flooding and evaporation processes of the retained water were demonstrated through the time-series animation simulation in the digital twin platform. The system can update the twin scene in real time based on the dynamics of retained water monitored by remote sensing, providing an intuitive digital management method for saline-alkali land management.

[0097] Step S106: Determine the drainage design plan based on the retained water information in the digital twin model, analyze the accumulation area and water volume of the retained water, optimize the layout and depth of the drainage ditch, and select appropriate drainage materials and drainage methods to form a complete drainage design plan.

[0098] A digital twin model including information on the area of ​​retained water accumulation and water volume is obtained; the drainage ditch layout and depth parameters are optimized according to the topographical features and water volume distribution of the accumulation area; a hydraulic model including a preset number of types of drainage methods is constructed; a hydraulic simulation algorithm is adopted to simulate and analyze each drainage method using the hydraulic model to obtain drainage effect and efficiency; if the preset drainage requirements are met, the drainage method is determined; if not, the optimized drainage ditch layout and depth parameters are returned; the optimized drainage ditch layout, depth parameters, drainage materials and drainage methods are integrated to obtain a drainage design plan.

[0099] To further optimize the scheme, based on the digital twin model of retained water constructed in the previous step, the vector boundary and grid water depth data of the retained water accumulation area are extracted. Using the hydrological analysis module of GIS, based on the DEM terrain data and with the accumulation area boundary as a constraint, hydrological elements such as flow direction, flow accumulation, and confluence path are generated. Combined with the heat map of retained water distribution, areas with severe water accumulation and poor drainage are identified to optimize the location of drainage ditches. The reasonable depth of the drainage ditch is calculated through indicators such as terrain undulation and slope, and the geometric parameters of the drainage ditch are preliminarily determined. In the digital twin platform, a three-dimensional solid model of the drainage ditch is constructed, and three drainage methods such as open ditch, hidden ditch, and pipeline are set. A hydraulic model is established using computational fluid dynamics (CFD) software such as ANSYS Fluent. The geometric parameters of the drainage ditch, the amount of retained water, the soil permeability coefficient, etc. are input, and the volume offluid (VOF) method is used to simulate the free liquid surface to analyze the distribution of water velocity field and pressure field under different drainage methods. The maximum drainage flow of the open ditch, hidden ditch, and pipeline is 5m 3 / s、2m 3 / s、8m 3 / s, and the emptying time is 6h, 8h, and 12h respectively. If the requirement of 80% of the retained water is satisfied within 24h, the corresponding drainage method is selected, otherwise the optimized drainage ditch parameters are returned. Finally, in the digital twin scene, the 3D model of the drainage ditch is superimposed, and the optimized drainage ditch parameter table, drainage material list, and CFD simulation results are associated to generate a complete retained water drainage design plan, which intuitively presents the drainage layout and dynamic drainage effect.

[0100] like Figure 4 As shown, this embodiment also provides a saline-alkali land improvement control and analysis system based on digital twins, which mainly includes:

[0101] Data acquisition and preprocessing module, used to acquire and process saline-alkali land related data;

[0102] A retained water accumulation area prediction module is used to predict the retained water accumulation area;

[0103] A module for estimating the amount of retained water, used to estimate the amount of retained water in the accumulation area;

[0104] The dynamic change simulation module of retained water is used to simulate the dynamic change process of retained water;

[0105] A digital twin model building module for generating a digital twin model of saline-alkali land retention water;

[0106] The drainage design scheme formulation module is used to determine the appropriate drainage design scheme.

[0107] The above are only preferred specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A saline-alkali land improvement control and analysis method based on digital twin, characterized in that: The following steps are involved: Obtain topographic data, soil permeability data and precipitation data of saline-alkali land; The acquired data were preprocessed to obtain normalized topography, soil permeability and precipitation datasets; Constructing a retained water accumulation area prediction model through a support vector machine model, and calculating the normalized terrain, soil permeability and precipitation data set through the retained water accumulation area prediction model to obtain the retained water accumulation area; The amount of retained water in the accumulated area of ​​the retained water is estimated based on the Kriging interpolation method to obtain a spatial distribution map of the amount of retained water; The accumulation area of ​​the retained water and the spatial distribution map of the retained water volume are simulated by an artificial neural network algorithm to obtain a simulated and predicted dynamic change process of the retained water; Integrate the accumulation area of ​​the retained water, the spatial distribution map of the retained water volume, and the simulated predicted dynamic change process of the retained water into the digital twin model to generate a digital twin model of retained water in saline-alkali land; A complete drainage design plan is generated based on the digital twin model of the retained water in the saline-alkali land.

2. The digital twin-based saline-alkali land improvement control and analysis method according to claim 1 is characterized in that: The process of obtaining topographic data, soil permeability data and precipitation data of saline-alkali land includes: Obtain elevation and slope terrain parameter data of saline-alkali land areas, obtain soil texture, soil bulk density, soil porosity soil parameter data of saline-alkali land areas, and obtain daily and monthly precipitation meteorological data of saline-alkali land areas; Processing the terrain parameter data using a digital elevation model to extract terrain features and obtain the terrain data; Based on the soil moisture characteristic curve model, calculating the soil permeability coefficient corresponding to the soil parameter data to obtain the soil permeability data; The multi-year average precipitation of the meteorological data is calculated to obtain precipitation data.

3. The digital twin-based saline-alkali land improvement control and analysis method according to claim 1 is characterized in that: The process of preprocessing the acquired data to obtain a normalized terrain, soil permeability and precipitation data set includes: preprocessing the acquired data, removing outliers and noise data, and normalizing the data to obtain a normalized terrain, soil permeability and precipitation data set.

4. The digital twin-based saline-alkali land improvement control and analysis method according to claim 1 is characterized in that: The process of constructing a prediction model for the accumulation area of ​​retained water by using a support vector machine model includes: Build a support vector machine model and construct an input data set for the support vector machine model; Optimizing the parameters of the support vector machine model to obtain an optimized model; The optimization model is trained using the input data set to generate the retained water accumulation area prediction model.

5. The digital twin-based saline-alkali land improvement control and analysis method according to claim 4 is characterized in that: The process of optimizing the parameters of the support vector machine model to obtain the optimized model includes: A grid search method is used to optimize the hyperparameters of the support vector machine model, the model performance is evaluated through a cross-validation method, the model parameters with the highest prediction accuracy are selected, and the support vector machine model is optimized based on the model parameters with the highest accuracy to obtain an optimized model.

6. The digital twin-based saline-alkali land improvement control and analysis method according to claim 1 is characterized in that: The process of estimating the amount of retained water in the accumulated area of ​​retained water based on the Kriging interpolation method and obtaining a spatial distribution map of the amount of retained water includes: Determining the spatial distribution of sampling points based on the accumulation area of ​​the retained water, wherein the sampling points cover the entire accumulation area; Obtain the geographic coordinate information and measured retained water volume data of each sampling point to establish a sampling point data set; Analyze the sampling point data set, determine the distribution characteristics, trends and outliers of the data, and generate analysis results; Preprocessing the sampling point data set based on the analysis result to generate a preprocessed sampling point data set; Constructing a Kriging interpolation model based on the preprocessed sampling point data set; Optimizing the variogram and interpolation parameters of the Kriging interpolation model through cross-validation to generate an optimized Kriging interpolation model; Based on the optimized Kriging interpolation model, the spatial interpolation estimation of the entire retained water accumulation area is carried out to obtain the predicted value of retained water volume at each location in the area; A spatial distribution map of the retained water volume is obtained based on the predicted value of the retained water volume at each location in the area.

7. The digital twin-based saline-alkali land improvement control and analysis method according to claim 1 is characterized in that: The process of obtaining the simulated predicted dynamic change process of retained water includes: Obtain historical retained water detection datasets and build a recurrent neural network model; Preprocessing the historical retained water detection data set, extracting key features, and generating a training set and a test set; Inputting the training set into the recurrent neural network model, training the recurrent neural network model through forward propagation and back propagation algorithms, and obtaining a trained recurrent neural network model; Inputting the test set into the trained recurrent neural network model to obtain a test result; and calculating the prediction accuracy of the trained recurrent neural network model according to the test result; Adjusting the structure and parameters of the trained recurrent neural network model based on the test accuracy to generate a retained water prediction model; The accumulated area of ​​the retained water and the spatial distribution diagram of the retained water volume are simulated based on the retained water prediction model to generate the simulated predicted dynamic change process of the retained water.

8. The digital twin-based saline-alkali land improvement control and analysis method according to claim 1 is characterized in that: The process of generating the digital twin model of saline-alkali land retention water includes: Obtain remote sensing image data and geographic information data of saline-alkali land retention water; Performing image segmentation processing on the remote sensing image data to obtain vector boundaries of the retained water accumulation area; A digital twin model of the retained water in the saline-alkali land is constructed based on the vector boundary of the retained water accumulation area and the simulated and predicted dynamic change process of the retained water.

9. The digital twin-based saline-alkali land improvement control and analysis method according to claim 1 is characterized in that: The process of generating a complete drainage design scheme based on the digital twin model of saline-alkali land retention water includes: Based on the digital twin model of the retained water in the saline-alkali land, information on the accumulated area and amount of retained water is obtained; Optimizing drainage ditch layout and depth parameters according to the topographical features and water distribution of the accumulation area; Adopt hydraulic simulation algorithm and use hydraulic model to simulate and analyze various drainage methods to obtain drainage effect and efficiency; The complete drainage design scheme is constructed based on the drainage ditch layout and depth parameters, as well as the drainage effect and efficiency.

10. The analysis system of the digital twin-based saline-alkali land improvement control analysis method according to any one of claims 1 to 9, characterized in that: include: The data acquisition and preprocessing module is used to acquire the terrain data, soil permeability data and precipitation data of saline-alkali land, and preprocess the acquired data to obtain standardized terrain, soil permeability and precipitation data sets; A retained water accumulation area prediction module is used to construct a retained water accumulation area prediction model through a support vector machine model, and calculate the normalized terrain, soil permeability and precipitation data set through the retained water accumulation area prediction model to obtain the retained water accumulation area; A retained water volume estimation module is used to estimate the volume of retained water in the accumulated area of ​​the retained water based on the Kriging interpolation method to obtain a spatial distribution map of the retained water volume; A retained water dynamic change simulation module is used to simulate the accumulation area of ​​the retained water and the spatial distribution map of the retained water volume through an artificial neural network algorithm to obtain a simulated and predicted retained water dynamic change process; A digital twin model construction module is used to integrate the accumulation area of ​​the retained water, the spatial distribution map of the retained water volume, and the simulated predicted dynamic change process of the retained water into the digital twin model to generate a digital twin model of the retained water in the saline-alkali land; The drainage design scheme formulation module is used to generate a complete drainage design scheme based on the digital twin model of the saline-alkali land retained water.

Images