Model-based soil electric condition acquisition system and method

Abstract

The invention relates to the technical field of measurement and analysis of soil physicochemical properties, and discloses a model-based soil electric condition acquisition system and method, the system comprises a soil physicochemical property data acquisition module used for collecting soil samples and measuring soil physicochemical properties, including water content, organic matter content and heavy metal content; the model building module is used for building a soil electrokinetic condition prediction model based on the soil physicochemical property data, and the model comprises a data preprocessing module and a soil electrokinetic response prediction module; the model training and verification module is used for training the soil electric condition prediction model by using soil sample data and verifying the accuracy of the model so as to obtain a trained soil electric condition prediction model; and the electric condition acquisition module is used for inputting the physicochemical property data of the soil sample to be detected into the trained soil electric condition prediction model to acquire the electric condition of the soil. According to the invention, the accuracy and efficiency of soil electric condition prediction can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of measurement and analysis of soil physical and chemical properties, and in particular to a soil electrokinetic condition acquisition system and method based on a model. Background Art

[0002] In the field of soil pollution assessment and control, traditional soil physical and chemical property analysis methods have certain limitations, such as long analysis cycles, high costs, low efficiency, and difficulty in achieving rapid and accurate predictions of soil electrokinetic conditions. Existing soil electrokinetic condition acquisition technologies mostly rely on laboratory analysis and empirical judgment, lacking systematic and automated processing procedures, which to a certain extent limits the accuracy and timeliness of soil pollution assessments. In addition, the diversity and complexity of soil samples also pose challenges to the accurate prediction of electrokinetic conditions.

[0003] In the process of implementing the embodiments of the present invention, the inventors found that there are at least the following problems or defects in the prior art: first, the collection and processing of soil physical and chemical property data lacks standardization and automation, which makes it difficult to ensure the quality and consistency of the data; second, the prediction models of soil electrokinetic conditions are mostly empirical models, which lack sufficient data support and scientific verification, and the accuracy and reliability of the prediction results need to be improved; finally, when the prior art processes large-scale soil sample data, the complexity and computational cost of the model are high, which makes it difficult to meet the needs of practical applications. Summary of the invention

[0004] The embodiments of the present invention aim to provide a model-based soil electrokinetic condition acquisition system and method to solve the technical problems raised in the prior art.

[0005] The embodiment of the present invention solves the technical problem by adopting the following technical solution:

[0006] In a first aspect, a soil electrokinetic condition acquisition system based on a model is provided, comprising:

[0007] Soil physical and chemical properties data acquisition module, used to collect soil samples and measure soil physical and chemical properties, including moisture content, organic matter content and heavy metal content;

[0008] A model building module, used to build a soil electrokinetic condition prediction model based on soil physical and chemical property data, the model includes a data preprocessing module and a soil electrokinetic response prediction module, wherein the soil electrokinetic response prediction module is used to predict the electrokinetic response of the soil according to the soil physical and chemical properties;

[0009] A model training and verification module, used to train the soil electrokinetic condition prediction model using soil sample data and verify the model accuracy to obtain a trained soil electrokinetic condition prediction model;

[0010] And an electrokinetic condition acquisition module is used to input the physical and chemical property data of the soil sample to be tested into the trained soil electrokinetic condition prediction model to obtain the electrokinetic condition of the soil.

[0011] Furthermore, the data preprocessing module is used to preprocess the soil physicochemical property data using a specific algorithm before training the soil electric condition prediction model; and before inputting the physicochemical property data of the soil sample to be tested into the trained soil electric condition prediction model, use the specific algorithm to preprocess the physicochemical property data of the soil sample to be tested.

[0012] Furthermore, the preprocessing module includes a data cleaning submodule, a data standardization submodule and a feature selection submodule, wherein the data cleaning submodule is used to remove abnormal values ​​and missing values ​​in the soil physical and chemical property data;

[0013] The data standardization submodule is used to standardize the soil physical and chemical property data to eliminate the influence of different dimensions; and the feature selection submodule is used to select soil physical and chemical property indicators that have a significant impact on the prediction of soil electrokinetic conditions.

[0014] Furthermore, the data preprocessing module includes a principal component analysis submodule and a data dimension reduction submodule, wherein the principal component analysis submodule is used to perform principal component analysis on the soil physical and chemical property data to extract main features;

[0015] The data dimension reduction submodule is used to perform dimension reduction processing on the result of principal component analysis to reduce the complexity of the model and improve the prediction accuracy.

[0016] Further, the soil electrokinetic response prediction module includes a multi-layer perceptron submodule and a support vector machine submodule, wherein the multi-layer perceptron submodule is used to make a preliminary prediction of the soil electrokinetic condition based on the main characteristics of the soil physical and chemical properties;

[0017] The support vector machine submodule is used to optimize the prediction results of the multi-layer perceptron submodule to improve the prediction accuracy.

[0018] Furthermore, the soil electrokinetic condition prediction model is:

[0019]

[0020] Where: E represents the predicted soil electrokinetic response value; X i represents the ith soil physical and chemical property; β i It corresponds to X i The linear regression coefficient of ij It corresponds to X i and X j The regression coefficient of the interaction term, j>i; γi It corresponds to X i is the regression coefficient of the squared term, used to capture nonlinear effects; ∈ is the error term; and n is the total number of soil physical and chemical properties.

[0021] Furthermore, the model training and verification module is used to train the soil electrokinetic condition prediction model using soil sample data to calculate the prediction error, and calculate the model performance index by the following formula:

[0022]

[0023] Among them, J represents the mean square error, which is an indicator of model performance, N is the total number of samples, and E i is the electrokinetic response of the ith sample predicted by the model, E s,i is the actual measured electrodynamic response value of the ith sample.

[0024] Furthermore, the soil physical and chemical properties data acquisition module is used to obtain soil sample data of different soil types and different pollution degrees from a soil sample library, and divide the soil sample data of different soil types and different pollution degrees into a training set and a verification set at a predetermined ratio M%, wherein M% of the soil sample data of different soil types and different pollution degrees is used as the training set; and 1-M% of the soil sample data of different soil types and different pollution degrees is used as the verification set.

[0025] In a second aspect of the present invention, a method for obtaining soil electrokinetic conditions based on a model is provided, comprising:

[0026] Collect soil samples and measure soil physical and chemical properties, including moisture content, organic matter content, and heavy metal content;

[0027] A soil electrokinetic condition prediction model is established based on soil physical and chemical property data, wherein the model includes a data preprocessing module and a soil electrokinetic response prediction module, wherein the soil electrokinetic response prediction module is used to predict the electrokinetic response of the soil according to the soil physical and chemical properties;

[0028] Using soil sample data to train the soil electric condition prediction model and verify the model accuracy to obtain a trained soil electric condition prediction model;

[0029] The physical and chemical property data of the soil sample to be tested are input into the trained soil electrokinetic condition prediction model to obtain the electrokinetic condition of the soil.

[0030] Furthermore, the method further comprises:

[0031] Before training the soil electric condition prediction model, the soil physicochemical property data is preprocessed using a specific algorithm; and before inputting the physicochemical property data of the soil sample to be tested into the trained soil electric condition prediction model, the physicochemical property data of the soil sample to be tested is preprocessed using a specific algorithm.

[0032] The above-mentioned embodiments of the present invention have at least the following beneficial effects: the present invention can effectively collect and analyze the physical and chemical properties data of soil samples, including key indicators such as moisture content, organic matter content and heavy metal content, by establishing a model-based soil electrokinetic condition acquisition system. The system adopts a data preprocessing module, including submodules such as data cleaning, standardization and feature selection, to ensure the quality and accuracy of input data. In addition, through principal component analysis and data dimension reduction technology, the system can extract the main features in the soil sample data, reduce the complexity of the model, and thus improve the prediction accuracy. The system further integrates advanced prediction algorithms such as multi-layer perceptron and support vector machine to accurately predict the soil electrokinetic response, and uses actual soil sample data to train and verify the model through the model training and verification module to ensure the reliability of the prediction results. Further, the system of the present invention can quickly perform electrokinetic condition analysis on new soil samples according to the trained soil electrokinetic condition prediction model, which can improve the efficiency and accuracy of soil pollution assessment. This model-based method can not only reduce the dependence on laboratory analysis, reduce costs, but also improve the automation level of data processing, making the acquisition of soil electrokinetic conditions faster and more convenient. In addition, the design of the system allows for flexible adjustment and optimization to suit different types of soil samples and environmental conditions, with great versatility and adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] One or more embodiments are exemplarily described by the figures in the corresponding drawings, and these exemplifications do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings represent similar elements, and unless otherwise stated, the figures in the drawings do not constitute proportional limitations.

[0034] Figure 1 It is a structural schematic diagram of the soil electrokinetic condition acquisition system based on the model provided by the present invention;

[0035] Figure 2 It is a flow chart of the method for acquiring soil electrokinetic conditions based on the model provided by the present invention. DETAILED DESCRIPTION

[0036] In order to facilitate the understanding of the present invention, the present invention is described in more detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that when an element is described as "connecting" another element, it can be directly on another element, or there can be one or more centered elements therebetween. The orientation or positional relationship indicated by the terms "upper", "lower", "left", "right", "upper end", "lower end", "top" and "bottom" used in this specification is based on the orientation or positional relationship shown in the accompanying drawings, only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present invention. In addition, the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0037] Unless otherwise defined, all technical and scientific terms used in this specification have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0038] Combine the following Figure 1 , the model-based soil electrokinetic condition acquisition system 100 provided in the embodiment of the present application is described in detail through a specific example.

[0039] Figure 1 : is a schematic diagram of the structure of the soil electrodynamic condition acquisition system based on the model provided by the present invention. The soil electrodynamic condition acquisition system based on the model provided by one embodiment of the present invention comprises:

[0040] Soil physical and chemical properties data acquisition module, used to collect soil samples and measure soil physical and chemical properties, including moisture content, organic matter content and heavy metal content;

[0041] A model building module, used to build a soil electrokinetic condition prediction model based on soil physical and chemical property data, the model includes a data preprocessing module and a soil electrokinetic response prediction module, wherein the soil electrokinetic response prediction module is used to predict the electrokinetic response of the soil according to the soil physical and chemical properties;

[0042] A model training and verification module, used to train the soil electrokinetic condition prediction model using soil sample data and verify the model accuracy to obtain a trained soil electrokinetic condition prediction model;

[0043] And an electrokinetic condition acquisition module is used to input the physical and chemical property data of the soil sample to be tested into the trained soil electrokinetic condition prediction model to obtain the electrokinetic condition of the soil.

[0044] It should be noted that the system is a comprehensive solution that achieves the acquisition of soil electrokinetic conditions through a series of modules and steps. The core of this system is the collection and analysis of soil physical and chemical property data, and the establishment and application of prediction models based on these data.

[0045] Specifically, the soil physical and chemical properties data acquisition module involves the use of various instruments and methods to measure indicators such as moisture content, organic matter content, and heavy metal content of soil samples. The collection of these data can be completed through field sampling and laboratory testing, such as using a moisture meter to measure moisture content and using spectral analysis technology to measure organic matter content and heavy metal content. The model building module is based on these data, through data mining and machine learning technology, to build a mathematical model that can predict the electrokinetic response of the soil.

[0046] Preferably, the data preprocessing module can use a variety of algorithms to improve data quality, such as using mean filling or interpolation methods to handle missing values, and using standard deviation thresholding to identify and remove outliers. In addition, data standardization can be achieved through Z-score standardization or Min-Max standardization to ensure that data of different dimensions have the same weight in model training. The feature selection submodule can select indicators that have a significant impact on the prediction results based on correlation analysis or model-based feature importance evaluation.

[0047] More specifically, the model training and validation module can use a cross-validation approach to evaluate the generalization ability of the model, such as using K-fold cross-validation to ensure that the model maintains stable prediction performance on different data subsets. In the electrokinetic condition acquisition module, the trained model can be deployed to a cloud platform or mobile device to quickly obtain the electrokinetic conditions of soil samples on site without sending the samples back to the laboratory. In addition, in order to improve the adaptability and flexibility of the model, it is possible to consider introducing a model update mechanism to regularly use new soil sample data to update and optimize model parameters.

[0048] In some embodiments, the data preprocessing module is used to preprocess the soil physicochemical property data using a specific algorithm before training the soil electric condition prediction model; and before inputting the physicochemical property data of the soil sample to be tested into the trained soil electric condition prediction model, use the specific algorithm to preprocess the physicochemical property data of the soil sample to be tested.

[0049] It should be noted that the data preprocessing module is a key step in the model building process, which ensures that the data input to the model is clean, standardized and valid. This module optimizes the data quality through specific algorithms, providing a solid foundation for the subsequent soil electrokinetic condition prediction.

[0050] Specifically, the data preprocessing module includes a data cleaning submodule, a data standardization submodule, and a feature selection submodule. The purpose of the data cleaning submodule is to remove outliers and missing values ​​in the data, which can be identified by setting a threshold. For example, if the value of a certain attribute exceeds three times the standard deviation of the normal range, it is considered an outlier. The data standardization submodule usually uses the Z-score standardization method to convert the data into a standard normal distribution with a mean of 0 and a standard deviation of 1 to eliminate the influence of different dimensions. The feature selection submodule uses statistical tests or machine learning algorithms to identify soil physical and chemical property indicators that have a significant impact on the prediction results.

[0051] Preferably, the data preprocessing module also includes a principal component analysis submodule and a data dimension reduction submodule. The principal component analysis submodule is used to extract the main features in the soil physical and chemical property data, which can be achieved by calculating the covariance matrix of the data and performing eigenvalue decomposition. The data dimension reduction submodule performs dimensionality reduction processing on the results of the principal component analysis to reduce the complexity of the model and improve the prediction accuracy, which can be achieved by selecting the first few principal components with the largest variance contribution.

[0052] More specifically, during data preprocessing, an automated feature engineering process can be introduced to automatically identify and transform useful information in the data. For example, an automated process can be set up to detect and handle missing values, using interpolation techniques to fill in missing data points. In addition, feature selection can be iteratively optimized based on model feedback to ensure that the selected features contribute the most to model performance. In terms of data dimensionality reduction, in addition to principal component analysis, other dimensionality reduction techniques such as linear discriminant analysis (LDA) or t-distributed stochastic neighbor embedding (t-SNE) can also be considered to adapt to different types of data distribution characteristics.

[0053] In some embodiments, the preprocessing module includes a data cleaning submodule, a data standardization submodule and a feature selection submodule, wherein the data cleaning submodule is used to remove abnormal values ​​and missing values ​​in the soil physical and chemical property data;

[0054] The data standardization submodule is used to standardize the soil physical and chemical property data to eliminate the influence of different dimensions; and the feature selection submodule is used to select soil physical and chemical property indicators that have a significant impact on the prediction of soil electrokinetic conditions.

[0055] It should be noted that the refinement function of the data preprocessing module is to improve the prediction accuracy and robustness of the model. Through data cleaning, standardization and feature selection, the system can handle the inconsistency and noise in the raw data, thus laying the foundation for building an accurate prediction model.

[0056] Specifically, the data cleaning submodule is responsible for identifying and processing outliers and missing values ​​in the data set. Outliers can be identified by the interquartile range (IQR) of the box plot, and points that exceed 1.5 times the IQR are usually considered outliers. Missing values ​​can be filled by the mean, median, or mode, or by more advanced interpolation techniques such as time series filling or model-based prediction filling. The data standardization submodule uses Z-score standardization or Min-Max standardization methods to convert data to a uniform scale to facilitate comparison and weighting between different features. The feature selection submodule selects features that have a significant impact on the prediction target based on statistical tests (such as analysis of variance ANOVA) or model-based feature importance assessment (such as random forest feature importance).

[0057] Preferably, the operation steps of the data preprocessing module can be further refined or alternatives can be provided. For example, in the data cleaning stage, in addition to the IQR-based method, a density-based method or a distance-based method can also be used to identify outliers. In terms of data standardization, in addition to Z-score and Min-Max standardization, Robust Scaling can also be considered, which has better robustness to outliers. In the feature selection process, in addition to statistical tests and model-based methods, feature selection algorithms based on model selection can also be used, such as recursive feature elimination (RFE) or tree model-based feature selection methods.

[0058] More specifically, the operation of the feature selection submodule can be further optimized, for example, by evaluating the stability and effectiveness of different feature selection methods through cross-validation. In addition, regularization techniques, such as L1 or L2 regularization, can be introduced to automatically perform feature selection during model training and reduce the complexity of the model. In terms of data dimensionality reduction, in addition to principal component analysis (PCA), alternative techniques such as singular value decomposition (SVD) or non-negative matrix factorization (NMF) can also be considered to adapt to different data characteristics and analysis requirements.

[0059] In some embodiments, the data preprocessing module includes a principal component analysis submodule and a data dimension reduction submodule, wherein the principal component analysis submodule is used to perform principal component analysis on the soil physical and chemical property data to extract main features;

[0060] The data dimension reduction submodule is used to perform dimension reduction processing on the result of principal component analysis to reduce the complexity of the model and improve the prediction accuracy.

[0061] It should be noted that further refinement of the data preprocessing module is achieved through principal component analysis (PCA) and data dimension reduction techniques, which helps to extract key features in soil physical and chemical property data and reduce the complexity of model training.

[0062] Specifically, the purpose of the principal component analysis submodule is to transform the original data into a new coordinate system, which is composed of linear combinations of the original data, and these combinations are sorted by the size of the variance. The first principal component has the largest variance, and each subsequent principal component is orthogonal to the previous one and has the second largest variance. The data dimensionality reduction submodule is based on the results of principal component analysis and reduces the dimensionality of the data by selecting the first few principal components. These principal components can explain most of the data variance, thereby reducing the complexity of the model while retaining the main information of the data.

[0063] Preferably, the principal component analysis submodule can adjust its performance by setting different parameters, for example, a threshold can be set to determine how many principal components to retain, and this threshold can be a percentage of the total variance, such as 95%. In addition, the data dimension reduction submodule can use different techniques to select principal components, for example, principal components can be selected based on the cumulative variance contribution rate until a preset threshold is reached.

[0064] More specifically, in addition to principal component analysis, the data dimensionality reduction submodule can also consider using other dimensionality reduction techniques, such as linear discriminant analysis (LDA) or t-distributed stochastic neighbor embedding (t-SNE), which can reduce the dimensionality while maintaining the classification information of the data. When selecting the principal components, a stepwise regression method can be used, which is an automated feature selection process that optimizes the performance of the model by gradually adding or removing features. In addition, in order to further improve the prediction accuracy of the model, the knowledge of domain experts can be combined to manually select some key features that have a direct impact on the soil electrokinetic response.

[0065] In some embodiments, the soil electrokinetic response prediction module includes a multi-layer perceptron submodule and a support vector machine submodule, wherein the multi-layer perceptron submodule is used to make a preliminary prediction of the soil electrokinetic condition based on the main features of the soil physical and chemical properties;

[0066] The support vector machine submodule is used to optimize the prediction results of the multi-layer perceptron submodule to improve the prediction accuracy.

[0067] It should be noted that the soil electrodynamic response prediction module is the core component of the system, which uses two advanced machine learning algorithms, multi-layer perceptron (MLP) and support vector machine (SVM), to predict the electrodynamic response of soil.

[0068] Specifically, the multi-layer perceptron submodule makes preliminary predictions by building a neural network with multiple hidden layers. Each hidden layer consists of multiple neurons that connect input features through weights and biases and transform their outputs through activation functions. The support vector machine submodule implements classification or regression tasks by finding the optimal boundary between data points. It improves the generalization ability of the model by maximizing the boundary. In practical applications, the parameter settings of these modules, such as the learning rate, the number of hidden layers, and the number of neurons, will directly affect the performance of the model.

[0069] Preferably, the operation steps of the soil electrokinetic response prediction module can be further refined. For example, in a multi-layer perceptron, different activation functions, such as ReLU or Sigmoid, can be selected to optimize the nonlinear mapping ability of the network. In a support vector machine, the complexity and fitting ability of the model can be controlled by adjusting the penalty parameter C and the parameters of the kernel function. In addition, the best parameter combination can be selected by cross-validation to ensure that the model can maintain stable prediction performance on different data sets.

[0070] More specifically, in addition to the standard multi-layer perceptron and support vector machine, you can also consider using their variants or ensemble methods to improve the accuracy of predictions. For example, you can use a deep neural network (DNN) to increase the depth of the model, or a convolutional neural network (CNN) to process data with spatial structure. For support vector machines, you can consider using kernel tricks to handle nonlinear problems, or using random vector machines (SVM with stochastic processes) to improve the computational efficiency of the model. In addition, you can combine the prediction results of multiple models through ensemble learning techniques, such as bagging or boosting, to improve the overall prediction performance and robustness.

[0071] In some embodiments, the soil electrokinetic condition prediction model is:

[0072]

[0073] Where: E represents the predicted soil electrokinetic response value; X i represents the ith soil physical and chemical properties (such as water content H, organic matter content W, heavy metal content O and activator addition amount M); β i It corresponds to X i The linear regression coefficient of ij It corresponds to X i and X j The regression coefficient of the interaction term, j>i; γ i It corresponds to X iis the regression coefficient of the squared term, which is used to capture nonlinear effects; ∈ is the error term, which may contain random effects, such as specific effects of different soil types or different sampling locations; n is the total number of soil physical and chemical properties.

[0074] It should be noted that the soil electrokinetic condition prediction model is a complex regression equation that takes into account the linear effect, interaction effect, and nonlinear effect of soil physical and chemical properties. E in the model represents the predicted soil electrokinetic response value, which is a continuous value that can be voltage, current, or other related electrical parameters. i Represents the ith soil physical and chemical properties, such as moisture content H, organic matter content W, heavy metal content O and activator addition amount M, etc. i It corresponds to X i The linear regression coefficient, which quantifies the i Direct impact on the electrokinetic response of the soil. ij It corresponds to X i and X j The regression coefficient of the interaction term, j>i, which captures the interaction effect between different soil properties on the electrokinetic response. i It corresponds to X i The regression coefficient of the squared term is used to capture nonlinear effects. ∈ is the error term, which represents random variation that cannot be explained by the model and may include specific effects of different soil types or sampling locations. n is the total number of soil physical and chemical properties.

[0075] Preferably, the prediction model can be parameterized and optimized in a variety of ways. For example, the least squares method can be used to estimate the linear regression coefficients, and the maximum likelihood estimation can be used to determine the regression coefficients of the interaction term and the square term. In addition, regularization techniques such as Ridge Regression or Lasso Regression can be used to prevent overfitting of the model and improve the generalization ability of the model.

[0076] More specifically, the implementation of the model can be further refined, for example, by identifying and constructing interaction terms and square terms through automated feature engineering. During model training, parameter tuning techniques such as Grid Search or Random Search can be used to find the optimal regression coefficients. In addition, cross-validation can be introduced to evaluate the predictive performance of the model, and criteria such as Akaike Information Criterion (AIC) or Bayesian Information Criterion (BIC) can be used to select the optimal model complexity. As an alternative, other types of regression models such as decision tree regression, random forest regression, or gradient boosted tree regression can be considered, which are more suitable for handling complex nonlinear relationships.

[0077] In some embodiments, the model training and verification module is used to train the soil electrokinetic condition prediction model using soil sample data to calculate the prediction error, and calculate the model performance index using the following formula:

[0078]

[0079] Among them, J represents the mean square error, which is an indicator of model performance, N is the total number of samples, and E i is the electrokinetic response of the ith sample predicted by the model, E s,i is the actual measured electrodynamic response value of the ith sample.

[0080] It should be noted that the model training and validation module is a key part for evaluating and optimizing the performance of the soil electrokinetic condition prediction model. This module ensures the accuracy and reliability of the model by calculating the prediction error and model performance indicators.

[0081] Specifically, during the model training process, the system will use soil sample data to train the prediction model to obtain the optimal estimate of the model parameters. The prediction error is calculated by comparing the electric response value E predicted by the model i Compared with the actual measured electrodynamic response value E s,i The mean square error (MSE) is an indicator of model performance. It is the average of the squares of all sample prediction errors and reflects the average deviation between the model prediction value and the actual value. The total number of samples N represents the number of data points involved in model training and validation.

[0082] Preferably, the model training and validation module can use a variety of techniques to improve the performance of the model. For example, K-fold cross validation can be used to evaluate the performance of the model on different data subsets to ensure the stability and generalization ability of the model. In addition, the model training process can be optimized by adjusting the model's hyperparameters, such as the learning rate, the number of hidden layers, etc. When calculating the mean square error, weights can be introduced to deal with the problem that the importance of different samples may be different.

[0083] More specifically, the evaluation of model performance can be further refined. For example, in addition to the mean square error, other performance indicators such as the coefficient of determination (R-squared), root mean squared error (RMSE) or mean absolute error (MAE) can be considered to provide a more comprehensive evaluation of model performance. During the model validation process, the confusion matrix can be used to evaluate the performance of the classification model, or curve analysis such as the ROC curve or the PR curve can be used to evaluate the prediction threshold of the model. In addition, a model update mechanism can be introduced to regularly use new data to update the model parameters to adapt to changes in soil sample characteristics. As an alternative, the use of ensemble learning methods such as bagging or boosting can be considered to combine the predictions of multiple models to improve the overall prediction performance and robustness.

[0084] In some embodiments, the soil physical and chemical properties data acquisition module is used to obtain soil sample data of different soil types and different pollution degrees from a soil sample library, and divide the soil sample data of different soil types and different pollution degrees into a training set and a validation set at a predetermined ratio M%, wherein M% of the soil sample data of different soil types and different pollution degrees is used as a training set; and 1-M% of the soil sample data of different soil types and different pollution degrees is used as a validation set.

[0085] It should be noted that the soil physical and chemical properties data acquisition module is the basis of the entire system. It is responsible for collecting and organizing data from the soil sample library and providing necessary input for model training and verification.

[0086] Specifically, the soil sample library is a collection of samples of various soil types and different degrees of contamination. These sample data can come from laboratory tests, field sampling, or historical data records. The predetermined ratio M% is a value between 0 and 1, which is used to determine the division ratio of the training set and the validation set. For example, if M% is set to 0.7, 70% of the sample data will be used as a training set and the remaining 30% will be used as a validation set. This division method helps ensure that the model can be effectively validated on unseen data after training.

[0087] Preferably, the operation of the soil physical and chemical properties data acquisition module can be further refined. For example, in the process of collecting sample data, a standardized sampling protocol can be used to ensure the consistency and comparability of the data. When dividing the data, a stratified sampling method can be used to ensure that the sample distribution of various soil types and pollution levels in the training set and the validation set is consistent with the distribution in the overall sample library. In addition, data enhancement techniques such as rotation, scaling or adding noise can be introduced to increase the diversity of training data and improve the generalization ability of the model.

[0088] More specifically, the implementation of the soil physical and chemical properties data acquisition module can include the following steps: first, classify and label soil samples according to soil type and pollution degree; second, use automated equipment or software tools to measure the physical and chemical properties of soil samples, such as moisture content, organic matter content, and heavy metal content; then, divide the sample data into training sets and validation sets by random sampling or stratified sampling according to a predetermined ratio M%; finally, perform necessary cleaning and preprocessing on the data to ensure data quality. As an alternative, consider using semi-supervised learning methods to utilize unlabeled data, or use transfer learning to utilize pre-trained models in other fields or tasks to improve the training efficiency and prediction performance of the model.

[0089] The above-mentioned embodiments of the present invention have at least the following beneficial effects: The present invention provides an efficient and automated solution for obtaining soil electrokinetic conditions. This system can accurately measure the physical and chemical properties of soil samples and use advanced data processing and machine learning techniques to predict the electrokinetic response of the soil, which can improve the speed and accuracy of soil contamination assessment and reduce dependence on traditional laboratory tests, thereby reducing time and cost. In addition, the system can reduce human errors and improve the consistency and reliability of data processing through automated data preprocessing and model training processes. Furthermore, the system of the present invention can ensure the quality of model input data by comprehensively considering the diversity and complexity of soil samples and adopting preprocessing steps such as data cleaning, standardization, feature selection and dimensionality reduction. By introducing prediction algorithms such as multi-layer perceptrons and support vector machines, the system can capture the complex relationship between soil physical and chemical properties and electrokinetic responses, including linear and nonlinear effects. In addition, through the model training and verification modules, the system can evaluate and optimize model performance to ensure the accuracy and reliability of prediction results.

[0090] Please also read Figure 2 , Figure 2 The figure is a flow chart of the method for obtaining soil electrokinetic conditions based on the model provided by the present invention, comprising:

[0091] Step 201, collecting soil samples and measuring soil physical and chemical properties, including moisture content, organic matter content, and heavy metal content;

[0092] Step 202, establishing a soil electrokinetic condition prediction model based on soil physical and chemical property data, the model comprising a data preprocessing module and a soil electrokinetic response prediction module, wherein the soil electrokinetic response prediction module is used to predict the electrokinetic response of the soil according to the soil physical and chemical properties;

[0093] Step 203, using soil sample data to train the soil electrokinetic condition prediction model and verifying the model accuracy to obtain a trained soil electrokinetic condition prediction model;

[0094] Step 204: input the physical and chemical property data of the soil sample to be tested into the trained soil electrokinetic condition prediction model to obtain the electrokinetic condition of the soil.

[0095] It should be particularly noted that the model-based soil electric condition acquisition method 200 provided in the embodiment of the present invention only shows the part related to the technical problem to be solved by the embodiment of the present invention. It can be understood that the model-based soil electric condition acquisition method 200 provided in the embodiment of the present invention also includes other structures for realizing the functions of the model-based soil electric condition acquisition method 200, which will not be described one by one here.

[0096] In some embodiments, the method further comprises:

[0097] Before training the soil electric condition prediction model, the soil physicochemical property data is preprocessed using a specific algorithm; and before inputting the physicochemical property data of the soil sample to be tested into the trained soil electric condition prediction model, the physicochemical property data of the soil sample to be tested is preprocessed using a specific algorithm.

[0098] It should be noted that the model-based soil electrokinetic condition acquisition method enhances the model's predictive ability and accuracy by applying specific data preprocessing algorithms before and after model training.

[0099] Specifically, data preprocessing algorithms are a series of techniques used to improve the quality of raw soil physical and chemical property data, including data cleaning, standardization, feature selection, principal component analysis, and data dimensionality reduction. For example, data cleaning can remove or fill missing values, standardization can adjust the data to a uniform scale, feature selection can identify the features that have the greatest impact on the prediction target, and principal component analysis and data dimensionality reduction can reduce the complexity of the data and extract the main features. The parameter settings of these preprocessing steps, such as thresholds, standardization methods, feature selection criteria, etc., will directly affect the quality of the data and the performance of the model.

[0100] Preferably, the selection of data preprocessing algorithms and parameter settings can be customized according to the specific characteristics of the soil sample data. For example, if there are a large number of outliers in the data set, an IQR-based method can be selected for cleaning; if the data has non-normal distribution characteristics, Robust Scaling can be considered for standardization. In addition, feature selection can be performed through univariate statistical tests or model-based feature importance assessment. In principal component analysis, the number of principal components to be retained can be determined based on the cumulative variance contribution rate, while in data dimensionality reduction, different dimensionality reduction techniques, such as PCA or LDA, can be selected to adapt to the specific structure of the data.

[0101] More specifically, the embodiments may be further refined or provide alternatives. For example, in the data cleaning stage, the knowledge of domain experts can be combined to identify and process specific outliers related to soil properties. In the feature selection process, recursive feature elimination (RFE) or model-based cross-validation can be used to select the most stable feature set. In principal component analysis, domain knowledge can be combined to interpret and select principal components to ensure that they have practical physical meaning. In addition, the use of deep learning methods for data preprocessing and feature extraction can be explored, including the use of autoencoders to learn low-dimensional representations of the data. Alternatives include using tree-based models for feature selection, or using integrated methods to combine the advantages of multiple preprocessing techniques.

[0102] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them. Under the concept of the present invention, the technical features in the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other changes in different aspects of the present invention as above, which are not provided in detail for the sake of simplicity. Although the present invention is described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A model-based soil electrokinetic condition acquisition system, characterized in that: include: Soil physical and chemical properties data acquisition module, used to collect soil samples and measure soil physical and chemical properties, including moisture content, organic matter content and heavy metal content; A model building module, used to build a soil electrokinetic condition prediction model based on soil physical and chemical property data, the model includes a data preprocessing module and a soil electrokinetic response prediction module, wherein the soil electrokinetic response prediction module is used to predict the electrokinetic response of the soil according to the soil physical and chemical properties; A model training and verification module, used to train the soil electrokinetic condition prediction model using soil sample data and verify the model accuracy to obtain a trained soil electrokinetic condition prediction model; And an electrokinetic condition acquisition module is used to input the physical and chemical property data of the soil sample to be tested into the trained soil electrokinetic condition prediction model to obtain the electrokinetic condition of the soil.

2. The model-based soil electrokinetic condition acquisition system according to claim 1, characterized in that: A data preprocessing module is used to preprocess the soil physicochemical property data using a specific algorithm before training the soil electric condition prediction model; and before inputting the physicochemical property data of the soil sample to be tested into the trained soil electric condition prediction model, use the specific algorithm to preprocess the physicochemical property data of the soil sample to be tested.

3. The model-based soil electrokinetic condition acquisition system according to claim 2, characterized in that: The preprocessing module includes a data cleaning submodule, a data standardization submodule and a feature selection submodule, wherein the data cleaning submodule is used to remove abnormal values ​​and missing values ​​in the soil physical and chemical property data; The data standardization submodule is used to standardize the soil physical and chemical property data to eliminate the influence of different dimensions; and the feature selection submodule is used to select soil physical and chemical property indicators that have a significant impact on the prediction of soil electrokinetic conditions.

4. The model-based soil electrokinetic condition acquisition system according to claim 2, characterized in that: The data preprocessing module includes a principal component analysis submodule and a data dimension reduction submodule, wherein the principal component analysis submodule is used to perform principal component analysis on soil physical and chemical property data to extract main features; The data dimension reduction submodule is used to perform dimension reduction processing on the result of principal component analysis to reduce the complexity of the model and improve the prediction accuracy.

5. The model-based soil electrokinetic condition acquisition system according to claim 4, characterized in that: The soil electrokinetic response prediction module includes a multi-layer perceptron submodule and a support vector machine submodule, wherein the multi-layer perceptron submodule is used to make a preliminary prediction of the soil electrokinetic condition based on the main characteristics of the soil physical and chemical properties; The support vector machine submodule is used to optimize the prediction results of the multi-layer perceptron submodule to improve the prediction accuracy.

6. The soil electrokinetic condition acquisition system based on the model according to claim 5, characterized in that: The soil electrokinetic condition prediction model is: Where: E represents the predicted soil electrokinetic response value; X i represents the ith soil physical and chemical property; β i It corresponds to X i The linear regression coefficient of ij It corresponds to X i and X j The regression coefficient of the interaction term, j>i; γ i It corresponds to X i is the regression coefficient of the squared term, used to capture nonlinear effects; ∈ is the error term; and n is the total number of soil physical and chemical properties.

7. The model-based soil electrokinetic condition acquisition system according to claim 5, characterized in that: The model training and verification module is used to train the soil electrokinetic condition prediction model using soil sample data to calculate the prediction error, and calculate the model performance index using the following formula: Among them, J represents the mean square error, which is an indicator of model performance, N is the total number of samples, and E i is the electrokinetic response of the ith sample predicted by the model, E s,i is the actual measured electrodynamic response value of the ith sample.

8. The model-based soil electrokinetic condition acquisition system according to any one of claims 1 to 7, characterized in that: The soil physical and chemical property data acquisition module is used to obtain soil sample data of different soil types and different pollution degrees from a soil sample library, and divide the soil sample data of different soil types and different pollution degrees into a training set and a verification set at a predetermined ratio M%, wherein M% of the soil sample data of different soil types and different pollution degrees is used as the training set; and 1-M% of the soil sample data of different soil types and different pollution degrees is used as the verification set.

9. A method for obtaining soil electrokinetic conditions based on a model, characterized in that: include: Collect soil samples and measure soil physical and chemical properties, including moisture content, organic matter content, and heavy metal content; A soil electrokinetic condition prediction model is established based on soil physical and chemical property data, wherein the model includes a data preprocessing module and a soil electrokinetic response prediction module, wherein the soil electrokinetic response prediction module is used to predict the electrokinetic response of the soil according to the soil physical and chemical properties; Using soil sample data to train the soil electric condition prediction model and verify the model accuracy to obtain a trained soil electric condition prediction model; The physical and chemical property data of the soil sample to be tested are input into the trained soil electrokinetic condition prediction model to obtain the electrokinetic condition of the soil.

10. The method for acquiring soil electrokinetic conditions based on a model according to claim 9, characterized in that: The method further comprises: Before training the soil electric condition prediction model, the soil physicochemical property data is preprocessed using a specific algorithm; and before inputting the physicochemical property data of the soil sample to be tested into the trained soil electric condition prediction model, the physicochemical property data of the soil sample to be tested is preprocessed using a specific algorithm.

Images