Saline-alkali soil underground infiltrating irrigation water control and salt inhibition system
By using multi-layer perceptrons and regularization technology to build an irrigation prediction model underground in saline-alkali land, the problem of poor ventilation and permeability of saline-alkali land is solved, precise irrigation control is achieved, and water resource utilization efficiency and soil health status are improved.
Patent Information
- Application Number
- CN202411253569.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-09
- Publication Date
- 2024-12-27
AI Technical Summary
The soil in saline-alkali land has poor ventilation and water permeability due to colloid adsorption of sodium ions, and the soil is heavy and thick, shrinks and cracks when dry, muddy when wet, and has extremely poor tillage. An effective underground seepage irrigation water control and salt suppression system is needed.
A saline-alkali land underground seepage irrigation water and salt suppression system including data acquisition, data processing, model construction and control subsystem was designed. The irrigation prediction model was constructed using multi-layer perceptrons and regularization technology, and soil and meteorological data were obtained through sensors. Perform data integration and feature extraction to generate irrigation solutions to achieve precise control of irrigation time and water volume.
Improve water resource utilization efficiency, maintain a healthy soil state, reduce manual intervention, improve operational efficiency and accuracy, and avoid soil degradation caused by excessive or insufficient irrigation.
Smart Images
Figure CN119202692A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of water control and salt suppression, and in particular relates to a saline-alkali land underground infiltration irrigation water control and salt suppression system. Background Art
[0003] Since saline-alkali soil colloids absorb more sodium ions, they have poor ventilation and water permeability, and the soil is sticky, heavy, and compacted. When dry, the surface shrinks and cracks, and when wet, it becomes muddy and has extremely poor tillage properties. Therefore, there is an urgent need for a saline-alkali soil underground infiltration irrigation, water control, and salt inhibition system to solve the above problems. Summary of the invention
[0004] In order to solve the above technical problems, the present invention proposes a saline-alkali land underground infiltration irrigation water control and salt suppression system to solve the above problems existing in the prior art.
[0005] To achieve the above-mentioned purpose, the present invention provides a saline-alkali land underground infiltration irrigation water control and salt suppression system, comprising:
[0006] A data acquisition subsystem, used to acquire underground composition data and meteorological data of saline-alkali land through sensors;
[0007] A data processing subsystem, connected to the data acquisition subsystem, is used to integrate the underground composition data and meteorological data of the saline-alkali land to obtain a data set to be calculated;
[0008] A model building subsystem, connected to the data processing subsystem, is used to build a multi-layer perceptron, introduce regularization technology into the multi-layer perceptron for improvement and training, and generate an irrigation prediction model;
[0009] The control subsystem is connected to the model building subsystem and is used to input the data set to be calculated into the irrigation prediction model for calculation, obtain the prediction result, and generate an irrigation plan based on the prediction result.
[0010] Preferably, the data acquisition subsystem includes:
[0011] A sensor module for acquiring underground composition data of saline-alkali land;
[0012] A position positioning module, used for locating the position information of the sensor module based on a GPS sensor;
[0013] Wireless communication module, used to obtain meteorological data from the central database.
[0014] Preferably, the underground composition data includes soil moisture and salt content, groundwater level and temperature data.
[0015] Preferably, the data processing subsystem includes:
[0016] An exception processing module is used to perform data cleaning and fill abnormal values on the underground composition data and meteorological data in turn to generate abnormal processing data;
[0017] A feature extraction module, used to extract the features of the exception processing data into time window features and cross-domain features;
[0018] The integration module is used to integrate the time window features and the cross-domain features to obtain the data set to be calculated.
[0019] Preferably, the integration module comprises:
[0020] A normalization unit, used to robustly normalize the time window features and the cross-domain features to generate normalized feature data;
[0021] The association integration unit is used to integrate the normalized feature data based on the association degree to obtain the data set to be calculated.
[0022] Preferably, the model building subsystem includes:
[0023] A model building module, used to build a multi-layer perceptron, wherein the multi-layer perceptron includes an input layer, a hidden layer and an output layer;
[0024] Hidden improvement module, which is used to randomly select and discard some neurons in the hidden layer during each iterative forward propagation process based on the dropout method to generate an improved multilayer perceptron;
[0025] The training module is used to train the improved multi-layer perceptron through a sample set to generate the irrigation prediction model.
[0026] Preferably, the training module includes:
[0027] A sample amplification unit, used to obtain a sample set, re-label the sample set, and generate a training set and a test set;
[0028] A training unit, used for training the improved multilayer perceptron using the training set, randomly selecting and discarding some neurons during the training process to generate a training model, wherein the discarded neurons and their corresponding connections do not participate in calculations in forward propagation and back propagation;
[0029] A testing unit is used to test the training model through a test set to generate the irrigation prediction model, wherein all neurons are involved in the calculation during the model testing.
[0030] Preferably, the control subsystem comprises:
[0031] A calculation module, used for inputting the data set to be calculated into the irrigation prediction model for calculation to obtain a prediction result;
[0032] A scheme generation module, used for generating a water control and salt suppression scheme based on the prediction results;
[0033] The irrigation module is used for underground irrigation of saline-alkali land based on the water control and salt suppression scheme.
[0034] Preferably, the irrigation module comprises:
[0035] Irrigation water source, used to obtain water through pumps;
[0036] An underground drip irrigation pipeline for delivering water into the soil based on the water control and salt suppression scheme;
[0037] Drainage units to control groundwater levels and prevent soil salt accumulation.
[0038] Compared with the prior art, the present invention has the following advantages and technical effects:
[0039] The prediction model constructed by the present invention using multi-layer perceptron and regularization technology can accurately predict the impact of soil moisture demand and climate conditions on irrigation demand. By accurately controlling the irrigation time and water volume, the system not only improves the utilization efficiency of water resources, but also helps maintain the health of the soil and avoids soil degradation caused by excessive or insufficient irrigation. In addition, the present invention automatically converts the predicted data into an irrigation plan, greatly reducing manual intervention and improving operational efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] 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:
[0041] Figure 1 Schematic diagram of a system for underground irrigation, water control and salt suppression in saline-alkali land according to an embodiment of the present invention. DETAILED DESCRIPTION
[0042] 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.
[0043] 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.
[0044] Embodiment 1
[0045] like Figure 1 As shown, this embodiment provides a saline-alkali land underground infiltration irrigation water control and salt suppression system, including:
[0046] A data acquisition subsystem, used to acquire underground composition data and meteorological data of saline-alkali land through sensors;
[0047] A data processing subsystem, connected to the data acquisition subsystem, is used to integrate the underground composition data and meteorological data of the saline-alkali land to obtain a data set to be calculated;
[0048] A model building subsystem, connected to the data processing subsystem, is used to build a multi-layer perceptron, introduce regularization technology into the multi-layer perceptron for improvement and training, and generate an irrigation prediction model;
[0049] The control subsystem is connected to the model building subsystem and is used to input the data set to be calculated into the irrigation prediction model for calculation, obtain the prediction result, and generate an irrigation plan based on the prediction result.
[0050] Further optimizing the scheme, the data acquisition subsystem includes:
[0051] A sensor module for acquiring underground composition data of saline-alkali land;
[0052] A position positioning module, used for locating the position information of the sensor module based on a GPS sensor;
[0053] Wireless communication module, used to obtain meteorological data from the central database.
[0054] Sensor deployment module
[0055] Soil Sensors: Install multiple soil moisture sensors and soil conductivity (salinity) sensors to monitor soil moisture and salinity content.
[0056] Meteorological sensors: Deploy temperature, humidity, rainfall, wind speed, and solar radiation sensors to collect data on meteorological conditions.
[0057] Positioning: Use GPS sensors to determine the specific location of the sensor and accurately manage data collection points.
[0058] Data acquisition and transmission module
[0059] Data acquisition unit: responsible for collecting data from sensors and performing preliminary data formatting and cleaning.
[0060] Wireless transmission technology: Use wireless communication technology (such as LoRa, Wi-Fi, Bluetooth, 4G / 5G, etc.) to transmit data to a central database or cloud platform.
[0061] Data interface: Design standardized interfaces to ensure compatibility and interoperability between different types of sensors and devices
[0062] According to a further optimized solution, the underground component data include soil moisture and salt content, groundwater level and temperature data.
[0063] Further optimizing the scheme, the data processing subsystem includes:
[0064] An exception processing module is used to perform data cleaning and fill abnormal values on the underground composition data and meteorological data in turn to generate abnormal processing data;
[0065] A feature extraction module, used to extract the features of the exception processing data into time window features and cross-domain features;
[0066] The integration module is used to integrate the time window features and the cross-domain features to obtain the data set to be calculated.
[0067] Data preprocessing
[0068] Data cleaning: Remove erroneous records, such as data from sensor failures.
[0069] Missing value handling: Use appropriate methods to fill in missing values, such as linear interpolation or the average value based on neighboring time points.
[0070] Standardization: Normalize or standardize data to make it suitable for subsequent analysis.
[0071] Feature extraction and integration
[0072] Time window features: Depending on the analysis requirements, you may need to extract features from time series data, such as the average, maximum, and minimum values within a sliding window.
[0073] Cross-domain feature integration: Combine underground composition data with meteorological data and consider the correlation between them. For example, soil moisture may be correlated with rainfall and temperature.
[0074] Further optimizing the scheme, the integration module includes:
[0075] A normalization unit, used to robustly normalize the time window features and the cross-domain features to generate normalized feature data;
[0076] The expression of robust normalization is:
[0077]
[0078] Where Q1 and Q3 are the first and third quartiles respectively.
[0079] The association integration unit is used to integrate the normalized feature data based on the association degree to obtain the data set to be calculated.
[0080] Further optimizing the scheme, the model building subsystem includes:
[0081] A model building module, used to build a multi-layer perceptron, wherein the multi-layer perceptron includes an input layer, a hidden layer and an output layer;
[0082] Hidden improvement module, which is used to randomly select and discard some neurons in the hidden layer during each iterative forward propagation process based on the dropout method to generate an improved multilayer perceptron;
[0083] The training module is used to train the improved multi-layer perceptron through a sample set to generate the irrigation prediction model.
[0084] Multilayer Perceptron (MLP) is a basic artificial neural network, mainly used for classification and regression problems. It contains multiple layers of nodes, each node (or "neuron") calculates the output after receiving the input and passes it to the next layer. MLP mainly consists of the following structures:
[0085] 1. Structural composition
[0086] Input layer: Receives raw data input, and each input node represents a feature in the data set.
[0087] Hidden layer: One or more hidden layers, each containing multiple nodes. The hidden layer is the core of MLP and can learn nonlinear features in the input data.
[0088] Output layer: The number of nodes in the output layer depends on the specific task (for example, the number of categories in a classification problem). The output layer is used to summarize the information of the previous layer and generate the final prediction result.
[0089] 2. Activation Function
[0090] Activation function: used to introduce nonlinear factors so that the network can process complex data patterns. Common activation functions include ReLU, Sigmoid, and Tanh.
[0091] 3. Training methods
[0092] Backpropagation algorithm: It is used to train the MLP by calculating the gradient of the loss function with respect to each weight and using the gradient descent method to update the weights.
[0093] Further optimizing the scheme, the training module includes:
[0094] A sample amplification unit, used to obtain a sample set, re-label the sample set, and generate a training set and a test set;
[0095] A training unit, used for training the improved multilayer perceptron using the training set, randomly selecting and discarding some neurons during the training process to generate a training model, wherein the discarded neurons and their corresponding connections do not participate in calculations in forward propagation and back propagation;
[0096] Dropout is a regularization technique used to reduce overfitting of neural networks. This method randomly drops (i.e. temporarily removes) some neurons in the network (including their connections) during training. Doing so can reduce the complex co-adaptation relationship between neurons and force the network to learn more robust features that are useful in different subsets of the network.
[0097] Steps to implement the discard method
[0098] Step 1: Select the layer
[0099] Select the layers to apply dropout to: Typically, dropout is applied to one or more hidden layers in the network. It is not recommended to use dropout on the input or output layers.
[0100] Step 2: Set the drop rate
[0101] Determine the dropout rate: The dropout rate (usually denoted as pp) is the proportion of neurons that are randomly dropped during training. Common dropout rate values are 0.2 to 0.5. Choosing an appropriate dropout rate is important, because a dropout rate that is too high will cause the network to learn inefficiently, while a dropout rate that is too low may not effectively prevent overfitting.
[0102] Step 3: Training Adjustment
[0103] The dropout method is applied during training: During each iteration of the forward propagation, some neurons are randomly selected and discarded. The discarded neurons and their corresponding connections are not involved in the calculation in this forward propagation and subsequent backpropagation.
[0104] Disable dropout at test time: When evaluating or testing the model, all neurons are involved in the calculation. To compensate for dropout during training, the output of each neuron can be multiplied by 1-p (the complement of the dropout rate) to ensure that the expected value of the output is the same during training and testing.
[0105] A testing unit is used to test the training model through a test set to generate the irrigation prediction model, wherein all neurons are involved in the calculation during the model testing.
[0106] Further optimizing the scheme, the control subsystem includes:
[0107] A calculation module, used for inputting the data set to be calculated into the irrigation prediction model for calculation to obtain a prediction result;
[0108] A scheme generation module, used for generating a water control and salt suppression scheme based on the prediction results;
[0109] The irrigation module is used for underground irrigation of saline-alkali land based on the water control and salt suppression scheme.
[0110] Further optimizing the solution, the irrigation module includes:
[0111] Irrigation water source, used to obtain water through pumps;
[0112] An underground drip irrigation pipeline for delivering water into the soil based on the water control and salt suppression scheme;
[0113] Drainage units to control groundwater levels and prevent soil salt accumulation.
[0114] Irrigation water source: The system requires a stable irrigation water source, which can be surface water, groundwater or other available water resources.
[0115] Water pumps and pumping stations: used to extract and deliver irrigation water to designated irrigation areas.
[0116] Underground drip irrigation pipe: This is a drip irrigation system buried in the soil that can deliver water directly to the roots of crops, reducing water evaporation and improving water use efficiency.
[0117] Drainage system: A drainage system is essential to control the water table and prevent the accumulation of salt in the soil. This usually includes drain pipes and possibly a drainage pumping station to remove excess water when necessary.
[0118] Filtration and conditioning facilities: To ensure the smooth operation of the irrigation system, a filtration system is usually required to remove impurities from the irrigation water. It may also include some conditioning equipment to adjust the pH value and other chemical properties of the water to suit specific soil and crop requirements.
[0119] 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 underground infiltration irrigation water control and salt suppression system, characterized in that: include: A data acquisition subsystem, used to acquire underground composition data and meteorological data of saline-alkali land through sensors; A data processing subsystem, connected to the data acquisition subsystem, is used to integrate the underground composition data and meteorological data of the saline-alkali land to obtain a data set to be calculated; A model building subsystem, connected to the data processing subsystem, is used to build a multi-layer perceptron, introduce regularization technology into the multi-layer perceptron for improvement and training, and generate an irrigation prediction model; The control subsystem is connected to the model building subsystem and is used to input the data set to be calculated into the irrigation prediction model for calculation, obtain the prediction result, and generate an irrigation plan based on the prediction result.
2. The saline-alkali land underground infiltration irrigation water control and salt suppression system according to claim 1, characterized in that: The data acquisition subsystem includes: A sensor module for acquiring underground composition data of saline-alkali land; A position positioning module, used for locating the position information of the sensor module based on a GPS sensor; Wireless communication module, used to obtain meteorological data from the central database.
3. The saline-alkali land underground infiltration irrigation water control and salt suppression system according to claim 1, characterized in that: The subsurface composition data include soil moisture and salt content, groundwater level and temperature data.
4. The saline-alkali land underground infiltration irrigation water control and salt suppression system according to claim 1, characterized in that: The data processing subsystem comprises: An exception processing module is used to perform data cleaning and fill abnormal values on the underground composition data and meteorological data in turn to generate abnormal processing data; A feature extraction module, used to extract the features of the exception processing data into time window features and cross-domain features; The integration module is used to integrate the time window features and the cross-domain features to obtain the data set to be calculated.
5. The saline-alkali land underground infiltration irrigation water control and salt suppression system according to claim 4, characterized in that: The integration module includes: A normalization unit, used to robustly normalize the time window features and the cross-domain features to generate normalized feature data; The association integration unit is used to integrate the normalized feature data based on the association degree to obtain the data set to be calculated.
6. The saline-alkali land underground infiltration irrigation water control and salt suppression system according to claim 1, characterized in that: The model building subsystem includes: A model building module, used to build a multi-layer perceptron, wherein the multi-layer perceptron includes an input layer, a hidden layer and an output layer; Hidden improvement module, which is used to randomly select and discard some neurons in the hidden layer during each iterative forward propagation process based on the dropout method to generate an improved multilayer perceptron; The training module is used to train the improved multi-layer perceptron through a sample set to generate the irrigation prediction model.
7. The saline-alkali land underground infiltration irrigation water control and salt suppression system according to claim 6, characterized in that: The training module includes: A sample amplification unit, used to obtain a sample set, re-label the sample set, and generate a training set and a test set; A training unit, used for training the improved multilayer perceptron using the training set, randomly selecting and discarding some neurons during the training process to generate a training model, wherein the discarded neurons and their corresponding connections do not participate in calculations in forward propagation and back propagation; A testing unit is used to test the training model through a test set to generate the irrigation prediction model, wherein all neurons are involved in the calculation during the model testing.
8. The saline-alkali land underground infiltration irrigation water control and salt suppression system according to claim 1, characterized in that: The control subsystem comprises: A calculation module, used for inputting the data set to be calculated into the irrigation prediction model for calculation to obtain a prediction result; A scheme generation module, used for generating a water control and salt suppression scheme based on the prediction results; The irrigation module is used for underground irrigation of saline-alkali land based on the water control and salt suppression scheme.
9. The saline-alkali land underground infiltration irrigation water control and salt suppression system according to claim 8, characterized in that: The irrigation module comprises: Irrigation water source, used to obtain water through pumps; An underground drip irrigation pipeline for delivering water into the soil based on the water control and salt suppression scheme; Drainage units to control groundwater levels and prevent soil salt accumulation.