Farmland water efficiency data mining analysis method

Abstract

The invention provides a farmland water efficiency data mining analysis method, which comprises the following steps of: acquiring multiple types of soil moisture content information, irrigation water quantity data and crop growth index data corresponding to different depths on different point positions acquired in multiple monitoring periods in a target farmland area to form a dynamic monitoring data set; the method comprises the following steps: acquiring a dynamic monitoring data set, performing multi-dimensional correlation analysis on the dynamic monitoring data set to generate a farmland water efficiency correlation analysis result, performing strategy matching on the correlation analysis result based on a preset water efficiency optimization strategy network to generate a water efficiency optimization regulation and control scheme, and finally feeding back the water efficiency optimization regulation and control scheme to a farmland irrigation control system. Irrigation parameter adjusting operation is triggered, and optimization of farmland water efficiency and reasonable utilization of water resources are achieved.

Description

Technical Field

[0001] The present disclosure relates to the field of agricultural science and technology, and in particular to a method for mining and analyzing farmland water efficiency data. Background Art

[0002] The rational use of water resources has always been a crucial issue in agricultural production. Traditional irrigation methods often lack scientific evidence, resulting in significant water waste and low irrigation efficiency. Currently, the assessment and management of farmland water efficiency largely relies on empirical judgment and simple indicator monitoring.

[0003] Existing technologies often consider only one aspect of soil moisture, irrigation water volume, or crop growth indicators independently, without comprehensively analyzing these factors. For example, some methods determine irrigation water volume based solely on soil moisture, ignoring the varying water requirements of crops at different growth stages. Other methods focus solely on irrigation water volume without considering the soil's ability to retain water or the crop's actual water absorption.

[0004] Furthermore, existing farmland water efficiency management methods lack in-depth data mining and analysis, failing to accurately grasp the complex relationships between soil moisture, irrigation water volume, and crop growth indicators. Irrigation strategies often adopt a one-size-fits-all approach, failing to precisely tailor them to the specific conditions of individual farmland areas. This results in under-irrigation in some areas, impacting crop growth, while over-irrigation in others wastes water resources.

[0005] At the same time, existing technologies struggle to dynamically monitor and adjust farmland water efficiency in real time. During the irrigation process, irrigation parameters cannot be adjusted promptly based on changes in soil moisture, crop growth conditions, and environmental factors, making it difficult to effectively improve farmland water efficiency. Summary of the Invention

[0006] In view of the above-mentioned problems, in combination with the first aspect of the present invention, an embodiment of the present invention provides a method for mining and analyzing farmland water efficiency data, the method comprising:

[0007] Acquire a dynamic monitoring data set for a target farmland area, the dynamic monitoring data set including multiple types of soil moisture information, irrigation water volume data, and crop growth index data corresponding to different points and depths collected during multiple monitoring cycles, the soil moisture information being used to represent vertically layered soil moisture information for the farmland at a single point;

[0008] Performing multi-dimensional correlation analysis on the dynamic monitoring data set to generate a farmland water efficiency correlation analysis result for the target farmland area;

[0009] Based on a preset water efficiency optimization strategy network, the strategy matching process is performed on the farmland water efficiency correlation analysis results to generate a water efficiency optimization control plan for the target farmland area;

[0010] The water efficiency optimization control scheme is fed back to the farmland irrigation control system to trigger irrigation parameter adjustment operations.

[0011] On the other hand, an embodiment of the present invention also provides a farmland irrigation monitoring device, including a processor and a machine-readable storage medium, wherein the machine-readable storage medium is connected to the processor, the machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0012] Based on the above aspects, the farmland water efficiency data mining and analysis method realizes the multi-dimensional correlation of soil moisture information, irrigation water volume data and crop growth index data through comprehensive collection and in-depth analysis of the dynamic monitoring data set of the target farmland area, and can accurately reveal the nonlinear interaction law between these data. It performs strategy matching processing based on the preset water efficiency optimization strategy network, and combines the execution effect data of the historical control strategy for feasibility verification to ensure that the generated water efficiency optimization control plan not only meets the water-saving goal, but also adapts to the actual situation of the farmland and the growth needs of the crop. The water efficiency optimization control plan is fed back to the farmland irrigation control system, realizing the dynamic adjustment of irrigation parameters and forming a closed-loop optimization mechanism. It can adjust the irrigation strategy in real time according to the actual situation, effectively improving the farmland water efficiency, reducing the waste of water resources, and ensuring the healthy growth of crops.

[0013] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The accompanying drawings are used to provide a further understanding of the present disclosure and constitute a part of the specification. Together with the following detailed description, they are used to explain the present disclosure but do not constitute a limitation of the present disclosure. In the accompanying drawings:

[0015] Figure 1 It is a schematic diagram of the execution flow of the farmland water efficiency data mining and analysis method provided by an embodiment of the present invention.

[0016] Figure 2 Schematic diagram of exemplary hardware and software components of a farmland irrigation monitoring device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0018] The following describes the specific embodiments of the present disclosure in detail with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only used to illustrate and explain the present disclosure and are not intended to limit the present disclosure.

[0019] Step S110: Acquire a dynamic monitoring data set for the target farmland area. The dynamic monitoring data set includes multiple types of soil moisture information, irrigation water volume data, and crop growth indicator data corresponding to different points and depths collected during multiple monitoring cycles. The soil moisture information represents vertically layered soil moisture information at a single point in the farmland.

[0020] When conducting farmland water efficiency data mining and analysis, a dynamic monitoring data set of the target farmland area is obtained. In order to ensure the comprehensiveness and accuracy of the data, systematic data collection work will be carried out in the target farmland area.

[0021] To collect soil moisture information, multiple soil moisture monitoring points are strategically located throughout the farmland based on factors such as the farmland's topography and soil type distribution. Suppose there are A monitoring points, designated M1, M2, ..., MA. These monitoring points continuously record soil moisture conditions over B monitoring cycles. Each monitoring point generates a soil moisture information record during each monitoring cycle, ultimately resulting in a soil moisture information set consisting of A × B data points.

[0022] Irrigation water volume data collection relies on flow monitoring devices installed throughout the irrigation system. C flow monitoring devices, designated N1, N2, …, NC, are installed at key locations throughout the irrigation system, such as water source inlets and branch pipes. These devices record the volume of water passing through each location during each monitoring cycle. Similarly, over B monitoring cycles, a C × B data set of irrigation water volume data is generated.

[0023] Crop growth indicator data is acquired through a combination of various technical approaches. Remote sensing equipment installed above farmland regularly captures and analyzes crop images, obtaining macroscopic growth indicators such as canopy cover. Furthermore, specialized personnel regularly conduct field measurements in the fields to obtain microscopic growth indicators such as stem diameter and leaf count. Assuming D different crop growth indicators are acquired through these methods, each indicator will generate a series of data over B monitoring cycles, ultimately forming a crop growth indicator data set containing D × B data points.

[0024] By integrating these three data sets according to the monitoring cycle, we obtain a dynamic monitoring data set for the target farmland area. This set contains multiple types of soil moisture information, irrigation water data, and crop growth index data corresponding to different depths at different points collected during multiple monitoring cycles.

[0025] Step S120: performing multi-dimensional correlation analysis on the dynamic monitoring data set to generate a farmland water efficiency correlation analysis result for the target farmland area.

[0026] Step S121: performing time sequence alignment processing and standard ring conversion processing on the soil moisture information, the irrigation water quantity data and the crop growth index data in each monitoring cycle to generate a standardized monitoring data sequence.

[0027] After obtaining soil moisture information, irrigation water volume data, and crop growth indicator data for each monitoring cycle, time alignment is required because different types of data may be collected at different times. Using a unified timeline as a benchmark, the data within each monitoring cycle is time-calibrated.

[0028] For soil moisture information, the data collection time for each monitoring point may not be completely consistent. By finding the actual collection time corresponding to the data of each monitoring point, it is compared with the unified timeline. If the data collection time of a monitoring point deviates from that of a time point on the unified timeline, interpolation is used to adjust it. For example, if a monitoring point does not collect data at time t on the unified timeline, but has data at times t-Δt and t+Δt, the soil moisture information at time t is estimated based on the data at these two times through linear interpolation or other appropriate interpolation methods.

[0029] A similar approach is used to align irrigation water volume data and crop growth indicator data. Irrigation water volume data is adjusted based on the actual recording time of the flow monitoring equipment, while crop growth indicator data is calibrated based on the capture time of the remote sensing equipment and the field measurement time.

[0030] After completing the time series alignment, the next step is standard ring conversion. The purpose of standard ring conversion is to convert different types of data to a unified standard scale for subsequent analysis and processing. For soil moisture information, the value range of the original data is first determined, denoted as [Hmin, Hmax]. Then, using a transformation function, each soil moisture information is converted to a new standard range, assuming this standard range is [0, 1]. The transformation function is selected based on the distribution characteristics of the data. For example, if the data approximately follows a normal distribution, normalization can be used for the conversion.

[0031] A similar approach was used to convert irrigation water volume data and crop growth index data to the standard ring. For irrigation water volume data, the original value range [Wmin, Wmax] was determined and then converted to the standard range using an appropriate conversion function. For crop growth index data, the original value range [Gmin, Gmax] was determined for each growth index and then converted to the standard ring.

[0032] After time series alignment and standard cycle conversion, soil moisture information, irrigation water volume data, and crop growth indicator data for each monitoring cycle are converted into a standardized monitoring data sequence. The data in this sequence is aligned in time and uniform in scale, facilitating subsequent spatiotemporal feature extraction and correlation analysis.

[0033] Step S122: calling the pre-trained association analysis model to perform spatiotemporal feature extraction processing on the standardized monitoring data sequence to generate soil moisture change characteristics, irrigation water distribution characteristics and crop growth response characteristics corresponding to the monitoring period.

[0034] Step S1221: performing constrained spatial interpolation processing on the soil moisture information in the standardized monitoring data sequence in combination with the soil texture type distribution map to generate a soil moisture spatial distribution map.

[0035] After obtaining the standardized monitoring data series, the soil moisture information needs to be processed in conjunction with the soil texture type distribution map. The soil texture type distribution map reflects the soil texture conditions in different areas of farmland, and different soil textures have a significant impact on the distribution of soil moisture.

[0036] First, spatially match the soil moisture information in the standardized monitoring data series with the soil texture type distribution map. This means determining the soil texture type corresponding to each soil moisture monitoring point. Assume there are E soil texture types, denoted as T1, T2, …, TE. The soil moisture information at each monitoring point is labeled with the corresponding soil texture type.

[0037] Constrained spatial interpolation is then performed. The goal of spatial interpolation is to estimate soil moisture values at other locations within the field based on soil moisture information at known monitoring points. During interpolation, constraints based on soil texture type are considered. For example, different interpolation parameters or methods may be used for adjacent areas with different soil texture types.

[0038] For areas with the same soil texture type, an appropriate interpolation algorithm, such as Kriging, is used to interpolate soil moisture information from monitoring points within the area. For regions bordering different soil texture types, a special interpolation method is used to combine the data from monitoring points on both sides of the boundary and the differences in soil texture types to ensure the rationality of the interpolation results. This constrained spatial interpolation process ultimately generates a soil moisture spatial distribution map. This map reflects the spatial distribution of soil moisture within the entire target farmland area.

[0039] Step S1222: performing time gradient calculation on the soil moisture spatial distribution map based on a sliding time window to extract the soil moisture change characteristics.

[0040] After obtaining the soil moisture spatial distribution map, we use a sliding time window method to extract soil moisture variation characteristics. First, determine the size and step size of the sliding time window. Assume that the sliding time window size is F monitoring cycles and the step size is G monitoring cycles. Starting from the first monitoring cycle, select the soil moisture spatial distribution map of F consecutive monitoring cycles as a window.

[0041] Within each window, the temporal gradient of the soil moisture spatial distribution map is calculated. For each location in the field, the rate of change of soil moisture at that location between different monitoring cycles within the window is calculated. For example, for a location P, the difference in soil moisture between the first and last monitoring cycles within the window is calculated and then divided by the window's time span to obtain the temporal gradient of soil moisture at that location within that window.

[0042] As the windows slide sequentially with a step size G, the same temporal gradient calculation is performed for each window. Ultimately, a set of soil moisture temporal gradients for each location within different time windows is obtained. These temporal gradients reflect the temporal variation of soil moisture and are used as soil moisture variation characteristics.

[0043] Step S1223: Perform sliding window outlier detection processing on the irrigation water volume data, eliminate instantaneous water flow fluctuation noise and then perform irrigation event segmentation processing, identify the starting time point and water volume distribution curve of each irrigation operation, and extract the irrigation water volume distribution characteristics in combination with the irrigation equipment response delay parameter.

[0044] For the irrigation water volume data in the standardized monitoring data series, we first perform a sliding window outlier detection process. The size of a sliding window is set to H data points. Starting from the first data point, H consecutive irrigation water volume data are selected as a window.

[0045] Within each window, statistical methods are used to determine whether there are outliers. For example, the mean and standard deviation of the data within the window can be calculated, and data points that deviate from the mean by more than a set multiple of the standard deviation are identified as outliers. When an outlier is detected, appropriate methods are used to address it, such as replacing it with the mean or median of the other data points within the window to eliminate instantaneous flow fluctuation noise.

[0046] After outlier detection and processing, irrigation event segmentation is performed. By analyzing the changing trends of irrigation water volume data, the start and end time points of each irrigation operation are identified. For example, when the irrigation water volume data suddenly rises from a low, stable value to a higher value, the time of the rise is used as the start time point of the irrigation operation; when the irrigation water volume data drops from a high value to a low, stable value, the time of the drop is used as the end time point of the irrigation operation. After determining the start and end time points of each irrigation operation, the water volume distribution curve for each irrigation operation is extracted. This curve reflects the change in irrigation water volume over time during an irrigation operation.

[0047] The response delay parameter of the irrigation equipment is also considered. After receiving the start signal, the irrigation equipment may take some time to reach a stable irrigation flow rate. When extracting the irrigation water distribution characteristics, the water distribution curve is corrected to compensate for the response delay of the irrigation equipment. For example, the starting time of the water distribution curve is postponed by a time associated with the response delay parameter. Ultimately, the starting time of each irrigation operation, the water distribution curve, and this corrected information are used as the irrigation water distribution characteristics.

[0048] Step S1224: dividing the crop growth index data into growth stages, combining the divided growth stage data with the soil moisture change characteristics of the corresponding time interval, calculating the response coefficient of the crop growth rate and soil moisture, and generating the crop growth response characteristics.

[0049] For the crop growth indicator data in the standardized monitoring data series, growth stage classification is first performed. Based on the crop's growth patterns and biological characteristics, the crop's growth process is divided into one growth stage, designated S1, S2, ..., S1. By analyzing the changing trends of the crop growth indicator data, the crop growth stage corresponding to each monitoring cycle is determined. For example, when a plant's growth indicator, such as plant height or leaf count, reaches a set threshold, the crop is determined to have entered the next growth stage.

[0050] After completing the growth stage division, the divided growth stage data is combined with the soil moisture variation characteristics of the corresponding time interval. For each growth stage, the soil moisture variation characteristics and crop growth index data within the time interval corresponding to that stage are selected.

[0051] Calculate the response coefficient between crop growth rate and soil moisture. For a specific crop growth indicator, such as plant height, calculate the plant height growth rate within each growth stage. Then, analyze the relationship between this growth rate and the soil moisture variation characteristics within the corresponding time interval. Using statistical analysis methods such as regression analysis, calculate the response coefficient between crop growth rate and soil moisture. This response coefficient reflects the sensitivity of crop growth to changes in soil moisture. The set of response coefficients between crop growth rate and soil moisture for each growth stage is used as the crop growth response characteristic.

[0052] Step S1225: performing time stamp alignment processing on the soil moisture change characteristics, the irrigation water distribution characteristics, and the crop growth response characteristics to form a spatiotemporal correlation feature set.

[0053] After obtaining the soil moisture change characteristics, irrigation water distribution characteristics, and crop growth response characteristics, timestamp alignment is required because the timestamps of these characteristics may be different.

[0054] The timestamps of these three features were compared and adjusted using a unified timeline as the benchmark. For the soil moisture change feature, each feature value corresponds to a time window, and the center time point of this time window is used as the timestamp of the feature value. For the irrigation water distribution feature, the start time point of each irrigation operation is used as the timestamp of the relevant features of the irrigation operation. For the crop growth response feature, the midpoint of each growth stage is used as the timestamp of the relevant features of that growth stage.

[0055] By searching and matching these timestamps, we align the characteristics of soil moisture variation, irrigation water distribution, and crop growth response in time. Feature values with identical or similar timestamps are combined to form a spatiotemporal correlation feature set. The features in this spatiotemporal correlation feature set are correlated in both time and space, providing more comprehensive information for subsequent correlation analysis.

[0056] Step S123: constructing a first correlation feature vector based on the dynamic coupling relationship between the soil moisture change characteristics and the irrigation water distribution characteristics.

[0057] Step S1231: determining a time lag correlation coefficient between a moisture accumulation gradient in the soil moisture variation characteristic and an irrigation water pulse in the irrigation water distribution characteristic.

[0058] After obtaining the soil moisture variation characteristics and irrigation water distribution characteristics, it is necessary to analyze the dynamic coupling relationship between them. First, we focus on the moisture accumulation gradient in the soil moisture variation characteristics and the irrigation water pulse in the irrigation water distribution characteristics. The moisture accumulation gradient reflects the cumulative change in soil moisture over a period of time, while the irrigation water pulse indicates the sudden increase in irrigation water during each irrigation operation.

[0059] To determine the time-lagged correlation coefficient between them, a correlation analysis is performed on the humidity cumulative gradient and irrigation water pulse data at different time lags. The time lag is assumed to range from -J monitoring cycles to +J monitoring cycles. For each time lag value k, the humidity cumulative gradient data and irrigation water pulse data are time-shifted accordingly, and the correlation coefficient between them is calculated.

[0060] For example, when the time lag value k = 0, the correlation coefficient between the current humidity cumulative gradient and the irrigation water pulse is directly calculated. When the time lag value k = 1, the humidity cumulative gradient data is offset backward by one monitoring period, and the correlation coefficient is then calculated with the current irrigation water pulse data. When the time lag value k = -1, the humidity cumulative gradient data is offset forward by one monitoring period, and the correlation coefficient is then calculated with the current irrigation water pulse data. By traversing all time lag values, a set of time lag correlation coefficients is obtained. These coefficients reflect the strength of the correlation between the humidity cumulative gradient and the irrigation water pulse at different time lags.

[0061] Step S1232: performing irrigation effect correction processing on the soil moisture change characteristic according to the time lag correlation coefficient to generate a corrected soil moisture response characteristic.

[0062] After obtaining the time-lagged correlation coefficients, these coefficients are used to correct the irrigation effect of the soil moisture variation characteristics. For each moisture cumulative gradient value in the soil moisture variation characteristics, it is adjusted according to its corresponding time-lagged correlation coefficient.

[0063] Assume that the time-lagged correlation coefficient corresponding to a certain humidity cumulative gradient value is r, and the humidity cumulative gradient value is h. h is modified based on the magnitude and sign of r. If r is positive and large, it indicates that the irrigation water pulse has a strong positive impact on the humidity cumulative gradient, and the value of h is appropriately increased. If r is negative and large, it indicates that the irrigation water pulse has a strong negative impact on the humidity cumulative gradient, and the value of h is appropriately decreased.

[0064] By performing such correction processing on each moisture accumulation gradient value in the soil moisture change characteristic, a corrected soil moisture response characteristic is generated, which more accurately reflects the impact of irrigation operations on soil moisture changes.

[0065] Step S1233: The spatial distribution map of the corrected soil moisture response characteristics and the time series data of the irrigation water distribution characteristics are processed by a spatiotemporal graph convolutional network, and the soil-irrigation dynamic coupling characteristics are extracted by fusing the spatial adjacency relationship and the time-dependent feature interaction.

[0066] The spatial distribution map of the corrected soil moisture response characteristics and the time series data of irrigation water distribution characteristics were input into the spatiotemporal graph convolutional network (STGCN), a deep learning model that can process both spatial and temporal information.

[0067] In the spatiotemporal graph convolutional network, a spatial convolution operation is first performed on the spatial distribution map of the corrected soil moisture response characteristics. This operation considers the spatial adjacency of soil moisture and extracts spatial features by sliding the convolution kernel across the spatial distribution map. Simultaneously, a temporal convolution operation is performed on the time series data of irrigation water volume distribution characteristics. This operation considers the temporal dependence of irrigation water volume and extracts temporal features by sliding the convolution kernel across the time series.

[0068] Then, the spatial and temporal features are fused. This fusion process considers the interaction between space and time. Through network structure and computational rules, the spatial and temporal features are combined and transformed to produce a soil-irrigation dynamic coupling feature. This soil-irrigation dynamic coupling feature reflects the dynamic coupling relationship between soil moisture and irrigation water volume in space and time.

[0069] Step S1234: performing feature dimensionality reduction processing on the soil-irrigation dynamic coupling feature to generate the first correlation feature vector.

[0070] After obtaining the soil-irrigation dynamic coupling feature, since the dimension of the feature may be high, feature dimensionality reduction is required to facilitate subsequent analysis and processing.

[0071] Methods such as principal component analysis (PCA) can be used for feature dimensionality reduction. The basic idea of principal component analysis is to transform the original high-dimensional features into a set of new low-dimensional features through linear transformation. These low-dimensional features are linear combinations of the original features and can retain most of the information of the original features.

[0072] First, the covariance matrix of the soil-irrigation dynamic coupling feature is calculated. The covariance matrix reflects the correlation between features. Then, the covariance matrix is subjected to eigenvalue decomposition to obtain eigenvalues and eigenvectors. The first K eigenvectors with the largest eigenvalues are selected, and the soil-irrigation dynamic coupling feature is projected into the low-dimensional space formed by these K eigenvectors to obtain the reduced-dimensional feature. The reduced-dimensional feature is used as the first correlation eigenvector. This first correlation eigenvector is used to characterize the dynamic impact of irrigation operations on the spatial distribution of soil moisture.

[0073] Step S124: generating a water efficiency correlation sub-result of the monitoring period according to the mapping relationship between the crop growth response characteristics and the first correlation feature vector.

[0074] Step S1241: performing a nonlinear regression analysis with regularization constraints on the growth rate data in the crop growth response characteristics and the first associated eigenvector, and determining the response function of crop growth to the soil-irrigation coupling effect in combination with the crop physiological water requirement threshold.

[0075] The growth rate data and the first associated eigenvector in the crop growth response feature are subjected to nonlinear regression analysis with regularization constraints. The goal of nonlinear regression analysis is to find a nonlinear function that can describe the relationship between the growth rate data and the first associated eigenvector as accurately as possible.

[0076] Regularization constraints are added during regression analysis. Regularization constraints prevent overfitting and improve model generalization. Common regularization methods include L1 regularization and L2 regularization.

[0077] The crop physiological water requirement threshold is also considered. This threshold represents the critical value of a crop's water demand during growth. When determining the response function, the change in crop growth rate when the soil-irrigation coupling effect reaches the threshold is considered.

[0078] By continuously adjusting the parameters of the nonlinear regression model, the model's predictions were made as close as possible to the actual growth rate data while satisfying regularization constraints. After multiple iterations and optimizations, a crop growth response function to soil-irrigation coupling was ultimately determined. This response function describes how crop growth rates vary under varying soil-irrigation coupling conditions and takes into account the limiting effect of crop physiological water requirement thresholds on crop growth.

[0079] Step S1242: Calculating crop growth simulation values under different irrigation water distributions based on the response function, and performing residual analysis on the crop growth index data actually collected.

[0080] After obtaining a response function for crop growth to the soil-irrigation coupling effect, we used this function to calculate simulated crop growth values under different irrigation water distributions. First, different irrigation water distribution characteristics were input into the response function. Based on the response function's operational rules, corresponding simulated crop growth values were obtained. These simulated values reflect the predicted crop growth rate under a given irrigation water distribution.

[0081] Then, residual analysis is performed between the calculated crop growth simulation values and the actual crop growth index data collected. The purpose of residual analysis is to evaluate the prediction accuracy of the response function. For each time point or monitoring period, the difference between the crop growth simulation value and the actual crop growth index data collected is calculated. This difference is the residual. By analyzing the distribution of the residuals, such as the mean and variance of the residuals, the degree of prediction deviation of the response function can be understood. If the mean of the residuals is close to zero and the variance is small, it indicates that the prediction accuracy of the response function is high; conversely, if the mean of the residuals deviates significantly from zero and the variance is large, it indicates that the prediction of the response function has a large deviation and requires further adjustment.

[0082] Step S1243: adjusting the parameter weights of the response function according to the residual analysis results to generate an optimized water efficiency response model.

[0083] Adjust the parameter weights of the response function based on the results of the residual analysis. If the residual analysis indicates that the response function has large prediction deviations under certain irrigation water distribution conditions, it is necessary to adjust the parameter weights that affect the response function output in these cases. This adjustment can be done using optimization algorithms such as gradient descent.

[0084] The basic idea of gradient descent is to calculate the gradient of the residual with respect to the parameter weights of the response function, and then update the parameter weights in the opposite direction of the gradient, so that the residual gradually decreases. Specifically, the partial derivative of the residual with respect to each parameter weight is first calculated to obtain the gradient vector. Then, according to a preset learning rate, the parameter weights are updated in the opposite direction of the gradient. The learning rate controls the step size of the parameter weight update. A learning rate that is too large may cause the parameter weights to update too quickly and fail to converge to the optimal value; a learning rate that is too small will result in slow convergence.

[0085] The parameter weights are updated through multiple iterations until the residuals meet preset convergence criteria, such as the residual mean and variance being less than a certain threshold. At this point, the optimized response function is obtained and used as the optimized water efficiency response model. This model can more accurately describe the relationship between crop growth and irrigation water distribution, providing a more reliable basis for subsequent water efficiency correlation analysis.

[0086] Step S1244: performing joint encoding processing on the optimized water efficiency response model and the first correlation feature vector to generate a water efficiency correlation sub-result of the monitoring period.

[0087] The optimized water effect response model and the first associated eigenvector are jointly encoded to integrate the information of the two to form a more comprehensive and representative feature representation.

[0088] First, feature extraction is performed on the optimized water efficiency response model. Representative features can be extracted from the model's parameters and outputs. For example, the coefficients of crop growth rates corresponding to different irrigation water distributions in the model can be extracted as features.

[0089] The extracted water response model features are then concatenated with the first associated feature vector. This concatenation can be accomplished by arranging the features of both models in a predetermined order to form a new feature vector. During the concatenation process, it is important to ensure that the features are dimensionally aligned and their dimensions are consistent. If the dimensions of the features differ, dimensionality adjustment may be necessary, such as reducing or increasing the dimensionality of certain features. If the dimensions differ, normalization or standardization may be necessary to ensure that all features are compared and analyzed on the same scale.

[0090] After joint coding, a new feature vector is generated, which serves as the water efficiency correlation sub-result for the monitoring period. This result includes information such as the sensitivity index of crop growth to irrigation strategy and the dynamic balance threshold of soil moisture. The sensitivity index reflects the sensitivity of crop growth rate to different irrigation water distributions, while the dynamic balance threshold of soil moisture indicates the equilibrium state of soil moisture required to maintain normal crop growth.

[0091] Step S125: Aggregate the water efficiency correlation sub-results of each monitoring period to generate a farmland water efficiency correlation analysis result for the target farmland area.

[0092] The farmland water efficiency correlation analysis results are used to characterize the nonlinear interaction patterns among the soil moisture information, irrigation water volume data, and crop growth index data.

[0093] After obtaining the water efficiency correlation sub-results for each monitoring period, these results need to be aggregated to generate the farmland water efficiency correlation analysis results for the target farmland area. The purpose of aggregation is to comprehensively consider information from multiple monitoring periods and more comprehensively reflect the nonlinear interactions between soil moisture information, irrigation water data, and crop growth indicator data.

[0094] A weighted average approach can be used for aggregation. First, a weight is assigned to each water efficiency-related sub-result for each monitoring period. Weight assignment can be determined based on factors such as the importance of the monitoring period and the reliability of the data. For example, a higher weight can be assigned to the most recent monitoring period, as recent data better reflects the current farmland conditions. Similarly, higher weights can be assigned to monitoring periods with higher data quality.

[0095] Next, multiply the water efficiency sub-results for each monitoring period by their corresponding weights and sum these weighted results. During the summation process, ensure that the characteristic dimensions and dimensions of each water efficiency sub-result are consistent. If there are any inconsistencies in dimensions or dimensions, adjust accordingly before aggregation.

[0096] Through weighted average aggregation, a comprehensive feature vector is obtained, which is used as the result of the farmland water efficiency correlation analysis in the target farmland area. This result can reflect the nonlinear interaction between soil moisture information, irrigation water data, and crop growth index data over multiple monitoring cycles.

[0097] Step S130: Based on a preset water efficiency optimization strategy network, a strategy matching process is performed on the farmland water efficiency association analysis result to generate a water efficiency optimization control plan for the target farmland area.

[0098] Step S131: inputting the soil-irrigation dynamic coupling characteristics and crop growth sensitivity index in the farmland water efficiency association analysis results into the first strategy layer of the water efficiency optimization strategy network to generate irrigation time window optimization suggestions.

[0099] The pre-set water efficiency optimization strategy network is a trained neural network model that generates optimized water efficiency control strategies based on the input farmland water efficiency correlation analysis results. The network consists of multiple strategy layers, each responsible for processing different types of information and generating corresponding optimization recommendations.

[0100] The dynamic coupling characteristics of soil and irrigation and crop growth sensitivity indicators from the farmland water efficiency correlation analysis are input into the first strategy layer of the water efficiency optimization strategy network. The first strategy layer is a neural network layer with initialized structure and parameters, which processes and analyzes the input features.

[0101] In the first strategy layer, soil-irrigation dynamic coupling characteristics and crop growth sensitivity indicators are first linearly combined and nonlinearly transformed through a series of neurons. The linear combination of neurons multiplies the input features by the corresponding weights and sums them to obtain an intermediate result. This intermediate result is then transformed through a nonlinear activation function to obtain the neuron output.

[0102] Nonlinear activation functions introduce nonlinear factors, enabling the network to handle complex nonlinear relationships. Examples of nonlinear activation functions include the sigmoid function and the ReLU function. Through multiple linear combinations and nonlinear transformations, the first strategy layer extracts useful information from the input features and generates optimized irrigation time window recommendations based on this information.

[0103] Irrigation time window optimization recommendations determine when irrigation can best meet crop growth needs while improving water use efficiency. For example, based on soil-irrigation dynamic coupling characteristics and crop growth sensitivity indicators, the first strategy layer may recommend irrigation during a specific period of crop growth, when soil moisture is low and crop water demand is high.

[0104] Step S132: Based on the soil moisture dynamic balance threshold in the farmland water efficiency association analysis result and the current irrigation water distribution characteristics, the second strategy layer of the water efficiency optimization strategy network is called to calculate the irrigation water adjustment gradient.

[0105] Next, based on the soil moisture dynamic balance threshold and current irrigation water distribution characteristics from the farmland water efficiency correlation analysis results, the second strategy layer of the water efficiency optimization strategy network is called. This second strategy layer is also a neural network layer that receives the soil moisture dynamic balance threshold and current irrigation water distribution characteristics as input.

[0106] In the second strategy layer, the soil moisture dynamic balance threshold and the current irrigation water distribution characteristics are first extracted and processed. Through a series of neural network operations, the input features are transformed and combined to obtain more representative intermediate features.

[0107] Then, an irrigation water adjustment gradient is calculated based on these intermediate features. The irrigation water adjustment gradient indicates the direction and magnitude of the adjustment required to adjust the current irrigation water flow in order to bring the soil moisture content to a dynamic equilibrium threshold. For example, if the current soil moisture content is below the dynamic equilibrium threshold, the irrigation water adjustment gradient may indicate a need for an increase in irrigation water flow; if the current soil moisture content is above the dynamic equilibrium threshold, the irrigation water adjustment gradient may indicate a need for a decrease in irrigation water flow.

[0108] The calculation of the irrigation water adjustment gradient takes into account factors such as soil water holding capacity and crop water requirements. The second strategy layer comprehensively analyzes the input features based on these factors and, through the learning and reasoning capabilities of the neural network, derives a reasonable irrigation water adjustment gradient.

[0109] Step S133: The irrigation time window optimization suggestion and the irrigation water volume adjustment gradient are jointly solved by a multi-objective optimization algorithm to generate a preliminary control strategy that meets the Pareto optimal condition.

[0110] After obtaining the irrigation time window optimization recommendations and irrigation water adjustment gradients, they need to be jointly solved using a multi-objective optimization algorithm to generate a preliminary control strategy that meets Pareto optimality. The goal of a multi-objective optimization algorithm is to find an optimal solution among multiple conflicting objectives.

[0111] In this scenario, the objectives include optimizing irrigation time windows and adjusting irrigation water volumes to achieve the dual goals of improving water use efficiency and meeting crop growth requirements. Pareto optimality refers to a solution where one objective cannot be further improved without also decreasing other objectives.

[0112] The multi-objective optimization algorithm takes the irrigation time window optimization recommendations and the irrigation water adjustment gradient as input, taking into account the interrelationships and constraints between them. For example, the choice of irrigation time window affects the irrigation water demand, while the adjustment of irrigation water quantity also affects soil moisture and crop growth.

[0113] The algorithm iteratively searches for possible control strategies and evaluates each strategy's performance on various objectives. In each iteration, the algorithm generates new control strategies based on the set rules and calculates the scores of these strategies on various objectives. Through continuous iteration and comparison, a set of preliminary control strategies that meet Pareto optimality conditions is eventually found. These preliminary control strategies may include multiple different combinations, each of which balances the optimization of irrigation time windows and irrigation water volume to varying degrees. For example, a preliminary control strategy may recommend irrigation within a specific time period, adjusting the water volume according to a set gradient, to improve water efficiency while ensuring normal crop growth.

[0114] Step S134: Perform feasibility verification processing on the preliminary control strategy based on the execution effect data of the historical control strategy, and screen a set of feasible strategies that meet the preset water-saving target.

[0115] Step S1341: Acquire the irrigation control strategies implemented in the historical period and their corresponding soil moisture change data and crop growth index deviation data.

[0116] To verify the feasibility of the preliminary control strategy, we first need to obtain historical data on irrigation control strategies implemented over time, along with corresponding soil moisture change data and crop growth indicator deviation data. This historical data can be data from multiple past monitoring periods, documenting the actual farmland conditions under different irrigation control strategies.

[0117] The data collection system collects the specific details of each irrigation control strategy over the historical period, including information such as irrigation time, irrigation water volume, and irrigation method. It also collects corresponding soil moisture change data. This data, acquired through soil moisture sensors, reflects the dynamic changes in soil moisture after the irrigation control strategy was implemented.

[0118] Additionally, crop growth indicator deviation data is collected. This refers to the difference between actual and expected crop growth indicators. Expected crop growth indicators can be predicted based on crop growth models and historical data, while actual crop growth indicators are obtained through field measurements or remote sensing monitoring.

[0119] These historical data are collated and stored to form a historical data set for subsequent feasibility verification and analysis.

[0120] Step S1342: constructing a strategy effect evaluation matrix, wherein the rows of the strategy effect evaluation matrix represent different strategy types, and the columns represent water-saving efficiency, crop growth impact, and soil water retention stability indicators.

[0121] Based on the historical data obtained, a strategy effect evaluation matrix was constructed. The strategy effect evaluation matrix is a two-dimensional matrix, whose rows represent different irrigation control strategy types and columns represent water-saving efficiency, crop growth impact, and soil water retention stability indicators.

[0122] For each irrigation control strategy type, its performance in terms of water conservation efficiency, crop growth impact, and soil water retention stability is calculated based on historical data. Water conservation efficiency can be measured by calculating the ratio of water saved after implementing the strategy to total water consumption; crop growth impact can be assessed by using data on deviations in crop growth indicators, such as changes in plant height and yield; and soil water retention stability can be analyzed using data on soil moisture changes, such as the range of soil moisture fluctuations and how long it lasts.

[0123] The performance values of each strategy type on each indicator are entered into the strategy effect evaluation matrix to form a complete evaluation matrix. This matrix can intuitively reflect the effects of different irrigation control strategies in multiple aspects and provide a basis for subsequent strategy screening.

[0124] Step S1343: Map the preliminary control strategy to the strategy effect evaluation matrix, and calculate its similarity score with historical successful strategies.

[0125] The preliminary control strategy is mapped to a strategy effectiveness evaluation matrix to calculate its similarity score with historically successful strategies. First, it is necessary to determine the expected performance of the preliminary control strategy in terms of water conservation efficiency, crop growth impact, and soil water retention stability. Based on the specific content of the preliminary control strategy, combined with the actual farmland conditions and crop growth characteristics, its performance on each indicator can be predicted.

[0126] The expected performance of the initial control strategy on each indicator is then compared with the corresponding indicator values of historically successful strategies in the strategy effectiveness evaluation matrix. Historically successful strategies are irrigation control strategies that have performed well in terms of water conservation efficiency, crop growth impact, and soil water retention stability during past implementation.

[0127] Calculate the similarity score between the initial control strategy and the historically successful strategy using a similarity calculation method, such as Euclidean distance or cosine similarity. A higher similarity score indicates a more similar initial control strategy to the historically successful strategy, and a higher feasibility score.

[0128] Step S1344: Determine the adaptability level of the preliminary control strategy based on the similarity score and the crop growth stage requirements of the current farmland area.

[0129] The adaptability of the initial regulation strategy is determined based on the calculated similarity score and the crop growth stage requirements of the current farmland area. Different crops have different water requirements at different growth stages, so the current crop growth stage needs to be considered when evaluating the feasibility of the initial regulation strategy.

[0130] First, based on the characteristics of the crop growth stage, determine the key indicators for water conservation efficiency, crop growth impact, and soil water retention stability. For example, in the early stages of crop growth, more attention may be paid to soil water retention stability to ensure good root development; in the later stages of crop growth, more attention may be paid to water conservation efficiency and crop growth impact to increase yield and conserve water resources.

[0131] Then, combined with the similarity score, the adaptability of the preliminary control strategy to the current crop growth stage is comprehensively evaluated. If the preliminary control strategy has a high similarity score with the historically successful strategy and performs well on the key indicators of the current crop growth stage, it can be given a higher adaptability grade. Conversely, if the similarity score is low or the performance is poor on the key indicators of the current crop growth stage, it will be given a lower adaptability grade.

[0132] Step S1345: Filtering strategies whose adaptability levels are higher than a preset threshold and whose water-saving efficiency reaches the target range to generate the feasible strategy set. The feasibility verification process includes soil water migration simulation and crop transpiration effect compensation calculation.

[0133] Strategies with adaptability levels above a preset threshold and water-saving efficiencies within a target range are screened to generate a set of feasible strategies. The preset threshold is a standard set based on actual needs and experience to select strategies with good adaptability. The target range is a reasonable range of water-saving efficiencies that should be achieved, neither too low to waste water resources nor too high to affect normal crop growth.

[0134] During the screening process, feasibility verification processes such as soil moisture transport simulation and crop transpiration compensation calculations are also conducted. Soil moisture transport simulation establishes a soil moisture transport model to simulate the distribution and changes in soil moisture after the implementation of the initial control strategy to ensure that soil moisture can meet the growth needs of crops. Crop transpiration compensation calculations consider the water lost through transpiration during crop growth and calculate the amount of water needed to maintain crop water balance based on factors such as crop transpiration rate and meteorological conditions.

[0135] Through these feasibility verification processes, we screen out strategies with adaptability levels above a preset threshold and water-saving efficiencies within the target range. These strategies are then organized into a feasible strategy set. The strategies in this set are theoretically highly feasible and effective, achieving water-saving goals while ensuring crop growth.

[0136] Step S135: Generate the water efficiency optimization control plan based on the strategy priority ranking result in the feasible strategy set, wherein the water efficiency optimization control plan includes irrigation parameter adjustment instructions implemented in stages and expected water-saving benefit indicators.

[0137] Step S1351: Determine the water-saving benefit weight, crop growth risk coefficient, implementation cost parameter, and equipment compatibility index of each feasible strategy to form a multi-dimensional evaluation vector.

[0138] After obtaining a set of feasible strategies, a comprehensive evaluation is conducted on each strategy in the set to determine their priority. First, the water-saving benefit weight, crop growth risk factor, implementation cost parameter, and equipment compatibility index of each feasible strategy are determined to form a multidimensional evaluation vector.

[0139] The water-saving benefit weight refers to the importance of the strategy in terms of water conservation, which can be determined based on factors such as the water resource status of the farmland and the water conservation target. For example, in areas with relatively scarce water resources, the water-saving benefit weight may be set higher.

[0140] The crop growth risk factor indicates the degree of risk to crop growth resulting from implementing a strategy. This factor can be used to assess the likelihood of adverse impacts on crop growth based on the strategy's impact on factors such as soil moisture and irrigation timing, combined with crop growth characteristics. The implementation cost parameter includes the human, material, and financial resources required to implement the strategy. For example, strategies that require replacing irrigation equipment may have high implementation costs.

[0141] The device compatibility index refers to the compatibility of the policy with the current farmland irrigation equipment. If the policy requires the use of specific irrigation equipment, but the current farmland equipment does not meet the requirements, then the policy's device compatibility is low.

[0142] The performance values of each feasible strategy in terms of water-saving benefit weight, crop growth risk coefficient, implementation cost parameter, and equipment compatibility index are combined to form a multi-dimensional evaluation vector that can comprehensively reflect the comprehensive performance of each feasible strategy.

[0143] Step S1352: performing weighted summation on the water-saving benefit weight, the crop growth risk coefficient, and the implementation cost parameter according to a multi-objective optimization algorithm to generate a comprehensive strategy score.

[0144] A multi-objective optimization algorithm is used to weight the water-saving benefit weights, crop growth risk factors, and implementation cost parameters in the multidimensional evaluation vector to generate a comprehensive strategy score. The multi-objective optimization algorithm assigns a weight to each indicator, reflecting its importance in the comprehensive evaluation.

[0145] For example, assuming the water-saving benefit weight is a, the crop growth risk coefficient weight is b, and the implementation cost parameter weight is c, and a + b + c = 1. For each feasible strategy, its water-saving benefit weight value is multiplied by a, its crop growth risk coefficient value is multiplied by b, and its implementation cost parameter value is multiplied by c. These three products are then added together to obtain a preliminary score for the strategy.

[0146] However, since the dimensions of these three indicators may differ, they need to be normalized or standardized before performing the weighted summation to ensure the rationality of the calculation results. For the water-saving benefit weight, its value can be mapped to a specific range, such as the [0, 1] interval. The specific method can be to divide the water-saving benefit weight value of the strategy by the maximum water-saving benefit weight of all feasible strategies. For the crop growth risk coefficient, its value is also normalized to be in the [0, 1] interval. The higher the risk coefficient, the closer the normalized value is to 1. For the implementation cost parameter, a similar normalization operation is performed to convert its value to the [0, 1] interval. The higher the cost, the closer the normalized value is to 1.

[0147] After normalization, the normalized water-saving benefit weights, crop growth risk coefficients, and implementation cost parameters are weighted and summed according to the weighted summation method described above to obtain a comprehensive strategy score for each feasible strategy. This score comprehensively considers multiple factors, including water-saving benefits, crop growth risks, and implementation costs, and can more comprehensively reflect the advantages and disadvantages of each strategy.

[0148] Step S1353: Arrange the feasible strategy set in descending order according to the strategy comprehensive scores to generate a strategy priority sequence.

[0149] After obtaining the comprehensive score of each feasible strategy, the set of feasible strategies is sorted in descending order according to these scores. The purpose of descending order is to put strategies with higher comprehensive scores at the top. These strategies have better comprehensive performance in terms of water conservation benefits, crop growth risks, and implementation costs, and thus have higher priority.

[0150] The sorting process can use common sorting algorithms, such as bubble sort and quick sort. Taking bubble sort as an example, starting with the first strategy in the set of feasible strategies, the comprehensive scores of two adjacent strategies are compared in sequence. If the score of the first strategy is lower than the score of the second strategy, the positions of the two strategies are swapped. After multiple iterations and comparisons, the set of feasible strategies is finally sorted from highest to lowest comprehensive score, generating a strategy priority sequence.

[0151] Step S1354: Select the top N strategies in the priority sequence as candidate control strategies, where N is dynamically determined according to the adjustment capability of the current irrigation system.

[0152] According to the generated strategy priority sequence, the top N strategies are selected as candidate control strategies. Here, N is not a fixed value, but is dynamically determined according to the adjustment capacity of the current irrigation system.

[0153] The current irrigation system's adjustment capability is influenced by multiple factors, including equipment performance, operator skill, and capital investment. If the irrigation system has advanced equipment, highly skilled operators, and sufficient funding, then the system's adjustment capability is strong, and the value of N can be relatively large, allowing for the selection of more candidate control strategies for further evaluation and implementation. Conversely, if the irrigation system's adjustment capability is weak, the value of N should be reduced accordingly to ensure that the candidate control strategies selected are actually implementable by the irrigation system.

[0154] A specific method for determining N can be to establish an evaluation model based on various irrigation system adjustment capacity indicators. For example, indicators such as equipment performance, operator skill level, and capital investment can be quantified. Then, an adjustment capacity score can be calculated based on these quantitative indicators. Based on this score, combined with pre-defined rules, the value of N can be determined.

[0155] Step S1355: Perform time conflict detection and resource constraint verification on the candidate control strategies to generate a conflict-free strategy subset.

[0156] After the candidate control strategies are selected, time conflict detection and resource constraint verification need to be performed on these strategies to ensure that they can be successfully implemented in practice.

[0157] Temporal conflict detection checks for conflicts in the implementation timeframes of candidate control strategies. Each candidate control strategy has specific implementation timeframes. For example, some strategies may require irrigation operations to occur within a specific timeframe. If the implementation times of two or more candidate control strategies overlap, a temporal conflict arises. To perform temporal conflict detection, the implementation timeframes of each candidate control strategy must be collated and analyzed. By comparing the timeframes of different strategies, conflicting strategy combinations can be identified.

[0158] Resource constraint verification checks whether candidate control strategies exceed available resource limits in their resource usage. Resources include water, electricity, and human resources. For example, some candidate control strategies may require significant amounts of water for irrigation. If the current water supply cannot meet these requirements, these strategies are not feasible. Resource constraint verification involves a detailed analysis of each candidate control strategy's resource requirements, combined with the currently available resources, to determine whether the strategy satisfies the resource constraints.

[0159] Candidate control strategies that have time conflicts or do not meet resource constraints are removed from the candidate set, ultimately generating a conflict-free strategy subset. This subset contains strategies that do not conflict with each other in time and do not exceed the actual available resource limits, making it more feasible.

[0160] Step S1356: Combining and optimizing the non-conflicting strategy subsets according to the implementation timeline to generate the water efficiency optimization control plan, which includes emergency control instructions and long-term optimization recommendations.

[0161] After obtaining a subset of conflict-free strategies, the strategies in the subset are combined and optimized according to the implementation timeline to generate a water efficiency optimization control plan.

[0162] First, the subset of non-conflicting strategies is arranged in chronological order based on the implementation time requirements of each strategy. Then, strategies are combined, considering their interrelationships and synergies. For example, some strategies might create better conditions for subsequent strategies when implemented earlier, or some strategies might produce better water efficiency optimization results when combined.

[0163] The combination optimization process also requires consideration of the actual farmland conditions and crop growth stages. The strategy combination is adjusted based on the soil type, climate conditions, and the water requirements of crops at different growth stages to ensure that the combined strategy best meets the needs of the farmland and crops.

[0164] After combined optimization, the resulting water efficiency optimization plan consists of two parts: emergency control instructions and long-term optimization recommendations. Emergency control instructions are immediate response measures for potential farmland emergencies, such as a sudden drop in soil moisture or crop drought. These instructions are typically highly targeted and timely, effectively alleviating critical farmland situations within a short period of time.

[0165] Long-term optimization recommendations plan and guide farmland irrigation strategies from a long-term perspective. These recommendations take into account overall water efficiency improvements and sustainable crop growth, including long-term planning of irrigation schedules, rational allocation of irrigation water volumes, and upgrades to irrigation equipment. By implementing these recommendations, farmland water utilization efficiency can be gradually improved, achieving sustainable water resource utilization.

[0166] Step S140: Feedback the water efficiency optimization control scheme to the farmland irrigation control system to trigger an irrigation parameter adjustment operation.

[0167] Step S141: converting the irrigation time window optimization suggestion in the water efficiency optimization control scheme into a timing trigger instruction of the irrigation controller, and performing a drip irrigation belt pressure calibration operation during the non-irrigation period.

[0168] After generating a water efficiency optimization control plan, the irrigation time window optimization recommendations in the plan need to be fed back to the farmland irrigation control system to trigger the corresponding irrigation parameter adjustment operations. First, the irrigation time window optimization recommendations are converted into timing trigger instructions for the irrigation controller.

[0169] Irrigation time window optimization recommendations are typically given in the form of time ranges, such as suggesting irrigation within a certain time period. In order for the irrigation controller to accurately execute these recommendations, the time ranges need to be converted into specific timing trigger instructions. This can be achieved by interfacing with the irrigation controller's time system, converting the recommended irrigation time ranges into time points and durations that the controller can recognize. For example, if irrigation is recommended from 9:00 AM to 11:00 AM, this time range can be converted into instructions for the controller to trigger irrigation at 9:00 AM for two hours.

[0170] Perform drip tape pressure calibration during non-irrigation periods. Drip tape pressure significantly impacts irrigation effectiveness. Unstable or inaccurate pressure can lead to uneven irrigation and affect crop growth. During non-irrigation periods, when the irrigation system is idle, this is an appropriate time to perform drip tape pressure calibration.

[0171] Pressure calibration is achieved using a pressure sensor installed on the drip irrigation tape. The pressure sensor monitors the pressure inside the drip irrigation tape in real time and feeds these values back to the irrigation control system. The irrigation control system then adjusts the pressure in the drip irrigation tape according to a preset standard pressure. If the monitored pressure is higher than the standard, the system reduces the pressure by adjusting valves or other means. If the monitored pressure is lower than the standard, the system increases the pressure until it reaches the standard range.

[0172] Step S142: generating a pump station flow control parameter and a valve opening adjustment curve according to the irrigation water volume adjustment gradient.

[0173] Based on the irrigation water flow adjustment gradient in the water efficiency optimization control plan, pump station flow control parameters and valve opening adjustment curves are generated. The irrigation water flow adjustment gradient indicates the direction and magnitude of the adjustment required to achieve the goal of water efficiency optimization.

[0174] The pump station flow control parameter refers to the flow rate the pump station is required to deliver during irrigation. The flow rate the pump station needs to provide is calculated based on the irrigation water adjustment gradient and the actual needs of the farmland. This calculation takes into account multiple factors, such as farmland size, crop water requirements, and irrigation method. For example, if the irrigation water adjustment gradient indicates a need for an increase in irrigation volume, the pump station's flow control parameter should be increased accordingly; conversely, if a decrease in irrigation volume is required, the pump station's flow control parameter should be decreased.

[0175] The valve opening adjustment curve describes how valve opening changes over time. Valve opening directly affects the amount of irrigation water delivered. By adjusting the valve opening, irrigation water volume can be precisely controlled. The valve opening adjustment curve is generated based on the irrigation water volume adjustment gradient and the pump station's flow control parameters. During irrigation, the irrigation control system adjusts the valve opening in real time based on this curve to ensure that irrigation water volume changes according to the desired adjustment gradient.

[0176] Step S143: writing the adjusted irrigation parameters into the irrigation equipment control interface, executing the nozzle angle optimization sub-operation, and starting the parameter verification mechanism.

[0177] After generating adjusted irrigation parameters, such as pump station flow control parameters and valve opening adjustment curves, these parameters are written to the irrigation equipment control interface. The irrigation equipment control interface is the channel for information exchange between the irrigation control system and the irrigation equipment. By writing the adjusted irrigation parameters to this interface, the irrigation equipment can receive the new control instructions and operate according to them.

[0178] After the parameters are written, the parameter verification mechanism is activated. The purpose of this parameter verification mechanism is to ensure the accuracy and rationality of the written irrigation parameters. The irrigation control system will check the written parameters to verify whether they are within the device's allowable range and whether they meet the requirements of the water efficiency optimization control plan. For example, it checks whether the pump station flow control parameters exceed the pump station's maximum output flow rate and whether the valve opening adjustment curve is within the valve's adjustable range.

[0179] If the parameter verification passes, it means that the written irrigation parameters are valid and the irrigation equipment can perform normal irrigation operations according to these parameters; if the parameter verification fails, the irrigation control system will issue an alarm and prompt the operator to correct the parameters until the parameter verification passes.

[0180] Step S144: deploy flow sensors at key nodes of the irrigation network to collect actual irrigation water volume data for each region.

[0181] To monitor actual water consumption during irrigation operations, flow sensors are deployed at key nodes in the irrigation network. These include locations such as irrigation water source inlets and branch pipe junctions. Flow sensors measure the volume of water flowing through these nodes in real time. By deploying flow sensors at multiple key nodes, actual irrigation water consumption data can be collected for each region.

[0182] Each flow sensor transmits collected water volume data to the irrigation control system in real time. The control system compiles and analyzes this data to determine the actual water consumption for each area during the irrigation process. For example, by analyzing data from flow sensors on different branch pipes, the distribution of irrigation water within each branch can be determined.

[0183] Step S145: Acquire the regulated soil moisture spatial distribution change data through the soil moisture sensor network.

[0184] In addition to collecting actual irrigation water volume data, a soil moisture sensor network is also used to obtain data on the spatial distribution of soil moisture after regulation. This network consists of multiple soil moisture sensors distributed across the farmland. These sensors monitor soil moisture in real time and transmit this data to the irrigation control system.

[0185] During irrigation operations, soil moisture changes as the amount of irrigation water varies. A network of soil moisture sensors can be used to obtain data on soil moisture variations at different locations on the farmland, thereby understanding the spatial distribution of soil moisture. For example, by analyzing data from soil moisture sensors at different locations, it is possible to identify areas with faster and slower increases in soil moisture, allowing for timely adjustments to irrigation strategies.

[0186] Step S146: Calculating actual water-saving efficiency indicators, where the actual water-saving efficiency indicators include a water consumption reduction rate per unit output and an improvement degree of soil water use efficiency.

[0187] Based on the collected data on actual irrigation water volume and spatial distribution changes in soil moisture, the actual water-saving efficiency index is calculated. The actual water-saving efficiency index includes the reduction rate of water consumption per unit of output and the improvement degree of soil water use efficiency.

[0188] The water consumption reduction rate per unit of yield refers to the percentage decrease in water consumption per unit of yield after implementing a water efficiency optimization program. Calculating the water consumption reduction rate per unit of yield requires statistically analyzing crop yields and irrigation water usage before and after the program's implementation. By comparing water consumption per unit of yield before and after implementation, the percentage decrease is calculated.

[0189] Improved soil water use efficiency (SWUE) refers to the percentage increase in the effective utilization of soil water by crops after implementing a control plan, relative to the pre-implementation rate. Calculating SWUE requires a comprehensive consideration of both soil moisture changes and crop growth. For example, analyzing soil moisture trends and crop growth indicators can be used to assess whether crop absorption and utilization efficiency has improved.

[0190] Step S147: After performing time lag compensation on the actual water-saving efficiency index according to the soil type permeability parameter, a dynamic difference analysis is performed with the expected index in the water efficiency optimization control scheme.

[0191] Because different soil types have different permeabilities, the speed at which water is transported through the soil also varies. This can cause a time lag in the actual water-saving efficiency indicator. To accurately assess actual water-saving efficiency, it is necessary to compensate for this time lag based on the soil type permeability parameter.

[0192] The soil permeability parameter reflects the water penetration capacity of different soil types. For soils with higher permeability, water transfers faster and with a relatively smaller time lag. For soils with lower permeability, water transfers slower and with a relatively larger time lag. Based on the soil permeability parameter, the calculation time of the actual water-saving efficiency indicator is adjusted to compensate for the effect of this time lag.

[0193] After time lag compensation, a dynamic difference analysis is performed between the actual water-saving efficiency index and the expected index in the water efficiency optimization control plan. Dynamic difference analysis compares the degree of difference between the actual and expected indicators and analyzes how this difference changes over time. This dynamic difference analysis can promptly identify deviations between actual irrigation results and expected targets so that appropriate adjustments can be taken.

[0194] Step S148: When the difference exceeds the preset threshold, a strategy adjustment request is generated and the re-execution of the farmland water efficiency association analysis process is triggered, and the parameter weights of the water efficiency optimization strategy network are updated according to the re-generated farmland water efficiency association analysis results to form a closed-loop optimization mechanism.

[0195] Based on the results of the dynamic difference analysis, when the difference between the actual water-saving efficiency index and the expected index exceeds the preset threshold, it indicates that there may be problems with the current irrigation strategy and that it needs to be adjusted. At this time, a strategy adjustment request is generated.

[0196] The policy adjustment request will include information such as the difference between the actual water-saving efficiency index and the expected index, the current irrigation parameters, etc. The irrigation control system will send the request to the relevant decision-making module, triggering the re-execution of the farmland water efficiency correlation analysis process.

[0197] When re-running the farmland water efficiency correlation analysis, a new set of dynamic monitoring data for the target farmland area is acquired, including soil moisture information, irrigation water volume data, and crop growth indicator data. This data is then analyzed according to the previous multi-dimensional correlation analysis steps to generate new farmland water efficiency correlation analysis results. Based on these new analysis results, the water efficiency optimization strategy network is re-invoked for strategy matching, generating a new water efficiency optimization control plan to correct deviations in the current irrigation strategy and improve water efficiency.

[0198] After re-running the farmland water efficiency correlation analysis and generating new farmland water efficiency correlation analysis results, the parameter weights of the water efficiency optimization strategy network are updated based on these results. The water efficiency optimization strategy network is a trained neural network model, and its parameter weights determine the model's performance and output results.

[0199] The water efficiency optimization strategy network is retrained using the new farmland water efficiency correlation analysis results as training data. During the training process, the network's parameter weights are adjusted so that the network can better generate accurate water efficiency optimization control plans based on the input farmland water efficiency correlation analysis results.

[0200] By continuously monitoring actual irrigation results, performing variance analysis, re-performing correlation analysis, and updating network parameter weights, a closed-loop optimization mechanism is formed. This closed-loop optimization mechanism can adjust irrigation strategies in real time based on actual farmland conditions and feedback on irrigation results, continuously improving farmland water use efficiency, achieving efficient water resource utilization and high-quality crop growth.

[0201] Step S149: When it is detected that the deviation between the actual data and the expected indicator exceeds the tolerance threshold, a dynamic retrospective adjustment of the control strategy is triggered.

[0202] The present disclosure also provides a device for mining and analyzing farmland water efficiency data, including:

[0203] processor, machine-readable storage medium;

[0204] The machine-readable storage medium is connected to the processor, and the machine-readable storage medium is used to store programs, instructions or codes. The processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0205] Figure 2The farmland water efficiency data mining and analysis device 100 shown includes: a processor 1001 and a memory 1003. The processor 1001 and the memory 1003 are connected, such as through a bus 1002. Optionally, the farmland water efficiency data mining and analysis device 100 may also include a communication component 1004, which can be used for data interaction between the device 100 and other devices, such as data transmission and / or data reception. It should be noted that in actual scheduling, the communication component 1004 is not limited to one, and the structure of the farmland water efficiency data mining and analysis device 100 does not constitute a limitation on the embodiments of the present application.

[0206] The processor 1001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It may implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. The processor 1001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like.

[0207] The bus 1002 may include a path for transmitting information between the above components. The bus 1002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 1002 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 2 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0208] The memory 1003 can be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic disk storage media, other magnetic storage devices, or any other medium that can be used to carry or store program code and can be read by a computer, without limitation here.

[0209] The memory 1003 is used to store program codes for executing the embodiments of the present disclosure, and the execution is controlled by the processor 1001. The processor 1001 is used to execute the program codes stored in the memory 1003 to implement the steps shown in the above-mentioned farmland water efficiency data mining and analysis method embodiment.

[0210] The preferred embodiments of the present disclosure are described in detail above in conjunction with the accompanying drawings. However, the present disclosure is not limited to the specific details in the above embodiments. Within the technical concept of the present disclosure, various changes, modifications, replacements and variations can be made to these embodiments, and these changes, modifications, replacements and variations all fall within the scope of protection of the present disclosure.

[0211] It should also be noted that the various specific technical features described in the above specific embodiments may be combined in any suitable manner, unless there is any contradiction, and these combinations shall also be considered as the contents disclosed in this disclosure. To avoid unnecessary repetition, this disclosure will not further describe various possible combinations. The technical scope of this application is not limited to the contents of the specification and must be determined based on the scope of the claims.

Claims

1. A method for mining and analyzing farmland water efficiency data, characterized in that: Applied to a farmland irrigation monitoring device, the method includes: Obtaining a dynamic monitoring data set for a target farmland area, the dynamic monitoring data set including multiple types of soil moisture information, irrigation water volume data, and crop growth indicator data corresponding to different points and depths collected during multiple monitoring cycles, wherein the soil moisture information is used to represent vertically layered soil moisture information of the farmland at a single point; Performing multi-dimensional correlation analysis on the dynamic monitoring data set to generate a farmland water efficiency correlation analysis result for the target farmland area; Based on a preset water efficiency optimization strategy network, the strategy matching process is performed on the farmland water efficiency correlation analysis results to generate a water efficiency optimization control plan for the target farmland area; The water efficiency optimization control scheme is fed back to the farmland irrigation control system to trigger irrigation parameter adjustment operations.

2. The farmland water efficiency data mining and analysis method according to claim 1, characterized in that: The performing of multi-dimensional correlation analysis on the dynamic monitoring data set to generate a farmland water efficiency correlation analysis result for the target farmland area includes: For the soil moisture information, the irrigation water quantity data, and the crop growth index data in each monitoring period, performing time series alignment processing and standard ring conversion processing to generate a standardized monitoring data sequence; Calling a pre-trained association analysis model to perform spatiotemporal feature extraction processing on the standardized monitoring data sequence to generate soil moisture change characteristics, irrigation water distribution characteristics, and crop growth response characteristics corresponding to the monitoring period; constructing a first correlation feature vector based on a dynamic coupling relationship between the soil moisture change characteristic and the irrigation water distribution characteristic; generating a water efficiency correlation sub-result of the monitoring period according to a mapping relationship between the crop growth response characteristic and the first correlation feature vector; Aggregating the water efficiency correlation sub-results of each monitoring period to generate a farmland water efficiency correlation analysis result for the target farmland area; The farmland water efficiency correlation analysis results are used to characterize the nonlinear interaction patterns among the soil moisture information, irrigation water volume data, and crop growth index data.

3. The farmland water efficiency data mining and analysis method according to claim 2, characterized in that: The calling of the pre-trained association analysis model to perform spatiotemporal feature extraction processing on the standardized monitoring data sequence to generate soil moisture change characteristics, irrigation water distribution characteristics, and crop growth response characteristics corresponding to the monitoring period includes: Performing constrained spatial interpolation processing on the soil moisture information in the standardized monitoring data sequence in combination with the soil texture type distribution map to generate a soil moisture spatial distribution map; Performing time gradient calculation on the soil moisture spatial distribution map based on a sliding time window to extract soil moisture change characteristics; Performing sliding window outlier detection processing on the irrigation water volume data, eliminating instantaneous water flow fluctuation noise, and then performing irrigation event segmentation processing to identify the starting time point and water volume distribution curve of each irrigation operation, and extracting the irrigation water volume distribution characteristics in combination with the irrigation equipment response delay parameter; dividing the crop growth index data into growth stages, combining the divided growth stage data with the soil moisture variation characteristics in the corresponding time interval, calculating the response coefficient of the crop growth rate and soil moisture, and generating the crop growth response characteristics; The soil moisture change characteristics, the irrigation water distribution characteristics and the crop growth response characteristics are timestamp aligned to form a spatiotemporal correlation feature set.

4. The farmland water efficiency data mining and analysis method according to claim 3, characterized in that: The constructing of a first correlation feature vector based on the dynamic coupling relationship between the soil moisture change characteristic and the irrigation water distribution characteristic includes: determining a time-lagged correlation coefficient between a moisture accumulation gradient in the soil moisture variation characteristic and an irrigation water pulse in the irrigation water distribution characteristic; performing irrigation effect correction processing on the soil moisture change characteristic according to the time lag correlation coefficient to generate a corrected soil moisture response characteristic; The spatial distribution map of the modified soil moisture response characteristics and the time series data of the irrigation water distribution characteristics are processed by a spatiotemporal graph convolutional network, and the soil-irrigation dynamic coupling characteristics are extracted by fusing the spatial adjacency relationship and the time-dependent feature interaction; Performing feature dimensionality reduction processing on the soil-irrigation dynamic coupling feature to generate the first correlation feature vector, wherein the first correlation feature vector is used to characterize the dynamic impact pattern of the irrigation operation on the spatial distribution of soil moisture.

5. The farmland water efficiency data mining and analysis method according to claim 4, characterized in that: Generating the water efficiency correlation sub-result of the monitoring period according to the mapping relationship between the crop growth response characteristic and the first correlation feature vector includes: Performing a nonlinear regression analysis with regularization constraints on the growth rate data in the crop growth response characteristics and the first associated eigenvector, and determining a response function of crop growth to the soil-irrigation coupling effect in combination with a crop physiological water requirement threshold; Calculating crop growth simulation values under different irrigation water distributions based on the response function, and performing residual analysis on the crop growth index data actually collected; Adjusting the parameter weights of the response function according to the residual analysis results to generate an optimized water efficiency response model; Performing joint encoding processing on the optimized water efficiency response model and the first correlation feature vector to generate a water efficiency correlation sub-result of the monitoring period; The water efficiency correlation sub-results include the sensitivity index of crop growth to irrigation strategy and the soil moisture dynamic balance threshold.

6. The farmland water efficiency data mining and analysis method according to claim 1, characterized in that: The water efficiency optimization strategy network based on the preset water efficiency optimization strategy is used to perform strategy matching processing on the farmland water efficiency association analysis results to generate a water efficiency optimization control plan for the target farmland area, including: Inputting the soil-irrigation dynamic coupling characteristics and crop growth sensitivity indicators in the farmland water efficiency association analysis results into the first strategy layer of the water efficiency optimization strategy network to generate irrigation time window optimization suggestions; According to the soil moisture dynamic balance threshold value in the farmland water efficiency association analysis result and the current irrigation water distribution characteristics, calling the second strategy layer of the water efficiency optimization strategy network to calculate the irrigation water adjustment gradient; The irrigation time window optimization suggestion and the irrigation water adjustment gradient are jointly solved by a multi-objective optimization algorithm to generate a preliminary control strategy that meets the Pareto optimal conditions. Performing feasibility verification on the preliminary control strategy based on historical control strategy execution effect data to screen a set of feasible strategies that meet the preset water-saving goals, wherein the feasibility verification includes soil water movement simulation and crop transpiration effect compensation calculation; The water efficiency optimization and control scheme is generated according to the strategy priority ranking result in the feasible strategy set, wherein the water efficiency optimization and control scheme includes irrigation parameter adjustment instructions implemented in stages and expected water-saving benefit indicators.

7. The farmland water efficiency data mining and analysis method according to claim 6, characterized in that: The feasibility verification process of the preliminary control strategy is performed based on the execution effect data of the historical control strategy, and a feasible strategy set that meets the preset water-saving target is selected, including: Obtain the irrigation control strategies implemented in the historical period and their corresponding soil moisture change data and crop growth index deviation data; Constructing a strategy effect evaluation matrix, wherein the rows of the strategy effect evaluation matrix represent different strategy types, and the columns represent water-saving efficiency, crop growth impact, and soil water retention stability indicators; Mapping the preliminary control strategy to the strategy effect evaluation matrix and calculating its similarity score with historical successful strategies; determining an adaptability level of the preliminary regulation strategy based on the similarity score and the crop growth stage requirements of the current farmland area; Strategies whose adaptability levels are higher than a preset threshold and whose water-saving efficiency reaches a target range are screened to generate the feasible strategy set.

8. The farmland water efficiency data mining and analysis method according to claim 7, characterized in that: Generating the water efficiency optimization control scheme according to the strategy priority ranking result in the feasible strategy set includes: Determine the water-saving benefit weight, crop growth risk coefficient, implementation cost parameter, and equipment compatibility index of each feasible strategy to form a multi-dimensional evaluation vector; Performing a weighted summation of the water-saving benefit weight, the crop growth risk coefficient, and the implementation cost parameter according to a multi-objective optimization algorithm to generate a comprehensive strategy score; Arrange the feasible strategy set in descending order according to the comprehensive strategy scores to generate a strategy priority sequence; The top N strategies in the priority sequence are selected as candidate control strategies, where N is dynamically determined based on the adjustment capacity of the current irrigation system; Performing time conflict detection and resource constraint verification on the candidate control strategies to generate a conflict-free strategy subset; The conflict-free strategy subsets are combined and optimized according to the implementation timeline to generate the water efficiency optimization control plan, wherein the water efficiency optimization control plan includes two parts: emergency control instructions and long-term optimization suggestions.

9. The farmland water efficiency data mining and analysis method according to claim 6, characterized in that: The irrigation parameter adjustment operation includes drip irrigation belt pressure calibration and nozzle angle optimization sub-operations. Feedback of the water efficiency optimization control scheme to the farmland irrigation control system to trigger the irrigation parameter adjustment operation includes: Convert the irrigation time window optimization suggestion in the water efficiency optimization control scheme into a timing trigger instruction of the irrigation controller, and perform the drip irrigation belt pressure calibration operation during the non-irrigation period; generating a pump station flow control parameter and a valve opening adjustment curve according to the irrigation water adjustment gradient; Writing the adjusted irrigation parameters into the irrigation equipment control interface, executing the nozzle angle optimization sub-operation, and starting the parameter verification mechanism; Deploy flow sensors at key nodes of the irrigation network to collect actual irrigation water volume data for each region; The soil moisture sensor network is used to obtain the spatial distribution change data of the regulated soil moisture and calculate the actual water-saving efficiency index, which includes the reduction rate of water consumption per unit of output and the improvement degree of soil water use efficiency. After time lag compensation is performed on the actual water-saving efficiency index according to the soil type permeability parameter, a dynamic difference analysis is performed with the expected index in the water efficiency optimization control scheme; When the difference exceeds a preset threshold, a strategy adjustment request is generated and the re-execution of the farmland water efficiency association analysis process is triggered, and the parameter weights of the water efficiency optimization strategy network are updated according to the re-generated farmland water efficiency association analysis results to form a closed-loop optimization mechanism; When it is detected that the deviation between actual data and expected indicators exceeds the tolerance threshold, dynamic retrospective adjustment of the control strategy is triggered.

10. A farmland irrigation monitoring device, characterized in that: include: processor, machine-readable storage medium; The machine-readable storage medium is connected to the processor, and the machine-readable storage medium is used to store programs, instructions or codes. The processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

Images