A precise irrigation decision analysis method and system based on crop water demand

Abstract

The invention discloses a method and system for analyzing precise irrigation decision-making based on crop water demand, and specifically relates to the technical field of crop irrigation. The method divides the monitoring area and installs soil moisture sensors at different depths to monitor the moisture status data of the root zone in real time. The moisture gradient change characteristics and the moisture retention time deviation characteristics are extracted and analyzed to calculate the moisture fluctuation abnormality index, and the index is compared with a threshold value to divide the abnormal groundwater fluctuation into three levels: high, medium and low. Corresponding irrigation adjustment measures are taken according to the fluctuations at different levels: for high-level abnormal fluctuations, drainage measures are taken to reduce excessive soil moisture; for medium-level fluctuations, crop water demand model parameters are dynamically adjusted; for low-level fluctuations, normal irrigation strategies are maintained, which can effectively avoid the negative impact of groundwater fluctuations and excessive irrigation on the crop root zone and soil structure, maintain soil permeability and water storage capacity, ensure the long-term productivity of farmland, and improve irrigation efficiency and water resource utilization.

Description

Technical Field

[0001] The present invention relates to the technical field of crop irrigation, and in particular to a method and system for analyzing precise irrigation decisions based on crop water demand. Background Art

[0002] Precision irrigation decision analysis based on crop water requirements is an advanced agricultural method that combines crop growth needs and water resource management. The core is to understand the specific water requirements of crops at different growth stages through real-time monitoring and data analysis, and then use this information to optimize irrigation arrangements. This decision analysis process usually involves sensor technology and data processing systems. Soil moisture and meteorological sensors installed on the farm can collect environmental data in real time. After transmitting this data to the analysis system, the system will evaluate the current moisture status of the crop according to the set crop water requirement model to provide irrigation recommendations. This irrigation plan based on precise data helps to improve the efficiency of water resource utilization, meet the water requirements of crops, thereby increasing the yield and quality of crops while protecting the environment.

[0003] The prior art has the following deficiencies:

[0004] In some areas with high groundwater levels, rising or fluctuating groundwater levels can cause roots to obtain water without irrigation. Existing crop water demand models usually do not take into account seasonal or artificial fluctuations in groundwater levels and may overestimate water requirements. If irrigation is carried out when the groundwater level has already risen, it is easy to cause excess water in the roots, causing root rot or suffocation of crop roots. At the same time, continued excessive irrigation combined with rising groundwater may cause soil particles to be filled and saturated with water, resulting in soil compaction and poor drainage. Saturated soil is prone to lose oxygen and produce an anaerobic environment, which damages the root system and affects the ability of crops to absorb water and fertilizer. In addition, after the soil is compacted, its permeability and water storage capacity gradually decrease. Even after the groundwater level drops or the irrigation strategy is adjusted, the soil is difficult to recover, affecting the long-term productivity of the field. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for analyzing precise irrigation decisions based on crop water demand, so as to solve the deficiencies in the background technology.

[0006] In order to achieve the above object, the present invention provides the following technical solution: a method for analyzing precise irrigation decision-making based on crop water demand, comprising the following steps:

[0007] S1: Divide the crop planting area into several monitoring areas, and monitor at different depths through soil moisture sensors installed in each monitoring area to capture the moisture status data of the root zone;

[0008] S2: preprocessing the root zone water status data of crops at different depths, and extracting features of the preprocessed root zone water status data at different depths, respectively extracting water gradient change features and water retention time deviation features in the root zone water status data;

[0009] S3: Determine the weight assignment of the water status data of the root zone at different depths according to the water gradient change characteristics and water retention time deviation characteristics in the extracted water status data of the root zone, and calculate the water fluctuation abnormality index by weighted summation of the weight assignment of the water status data of the root zone at different depths;

[0010] S4: After comparing and analyzing the calculated water fluctuation anomaly index with the gradient standard threshold, the abnormal degree of groundwater fluctuation is divided into different levels, and the abnormal degree of groundwater fluctuation is divided into high-level abnormal fluctuation, medium-level abnormal fluctuation and low-level abnormal fluctuation;

[0011] S5: According to the abnormal degree of groundwater fluctuation, take corresponding irrigation adjustment measures for different levels of fluctuation. For high-level abnormal fluctuation, take drainage measures immediately to reduce soil moisture. For medium-level abnormal fluctuation, dynamically adjust the parameters of the crop water demand model. For low-level abnormal fluctuation, implement irrigation strategy normally.

[0012] Preferably, in S2, the moisture gradient change characteristics in the extracted root zone moisture status data are analyzed to generate a moisture gradient change rate drift index, and the method for obtaining the moisture gradient change rate drift index is:

[0013] From the soil moisture data collected at different depths, the moisture gradient change rate at each monitoring depth is calculated and expressed as a time series matrix X, where X∈R n×m :R is the real number domain, that is, the set of all real numbers, n is the number of sample points of time series data, m is the depth layer of soil monitoring, each column in X is standardized, and the standardized moisture gradient change rate data matrix X is calculated std The covariance matrix C of is expressed as: Where C∈R n×m represents the covariance of the rate of change of water gradient between depths, T is the matrix transpose, and the covariance matrix C is subjected to eigendecomposition to obtain the eigenvalues ​​and eigenvectors, expressed as: Cv i =λ i v i ; In the formula, λ i is the i-th eigenvalue of the covariance matrix, v i is the corresponding eigenvector, arrange the eigenvalues ​​in descending order, select the eigenvectors corresponding to the first k largest eigenvalues, and form the principal component matrix V k , V k=v1,v2,…,v k ]; The standardized data Projection to the principal component matrix V k On the basis of the above, we get the principal component score matrix Z, which is expressed as: Z = X std ·V k ; Z∈R n×m It is the principal component score matrix, each column is a principal component score; the absolute mean of the score of the first principal component is used as the moisture gradient change rate drift index, and the expression is: Where WD is the drift index of the moisture gradient change rate.

[0014] Preferably, in S2, the water retention time deviation characteristics in the extracted root zone water status data are analyzed to generate a water retention time deviation index, and the method for obtaining the water retention time deviation index is:

[0015] For each time point t and depth layer m, the water retention time data series S t,m , the calculation expression of exponential weighted moving average is: EWMA t,m =αS t,m +(1-α)EWMA t-1,m ; In the formula, EWMA t,m is the exponentially weighted moving average calculated at the current time t and depth level m, EWMA t-1,m is the exponentially weighted moving average of the previous time t-1 and depth layer m, α is the smoothing coefficient, 0<α≤1, and the initial condition is set to EWMA 0,m =S 0,m ; Calculate the residence time deviation of each depth layer m, the expression is: ΔS t,m =S t,m -EWMA t,m ;Deviation ΔS t,m It represents the difference between the current water retention time and the smoothed retention time. The water retention time deviation index is defined as the absolute mean of the retention time deviations of each depth layer. The expression is: D is the total number of monitored soil depth layers, and WT is the water residence time deviation index.

[0016] Preferably, in S3, the moisture gradient change rate drift index and the moisture retention time deviation index are converted into a first eigenvector, and the first eigenvector is used as the input of the machine learning model. The machine learning model uses the weighted assignment labels of the moisture status data of the root zone at different depths predicted by each group of first eigenvectors as the prediction target, and takes minimizing the sum of the prediction errors of the weighted assignment labels of the moisture status data of the root zone at different depths as the training target. The machine learning model is trained until the sum of the prediction errors converges and the model training is stopped. The weight assignment of the moisture status data of the root zone at different depths is determined according to the output results of the model, wherein the machine learning model is a polynomial regression model, and the moisture fluctuation abnormality index is obtained by weighted summing the weight assignments of the moisture status data of the root zone at different depths.

[0017] Preferably, in S4, after comparing and analyzing the calculated water fluctuation anomaly index with the gradient standard threshold, the abnormal degree of groundwater fluctuation is divided into different levels, specifically:

[0018] The obtained water fluctuation anomaly index is compared with the gradient standard threshold, the gradient standard threshold includes a first standard threshold and a second standard threshold, and the first standard threshold is less than the second standard threshold, and the water fluctuation anomaly index is compared with the first standard threshold and the second standard threshold respectively;

[0019] If the water fluctuation anomaly index is greater than the second standard threshold, it means that the abnormal degree of groundwater fluctuation is high, and a high fluctuation signal is generated at this time, and the abnormal degree of groundwater fluctuation is classified as high-level abnormal fluctuation;

[0020] If the water fluctuation anomaly index is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it means that the abnormal degree of groundwater fluctuation is medium. At this time, a medium fluctuation signal is generated, and the abnormal degree of groundwater fluctuation is classified as medium-level abnormal fluctuation.

[0021] If the moisture fluctuation anomaly index is less than the first standard threshold, it means that the abnormal degree of groundwater fluctuation is low. At this time, a low fluctuation signal is generated, and the abnormal degree of groundwater fluctuation is classified as low-level abnormal fluctuation.

[0022] Preferably, in S5, according to the abnormal degree of groundwater fluctuation, corresponding irrigation adjustment measures are taken for different levels of fluctuation. For high-level abnormal fluctuation, drainage measures are immediately taken to reduce soil moisture. For medium-level abnormal fluctuation, the parameters of the crop water demand model are dynamically adjusted. For low-level abnormal fluctuation, the irrigation strategy is normally implemented, specifically:

[0023] When a high-level abnormal fluctuation is detected, that is, the moisture fluctuation abnormal index is greater than the second standard threshold T2, drainage measures need to be taken immediately. The drainage rate formula is: D rate =k drain×(WAI-T2); where D rate is the drainage rate, which indicates the amount of water that needs to be discharged per unit time, k drain is the drainage coefficient, which reflects the efficiency of the drainage system and the soil drainage capacity, WAI is the water fluctuation anomaly index, and T2 is the second standard threshold;

[0024] For medium-level abnormal fluctuations, that is, the water fluctuation abnormal index WAI is greater than or equal to the first standard threshold T1 and less than or equal to the second standard threshold T2, the crop water requirement model adjustment formula is dynamically adjusted by adjusting the parameters of the crop water requirement model: Where I is the adjusted irrigation amount, which indicates the water demand after dynamic adjustment according to groundwater fluctuations; I0 is the initial irrigation amount, which indicates the target irrigation amount set in the crop water demand model under normal circumstances; q is the irrigation adjustment coefficient, which is used to control the adjustment range; and T1 is the first standard threshold.

[0025] The present invention also provides a precise irrigation decision analysis system based on crop water demand, including a data acquisition module, a feature extraction module, a moisture fluctuation abnormal index calculation module, an abnormal fluctuation grade classification module and an irrigation adjustment strategy module;

[0026] Data collection module: The crop planting area is divided into several monitoring areas, and soil moisture sensors installed in each monitoring area are used to monitor at different depths to capture the moisture status data of the root zone;

[0027] Feature extraction module: preprocess the root zone water status data of crops at different depths, and perform feature extraction on the preprocessed root zone water status data at different depths, respectively extracting the water gradient change characteristics and water retention time deviation characteristics in the root zone water status data;

[0028] Water fluctuation abnormality index calculation module: Determine the weight assignment of water status data in the root zone at different depths according to the water gradient change characteristics and water retention time deviation characteristics in the extracted water status data of the root zone, and perform weighted summation calculation on the weight assignment of water status data in the root zone at different depths to obtain the water fluctuation abnormality index;

[0029] Abnormal fluctuation level classification module: after comparing and analyzing the calculated water fluctuation abnormal index with the gradient standard threshold, the abnormal degree of groundwater fluctuation is divided into different levels, and the abnormal degree of groundwater fluctuation is divided into high-level abnormal fluctuation, medium-level abnormal fluctuation and low-level abnormal fluctuation;

[0030] Irrigation adjustment strategy module: According to the abnormal degree of groundwater fluctuation, corresponding irrigation adjustment measures are taken for different levels of fluctuation. For high-level abnormal fluctuations, drainage measures are immediately taken to reduce soil moisture. For medium-level abnormal fluctuations, the parameters of the crop water demand model are dynamically adjusted. For low-level abnormal fluctuations, the irrigation strategy is implemented normally.

[0031] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0032] 1. The present invention solves the problem of traditional irrigation models overestimating water demand and ignoring groundwater fluctuations by comprehensively analyzing the dynamic impact of groundwater levels on the moisture status of the root zone. Multi-layer soil moisture sensors are used to monitor the moisture status of the root zone, and moisture gradient changes and retention time characteristics are collected and analyzed at different depths to effectively identify the impact of groundwater level fluctuations on crop water demand. By calculating the moisture fluctuation anomaly index, groundwater fluctuation anomalies are divided into three levels: high, medium, and low, providing a reliable data basis for scientific irrigation. The introduction of the polynomial regression model makes the weight assignment more accurate, so that irrigation decisions are based on actual soil conditions and adapt to different environmental changes.

[0033] 2. The present invention adopts graded control measures for abnormal groundwater fluctuations of different levels. In case of high-level fluctuations, drainage is immediately adopted. In case of medium-level fluctuations, the irrigation amount is optimized by adjusting the parameters of the water demand model. In case of low-level fluctuations, the normal irrigation strategy is maintained, thereby avoiding problems such as root rot and soil compaction caused by excess water. By dynamically adjusting the irrigation strategy, the irrigation efficiency and water resource utilization rate are greatly improved, the healthy growth of crops is guaranteed, and the soil structure and long-term productivity of farmland are effectively maintained. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0035] Figure 1 The present invention is a flow chart of the method.

[0036] Figure 2 It is a system module diagram of the present invention. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are 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 creative work are within the scope of protection of the present invention.

[0038] Example 1, please refer to Figure 1 and Figure 2 As shown, the precise irrigation decision analysis method based on crop water requirement described in this embodiment includes the following steps:

[0039] S1: Divide the crop planting area into several monitoring areas, and monitor at different depths through soil moisture sensors installed in each monitoring area to capture the moisture status data of the root zone;

[0040] S2: preprocessing the root zone water status data of crops at different depths, and extracting features of the preprocessed root zone water status data at different depths, respectively extracting water gradient change features and water retention time deviation features in the root zone water status data;

[0041] S3: Determine the weight assignment of the water status data of the root zone at different depths according to the water gradient change characteristics and water retention time deviation characteristics in the extracted water status data of the root zone, and calculate the water fluctuation abnormality index by weighted summation of the weight assignment of the water status data of the root zone at different depths;

[0042] S4: After comparing and analyzing the calculated water fluctuation anomaly index with the gradient standard threshold, the abnormal degree of groundwater fluctuation is divided into different levels, and the abnormal degree of groundwater fluctuation is divided into high-level abnormal fluctuation, medium-level abnormal fluctuation and low-level abnormal fluctuation;

[0043] S5. According to the abnormal degree of groundwater fluctuation, corresponding irrigation adjustment measures are taken for different levels of fluctuation. For high-level abnormal fluctuation, drainage measures are immediately taken to reduce soil moisture. For medium-level abnormal fluctuation, the parameters of the crop water demand model are dynamically adjusted. For low-level abnormal fluctuation, the irrigation strategy is implemented normally.

[0044] In S1, the crop planting area is divided into several monitoring areas, and soil moisture sensors installed in each monitoring area are used to monitor at different depths to capture the moisture status data of the root zone, specifically:

[0045] According to factors such as soil type, topography, groundwater level, meteorological conditions, and crop types in the crop planting area, the planting area is divided into several monitoring areas so that each monitoring area has relatively consistent characteristics in these factors. The size of the monitoring area can be determined based on the growth requirements of the crop and the size of the farmland. Smaller monitoring areas facilitate more detailed data collection, while larger monitoring areas are more economical. Ensure that each divided area can represent the soil and environmental conditions in the area, especially in farmland areas where soil changes significantly, the division can be based on the location where the soil profile changes significantly. Specifically, the division can be based on the following main factors:

[0046] Differences in soil texture: Identify the texture changes of soil in different areas (such as sandy soil, clay soil, loam, etc.). Because soils of different textures have different water permeability and water retention, these areas with obvious differences should be divided separately when dividing the area.

[0047] Soil organic matter content: Areas with higher organic matter content in the soil generally have better water retention capacity, while areas with lower organic matter content tend to drain more easily. Monitoring different areas of organic matter content separately based on differences in organic matter distribution can help control irrigation more accurately.

[0048] Topography and slope: In areas with slopes or undulating terrain, water flow and infiltration characteristics can vary significantly, so flat and sloping land can be divided into different monitoring areas to more accurately manage water needs on different terrains.

[0049] Groundwater level: Areas with significantly different groundwater levels can affect the natural water supply to the root zone. Dividing regions based on changes in groundwater levels can help identify the impact of dynamic changes in water at different depths on crops.

[0050] Determine the number of soil moisture sensors based on the area and crop types of each monitoring area. Install at least one soil moisture sensor at the center or representative location of each area to capture moisture data in the root zone. Arrange sensors in layers at different soil depths in each monitoring area. For example, moisture sensors can be set in the surface layer (10 cm), middle layer (30 cm), and deep layer (50 cm or deeper) to capture the moisture status at different depths within the growth range of crop roots. This layered design can reflect the vertical distribution of soil moisture content and help identify the impact of rising groundwater on soil moisture at each layer.

[0051] Choose a soil moisture sensor that is suitable for the target soil type and environmental conditions. For example, resistive or time domain reflectometry (TDR) sensors are suitable for conventional soils, while heavier or sandy soils may require different types of sensors. The sensor installation depth should be as accurate as possible to avoid human installation errors affecting the monitoring depth. At the same time, the sensor should be in close contact with the soil to avoid gaps or excessive digging around the sensor to ensure the accuracy of the moisture data. After the sensor is installed, connect it to the data acquisition system (usually a data acquisition terminal or gateway) to transmit the data to the central monitoring system. The system can transmit through a wired or wireless network (such as LoRa or NB-IoT) to ensure that the data can be uploaded in real time.

[0052] The soil moisture sensor collects data according to a preset sampling frequency (e.g. every hour or every half hour) to capture real-time moisture changes at different depths. The collected data can be stored in the cloud or local server for subsequent analysis and historical data comparison. Through long-term data accumulation, the changing trend of root zone moisture in different seasons and climate conditions can be evaluated.

[0053] In the monitoring system, the soil moisture data of different depths in each area can be viewed in real time. If the moisture data at certain depths is found to be abnormal (such as continuously too high or too low), it can be judged that there may be groundwater level fluctuations or other soil problems. The collected data is analyzed to calculate the vertical moisture gradient and moisture retention time of each monitoring area, and the soil moisture status is evaluated in combination with meteorological data to provide support for subsequent irrigation decisions.

[0054] Soil moisture sensors may drift over time and require regular calibration to ensure data accuracy. Sensors in the monitoring area should be checked regularly to remove soil debris around them to avoid sensor damage or interruptions in data collection.

[0055] S2: Preprocess the root zone moisture status data of the crop at different depths, and perform feature extraction on the preprocessed root zone moisture status data at different depths, respectively extracting moisture gradient change characteristics and moisture retention time deviation characteristics from the root zone moisture status data.

[0056] Check the outliers in the data (such as extremely high or low humidity values) and remove possible erroneous data in combination with the reasonable range of soil moisture (such as 0-100%). The triple standard deviation method or box plot method can be used to identify outliers. For missing data, linear interpolation, mean interpolation and other methods can be used to fill in to ensure the integrity of the data. Soil moisture data may be interfered with by the external environment (such as sensor failure or sudden weather). Moving average or exponential smoothing can be used to remove random fluctuations and reduce noise. Sliding window smoothing technology is applied to reduce the impact of irregular fluctuations on data trends and highlight the changing trend of soil moisture. In order to facilitate the comparison and feature extraction of data at different depths, the data at different depths are normalized or standardized. The minimum-maximum normalization can be used to scale the data to a range of 0 to 1, or z-score normalization can be used to eliminate the differences in scale between different depths. If the time trend of humidity data needs to be analyzed, the time series data can be differentially processed or moved averaged to better extract the changing trend of moisture status at different depths. After preprocessing the data at different depths (such as surface, middle, and deep layers) respectively, the layered data are integrated to generate preprocessed moisture data sequences at each depth as input for subsequent feature extraction and analysis.

[0057] The pre-processed root zone water status data at different depths are subjected to feature extraction, and the water gradient change characteristics in the extracted root zone water status data are analyzed to generate a water gradient change rate drift index. The method for obtaining the water gradient change rate drift index is as follows:

[0058] From the soil moisture data collected at different depths, the moisture gradient change rate at each monitoring depth is calculated and expressed as a time series matrix X, where X∈R n×m :R is the real number domain, that is, the set of all real numbers, n is the number of sample points of time series data (such as sampling once a day for 30 days, then n = 30). m is the depth layer of soil monitoring (such as monitoring 3 depths, then m = 3). Each column in X (the data of moisture gradient change rate at different depths) is standardized so that the mean of each column of data is 0 and the variance is 1 to ensure the comparability of data at different depths.

[0059] Calculate the standardized moisture gradient change rate data matrix X std The covariance matrix C of is expressed as: Where C∈R n×m represents the covariance of the rate of change of water gradient between depths, T is the matrix transpose, and the covariance matrix C is subjected to eigendecomposition to obtain the eigenvalues ​​and eigenvectors, expressed as: Cv i =λ i v i ; In the formula, λ iis the i-th eigenvalue of the covariance matrix, indicating the variance of the moisture gradient change rate in the direction of the principal component, v i is the corresponding eigenvector, indicating the main direction of the moisture gradient change rate; the eigenvalues ​​are arranged in descending order, and the eigenvectors corresponding to the first k largest eigenvalues ​​are selected to form the main component matrix V k , V k =[v1,v2,…,v k ]; k is usually selected to explain more than 85% of the total variance. Projection to the principal component matrix V k On the basis of the above, we get the principal component score matrix Z, which is expressed as: Z = X std ·V k ; Z∈R n×m It is the principal component score matrix, and each column is a principal component score. The absolute mean of the score of the first principal component is used as the moisture gradient change rate drift index to measure the overall drift degree of the moisture gradient change rate. The expression is: Where WD is the drift index of the moisture gradient change rate.

[0060] The larger the drift index of the moisture gradient change rate, the greater the gradient of moisture change in the root zone at different depths, and the unstable soil moisture condition. A larger drift index indicates that the difference in moisture change rate between the deep and shallow layers of the soil is significant, which usually means that the groundwater level is fluctuating significantly or rising rapidly, causing the moisture in the deep soil to increase and penetrate into the surface. At this time, the abnormality of groundwater fluctuation is high, and the root system may face the risk of hypoxia due to excessive water inflow, which may further lead to root suffocation or root rot. In order to prevent the risks brought by excess water, it is necessary to reduce irrigation or take drainage measures to prevent the soil from being oversaturated for a long time.

[0061] On the contrary, when the drift index of the moisture gradient change rate is small, it means that the soil moisture gradient at different depths tends to be stable, the abnormal degree of groundwater fluctuation is low, and the impact of groundwater on the moisture in the root zone is small. A lower drift index means that the soil moisture is more uniform, indicating that the groundwater level is basically stable, and the difference in moisture content between the deep and surface soil is small. In this case, a normal irrigation strategy can maintain a healthy water supply for crops without adjusting the irrigation amount due to groundwater fluctuations. However, to ensure a balanced water supply, the drift index should be monitored regularly to cope with possible groundwater fluctuations in the future.

[0062] The pre-processed root zone water status data at different depths are subjected to feature extraction, and the water retention time deviation characteristics in the extracted root zone water status data are analyzed to generate a water retention time deviation index. The method for obtaining the water retention time deviation index is as follows:

[0063] For each time point t and depth layer m, the water retention time data series S t,m , the calculation expression of exponential weighted moving average is: EWMA t,m =αS t,m +(1-α)EWMA t-1,m ; In the formula, EWMA t,m is the exponentially weighted moving average calculated at the current time t and depth level m, EWMA t-1,m is the exponentially weighted moving average of the previous time t-1 and depth layer m, α is the smoothing coefficient, 0<α≤1, which is used to control the weight of the latest data. A larger α value indicates greater sensitivity to the current data point, while a smaller α value makes historical data more influential. For the initial condition (i.e. t=0), you can set EWMA 0,m =S 0,m ; Calculate the residence time deviation of each depth layer m, the expression is: ΔS t,m =S t,m -EWMA t,m ;Deviation ΔS t,m It represents the difference between the current water retention time and the smoothed retention time, which can be used to evaluate the fluctuation of the retention time. The water retention time deviation index is defined as the absolute mean of the retention time deviations of each depth layer, and the expression is: D is the total number of monitored soil depth layers, and WT is the water residence time deviation index.

[0064] The larger the water retention time deviation index, the greater the deviation of soil water retention time at different depths from the smoothed value, indicating a high degree of abnormal groundwater fluctuation. A high water retention time deviation index usually means that the groundwater level changes rapidly in a short period of time, resulting in significant differences in water retention time between deep and shallow soils. Such fluctuations can affect the stability of the water environment in the root zone, and the roots may suffer from hypoxia due to excessive soil moisture, or even cause root rot. In addition, a high water retention time deviation index indicates that the rise or fluctuation of groundwater has significantly affected soil moisture distribution, and it is recommended to reduce irrigation or increase drainage to prevent over-saturation of the soil and damage to crop health.

[0065] On the contrary, when the water retention time deviation index is small, it means that the deviation of water retention time is low, indicating that the water status at different depths of the soil is relatively balanced and the abnormality of groundwater fluctuation is low. A low water retention time deviation index usually indicates that the groundwater level is stable, the water retention time of deep soil and shallow soil is relatively close, and the water environment in the root zone is stable, which is suitable for healthy growth of crops. In this case, conventional irrigation strategies can meet crop needs without special adjustments to irrigation due to groundwater fluctuations.

[0066] The moisture gradient change rate drift index and the moisture retention time deviation index are converted into the first eigenvector, and the first eigenvector is used as the input of the machine learning model. The machine learning model uses the weighted assignment labels of the moisture status data of the root zone at different depths predicted by each group of first eigenvectors as the prediction target, and takes minimizing the sum of the prediction errors of the weighted assignment labels of the moisture status data of the root zone at different depths as the training target. The machine learning model is trained until the sum of the prediction errors converges and the model training is stopped. The weight assignment of the moisture status data of the root zone at different depths is determined according to the output results of the model. The machine learning model is a polynomial regression model, and the moisture fluctuation anomaly index is obtained by weighted summation of the weight assignments of the moisture status data of the root zone at different depths.

[0067] The method for obtaining the weight assignment of the water status data of the root zone at different depths is as follows: from the first eigenvector training data of the trained machine learning model, the corresponding function expression is obtained: CH=F(WD, WT); where F is the output function of the model, WD is the drift index of the water gradient change rate, WT is the water residence time deviation index, and CH is the weight assignment of the water status data of the root zone at different depths.

[0068] S4: After comparing and analyzing the calculated moisture fluctuation anomaly index with the gradient standard threshold, the abnormal degree of groundwater fluctuation is divided into different levels, and the abnormal degree of groundwater fluctuation is divided into high-level abnormal fluctuation, medium-level abnormal fluctuation and low-level abnormal fluctuation.

[0069] The obtained water fluctuation anomaly index is compared with the gradient standard threshold, the gradient standard threshold includes a first standard threshold and a second standard threshold, and the first standard threshold is less than the second standard threshold, and the water fluctuation anomaly index is compared with the first standard threshold and the second standard threshold respectively;

[0070] If the water fluctuation anomaly index is greater than the second standard threshold, it means that the abnormal degree of groundwater fluctuation is high, and a high fluctuation signal is generated at this time, and the abnormal degree of groundwater fluctuation is classified as high-level abnormal fluctuation;

[0071] If the water fluctuation anomaly index is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it means that the abnormal degree of groundwater fluctuation is medium. At this time, a medium fluctuation signal is generated, and the abnormal degree of groundwater fluctuation is classified as medium-level abnormal fluctuation.

[0072] If the moisture fluctuation anomaly index is less than the first standard threshold, it means that the abnormal degree of groundwater fluctuation is low. At this time, a low fluctuation signal is generated, and the abnormal degree of groundwater fluctuation is classified as low-level abnormal fluctuation.

[0073] S5. According to the abnormal degree level of groundwater fluctuations, corresponding irrigation adjustment measures are taken for fluctuations of different levels. For high-level abnormal fluctuations, drainage measures are immediately taken to reduce soil moisture. For medium-level abnormal fluctuations, the parameters of the crop water requirement model are dynamically adjusted. For low-level abnormal fluctuations, the irrigation strategy is normally implemented.

[0074] When a high-level abnormal fluctuation is detected, that is, when the moisture fluctuation abnormal index is greater than the second standard threshold T2, the soil moisture content may have seriously exceeded the standard. Therefore, drainage measures need to be taken immediately. The drainage rate formula is: D rate = k drain ×(WAI - T2); where D rate is the drainage rate, indicating the amount of water that needs to be drained per unit time to reduce soil moisture. k drain is the drainage coefficient, which is an adjustable parameter reflecting the efficiency of the drainage system and the soil drainage capacity. The larger the value, the higher the drainage rate, usually set according to the soil type and the capacity of the drainage system. WAI is the moisture fluctuation abnormal index, used to measure the degree to which the current soil moisture exceeds the normal range. T2 is the second standard threshold, which is the critical value for judging high-level abnormal fluctuations, indicating the impact of severe groundwater fluctuations on the root zone. Ensuring that the drainage rate is proportional to the degree of soil moisture exceeding the standard can quickly reduce soil moisture and prevent crop roots from suffocating or rotting due to excessive moisture.

[0075] For medium-level abnormal fluctuations, that is, when the moisture fluctuation abnormal index WAI is greater than or equal to the first standard threshold T1 and less than or equal to the second standard threshold T2, the groundwater fluctuation has begun to affect the root zone moisture but has not reached a severe level. At this time, the parameters of the crop water requirement model can be dynamically adjusted to appropriately reduce the irrigation amount to avoid further increasing soil humidity. The crop water requirement model adjustment formula is: where I is the adjusted irrigation amount, indicating the water requirement dynamically adjusted according to the groundwater fluctuation situation. I0 is the initial irrigation amount, indicating the target irrigation amount set in the crop water requirement model under normal circumstances. q is the irrigation adjustment coefficient (0 < q < 1), used to control the adjustment range, usually set according to the crop water tolerance and the soil drainage capacity. The larger the value, the greater the irrigation adjustment range. T1 is the first standard threshold, which is the lower limit for judging medium-level abnormal fluctuations, reflecting the impact of mild groundwater fluctuations. By dynamically reducing the irrigation amount, the irrigation decision is matched with the actual soil moisture condition, effectively avoiding soil over-wetting, providing a suitable moisture environment for the crop, and thus optimizing the accuracy and environmental adaptability of the water requirement model. At the same time, for low-level abnormal fluctuations, the irrigation strategy is normally implemented.

[0076] In this embodiment, precise control of the moisture status of the crop root zone is achieved through refined monitoring and hierarchical management. First, the crop planting area is divided into multiple monitoring areas, and soil moisture sensors of different depths are installed in each area to collect root zone moisture status data in real time. Subsequently, the moisture data at different depths are preprocessed and feature extracted, and the moisture gradient change characteristics and moisture retention time deviation characteristics are extracted respectively. According to these characteristics, corresponding weights are assigned to the moisture data at different depths, and the moisture fluctuation anomaly index is calculated by weighted summation. The index is compared with the preset gradient standard threshold and divided into three levels of groundwater fluctuation anomaly: high, medium and low. For different levels, corresponding irrigation adjustment measures are formulated: immediate drainage in case of high-level fluctuations, dynamic adjustment of water demand model parameters in case of medium-level fluctuations, and normal execution of irrigation strategies in case of low-level fluctuations. This method realizes intelligent management of irrigation strategies that accurately adjust irrigation strategies according to groundwater fluctuations.

[0077] Embodiment 2, a precise irrigation decision analysis system based on crop water demand described in this embodiment includes a data acquisition module, a feature extraction module, a moisture fluctuation abnormal index calculation module, an abnormal fluctuation level classification module and an irrigation adjustment strategy module;

[0078] Data collection module: The crop planting area is divided into several monitoring areas, and soil moisture sensors installed in each monitoring area are used to monitor at different depths to capture the moisture status data of the root zone;

[0079] Feature extraction module: preprocess the root zone water status data of crops at different depths, and perform feature extraction on the preprocessed root zone water status data at different depths, respectively extracting the water gradient change characteristics and water retention time deviation characteristics in the root zone water status data;

[0080] Water fluctuation abnormality index calculation module: Determine the weight assignment of water status data in the root zone at different depths according to the water gradient change characteristics and water retention time deviation characteristics in the extracted water status data of the root zone, and perform weighted summation calculation on the weight assignment of water status data in the root zone at different depths to obtain the water fluctuation abnormality index;

[0081] Abnormal fluctuation level classification module: after comparing and analyzing the calculated water fluctuation abnormal index with the gradient standard threshold, the abnormal degree of groundwater fluctuation is divided into different levels, and the abnormal degree of groundwater fluctuation is divided into high-level abnormal fluctuation, medium-level abnormal fluctuation and low-level abnormal fluctuation;

[0082] Irrigation adjustment strategy module: According to the abnormal degree of groundwater fluctuation, corresponding irrigation adjustment measures are taken for different levels of fluctuation. For high-level abnormal fluctuations, drainage measures are immediately taken to reduce soil moisture. For medium-level abnormal fluctuations, the parameters of the crop water demand model are dynamically adjusted. For low-level abnormal fluctuations, the irrigation strategy is implemented normally.

[0083] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0084] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0085] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0086] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A precision irrigation decision analysis method based on crop water demand, characterized by: The steps include: S1: Divide the crop planting area into several monitoring areas, and monitor at different depths through soil moisture sensors installed in each monitoring area to capture the moisture status data of the root zone; S2: preprocessing the root zone water status data of crops at different depths, and extracting features of the preprocessed root zone water status data at different depths, respectively extracting water gradient change features and water retention time deviation features in the root zone water status data; S3: Determine the weight assignment of the water status data of the root zone at different depths according to the water gradient change characteristics and water retention time deviation characteristics in the extracted water status data of the root zone, and calculate the water fluctuation abnormality index by weighted summation of the weight assignment of the water status data of the root zone at different depths; S4: After comparing and analyzing the calculated water fluctuation anomaly index with the gradient standard threshold, the abnormal degree of groundwater fluctuation is divided into different levels, and the abnormal degree of groundwater fluctuation is divided into high-level abnormal fluctuation, medium-level abnormal fluctuation and low-level abnormal fluctuation; S5: According to the abnormal degree of groundwater fluctuation, take corresponding irrigation adjustment measures for different levels of fluctuation. For high-level abnormal fluctuation, take drainage measures immediately to reduce soil moisture. For medium-level abnormal fluctuation, dynamically adjust the parameters of the crop water demand model. For low-level abnormal fluctuation, implement irrigation strategy normally.

2. The method for analyzing precise irrigation decision based on crop water demand according to claim 1, characterized in that: In S2, the moisture gradient change characteristics in the extracted root zone moisture status data are analyzed to generate a moisture gradient change rate drift index. The method for obtaining the moisture gradient change rate drift index is: From the soil moisture data collected at different depths, the moisture gradient change rate at each monitoring depth is calculated and expressed as a time series matrix X, where X∈R n×m :R is the real number domain, that is, the set of all real numbers, n is the number of sample points of time series data, m is the depth layer of soil monitoring, each column in X is standardized, and the standardized moisture gradient change rate data matrix X is calculated std The covariance matrix C of is expressed as: Where C∈R n×m represents the covariance of the rate of change of water gradient between depths, T is the matrix transpose, and the covariance matrix C is subjected to eigendecomposition to obtain the eigenvalues ​​and eigenvectors, expressed as: Cv i =λ i v i ; In the formula, λ i is the i-th eigenvalue of the covariance matrix, v i is the corresponding eigenvector, arrange the eigenvalues ​​in descending order, select the eigenvectors corresponding to the first k largest eigenvalues, and form the principal component matrix V k , V k =v1,v2,…,v k ]; The standardized data Projection to the principal component matrix V k On the basis of the above, we get the principal component score matrix Z, which is expressed as: Z = X std ·V k ; Z∈R n×m It is the principal component score matrix, each column is a principal component score; the absolute mean of the score of the first principal component is used as the moisture gradient change rate drift index, and the expression is: Where WD is the drift index of the moisture gradient change rate.

3. The method for analyzing precise irrigation decision based on crop water requirement according to claim 2, characterized in that: In S2, the water retention time deviation characteristics in the extracted root zone water status data are analyzed to generate a water retention time deviation index. The method for obtaining the water retention time deviation index is: For each time point t and depth layer m, the water retention time data series S t,m , the calculation expression of exponential weighted moving average is: EWMA t,m =αS t,m +(1-α)EWMA t-1,m ; In the formula, EWMA t,m is the exponentially weighted moving average calculated at the current time t and depth level m, EWMA t-1,m is the exponentially weighted moving average of the previous time t-1 and depth layer m, α is the smoothing coefficient, 0<α≤1, and the initial condition is set to EWMA 0,m =S 0,m ; Calculate the residence time deviation for each depth layer m, expressed as: ΔS t,m =S t,m -EWMA t,m ;Deviation ΔS t,m It represents the difference between the current water retention time and the smoothed retention time. The water retention time deviation index is defined as the absolute mean of the retention time deviations of each depth layer. The expression is: D is the total number of monitored soil depth layers, and WT is the water residence time deviation index.

4. The method for analyzing precise irrigation decision based on crop water requirement according to claim 3, characterized in that: In S3, the moisture gradient change rate drift index and the moisture retention time deviation index are converted into the first eigenvector, and the first eigenvector is used as the input of the machine learning model. The machine learning model uses the weighted assignment labels of the moisture status data of the root zone at different depths predicted by each group of first eigenvectors as the prediction target, and takes minimizing the sum of the prediction errors of the weighted assignment labels of the moisture status data of the root zone at different depths as the training target. The machine learning model is trained until the sum of the prediction errors converges and the model training is stopped. The weight assignment of the moisture status data of the root zone at different depths is determined according to the output results of the model, wherein the machine learning model is a polynomial regression model, and the moisture fluctuation anomaly index is obtained by weighted summation of the weight assignments of the moisture status data of the root zone at different depths.

5. The method for analyzing precise irrigation decision based on crop water requirement according to claim 4, characterized in that: In S4, after comparing and analyzing the calculated water fluctuation anomaly index with the gradient standard threshold, the abnormal degree of groundwater fluctuation is divided into different levels, specifically: The obtained water fluctuation anomaly index is compared with the gradient standard threshold, the gradient standard threshold includes a first standard threshold and a second standard threshold, and the first standard threshold is less than the second standard threshold, and the water fluctuation anomaly index is compared with the first standard threshold and the second standard threshold respectively; If the water fluctuation anomaly index is greater than the second standard threshold, it means that the abnormal degree of groundwater fluctuation is high, and a high fluctuation signal is generated at this time, and the abnormal degree of groundwater fluctuation is classified as high-level abnormal fluctuation; If the water fluctuation anomaly index is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it means that the abnormal degree of groundwater fluctuation is medium. At this time, a medium fluctuation signal is generated, and the abnormal degree of groundwater fluctuation is classified as medium-level abnormal fluctuation. If the moisture fluctuation anomaly index is less than the first standard threshold, it means that the abnormal degree of groundwater fluctuation is low. At this time, a low fluctuation signal is generated, and the abnormal degree of groundwater fluctuation is classified as low-level abnormal fluctuation.

6. The method for analyzing precise irrigation decision based on crop water requirement according to claim 1, characterized in that: In S5, according to the abnormal degree of groundwater fluctuation, corresponding irrigation adjustment measures are taken for different levels of fluctuation. For high-level abnormal fluctuation, drainage measures are immediately taken to reduce soil moisture. For medium-level abnormal fluctuation, the parameters of the crop water demand model are dynamically adjusted. For low-level abnormal fluctuation, the irrigation strategy is implemented normally, specifically: When a high-level abnormal fluctuation is detected, that is, the moisture fluctuation abnormal index is greater than the second standard threshold T2, drainage measures need to be taken immediately. The drainage rate formula is: D rate =k drain ×(WAI-T2); where D rate is the drainage rate, which indicates the amount of water that needs to be discharged per unit time, k drain is the drainage coefficient, which reflects the efficiency of the drainage system and the soil drainage capacity, WAI is the water fluctuation anomaly index, and T2 is the second standard threshold; For medium-level abnormal fluctuations, that is, the water fluctuation abnormal index WAI is greater than or equal to the first standard threshold T1 and less than or equal to the second standard threshold T2, the crop water requirement model adjustment formula is dynamically adjusted by adjusting the parameters of the crop water requirement model: Where I is the adjusted irrigation amount, which indicates the water demand after dynamic adjustment according to groundwater fluctuations; I0 is the initial irrigation amount, which indicates the target irrigation amount set in the crop water demand model under normal circumstances; q is the irrigation adjustment coefficient, which is used to control the adjustment range; and T1 is the first standard threshold.

7. A precision irrigation decision analysis system based on crop water requirement, used to implement the precision irrigation decision analysis method based on crop water requirement according to any one of claims 1 to 6, characterized in that: It includes data acquisition module, feature extraction module, moisture fluctuation abnormal index calculation module, abnormal fluctuation grade classification module and irrigation adjustment strategy module; Data collection module: The crop planting area is divided into several monitoring areas, and soil moisture sensors installed in each monitoring area are used to monitor at different depths to capture the moisture status data of the root zone; Feature extraction module: preprocess the root zone water status data of crops at different depths, and perform feature extraction on the preprocessed root zone water status data at different depths, respectively extracting the water gradient change characteristics and water retention time deviation characteristics in the root zone water status data; Water fluctuation abnormality index calculation module: Determine the weight assignment of water status data in the root zone at different depths according to the water gradient change characteristics and water retention time deviation characteristics in the extracted water status data of the root zone, and perform weighted summation calculation on the weight assignment of water status data in the root zone at different depths to obtain the water fluctuation abnormality index; Abnormal fluctuation level classification module: after comparing and analyzing the calculated water fluctuation abnormal index with the gradient standard threshold, the abnormal degree of groundwater fluctuation is divided into different levels, and the abnormal degree of groundwater fluctuation is divided into high-level abnormal fluctuation, medium-level abnormal fluctuation and low-level abnormal fluctuation; Irrigation adjustment strategy module: According to the abnormal degree of groundwater fluctuation, corresponding irrigation adjustment measures are taken for different levels of fluctuation. For high-level abnormal fluctuations, drainage measures are immediately taken to reduce soil moisture. For medium-level abnormal fluctuations, the parameters of the crop water demand model are dynamically adjusted. For low-level abnormal fluctuations, the irrigation strategy is implemented normally.

Images