Farmland irrigation method of multilayer sensor model based on WOA-LSO algorithm

Abstract

The invention discloses a farmland irrigation method of a multilayer sensor model based on a WOA-LSO algorithm, and relates to the technical field of agricultural irrigation. The method comprises the following steps: acquiring meteorological data factors of a plurality of samples, and determining a plurality of factor combinations; a factor combination most related to the soil transpiration amount is screened out through a correlation test, and then the number of nodes of an input layer, an output layer and a hidden layer of a multi-layer sensor (MLP) model is determined; an MLP prediction model optimized based on a WOA-LSO algorithm is constructed, the optimal sequence is obtained by optimizing the weight and threshold sequence of the MLP, and the model is endowed with the optimal sequence; inputting a target meteorological data factor of a research area into the model, and predicting the soil transpiration in a future time period; inputting the soil effective water content and transpiration into a fuzzy inference system to obtain an irrigation coefficient, and calculating the irrigation amount required by the farmland according to the irrigation coefficient. According to the method, the problem of inaccurate prediction of the soil transpiration caused by scarcity and incompleteness of data in a plateau region is solved, and the prediction accuracy of the soil transpiration is remarkably improved.

Description

Technical Field

[0001] This application relates to the technical field of agricultural irrigation, and particularly to a farmland irrigation method based on a multi-layer perceptron model using the WOA-LSO algorithm. Background Art

[0002] Due to its unique geographical location and climatic conditions, the plateau region faces serious water resource shortage problems. Many plateau regions are usually classified as semi-arid areas, and their ecological environment is extremely fragile. The lack of water resources further exacerbates this situation. This water shortage has significantly restricted the development of local agriculture. In particular, the problem of irrigation water has become a key factor affecting agricultural production.

[0003] In the prior art, through a method for analyzing crop growth status based on satellite data and machine learning, by combining remote sensing data and ground measurement data, establishing the relationship between transpiration and vegetation index, and using machine learning methods to optimize the transpiration prediction model, real-time monitoring of crop growth conditions is achieved. However, this method fails to fully consider the complexity of various environmental factors and meteorological data in the plateau region, resulting in inaccurate transpiration prediction, thus affecting the effectiveness of irrigation decision-making.

[0004] Therefore, there is an urgent need for an intelligent irrigation method to accurately predict the irrigation water volume under the complex environmental factors and meteorological data in the plateau region. Summary of the Invention

[0005] Based on this, it is necessary to provide a farmland irrigation method based on a multi-layer perceptron model using the WOA-LSO algorithm to solve the problem of difficult accurate prediction of irrigation water volume under the complex environmental factors and meteorological data in the plateau region.

[0006] This specification adopts the following technical solutions: This specification provides a farmland irrigation method based on a multi-layer perceptron model using the WOA-LSO algorithm, including: Obtaining meteorological data factors; the meteorological data factors include: maximum temperature, average temperature, minimum temperature, average relative humidity, day ordinal number, sunshine hours, and wind speed; According to the correlation test between meteorological data factors and soil transpiration, determine the number of nodes in the input layer, hidden layer, and output layer of the multi-layer perceptron network, and construct a multi-layer perceptron prediction model optimized by the WOA-LSO algorithm to predict soil transpiration; for the multi-layer perceptron prediction model optimized by the WOA-LSO algorithm, encode the weight sequence and threshold sequence of the multi-layer perceptron model through the WOA-LSO algorithm, and iteratively calculate the fitness value of the WOA-LSO algorithm until reaching the maximum iteration number of the WOA-LSO algorithm or the preset value of the fitness value, to obtain the optimal weight sequence and optimal threshold sequence of the multi-layer perceptron model and assign them to the multi-layer perceptron model to obtain; Input the meteorological data factors into the multi-layer perceptron prediction model based on the WOA-LSO algorithm for training to obtain a multi-layer perceptron prediction model based on the WOA-LSO algorithm; and use the multi-layer perceptron prediction model based on the WOA-LSO algorithm to predict soil transpiration to obtain soil transpiration; Input the existing available soil water content and the predicted soil transpiration into the fuzzy inference system to obtain the irrigation coefficient; and use the irrigation coefficient to formulate an irrigation strategy to calculate the required irrigation amount for farmland.

[0007] Preferably, the determination of the number of nodes in the input layer, hidden layer, and output layer of the multi-layer perceptron network specifically includes: Determine the number of nodes in the input layer of the multi-layer perceptron network, including: Use the multi-layer perceptron model to test the correlation between each meteorological data factor and soil transpiration, determine the meteorological data factors participating in the final soil transpiration prediction, and determine the number of nodes in the input layer of the multi-layer perceptron model according to the number of meteorological data factors participating in the final soil transpiration prediction; Determine the number of nodes in the hidden layer of the multi-layer perceptron network, including: Determine the range of the number of nodes in the hidden layer through an empirical formula, and within this range, repeatedly test the network error of the multi-layer perceptron network corresponding to each number of nodes, compare the network errors of the multi-layer perceptron networks corresponding to each number of nodes, and select the number of nodes with the smallest error as the number of nodes in the hidden layer; Determine the number of nodes in the output layer of the multi-layer perceptron network, including: The number of prediction targets is the number of nodes in the output layer of the multi-layer perceptron network.

[0008] Preferably, the determination of the number of nodes in the input layer of the multi-layer perceptron network includes the following steps: Group the original meteorological data factors and input them into the existing multi-layer perceptron prediction model to predict soil transpiration to obtain the prediction result of soil transpiration; According to the prediction results of soil transpiration, a correlation test is carried out between each original meteorological data factor and soil transpiration, and based on the test results, the meteorological data factors finally participating in the prediction of soil transpiration are determined; Among them, the grouping of the original meteorological data factors includes: The first group of meteorological data factors includes: all meteorological data factors, meteorological data factors excluding the minimum temperature item, and meteorological data factors excluding the maximum temperature and the minimum temperature; The second group of meteorological data factors includes: on the basis of the meteorological data factors excluding the maximum temperature and the minimum temperature, and four items are randomly selected from the remaining five meteorological data factors for random permutation and combination, and all combinations thereof are the second group of meteorological data factors; The third group of meteorological data factors includes: on the basis of determining the best combination of the second group of meteorological data factors, one meteorological data factor is randomly excluded, and all combinations obtained are the third group of meteorological data factors.

[0009] Preferably, The first group of meteorological data factors is used to test the correlation between the maximum temperature, the average temperature and the minimum temperature and soil transpiration; The second group of meteorological data factors is used to test the correlation between the average temperature, the average relative humidity, the day sequence number, the wind speed and the sunshine duration and soil transpiration; The third group of meteorological data factors is used to determine the meteorological data factor with the strongest correlation.

[0010] Preferably, determining the number of nodes in the hidden layer of the multi-layer perceptron network specifically includes: According to the relationship between the number of nodes in the hidden layer, the number of nodes in the input layer and the number of nodes in the output layer, the formula is: ; In the formula, is the minimum value of the number of nodes in the hidden layer, is the maximum value of the number of nodes in the hidden layer, is the number of nodes in the input layer, is the number of nodes in the output layer, is the number of nodes in the hidden layer; Determine that the range of the number of nodes in the hidden layer is ; Within the range of the number of nodes in the hidden layer calculate the error of the multi-layer perceptron network in turn, and select the number of nodes corresponding to the minimum network error value as the number of nodes in the hidden layer.

[0011] Preferably, the WOA-LSO algorithm specifically includes: The WOA-LSO algorithm specifically includes: Initialize the algorithm parameters, including: Initialize the number of whale populations and the maximum number of iterations. The formula is: ; In the formula, is a linearly decreasing parameter used to control the balance between global search and local search; is the current iteration number, is the maximum number of iterations; ; ; ; In the formula, and are random numbers between [0, 1]. The parameter A is used to control the movement direction of the whale, and the parameter C is used to control the movement distance of the whale, p is used to select the predation strategy; Update the position of each whale and the current optimal solution to perform global search. The formula is: ; In the formula, is the current optimal solution at the -th update, is the optimal solution at the -th update, is the initialized random solution, is the initialized parameter, is the search range, which is the optimal solution at the -th update, Perform a random perturbation on the current optimal solution and update the current optimal solution again to perform local search. The formula is: ; In the formula, is the latest solution after random perturbation, is the current optimal solution at the -th update, is the perturbation coefficient, represents a random number between; Update the optimal solution. The formula is: ; In the formula, is the current optimal solution, is the latest solution after random perturbation, is the current optimal solution at the -th update, is The fitness value of is the fitness value of

[0012] Preferably, encoding the weight sequence and threshold sequence of the multi-layer perceptron model by the WOA-LSO algorithm specifically includes: Let the number of weights of the multi-layer perceptron model be n , and the number of thresholds be m , then the weight sequence of the model is , and the threshold sequence of the model is ; Encode the weight sequence and threshold sequence of the model into a n + m -dimensional vector, specifically: ; In the formula, is the encoding sequence of the model weights and thresholds, is the weight sequence of the model, is the threshold sequence of the model; Preferably, obtaining the optimal weight sequence and threshold sequence of the multi-layer perceptron model specifically includes: Initialize the population individual positions, population size, and maximum number of iterations of the WOA-LSO algorithm; the solutions corresponding to the positions of the population individuals are the encoding sequences of the weights and thresholds of the multi-layer perceptron model; Select the mean square error as the fitness function of the WOA-LSO algorithm, iteratively update the positions of the population individuals and their corresponding solutions, and iteratively calculate the fitness value of the WOA-LSO algorithm; When the fitness value of the WOA-LSO algorithm reaches the maximum number of iterations or the preset value of the fitness value, the iteration stops, and the optimal position of the population individual and its corresponding optimal solution are obtained, and the corresponding optimal solution is the optimal weight sequence and optimal threshold sequence of the multi-layer perceptron model.

[0013] Preferably, the fuzzy inference system includes: an input layer, a fuzzification layer, a rule layer, an aggregation layer, and a defuzzification layer; The input layer has two node numbers and is used to input the existing available soil water content and the predicted soil transpiration; In the fuzzification layer, the existing available soil water content and the predicted soil transpiration input are fuzzified using a Gaussian membership function, that is, the actual input set is transformed into a fuzzy set using the Gaussian membership function; Through the fuzzy rules predefined in the rule layer, reason about the fuzzy set to obtain a fuzzy output; Through the aggregation layer, aggregate all fuzzy outputs into a comprehensive fuzzy set; Through the defuzzification layer, the fuzzy output is clarified into the irrigation coefficient by the centroid method.

[0014] Preferably, using the irrigation coefficient, an irrigation strategy is formulated to calculate the required irrigation water volume for farmland, specifically including: Define the input variables of the fuzzy inference system and set the system output variable as the irrigation coefficient; Establish a fuzzy rule base, and use the trained fuzzy neural network to optimize the fuzzy rules and membership functions of the fuzzy inference system; Input the soil transpiration amount predicted in real time by the multi-layer perceptron model optimized by the WOA-LSO algorithm into the fuzzy inference system, predict the irrigation coefficient, and calculate the required irrigation water volume for the farmland in combination with the irrigation quota formula to assist the irrigation decision-making of the farmland.

[0015] The above at least one technical solution adopted in this specification can achieve the following beneficial effects: A farmland irrigation method based on a multi-layer perceptron model of the WOA-LSO algorithm provided in this specification constructs a multi-layer perceptron model optimized by the WOA-LSO algorithm. First, through the correlation test between meteorological data factors and soil transpiration amount, the number of nodes in the input layer, hidden layer, and output layer of the multi-layer perceptron network is determined. Since there are many meteorological factors affecting the prediction results and there are up to hundreds of input combination forms, therefore, the present invention only considers a type of meteorological data factors with extremely strong correlation with the soil transpiration amount as the model input, reducing the operation load of the model. By reducing the number of nodes in the network layer, the model is simplified and the prediction efficiency is improved. Secondly, the prediction model optimizes the weights and thresholds of the model through the WOA-LSO algorithm, and effectively avoids the local optimal solution problem encountered by traditional machine learning models in the case of the complexity of various environmental factors and meteorological data in the plateau region by using the global search ability of the WOA-LSO algorithm.

[0016] In summary, in the case of the complexity of various environmental factors and meteorological data in the plateau region, the present invention simplifies the network structure through the correlation test between meteorological data factors and soil transpiration amount, and at the same time optimizes the weights and thresholds of the model in combination with the WOA-LSO algorithm, solving the problem of inaccurate prediction of soil transpiration amount caused by the complexity of environmental factors and meteorological data in the plateau region, and significantly improving the prediction accuracy of soil transpiration amount. In addition, the present invention also constructs a fuzzy inference system, and on the basis of accurate prediction of soil transpiration amount, applies the predicted soil transpiration amount thereto, thereby calculating the required irrigation water volume for the farmland, providing strong support for the irrigation decision-making of farmland in the plateau region, effectively improving the irrigation efficiency and realizing the effective conservation of water resources, and laying a foundation for the sustainable development of agriculture. Description of the Drawings

[0017] The accompanying drawings described herein are used to provide a further understanding of the present application and form a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation of the present application. In the drawings: Figure 1 It is a schematic flow chart of a farmland irrigation method based on a multi-layer perceptron model using the WOA-LSO algorithm provided in this specification; Figure 2 It is a prediction flow chart of soil transpiration amount for a farmland irrigation method based on a multi-layer perceptron model using the WOA-LSO algorithm provided in this specification; Figure 3 It is the best network error graph provided in this specification; Figure 4 It is a modeling flow chart of a fuzzy irrigation decision-making system provided in this specification. Detailed implementation manners

[0018] To make the objectives, technical solutions, and advantages of this specification clearer, the technical solutions of the present application will be clearly and completely described below in conjunction with specific embodiments of this specification and the corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the specification, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present application.

[0019] The following will detail the technical solutions provided by each embodiment of the present application in conjunction with the drawings.

[0020] Figure 1 It is a schematic flow chart of a method based on a multi-layer perceptron model using the WOA-LSO algorithm in this specification, specifically including the following steps: S101: Obtain multiple sample meteorological data factors.

[0021] Specifically, the multiple sample meteorological data factors include: maximum temperature, average temperature, minimum temperature, average relative humidity, day ordinal number, sunshine hours, and wind speed.

[0022] Specifically, in this embodiment, the meteorological data from July to November 2024 in a certain plateau area and the corresponding soil transpiration amount obtained through the actual PM equation are used as meteorological factor data. The system hardware includes a sensor installation module, a data acquisition module, a data analysis module, an irrigation plan determination module, an irrigation execution module, an effect evaluation module, a maintenance module, and a user interaction module. It is characterized in that: the sensor installation module includes a soil temperature sensor, a soil moisture sensor, a soil nutrient sensor, a light intensity sensor, a wind speed and direction sensor, and an atmospheric pressure sensor, and all sensors can perform data transmission monitoring.

[0023] Additionally, design a visual decision support platform by yourself, providing an intuitive interface and data display to help users understand the decision-making logic and results of the system. This will improve user acceptance and operational convenience, and promote the popularization of the technology. Moreover, this cloud platform is connected to the data of the sensors, enabling users to clearly observe the data of all sensors.

[0024] S102: According to multiple sample meteorological data factors, determine multiple combinations of sample meteorological data factors, and respectively conduct correlation tests between each combination of sample meteorological data factors and soil transpiration to determine the target combination of sample meteorological data factors. According to the target combination of sample meteorological data factors, determine the number of nodes in the input layer, output layer, and hidden layer of the multi-layer perceptron model.

[0025] Optionally, multiple combinations of sample meteorological data factors include the first group of meteorological data factor combinations and the second group of meteorological data factor combinations; determining multiple combinations of sample meteorological data factors according to multiple sample meteorological data factors specifically includes: determining the first group of meteorological data factor combinations according to multiple sample meteorological data factors; the first group of meteorological data factor combinations includes all combinations of meteorological data factors, combinations of meteorological data factors with the lowest temperature item removed, and combinations of meteorological data factors with the highest and lowest temperatures removed; among the five meteorological data factors in the combination of meteorological data factors with the highest and lowest temperatures removed, select four for random permutation and combination, and determine all their combinations as the second group of meteorological data factor combinations.

[0026] Among them, the first group of meteorological data factors is used to test the correlation between the highest temperature, average temperature, and lowest temperature and soil transpiration; the second group of meteorological data factors is used to test the correlation between the average temperature, average relative humidity, day ordinal number, wind speed, and sunshine duration and soil transpiration; the third group of meteorological data factors is used to determine the meteorological data factors with the strongest correlation.

[0027] Optionally, respectively conduct correlation tests between each combination of sample meteorological data factors and soil transpiration to determine the target combination of sample meteorological data factors, including: respectively input each combination of meteorological data factors into the initial multi-layer perceptron model to obtain the predicted soil transpiration under each combination of meteorological data factors; according to the predicted soil transpiration and actual soil transpiration under each combination of meteorological data factors, determine the correlation between each combination of meteorological data factors and soil transpiration; obtain the candidate combination of meteorological data factors with the largest correlation in the second group of meteorological data factor combinations, and successively remove each meteorological data factor in the candidate combination of meteorological data factors to obtain the third group of meteorological data factor combinations; obtain the correlation between each combination of meteorological data factors and soil transpiration in the third group of meteorological data factor combinations; determine the combination of meteorological data factors with the largest correlation among all combinations of meteorological data factors as the target combination of sample meteorological data factors.

[0028] Optionally, according to the target sample meteorological data factor combination, determine the number of nodes in the input layer, output layer, and hidden layer of the multi-layer perceptron model.

[0029] Optionally, determine the number of nodes in the input layer as the number of sample meteorological data factors in the target sample meteorological data factor combination; determine the number of nodes in the output layer as 1; according to the number of nodes in the input layer and the number of nodes in the output layer, determine the number of nodes in the hidden layer.

[0030] Specifically, determining the number of nodes in the hidden layer according to the number of nodes in the input layer and the number of nodes in the output layer specifically includes: According to the relationship between the number of nodes in the hidden layer, the number of nodes in the input layer, and the number of nodes in the output layer, determine the range of the number of nodes in the hidden layer. The formula is: ; In the formula, is the minimum value of the number of nodes in the hidden layer, is the maximum value of the number of nodes in the hidden layer, is the number of nodes in the input layer, is the number of nodes in the output layer, is the number of nodes in the hidden layer; Within the range of the number of nodes in the hidden layer successively obtain the prediction accuracy of the multi-layer perceptron prediction model under each number of nodes in the hidden layer, and determine the number of nodes corresponding to the highest prediction accuracy as the number of nodes in the hidden layer.

[0031] Specifically, select a framework and tools to choose a suitable deep learning framework and then define the input layer of the model architecture: Define the number of input layer nodes according to the number of features. Hidden layer: Select 12 hidden layers and choose an appropriate number of nodes for each layer. Output layer: According to the prediction target, use one node to output the soil transpiration rate, and a linear activation function can be used. Use the training set to train the model, and at the same time use the validation set to monitor the model performance. Set appropriate epochs and batch_size, and select early stopping to prevent overfitting. Secondly, in the WOA-LSO algorithm, first set the size of the whale population to 30, the initial spectral population to 30, and at the same time set the maximum number of iterations to 100 to limit the running time of the algorithm. Next, the search space needs to be defined, specifically including the range of hyperparameters to be optimized, such as the learning rate and the number of hidden layer nodes. To evaluate the effects of different hyperparameter configurations, define a fitness function, and usually use the performance metrics (such as mean squared error, MSE) of the model on the validation set as the fitness value. This fitness function guides the whale population to adjust their positions in the search space to find the optimal solution by training the model according to the current hyperparameter configuration and calculating the loss on the validation set. Within the maximum number of iterations, update the positions of the whales, calculate the fitness, and keep the optimal solution. By iteratively optimizing the hyperparameters, find the optimal learning rate and the number of hidden layer nodes.

[0032] Specifically, there are many factors affecting the reference transpiration rate of crops. Generally, it is considered that the seven factors of the highest temperature, average temperature, lowest temperature, average relative humidity, day ordinal number, sunshine hours, and wind speed are the main influencing factors of its transpiration rate. The selection of the number of input layer nodes should be determined based on the criterion of the smallest error in the experimental results and the most accurate predicted value. However, due to the large number of influencing factors and up to hundreds of input combination forms, so, we consider that for a class of influencing factors with extremely strong correlation, by comparing their prediction results, only one of them can be selected as the input variable of the network. This can not only reduce the computational load of the network, reduce the number of input quantities, but also simplify the model and improve the prediction efficiency.

[0033] Specifically, according to the analysis, since the three factors of the highest temperature, average temperature, and lowest temperature have extremely strong correlation, so it should be confirmed that only one of the three factors of the highest temperature, average temperature, and lowest temperature can be retained as the input quantity to obtain a relatively accurate predicted value; secondly, further consider the magnitudes of the five influencing factors of average relative humidity, average temperature, sunshine hours, day ordinal number, and wind speed on the predicted value of transpiration rate; finally, according to the experimental results of the second step, we can eliminate two of the influencing factors determined in the previous step that have a greater impact on the prediction results, run the network, compare several groups of experiments, and observe whether the predicted results can meet the prediction requirements.

[0034] Specifically, compare the prediction results of the above three groups of input methods to determine the correlations of the highest temperature, average temperature, and lowest temperature in this network model, including: The second group: On the basis of the first group (3), examine the effects of the average temperature, average relative humidity, day ordinal number, wind speed, and sunshine duration on the output results. Select four out of the above five influencing factors for permutation and combination, and compare the experimental results. Specifically as follows: Establish a prediction model with the four influencing factors of average temperature, day ordinal number, sunshine duration, and wind speed as inputs, and observe the experimental results. Establish a prediction model with the four influencing factors of average relative humidity, day ordinal number, sunshine duration, and wind speed as inputs, and observe the experimental results. Establish a prediction model with the four influencing factors of average temperature, average relative humidity, day ordinal number, and sunshine duration as inputs, and observe the experimental results. Establish a prediction model with the four influencing factors of average temperature, average relative humidity, day ordinal number, and wind speed as inputs, and observe the experimental results. Establish a prediction model with the four influencing factors of average temperature, average relative humidity, sunshine duration, and wind speed as inputs, and observe the experimental results.

[0035] By comparing and analyzing the above experimental results, four factors that have a greater impact on the predicted value of transpiration can be determined.

[0036] The third group: Randomly remove one item on the basis of the above results, run the network, and observe the experimental results.

[0037] For convenience, the above 9 different combinations of meteorological factors are respectively denoted as Scheme 1, Scheme 2, Scheme 3, ……, Scheme 9, etc. By comprehensively comparing the experimental results of the above schemes, the input quantity combination that can meet the prediction accuracy requirements can be intuitively selected, and then the number of nodes in the input layer can be determined.

[0038] Specifically, according to meteorological data such as average temperature, average relative humidity, and wind speed, the number of nodes in the output layer for predicting transpiration is 1. Determine the number of nodes in the hidden layer, specifically including: First, determine a suitable range of the number of nodes in the hidden layer through an empirical formula, and then repeat the experiment one by one within this range, compare the network error sizes, and select the number of nodes in the hidden layer corresponding to the minimum error. Generally, there is the following relationship between the number of nodes in the hidden layer, the input layer, and the output layer. Let the number of nodes in the input layer, hidden layer, and output layer of a three-layer neural network be respectively taken as 、 、 。

[0039]

[0040] Then the number of nodes in the hidden layer should be obtained within the interval [a, b].

[0041] Accordingly, the appropriate range of the number of hidden layer nodes corresponding to different numbers of input and output layer nodes can be obtained. See Table 1. Table 1 Appropriate range of the number of hidden layer nodes corresponding to different numbers of input and output layer nodes

[0042] When the number of different input quantities or the same number of input quantities but different combination methods, experiments should be repeated in the above interval in sequence, and the number of hidden layer nodes corresponding to the minimum network error value should be taken.

[0043] Based on the meteorological data from July to November in 2024, seven influencing factors including the highest temperature, average temperature, lowest temperature, average relative humidity, day ordinal number, sunshine hours, and wind speed are used to establish a model prediction. At this time, the appropriate range of the number of hidden layer nodes is [4, 18]. Three experiments are carried out at each integer point in this interval in sequence, and the median of the error results of the three experiments is recorded as the experimental result. See Figure 3 , when the number of hidden layer nodes is 13, the network obtains the minimum error of 0.020307. Therefore, the number of hidden layer nodes should be 13. Repeat the results of Scheme 2 - 9 in this way. See Table 2: Table 2 Network prediction errors corresponding to different numbers of nodes

[0044] See Figure 2 , which is the flowchart for predicting soil transpiration. The establishment of the soil transpiration prediction model mainly includes two aspects: the programming of the MLP model and the optimization of the weights and thresholds of the model by the WOA - LSO algorithm. Generate a three - layer MLP grid structure through programming, use the weight and threshold sequences of the network as coding, and then use the WOA - LSO algorithm to optimize the weights and thresholds of the network. Assign the weights and thresholds that meet the accuracy requirements or reach the maximum number of evolutionary generations to the MLP model, and use the trained MLP model for prediction.

[0045] S103: Based on the multi - layer perceptron model with the determined number of nodes, construct a multi - layer perceptron prediction model optimized by the WOA - LSO algorithm; the multi - layer perceptron model optimized by the WOA - LSO algorithm is obtained by using the WOA - LSO algorithm to optimize the weight sequence and threshold sequence of the multi - layer perceptron model, and obtaining the optimal weight sequence and optimal threshold sequence and assigning them to the multi - layer perceptron model.

[0046] Optionally, the weight sequence and threshold sequence of the multi-layer perceptron model are optimized by the WOA-LSO algorithm, which specifically includes: initializing the population individual position, population size, and maximum number of iterations of the WOA-LSO algorithm; the population individual position is an encoded sequence obtained by encoding the weight sequence and threshold sequence of the multi-layer perceptron model; selecting the mean square error of the soil transpiration prediction by the multi-layer perceptron model as the fitness function value of the WOA-LSO algorithm, iteratively updating the population individual position, and iteratively calculating the fitness function value of the WOA-LSO algorithm; when the WOA-LSO algorithm reaches the maximum number of iterations or the fitness function value is greater than the preset threshold, the iteration stops, and the optimal population individual position is obtained. The optimal population individual position is the optimal encoded sequence, and the optimal encoded sequence is decoded to obtain the optimal weight sequence and optimal threshold sequence.

[0047] Specifically, the encoded sequence is: ; In the formula, is the encoded sequence, is the weight sequence of the multi-layer perceptron model, is the threshold sequence of the multi-layer perceptron model, n and m are the number of weight values and the number of thresholds of the multi-layer perceptron model, respectively.

[0048] Specifically, in the WOA-LSO algorithm, first set the size of the whale population to 30, the initial spectral population to 30, and at the same time set the maximum number of iterations to 100 to limit the running time of the algorithm. Next, the search space needs to be defined, which specifically includes the range of hyperparameters to be optimized, such as the learning rate and the number of hidden layer nodes. To evaluate the effects of different hyperparameter configurations, a fitness function is defined. Usually, the performance metric (such as mean square error, MSE) of the model on the validation set is used as the fitness value. This fitness function guides the whale population to adjust their positions in the search space to find the optimal solution by training the model according to the current hyperparameter configuration and calculating the loss on the validation set. Within the maximum number of iterations, update the positions of the whales, calculate the fitness, and keep the optimal solution. By iteratively optimizing the hyperparameters, the optimal learning rate and the number of hidden layer nodes are found.

[0049] S104: Input multiple target meteorological data factors of the study area into the multi-layer perceptron prediction model to obtain the soil transpiration in the future period.

[0050] Specifically, the inconsistent physical meanings and units of the input vectors will affect the training process of the network. Therefore, the input vectors and the expected output vector values are normalized according to the formula.

[0051] ; The anti-normalization formula is as follows: ; In the formula, are the maximum and minimum values in the original sample, are the original sample and the corresponding processed data. According to the characteristics of the reference crop evapotranspiration prediction problem, the following indicators are selected to evaluate the established MLP, standard deviation, and the formula is: ; The average value of the absolute relative error, and the formula is: ; The prediction qualification rate is set as the percentage of samples with the absolute value of the relative error in the total number of samples; is the coefficient of the linear regression equation; is the coefficient of determination, that is, the square of the correlation coefficient. To verify the stability of the model, we conduct independent repeated experiments and record the prediction relative error, prediction standard deviation, and prediction qualification rate of each experiment result.

[0052] Specifically, the input variables are seven influencing factors such as the highest temperature, average temperature, lowest temperature, average relative humidity, day number, sunshine hours, and wind speed. An MLP model is established for prediction, and the network structure is 7-13-1. The WOA-LSO algorithm is used to optimize the network weights and thresholds, and the fitness function is taken as the sum of the absolute values of the training errors. Then, the trained MLP network model is used to predict the reference crop evapotranspiration, and the evaluation indicators and prediction results are as follows: Standard deviation: ; The average value of the absolute relative error: ; The qualification rate is 94.36% when =0.987, and the square of the correlation coefficient ; In summary, Scheme 2-9 obtains the results as above and participates in Table 3: Table 3 Prediction accuracy results

[0053] From the prediction results of Comparative Schemes 1, 2, and 3, it can be seen that the prediction effects of the three schemes are all relatively good, and the values of various evaluation indicators are very close, indicating that there is a great correlation among the three influencing factors of the highest temperature, average temperature, and lowest temperature. Only the average temperature among the three needs to be selected, and combined with the other four influencing factors as the input quantity to obtain a relatively accurate prediction result. From the prediction results of Comparative Schemes 4, 5, 6, 7, and 8, it can be seen that on the basis of Scheme 3, when the input factors are reduced to four, the prediction results and indicators of Scheme 7 are almost close to those of Scheme 3, and are significantly better than those of Schemes 4, 5, 6, and 8. This shows that the day sequence number has little effect on the prediction result, while the four influencing factors of average temperature, average relative humidity, wind speed, and sunshine hours are all indispensable when predicting the reference evapotranspiration. Comparing the prediction results of Scheme 9 with those of Schemes 4, 5, 6, 7, and 8 further proves that the four influencing factors of average temperature, average relative humidity, wind speed, and sunshine hours are essential when predicting the crop reference evapotranspiration.

[0054] In summary, the prediction effects of Scheme 1, Scheme 2, Scheme 3, and Scheme 7 are better. Through comparative analysis, in order to reduce the network operation load, streamline the network structure, and improve the model operation efficiency, Scheme 7 is adopted as the model for predicting evapotranspiration. This scheme only needs to take the four influencing factors of average temperature, average relative humidity, sunshine hours, and wind speed as the input, and then a relatively accurate prediction value can be obtained through the MLP model. The prediction results of Scheme 7 show that the average value of the absolute value of the prediction relative error is 6.02%, the prediction qualification rate is 94.52%, and the standard deviation is 0.079.

[0055] S105: Input the available soil moisture content and soil evapotranspiration of the research area into the fuzzy inference system to obtain the irrigation coefficient, and calculate the required irrigation water volume for the farmland according to the irrigation coefficient.

[0056] Optionally, the fuzzy inference system includes: an input layer, a fuzzification layer, a rule layer, an aggregation layer, and a defuzzification layer; the step of inputting the available soil moisture content and soil evapotranspiration of the research area into the fuzzy inference system to obtain the irrigation coefficient specifically includes: inputting the available soil moisture content and soil evapotranspiration into the fuzzification layer through the input layer; in the fuzzification layer, using the Gaussian membership function to fuzzify the available soil moisture content and soil evapotranspiration to obtain the initial fuzzy sets; in the rule layer, according to the predefined fuzzy rules, reasoning about the fuzzy sets to obtain multiple fuzzy outputs; in the aggregation layer, aggregating all the fuzzy outputs into a target fuzzy set; in the defuzzification layer, using the centroid method to clarify the fuzzy outputs in the target fuzzy set to obtain the irrigation coefficient.

[0057] Optionally, calculating the required irrigation water volume for the farmland according to the irrigation coefficient includes: calculating the required irrigation water volume for the farmland according to the irrigation coefficient in combination with the irrigation quota formula, and the formula is: ; In the formula, is the irrigation water requirement of farmland, is the irrigation coefficient, is the soil transpiration, is the irrigation water use coefficient.

[0058] Specifically, to establish a model based on a Fuzzy Inference System (FIS), refer to Figure 4 , which is the modeling flowchart of the fuzzy irrigation decision-making system. To predict the irrigation coefficient α, the soil available water content (SAWC) and the actual transpiration (AT) of Poa pratensis will be taken as input variables and processed using a Fuzzy Neural Network (FNN). The following are the steps to establish this model: In this model, input and output variables are defined. The input variables include the soil available water content (SAWC) and the actual transpiration (AT) of the soil. Among them, the range of SAWC is [0, 100] mm, and the range of actual transpiration is [0, 10] mm / day. The output variable is the irrigation coefficient α, and its range is [0, 1]. For ease of understanding and analysis, we divide the input and output variables into linguistic variables. The soil available water content (SAWC) can be divided into three levels: Low, Medium, and High. The actual transpiration (AT) of the soil is also divided into Low, Medium, and High. Finally, the irrigation coefficient α is also divided into Low, Medium, and High. This division can help us better understand the relationships between different variables and their effects on the irrigation coefficient.

[0059] Specifically, in this model, the input variables, soil available water content (SAWC) and actual transpiration of the soil, are described by Gaussian membership functions, which can effectively express the fuzziness and uncertainty of the input variables within a specific range. On the other hand, the output variable, irrigation coefficient α, is represented by a triangular membership function. Here, a, b, and c are the three vertices of the triangle, defining different levels of the irrigation coefficient. Through this method, we can more clearly represent and analyze the variation of the irrigation coefficient under different input conditions. In this model, we use a fuzzy neural network (FNN) to construct a fuzzy inference system, combining the advantages of fuzzy logic and neural networks, and design it according to the following steps. First, the input layer has two nodes, corresponding to the soil available water content (SAWC) and the actual transpiration of the soil (AT) respectively. Next, in the fuzzification layer, the Gaussian membership function is used to fuzzify the input, converting the actual input value into a fuzzy set. Subsequently, in the rule layer, according to the predefined fuzzy rules (e.g., "if SAWC is high and AT is high, then α is high"), the fuzzy input is inferred to generate a fuzzy output. Then, in the aggregation layer, the fuzzy outputs of all rules are aggregated to generate a comprehensive fuzzy output. Finally, in the defuzzification layer, the centroid method is used to convert the aggregated fuzzy output into a clear numerical value α. To improve the user experience, we also provide a visualization interface where users can view the membership functions of the input and output variables through graphs and intuitively observe the fuzzy inference process and its output results under different input conditions.

[0060] In addition, when using the fuzzy neural network (FNN) to predict the irrigation coefficient α, it is first necessary to train and validate the model with historical data to ensure its accuracy. Once the training is completed and verified to be effective, the real-time soil available water content (SAWC) and the actual transpiration of Kentucky bluegrass (AT) can be input into the FNN model, and the model will output the irrigation coefficient α to assist in farmland irrigation management. To improve the intuitiveness of decision-making, the prediction results can be presented graphically and a sensitivity analysis can be conducted to observe the influence of different input variables on α. In addition, the model can be optimized by adjusting the parameters of the Gaussian membership function and expanding the fuzzy rule base. If more influencing factors (such as temperature, humidity, wind speed) are introduced, the prediction accuracy can be further improved. Through these methods, the FNN model can effectively support farmland management decisions.

[0061] Additionally, to address the complex irrigation problems in the Tibetan Plateau region, we designed an irrigation fuzzy decision-making system based on the irrigation coefficient fuzzy neural network, irrigation quota formula, and water balance equation. When selecting suitable experimental fields, the system comprehensively considers soil type, terrain, and water source conditions, conducts multiple experiments for different meteorological conditions (such as soil moisture changes on sunny days, cloudy days, and after rainfall), and records the irrigation effects and crop growth conditions, including growth rate, yield, and water resource utilization efficiency. During the construction of the fuzzy neural network, input variables (such as soil moisture, air temperature, humidity, and precipitation) are first defined, and the output variable is set as the irrigation coefficient α. Then, a fuzzy rule base is established, and the fuzzy neural network is trained using historical data to optimize the fuzzy rules and membership functions. Subsequently, the irrigation quota is calculated through the irrigation quota formula, covering key data such as the actual evapotranspiration of crops, precipitation, soil moisture leakage, and consumption. The water balance equation is used to evaluate soil moisture changes and determine the necessity of irrigation. Finally, real-time data is input into the fuzzy neural network to obtain the irrigation coefficient α, and the required irrigation volume is calculated in combination with the irrigation quota formula. Through the irrigation decision-making of this system, the effective soil moisture content can be stabilized within the expected range, not only ensuring the healthy growth of crops but also achieving effective water resource conservation. Using the irrigation fuzzy decision-making system to make irrigation decisions for the plateau region in November 2024, the results show that through the irrigation decision-making of this system, the effective soil moisture content can be stabilized within the expected set range, saving water resources while ensuring the healthy growth of crops.

[0062] In addition, based on the actual test results, the model and system are evaluated, and improvement suggestions are put forward to further improve the efficiency and effectiveness of irrigation management. By continuously optimizing the model parameters and decision rules, the adaptability and reliability of the system under different environmental conditions are ensured.

[0063] In summary, the present invention adopts a multi-layer perceptron model and nests the WOA-LSO algorithm. The global search ability effectively avoids the local optimal solution problem that may be encountered by traditional machine learning algorithms in the case of scarce or incomplete data. Moreover, the combination of the WOA and LSO algorithms can achieve complementary effects and make up for a series of problems in the prior art in data processing. The present invention introduces a fuzzy-processed crop coefficient, enabling the calculation of soil transpiration and available soil moisture to consider various uncertain factors. This method effectively reduces the risk brought by data uncertainty in traditional irrigation decision-making and lays a solid foundation for the next-step irrigation decision-making. By comprehensively discussing the accurate calculation of available moisture and water demand to determine irrigation requirements, water resource waste is reduced. The prediction accuracy of soil transpiration in data processing is higher and the speed is faster than that of traditional algorithms, enabling better irrigation decision-making and enhancing the adaptability and flexibility of irrigation decision-making. The present invention is specifically designed for the extreme climate and water resource shortage problems in the Qinghai-Tibet Plateau region, combined with efficient data acquisition means, ensuring feasibility and effectiveness in a specific environment. The WOA-LSO algorithm is far superior to other algorithms in terms of adaptability.

[0064] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

Claims

1. A farmland irrigation method based on a multilayer perceptron model of the WOA-LSO algorithm, characterized in that: include: Obtain multiple sample meteorological data factors; According to multiple sample meteorological data factors, multiple combinations of sample meteorological data factors are determined, and the correlation test between each sample meteorological data factor combination and soil evaporation is performed respectively to determine the target sample meteorological data factor combination, and according to the target sample meteorological data factor combination, the number of nodes in the input layer, output layer and hidden layer of the multilayer perceptron model is determined; Based on the multilayer perceptron model with the determined number of nodes, a multilayer perceptron prediction model optimized based on the WOA-LSO algorithm is constructed; the multilayer perceptron model optimized based on the WOA-LSO algorithm is obtained by optimizing the weight sequence and threshold sequence of the multilayer perceptron model through the WOA-LSO algorithm, obtaining the optimal weight sequence and the optimal threshold sequence and assigning them to the multilayer perceptron model; Input multiple target meteorological data factors of the study area into the multi-layer perceptron prediction model to obtain the soil evaporation in the future period; The effective soil moisture content and soil evaporation in the study area were input into the fuzzy inference system to obtain the irrigation coefficient, and the required irrigation amount of the farmland was calculated based on the irrigation coefficient.

2. The farmland irrigation method based on the multilayer perceptron model of the WOA-LSO algorithm as claimed in claim 1, characterized in that: The multiple sample meteorological data factors include: maximum temperature, average temperature, minimum temperature, average relative humidity, day number, sunshine hours and wind speed; the multiple sample meteorological data factor combinations include a first group of meteorological data factor combinations and a second group of meteorological data factor combinations; according to the multiple sample meteorological data factors, determining the multiple sample meteorological data factor combinations specifically includes: Determine a first group of meteorological data factor combinations according to multiple sample meteorological data factors; the first group of meteorological data factor combinations includes all meteorological data factor combinations, meteorological data factor combinations excluding the lowest temperature item, and meteorological data factor combinations excluding the highest temperature and the lowest temperature; Among the five meteorological data factors in the meteorological data factor combination excluding the highest temperature and the lowest temperature, four are selected for random arrangement and combination, and all their combinations are determined as the second group of meteorological data factor combinations.

3. The farmland irrigation method based on the multilayer perceptron model of the WOA-LSO algorithm as claimed in claim 2, characterized in that: The step of respectively performing a correlation test between each sample meteorological data factor combination and soil evaporation to determine a target sample meteorological data factor combination includes: Each combination of meteorological data factors is input into the initial multi-layer perceptron model to obtain the predicted soil evaporation under each combination of meteorological data factors; According to the predicted soil evaporation and the actual soil evaporation under each combination of meteorological data factors, the correlation between each combination of meteorological data factors and soil evaporation is determined; Obtain the candidate meteorological data factor combination with the greatest correlation in the second group of meteorological data factor combinations, and remove each meteorological data factor in the candidate meteorological data factor combination in turn to obtain a third group of meteorological data factor combinations; Obtain the correlation between each meteorological data factor combination in the third group of meteorological data factor combinations and soil evaporation; The meteorological data factor combination with the greatest correlation among all meteorological data factor combinations is determined as the target sample meteorological data factor combination.

4. The farmland irrigation method based on the multilayer perceptron model of the WOA-LSO algorithm as claimed in claim 1, characterized in that: The method of determining the number of nodes in the input layer, the output layer and the hidden layer in the multilayer perceptron model according to the combination of target sample meteorological data factors includes: The number of sample meteorological data factors in the target sample meteorological data factor combination is determined as the number of nodes in the input layer; Determine the number of nodes in the output layer to be 1; The number of nodes in the hidden layer is determined according to the number of nodes in the input layer and the number of nodes in the output layer.

5. The farmland irrigation method based on the multilayer perceptron model of the WOA-LSO algorithm as claimed in claim 4, characterized in that: Determining the number of nodes in the hidden layer according to the number of nodes in the input layer and the number of nodes in the output layer specifically includes: According to the relationship between the number of hidden layer nodes, the number of input layer nodes, and the number of output layer nodes, the range of the number of hidden layer nodes is determined by the formula: ； In the formula, is the minimum number of hidden layer nodes, is the maximum number of hidden layer nodes, is the number of input layer nodes, is the number of nodes in the output layer, is the number of hidden layer nodes; In the range of the number of hidden layer nodes The prediction accuracy of the multilayer perceptron prediction model under each hidden layer node number is obtained in turn, and the number of nodes corresponding to the highest prediction accuracy is determined as the number of hidden layer nodes.

6. The farmland irrigation method based on the multilayer perceptron model of the WOA-LSO algorithm as claimed in claim 1, characterized in that: The optimization of the weight sequence and threshold sequence of the multilayer perceptron model by the WOA-LSO algorithm specifically includes: Initialize the population individual position, population size and maximum number of iterations of the WOA-LSO algorithm; the population individual position is a coding sequence obtained by encoding the weight sequence and threshold sequence of the multilayer perceptron model; The mean square error of soil evaporation prediction by the multilayer perceptron model is selected as the fitness function value of the WOA-LSO algorithm, the positions of the individuals in the population are iteratively updated, and the fitness function value of the WOA-LSO algorithm is iteratively calculated; When the WOA-LSO algorithm reaches the maximum number of iterations or the fitness function value is greater than the preset threshold, the iteration stops and the optimal population individual position is obtained. The optimal population individual position is the optimal coding sequence, and the optimal coding sequence is decoded to obtain the optimal weight sequence and the optimal threshold sequence.

7. The farmland irrigation method based on the multilayer perceptron model of the WOA-LSO algorithm as claimed in claim 6, characterized in that: The coding sequence is: ； In the formula, is the coding sequence, is the weight sequence of the multilayer perceptron model, is the threshold sequence of the multilayer perceptron model, n and m are the number of weights and thresholds of the multilayer perceptron model respectively.

8. The farmland irrigation method based on the multilayer perceptron model of the WOA-LSO algorithm as claimed in claim 1, characterized in that: The fuzzy inference system includes: an input layer, a fuzzification layer, a rule layer, an aggregation layer and a defuzzification layer; the effective soil moisture content and soil transpiration of the study area are input into the fuzzy inference system to obtain the irrigation coefficient, which specifically includes: The effective soil water content and soil evaporation are input into the fuzzification layer through the input layer; In the fuzzification layer, the Gaussian membership function is used to fuzzify the soil effective water content and soil evaporation to obtain the initial fuzzy set. In the rule layer, the fuzzy set is reasoned according to the predefined fuzzy rules to obtain multiple fuzzy outputs; In the aggregation layer, all fuzzy outputs are aggregated into a target fuzzy set; In the defuzzification layer, the fuzzy output in the target fuzzy set is clarified using the centroid method to obtain the watering coefficient.

9. The farmland irrigation method based on the multilayer perceptron model of the WOA-LSO algorithm as claimed in claim 1, characterized in that: The calculation of the required irrigation amount for farmland according to the irrigation coefficient includes: According to the irrigation coefficient and the irrigation quota formula, the amount of irrigation required for farmland is calculated as follows: ； In the formula, The amount of water needed for irrigation of farmland, is the irrigation coefficient, is the soil evaporation, is the irrigation water utilization coefficient.

Images