Distributed improvement method and device for farmland crop irrigation water productivity

Abstract

The invention discloses a distributed improvement method and device for farmland crop irrigation water productivity. The method comprises the steps that soil physical indexes and nutrient indexes are obtained; performing management area division on the farmland by using a fuzzy clustering method based on the soil physical indexes and the nutrient indexes; sorting the contribution rate of each soil index to the crop irrigation water productivity in each management partition by using a partial least square discriminant analysis method to obtain a driving factor of the crop irrigation water productivity; using a boundary line analysis method to determine a regulation threshold value of a crop irrigation water productivity driving factor in each management area; calculating actual stage water consumption of crops in each management area, establishing an irrigation system optimization model, and performing multi-objective optimization solution on the irrigation system optimization model by applying a genetic algorithm to obtain an optimized irrigation system; and combining the regulation and control threshold value of the crop irrigation water productivity driving factor with the optimized irrigation system to obtain a farmland distributed crop management scheme, so that the crop irrigation water productivity is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of farmland crop management, and in particular relates to a distributed method and device for improving the productivity of farmland crop irrigation water. Background Art

[0002] Large-scale farmland management is developing rapidly in China today. It has the characteristics of high mechanization and labor saving that individual farmers do not have in decentralized management. However, in actual farmland crop management, a uniform and extensive management model has always been followed, without considering the spatial heterogeneity of soil, which leads to variability in crop yields within the farmland and low productivity of crop irrigation water. Therefore, how to "adapt to local conditions" to distributed management of crops, implement different management models in different regions, and improve irrigation water productivity under the premise of stable production is an urgent problem to be solved.

[0003] The productivity of crop irrigation water in farmland is mainly affected by factors such as soil physical properties, nutrients, and water and fertilizer management. Domestic and foreign scholars have used different methods to analyze the impact of different soil indicators on crop yield and irrigation water productivity, and conducted single-point experiments on the optimization of various water and fertilizer systems, but there is little research on regional distributed comprehensive improvement plans for crop irrigation water productivity. Summary of the invention

[0004] The technical problem to be solved by the present invention is how to formulate a regional distributed comprehensive improvement plan for crops, and to provide a distributed improvement method and device for farmland crop irrigation water productivity.

[0005] To achieve the above object, the present invention adopts the following technical solution:

[0006] A distributed method for improving farmland crop irrigation water productivity, characterized by comprising:

[0007] Step S1, obtaining soil physical indicators and nutrient indicators;

[0008] Step S2: Based on soil physical indicators and nutrient indicators, the fuzzy clustering method is used to divide the farmland into management areas; the partial least squares discriminant analysis method is used to rank the contribution rate of each soil indicator to the crop irrigation water productivity in each management area to obtain the driving factor of the crop irrigation water productivity; the boundary line analysis method is used to determine the control threshold of the driving factor of the crop irrigation water productivity in each management area;

[0009] Step S3, calculate the actual periodic water consumption of crops in each management area, establish an irrigation system optimization model, apply genetic algorithm to perform multi-objective optimization solution on the irrigation system optimization model, and obtain the optimized irrigation system; combine the control threshold of the crop irrigation water productivity driving factor with the optimized irrigation system to obtain a distributed crop management plan for farmland.

[0010] Preferably, in step S2, partial least squares discriminant analysis is used in each management zone to quantify and rank the contribution of each soil indicator to crop irrigation water productivity based on variable projection importance VIP, and indicators with VIP values ​​greater than 1 are screened out as driving factors of crop irrigation water productivity.

[0011] Preferably, soil physical indicators include: soil mechanical composition, soil bulk density; soil nutrient indicators include: soil organic matter, total nitrogen, total phosphorus, and total potassium.

[0012] Preferably, in step S3, a crop water production function is established, with the total irrigation water volume and the staged irrigation water volume as constraints, and two equally weighted objective functions of maximum crop irrigation water productivity and maximum yield are set to obtain an irrigation system optimization model; a genetic algorithm is used to perform multi-objective optimization on the irrigation system optimization model to obtain an optimized irrigation system.

[0013] The present invention also provides a distributed device for improving the productivity of farmland crop irrigation water, comprising:

[0014] The first processing module is used to obtain soil physical indicators and nutrient indicators;

[0015] The second processing module is used to divide farmland into management zones based on soil physical indicators and nutrient indicators using the fuzzy clustering method; to rank the contribution rate of each soil indicator to crop irrigation water productivity in each management zone using the partial least squares discriminant analysis method to obtain the driving factor of crop irrigation water productivity; and to determine the control threshold of the driving factor of crop irrigation water productivity in each management zone using the boundary line analysis method;

[0016] The third processing module is used to calculate the actual staged water consumption of crops in each management area, establish an irrigation system optimization model, and use genetic algorithms to perform multi-objective optimization solutions on the irrigation system optimization model to obtain an optimized irrigation system; the control threshold of the crop irrigation water productivity driving factor is combined with the optimized irrigation system to obtain a distributed crop management plan for farmland.

[0017] Preferably, the second processing module is used to use partial least squares discriminant analysis in each management area to quantify and rank the contribution rate of each soil indicator to crop irrigation water productivity based on the variable projection importance VIP, and screen out indicators with VIP values ​​greater than 1 as driving factors of crop irrigation water productivity.

[0018] Preferably, soil physical indicators include: soil mechanical composition, soil bulk density; soil nutrient indicators include: soil organic matter, total nitrogen, total phosphorus, and total potassium.

[0019] Preferably, the third processing module is used to establish a crop water production function, with the total irrigation water volume and the phased irrigation water volume as constraints, and set two equally weighted objective functions of maximum crop irrigation water productivity and maximum yield to obtain an irrigation system optimization model; a genetic algorithm is used to perform multi-objective optimization on the irrigation system optimization model to obtain an optimized irrigation system.

[0020] Compared with the existing technology, the present invention fully considers the spatial heterogeneity of soil physics and nutrient indicators to divide crop management zones, and quantifies the contribution rate and optimal control threshold of each indicator to crop irrigation water productivity in each zone. At the same time, the irrigation system is optimized based on the actual water consumption of crops in different management areas to obtain a distributed management plan for farmland crops, thereby achieving the purpose of distributed improvement of crop irrigation water productivity. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0022] Figure 1 It is a flow chart of the distributed comprehensive method for improving the productivity of farmland crop irrigation water;

[0023] Figure 2 Schematic diagram of boundary line analysis method. DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0026] Embodiment 1:

[0027] like Figure 1 As shown, an embodiment of the present invention provides a distributed method for improving the productivity of farmland crop irrigation water, comprising:

[0028] Step S1: Obtain soil physical indicators and nutrient indicators

[0029] Select the research area, obtain the selected farmland specifications and area, arrange sampling units and sampling points, and use this as a basis to obtain soil physical and nutrient indicators in each sampling unit, analyze NDVI images to invert crop yields, and collect irrigation water volume to calculate irrigation water productivity.

[0030] (1)

[0031] Where IWP is the irrigation water productivity, kg / m³; I i The amount of irrigation water (mm) at each growth stage.

[0032] Step S2: Based on soil physical indicators and nutrient indicators, the fuzzy c-means clustering method (FCM) is used to divide the farmland into management areas, and the fuzzy areas with smaller spatial heterogeneity of soil indicators are determined; the partial least squares discriminant analysis method (PLS-DA) is used in each management area to quantify the contribution rate of each soil indicator to crop irrigation water productivity according to variable projection importance (VIP) and sort them, and the indicators with VIP values ​​greater than 1 are selected as the driving factors of crop irrigation water productivity; the boundary line analysis method is used to determine the control threshold of the driving factors of crop irrigation water productivity in each management area; specifically, it includes:

[0033] Step 201: Considering soil physical indicators (soil mechanical composition, bulk density) and soil nutrients (organic matter, total nitrogen, total phosphorus, total potassium), the data is standardized to eliminate the dimensional differences between indicators, and the management area is divided using the fuzzy c-means clustering method. The basic idea is to find the iterative minimization of the objective function. The most commonly used objective function is:

[0034] (2)

[0035] Among them, U n×c is the membership matrix, n is the number of data, c is the number of categories, m is the fuzzy weighted index, u ik is the kth sample x in the data matrix X k Belongs to the i-th cluster center v in the cluster center matrix V i The membership value of ik ) 2 Equal to x k With v i The square of the distance on the eigenvector.

[0036] The fuzzy c-means clustering algorithm achieves classification through an iterative optimization process. First, the membership matrix is ​​initialized, and the initialization value is set randomly. Then, c cluster centers are calculated according to formula (3). The membership values ​​of each data point in the class are calculated by these cluster centers through formula (4). The membership matrix is ​​adjusted, and new c cluster centers are calculated based on the adjusted membership matrix. The changes of cluster centers are compared. If the change between two cycles is less than the preset threshold, the iterative calculation is stopped. Otherwise, the above steps are repeated until the change of cluster center is less than the set threshold.

[0037] (3)

[0038] (4)

[0039] The clustering effectiveness can be determined based on the fuzzy performance index (FPI) and normalized classification entropy (NCE) to determine the optimal number of clusters and test the clustering effect. The fuzzy performance index (FPI) can be defined as:

[0040] (5)

[0041] Normalized categorical entropy (NCE) can be defined as:

[0042] (6)

[0043] Among them, the smaller the FPI and NCE, the better the clustering effect.

[0044] Step 202: Apply the partial least squares discriminant analysis (PLS-DA) to quantitatively analyze the irrigation water productivity and calculate the variable projection importance (VIP) of each indicator using formula (7); VIP refers to the independent variable x in the partial least squares discriminant analysis. k Importance in explaining the dependent variable Y is defined as:

[0045] (7)

[0046] Among them, w jk Axis w j The kth component of k The interpretation of Y is through t h When the VIP value is greater than 1, it means that the independent variable has a more important effect on the dependent variable and is the main driving factor; when the VIP value is less than 0.8, the independent variable contributes less to the dependent variable. Based on this, the contribution rate of each independent variable to the dependent variable can be calculated by the VIP value of the independent variable and the multiple correlation coefficient of the PLS-DA regression model. The calculation method is as follows:

[0047] (8)

[0048] Among them, W i is the contribution rate of the ith factor to the dependent variable; VIP i is the VIP value of the i-th factor; R is the multiple correlation coefficient of the PLS-DA regression model; n is the number of independent variables in the PLS model. Finally, the key driving factors of crop water productivity (K1, K2, ..., K n ) and contribution rate (W1, W2, …, W n ).

[0049] Step 203: Apply boundary line analysis to determine the optimal control threshold of the key driving factor. Each data point corresponding to the X axis has a corresponding maximum Y value (the point with the highest irrigation water productivity corresponding to a certain indicator). This point is called a boundary point. The boundary points of all points are determined and fitted to obtain the boundary line, such as Figure 2 As shown in the figure, it is recommended that the irrigation water productivity reach 95% as the upper and lower limits of the field control thresholds of each indicator.

[0050] Step S3: Calculate the actual periodic water consumption of crops in each management area and establish an irrigation system optimization model; use genetic algorithms to perform multi-objective optimization on the irrigation system optimization model to obtain an optimized irrigation system; combine the control thresholds of the driving factors of crop irrigation water productivity in each management area to propose a distributed management plan for farmland crops to improve crop irrigation water productivity.

[0051] Step 301: Establish a crop water production function, take the total irrigation water volume and the staged irrigation water volume as constraints, set two equally weighted objective functions of maximum crop irrigation water productivity and maximum yield, and obtain an irrigation system optimization model

[0052] (1) Principle of crop water production function

[0053] The Crop Water Function (CWF) is a static model used to describe the macroscopic relationship between final yield (dry matter or grain yield) and water, without considering the microscopic mechanism of how dry matter accumulates during crop growth and development. The effects of irrigation time and irrigation amount on yield can be considered in stages, mainly based on the calculated water sensitivity coefficient to reflect the sensitivity of water deficit to crop yield at each stage.

[0054] (2) Construction of irrigation system optimization model

[0055] The irrigation schedule optimization model includes subject function, constraint conditions and objective function;

[0056] Main function:

[0057] (9)

[0058] Among them, y is the relative output, Y a is the actual yield (kg / ha), i is the growth stage during the growth period, ET ai is the actual transpiration of crops at each growth stage (mm), Y m is the maximum crop yield (kg / ha), ET mi is the maximum transpiration of crops at each growth stage (mm), λ i It is the moisture sensitivity coefficient of crops at different growth stages.

[0059] Objective function:

[0060] (10)

[0061] Where IWP is the irrigation water productivity, kg / m³; I i The amount of irrigation water (mm) at each growth stage.

[0062] Constraints:

[0063] 1) Phased irrigation volume constraints, , I imax and I imin is the maximum and minimum irrigation amount in the i-th growth period;

[0064] 2) Total irrigation volume constraint, ,∑I imax The maximum total amount of irrigation during the entire growth period;

[0065] 3) Output constraints, , Y min To set a minimum yield constraint for crops to meet production needs;

[0066] Step 302: Use genetic algorithm to solve the multi-objective optimization of the irrigation schedule optimization model to obtain the optimized irrigation schedule.

[0067] Genetic Algorithm (GA) is an optimization technique inspired by biological evolution and used to solve complex search and optimization problems. It simulates natural selection and genetic mechanisms, and finds the optimal or near-optimal solution by iteratively improving candidate solutions. In order to solve the multi-objective (yield, irrigation water productivity) optimization problem in irrigation schedule optimization, this paper uses the DEAP (Differential Evolutionary Algorithms in PurePython) library in the python programming language to implement the genetic algorithm, and combines the actual needs of agricultural irrigation to establish a fitness function and constraint processing mechanism. The details are as follows:

[0068] 1. Fitness function construction

[0069] Taking into account irrigation water productivity (IWP) and yield (Y a ) The optimization requirements of the two objectives are defined as the fitness function F:

[0070] F = w1*IWP + w2* (11)

[0071] Among them, w 1、 w2 is the equal weight coefficient, w1=0.5, w2=0.5; IWP is the irrigation water productivity shown in the objective function (10); Y a is the actual output; Y m For maximum output.

[0072] The fitness function takes into account both resource utilization efficiency and production benefits, which meets the requirements of sustainable agricultural development.

[0073] 2. Initialize the population:

[0074] Individual coding adopts real number coding, and each individual is represented by an n-dimensional vector, corresponding to the irrigation amount I at each growth stage. i (i=1, 2, ..., n), set the initial population size n=500.

[0075] 3. Constraint processing

[0076] In the genetic algorithm, the initialized population and the individuals after crossover mutation must be within certain constraints. The present invention uses the phased irrigation volume, total irrigation volume and yield for constraints, as follows:

[0077] Phased irrigation amount constraint: I imin ≤ I i ≤ I imax ;

[0078] Total irrigation amount constraint: ΣI i ≤ ∑I imax ;

[0079] Output constraint: Y a ≥ Y min .

[0080] In addition, the present invention uses a penalty function as a penalty for solutions that do not meet the constraints:

[0081] F' = F - penalty*Σ(constraint_violations) (12)

[0082] Among them, penalty=0.3 (penalty coefficient), constraint_violations includes the three constraints mentioned above, which ensures the feasibility of the optimization solution and avoids unreasonable irrigation schemes.

[0083] 4. Select an action:

[0084] The proportional selection method is used to retain individuals with higher fitness as parents.

[0085] 5. Crossover operation:

[0086] Implement a single-point crossover strategy to exchange some individual fragments to retain excellent individuals. p =(I p1 , I p2 , ..., I pn ), I q =(I q1 , I q2 , ..., I qn ), then offspring I r =(I r1 , I r2 , ..., I rn ) The generation rules are:

[0087] I ri = I pi * (1-α) + I qi * α, i ≤ k (13)

[0088] I ri = I qi * (1-β) + I pi * β, i > k (14)

[0089] Among them, α, β ∈ [0.7, 0.9]; k ∈ [1, n], are randomly selected crossover points.

[0090] 7. Mutation operation:

[0091] In order to avoid local solutions and expand the coverage of the population, for each gene of each offspring individual (i.e., the amount of irrigation at each growth stage I i ), with probability p m = 0.01 to perform mutation operation. The specific steps are as follows:

[0092] Randomly select a gene I i , with probability p m = 0.01 to mutate it; if mutation is triggered, the magnitude of mutation is calculated:

[0093] ΔIi = ε * (U(I imin , I imax )-I i ) (15)

[0094] Among them: ε ∈ [0.1, 0.3] is a uniformly distributed random number that controls the proportion of the variation amplitude; U(I imin , I imax ) represents the interval [I imin , I imax ] randomly sampled values ​​from ].

[0095] In this way, the variation operation ensures the reasonable adjustment of irrigation water volume and avoids excessive deviation from the actual agricultural operation range.

[0096] 8. Termination conditions:

[0097] The algorithm is terminated when the maximum number of iterations T = 300 is reached or the fitness converges.

[0098] The present invention fully considers the spatial variability of soil indicators and their contribution rate to crop irrigation water productivity, establishes the optimal threshold control model and irrigation system optimization model of the driving factors of zoned crop irrigation water productivity, obtains the main influencing indicators of crop irrigation water productivity and the crop optimized irrigation system in different management zones, and thus realizes the distributed comprehensive improvement of farmland crop irrigation water productivity.

[0099] Embodiment 2:

[0100] The embodiment of the present invention also provides a distributed device for improving the productivity of farmland crop irrigation water, comprising:

[0101] The first processing module is used to obtain soil physical indicators and nutrient indicators;

[0102] The second processing module is used to divide farmland into management zones based on soil physical indicators and nutrient indicators using the fuzzy clustering method; to rank the contribution rate of each soil indicator to crop irrigation water productivity in each management zone using the partial least squares discriminant analysis method to obtain the driving factor of crop irrigation water productivity; and to determine the control threshold of the driving factor of crop irrigation water productivity in each management zone using the boundary line analysis method;

[0103] The third processing module is used to calculate the actual staged water consumption of crops in each management area, establish an irrigation system optimization model, and use genetic algorithms to perform multi-objective optimization solutions on the irrigation system optimization model to obtain an optimized irrigation system; the control threshold of the crop irrigation water productivity driving factor is combined with the optimized irrigation system to obtain a distributed crop management plan for farmland.

[0104] As an implementation method of an embodiment of the present invention, the second processing module is used to use the partial least squares discriminant analysis method in each management area to quantify the contribution rate of each soil indicator to the crop irrigation water productivity according to the variable projection importance VIP and sort them, and screen out indicators with VIP values ​​greater than 1 as driving factors of crop irrigation water productivity.

[0105] As an implementation method of an embodiment of the present invention, soil physical indicators include: soil mechanical composition, soil bulk density; soil nutrient indicators include: soil organic matter, total nitrogen, total phosphorus, and total potassium.

[0106] As an implementation mode of an embodiment of the present invention, the third processing module is used to establish a crop water production function, with the total irrigation water volume and the staged irrigation water volume as constraints, setting two equally weighted objective functions of maximum crop irrigation water productivity and maximum yield, to obtain an irrigation system optimization model; a genetic algorithm is used to perform multi-objective optimization on the irrigation system optimization model to obtain an optimized irrigation system.

[0107] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A distributed method for improving the productivity of farmland crop irrigation water, characterized in that: include: Step S1, obtaining soil physical indicators and nutrient indicators; Step S2: Based on soil physical indicators and nutrient indicators, the fuzzy clustering method is used to divide the farmland into management areas; the partial least squares discriminant analysis method is used to rank the contribution rate of each soil indicator to the crop irrigation water productivity in each management area to obtain the driving factor of the crop irrigation water productivity; the boundary line analysis method is used to determine the control threshold of the driving factor of the crop irrigation water productivity in each management area; Step S3, calculate the actual periodic water consumption of crops in each management area, establish an irrigation system optimization model, apply genetic algorithm to perform multi-objective optimization solution on the irrigation system optimization model, and obtain the optimized irrigation system; combine the control threshold of the crop irrigation water productivity driving factor with the optimized irrigation system to obtain a distributed crop management plan for farmland.

2. The method for improving the productivity of farmland crop irrigation water in a distributed manner as claimed in claim 1, characterized in that: In step S2, the partial least squares discriminant analysis method is used in each management zone to quantify the contribution rate of each soil index to crop irrigation water productivity based on the variable projection importance VIP and rank them, and the indexes with VIP values ​​greater than 1 are screened out as the driving factors of crop irrigation water productivity.

3. The method for improving the productivity of farmland crop irrigation water in a distributed manner as claimed in claim 2, characterized in that: Soil physical indicators include: soil mechanical composition and soil bulk density; soil nutrient indicators include: soil organic matter, total nitrogen, total phosphorus and total potassium.

4. The method for improving the productivity of farmland crop irrigation water in a distributed manner as claimed in claim 3, characterized in that: In step S3, a crop water production function is established, and the total irrigation water volume and the staged irrigation water volume are used as constraints. Two equally weighted objective functions of maximum crop irrigation water productivity and maximum yield are set to obtain an irrigation schedule optimization model; a genetic algorithm is used to perform multi-objective optimization on the irrigation schedule optimization model to obtain an optimized irrigation schedule.

5. A distributed device for increasing the productivity of farmland crop irrigation water, characterized in that: include: The first processing module is used to obtain soil physical indicators and nutrient indicators; The second processing module is used to divide the farmland into management zones based on soil physical indicators and nutrient indicators using the fuzzy clustering method; the partial least squares discriminant analysis method is used to rank the contribution rate of each soil indicator to the crop irrigation water productivity in each management zone to obtain the driving factors of the crop irrigation water productivity; The boundary line analysis method was used to determine the regulatory thresholds of the driving factors of crop irrigation water productivity in each management area; The third processing module is used to calculate the actual staged water consumption of crops in each management area, establish an irrigation system optimization model, and use genetic algorithms to perform multi-objective optimization solutions on the irrigation system optimization model to obtain an optimized irrigation system; the control threshold of the crop irrigation water productivity driving factor is combined with the optimized irrigation system to obtain a distributed crop management plan for farmland.

6. The distributed device for increasing the productivity of farmland crop irrigation water according to claim 5, characterized in that: The second processing module is used to use the partial least squares discriminant analysis method in each management area to quantify and rank the contribution rate of each soil indicator to crop irrigation water productivity based on the variable projection importance VIP, and screen out indicators with VIP values ​​greater than 1 as driving factors of crop irrigation water productivity.

7. The distributed device for improving the productivity of farmland crop irrigation water according to claim 6, characterized in that: Soil physical indicators include: soil mechanical composition and soil bulk density; soil nutrient indicators include: soil organic matter, total nitrogen, total phosphorus and total potassium.

8. The distributed device for improving the productivity of farmland crop irrigation water according to claim 7, characterized in that: The third processing module is used to establish the crop water production function. With the total irrigation water volume and the staged irrigation water volume as constraints, two equally weighted objective functions of maximum crop irrigation water productivity and maximum yield are set to obtain an irrigation system optimization model; a genetic algorithm is used to perform multi-objective optimization on the irrigation system optimization model to obtain an optimized irrigation system.

Images