High-water-saving accurate drip irrigation optimization method based on population algorithm

Abstract

The invention discloses a high-water-saving accurate drip irrigation optimization method based on a population algorithm. The method comprises the following steps: S1, constructing a comprehensive input data set; s2, forming a multi-dimensional collaborative vector coding set; s3, constructing a multi-objective optimization model; s4, generating an initial improved shark search algorithm population; s5, in a global exploration stage, forming an evaluation result data set; s6, in a local development stage, recording a current optimal multi-target solution; s7, decoding the multi-dimensional collaborative vector code corresponding to the current optimal multi-target solution into a drip irrigation pipe network implementation data set and an irrigation scheduling implementation data set; and S8, generating an irrigation control instruction set based on the drip irrigation pipe network implementation data set and the irrigation scheduling implementation data set, and executing partition irrigation operation. According to the method, the pipe arrangement rationality and the water resource utilization efficiency can be optimized at the same time, and higher convergence and generalization ability are shown in complex terrains and heterogeneous crop distribution environments.

Description

Technical Field

[0001] The present invention relates to the field of drip irrigation technology, and in particular to a high-water-saving and precise drip irrigation optimization method based on a swarm algorithm. Background Art

[0002] With the development of agricultural water-saving technologies, drip irrigation systems, as an efficient and water-saving irrigation method, have been widely used in large-scale farmland irrigation, especially in arid or semi-arid areas. The advantages are particularly obvious. Drip irrigation systems achieve precise water supply to the crop root zone by controlling water flow and irrigation time, which not only improves the utilization efficiency of water resources, but also helps to control agricultural non-point source pollution. However, in actual applications, the design and scheduling of drip irrigation systems still face many challenges.

[0003] In existing technologies, the piping design of drip irrigation systems usually relies on engineers' experience or semi-automated software tools, which makes it difficult to fully adapt to the complexity of different terrains, crop structures and hydraulic conditions. Traditional drip irrigation piping solutions often adopt a regular layout, which lacks the ability to adapt to terrain undulations, crop distribution and soil heterogeneity, and easily causes uneven spatial distribution of water resources, thereby affecting the overall irrigation uniformity.

[0004] In recent years, some studies have begun to introduce swarm intelligence algorithms to optimize drip irrigation design and scheduling parameters, but there are the following shortcomings: on the one hand, the variable dimensions in the optimization process are diverse, and the coupling between the pipe layout structure parameters and the irrigation scheduling parameters is strong. Existing methods often lack overall modeling of the logical relationship between variables; on the other hand, the optimization goal setting is incomplete, and the evolution process is driven by only a single objective function, which fails to effectively integrate the multi-objective factors of irrigation uniformity, energy efficiency and crop water response, and the practicality of the optimization results is limited.

[0005] Therefore, it is urgent to propose a drip irrigation optimization method based on variable co-coding and subgroup evolution mechanism to systematically solve the current problems. Summary of the Invention

[0006] One purpose of the present invention is to propose a high-water-saving precision drip irrigation optimization method based on a swarm algorithm. The present invention can simultaneously optimize the rationality of pipe layout and the efficiency of water resource utilization, and show stronger convergence and generalization capabilities in complex terrain and heterogeneous crop distribution environments.

[0007] A high-water-saving precision drip irrigation optimization method based on a swarm algorithm according to an embodiment of the present invention includes the following steps:

[0008] S1. Collect the digital terrain model dataset and crop water requirement parameter dataset of the drip irrigation area to construct a comprehensive input dataset;

[0009] S2. Generate a set of drip irrigation network structure variables and a set of zoned irrigation scheduling variables based on the comprehensive input data set, encode the drip irrigation network structure variables into a first sub-vector, and encode the zoned irrigation scheduling variables into a second sub-vector, to form a multi-dimensional collaborative vector encoding set;

[0010] S3. Build a multi-objective optimization model and set constraints to form an optimization problem model with multiple objectives and multiple constraint boundaries;

[0011] S4. Initializing the population of the improved shark search algorithm for the multi-dimensional collaborative vector encoding set to generate an initial population of the improved shark search algorithm;

[0012] S5. In the global exploration phase, the multi-objective optimization model is used to calculate the objective function value for each multi-dimensional collaborative vector code to form an evaluation result dataset;

[0013] S6. During the local development phase, based on the multi-objective optimization model evaluation values ​​in the evaluation result dataset, a Pareto-sort-based fitness evaluation method is used to select dominant individuals. Lévy jump perturbations and soft constraint penalty mechanisms are then implemented on the selected individuals. The shark search algorithm population is then updated and improved, and the current optimal multi-objective solution is recorded.

[0014] S7. Encoding and decoding the multidimensional collaborative vector corresponding to the current optimal multi-objective solution into a drip irrigation network implementation dataset and an irrigation scheduling implementation dataset;

[0015] S8. Generate an irrigation control instruction set based on the drip irrigation network implementation dataset and the irrigation scheduling implementation dataset, and execute zoned irrigation operations.

[0016] Optionally, the S1 includes the following steps:

[0017] S11. Construct a digital terrain model dataset for the drip irrigation area. The digital terrain model dataset consists of multiple three-dimensional terrain sampling points. Each three-dimensional terrain sampling point consists of plane coordinates and corresponding elevation values. The digital terrain model dataset is used to represent the micro-topography undulation information of the drip irrigation area.

[0018] S12. Constructing a crop water requirement parameter dataset, the crop water requirement parameter dataset includes the crop evapotranspiration rate, root layer depth, and maximum allowable water deficit rate for each crop unit;

[0019] S13. Constructing a soil physical and chemical property information set, which includes the field water holding capacity, wilting point, initial soil moisture content, and effective permeability coefficient of each soil unit;

[0020] S14. Calculate the daily root zone water requirement of each crop unit based on the crop water requirement parameter dataset and the soil physical and chemical property information dataset;

[0021] S15. Integrate the digital terrain model dataset, the crop water requirement parameter dataset, the soil physical and chemical property information dataset, and the daily root zone water demand dataset to form a comprehensive input dataset.

[0022] Optionally, the S2 includes the following steps:

[0023] S21. Based on the digital terrain model dataset and irrigation area boundary information in the comprehensive input dataset, spatial grid division is performed on the drip irrigation area to establish a set of pipe layout analysis units;

[0024] S22. Construct a set of drip irrigation network structure variables. The set of drip irrigation network structure variables includes layout direction variables, layout spacing variables, branch pipe diameter variables, and branch pipe connection path variables corresponding to each pipe layout analysis unit. All variables correspond to each pipe layout analysis unit and are used to express the drip irrigation network structure.

[0025] S23. Construct an irrigation scheduling unit set based on the crop water requirement parameter dataset and the daily root zone water requirement set in the comprehensive input dataset, where each irrigation scheduling unit in the irrigation scheduling unit set spatially overlaps with one or more pipe layout analysis units;

[0026] S24. Construct a set of zoned irrigation scheduling variables. The set of zoned irrigation scheduling variables includes daily irrigation volume variables, zoned irrigation period variables, and pump station start / stop control variables corresponding to each irrigation scheduling unit. All parameters correspond to each irrigation scheduling unit and are used to describe the irrigation scheduling strategy.

[0027] S25. Arrange the drip irrigation network structure variable set in the order of layout direction variable, layout spacing variable, branch pipe diameter variable, and branch pipe connection path variable, and encode it into a first subvector. The first subvector contains the layout structure parameters corresponding to all the layout analysis units.

[0028] S26. Arrange the set of zoned irrigation scheduling variables in the order of daily irrigation amount variables, zoned irrigation period variables, and pump station start / stop control variables, and encode them into a second sub-vector. The second sub-vector contains the irrigation scheduling parameters corresponding to all irrigation scheduling units.

[0029] S27. Concatenate the first sub-vector and the second sub-vector to form a multi-dimensional collaborative vector encoding set.

[0030] Optionally, S3 includes the following steps:

[0031] S31. Taking maximizing irrigation uniformity as the main goal, construct the irrigation uniformity objective function F CU ,The irrigation uniformity objective function value is used to measure the outflow balance of the entire drip irrigation system;

[0032] S32. Taking the minimization of irrigation energy consumption per unit area as the auxiliary goal, construct the irrigation energy consumption objective function F per unit area. E ,The objective function value of irrigation energy consumption per unit area is used to evaluate the energy efficiency level of the entire irrigation process;

[0033] S33. Taking the root zone water deficit index as the auxiliary goal, construct the root zone water deficit objective function F SWDI , the root zone water deficit objective function value is used to measure the deviation between the water content of the crop root zone after irrigation and the target water state;

[0034] S34. Constructing a comprehensive objective function vector F by combining the irrigation uniformity objective function, the irrigation energy consumption per unit area objective function, and the root zone water deficit objective function;

[0035] S35. Construct a constraint function set G, where the constraint function set includes a pipe laying economic cost constraint, a maximum branch pipe connection length constraint, and an irrigation area boundary constraint;

[0036] S36. The objective function vector F and the constraint function set G constitute a complete multi-objective optimization model M. opt , the optimization variable is the multidimensional collaborative vector encoding set X total .

[0037] Optionally, the pipe laying economic cost constraint indicates that the total cost of the current pipe laying scheme cannot exceed the preset budget upper limit; the maximum branch pipe connection length constraint indicates that the total length of the branch pipe path in each pipe laying analysis unit cannot exceed the maximum connection length allowed by the area; the irrigation area boundary constraint indicates that the node coordinates of all pipe laying paths must be within the preset irrigation area boundary range, and there must be no crossing of the boundary or invalid laying area.

[0038] Optionally, the S4 includes the following steps:

[0039] S41. Construct an initial population coding matrix P composed of multiple multi-dimensional cooperative vectors (0) ,Each multi-dimensional collaborative vector represents a set of drip irrigation network structural parameters and zoning irrigation scheduling parameters to be optimized;

[0040] S42. Assign an initial velocity vector to each multidimensional cooperative vector in the initial population and define the dynamic inertia factor sequence ω for the improved shark search algorithm. (t) , each inertia factor of the dynamic inertia factor sequence is calculated by linearly decreasing from the maximum inertia factor to the minimum inertia factor;

[0041] S43. Introduce an adaptive step size adjustment mechanism, set the basic step size λ0, and update the step size factor corresponding to each variable dimension through the adaptive step size adjustment mechanism in each round of iteration

[0042] S44. The initial population encoding matrix P (0) According to the fields to which the pipe structure variables and irrigation scheduling variables belong in the multi-dimensional collaborative vector coding set, a logical division is performed to construct a subgroup set G = {G pipe ,G irri}, each subgroup in the subgroup set corresponds to a different variable subspace in the multidimensional cooperative vector encoding set:

[0043] Pipe structure variable subgroup G pipe represents the pipe layout behavior characteristics of the population in the drip irrigation network structure variable subspace; the irrigation scheduling variable subgroup G irri Represents the control strategy characteristics of the population in the subspace of the zoned irrigation scheduling variable;

[0044] In the pipe structure variable subgroup G pipe and irrigation scheduling variable subgroup G irri According to the reasonable objective function f pipe =F CU And the irrigation uniformity objective function f irri =w1·F SWDI +w2·F E The local guiding individuals with the corresponding fitness values ​​are selected. The local guiding individuals are used to guide the evolutionary direction of each subgroup in its own subspace, and the global optimal individual is selected from all the local guiding individuals as the global guiding individual.

[0045] S45. In all subgroups, local guide individuals are selected according to the complete multi-objective optimization model M. opt Perform multi-objective fitness evaluation on each local guide individual, and select the one with the best comprehensive performance of the objective function vector as the global guide individual

[0046] Optionally, the S4 includes the following steps:

[0047] S51. Encode each multidimensional cooperative vector in the current iteration population of the improved shark search algorithm Perform decoding operations;

[0048] S52. Calculate the irrigation uniformity objective function value Calculate the objective function value of irrigation energy consumption per unit area Calculate the root zone water deficit objective function value

[0049] S53. Set the irrigation uniformity objective function value Objective function value of irrigation energy consumption per unit area and the root zone water deficit objective function value Combined into the objective function vector of the current individual Each objective function vector is used to characterize the comprehensive performance of an individual in the multi-objective optimization model. The objective function vectors of all individuals in the current population are aggregated to form the evaluation result dataset of the current iteration round.

[0050] Optionally, the S6 includes the following steps:

[0051] S61. Based on the evaluation result dataset The objective function vector of each individual in Execute the non-dominated sorting method based on Pareto sorting on the current iterative population of the improved shark search algorithm to construct the Pareto non-dominated frontier set And all individuals that are not comprehensively inferior to other individuals are defined as dominant individuals;

[0052] S62. Based on the subgroup set G, perform local guided individual selection within the subgroups in the pipe layout variable subgroup and the irrigation scheduling variable subgroup;

[0053] S63. For each individual in the population Perform disturbance update operations on the variable dimensions of the pipe layout structure variable subgroup and the irrigation scheduling variable subgroup, respectively. Combined with the dynamic inertia factor and the adaptive step size factor, the Lévy jump mechanism is used to calculate the updated subvectors:

[0054] For the pipe structure variable dimension d∈G pipe :

[0055]

[0056] For the irrigation scheduling variable dimension d∈G irri :

[0057]

[0058] Where β is the jump scale factor, L(μ) is a random variable that satisfies the Lévy distribution, and μ is the distribution index. Indicates that in the pipe layout structure variable subgroup, the reasonable pipe layout objective function f pipe is the value of the local guided individual in the dth dimension obtained by screening the fitness evaluation index, Indicates that in the irrigation scheduling variable subgroup, the irrigation uniformity objective function f irri The value of the local guide individual in the dth dimension obtained by screening the fitness evaluation index;

[0059] S64. Restore the updated subgroup variables to a complete multidimensional synergy vector And for each Execute constraint condition detection, including pipe laying economic cost constraint, maximum branch connection length constraint and irrigation area boundary constraint. If there is a constraint violation, a soft constraint penalty mechanism is introduced to calculate the penalty item.

[0060] S65. Update the objective function vector of each individual and penalty items Add together to form a comprehensive penalty fitness vector

[0061] S66. Among all individuals that meet the constraints, select the individual with the best overall performance in the penalty fitness vector as the optimal multi-objective solution for the current iteration round.

[0062] Optionally, the S7 includes the following steps:

[0063] S71. Read the multi-dimensional collaborative vector code corresponding to the optimal multi-objective solution of the current iteration round And perform decoding operations on it;

[0064] S72. Generate a drip irrigation network implementation data set based on the layout direction variable, layout spacing variable, branch pipe diameter variable, and branch pipe connection path variable;

[0065] S73. Generate an irrigation scheduling implementation data set based on daily irrigation volume variables, zone irrigation period variables, and pump station start / stop control variables;

[0066] S74. Combine and output the drip irrigation network implementation dataset and the irrigation scheduling implementation dataset to generate a final data pair for guiding on-site precision irrigation implementation, which is used to directly deploy pipe layout plans and irrigation plans in actual irrigation district applications.

[0067] The beneficial effects of the present invention are:

[0068] (1) The present invention constructs a multi-dimensional collaborative vector encoding method to uniformly encode the drip irrigation network structure variables and the zoned irrigation scheduling variables, avoiding the global decoupling failure problem caused by the split modeling of structural design and scheduling strategy, and can simultaneously optimize the rationality of pipe layout and water resource utilization efficiency, showing stronger convergence and generalization capabilities in complex terrain and heterogeneous crop distribution environments.

[0069] (2) The present invention divides the initial group into pipe structure variable subgroups and irrigation scheduling variable subgroups, and guides each subgroup separately through an independent fitness function, guiding each subgroup to evolve independently in its corresponding subspace, effectively overcoming the problem of inconsistent evolutionary rhythms of different subproblems in the traditional full variable evolution process, and selecting the global optimal body as the global guide body among all local guide individuals of the subgroup, realizing knowledge transfer and cross-guidance between individuals, and significantly improving the optimization quality and search efficiency.

[0070] (3) The present invention introduces the Lévy jump distribution perturbation mechanism to enhance the long-range jump capability of individuals in the solution space, thereby avoiding falling into the local optimal solution; at the same time, combined with the soft constraint penalty strategy, the penalty items are dynamically assigned to individuals that violate the pipe laying cost upper limit, the maximum branch pipe length and the irrigation area boundary constraints, so that the boundary constraints can reflect moderate guidance while ensuring the integrity of the global search space, thereby improving the actual deployability of the solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0072] Figure 1 This is a flow chart of a high-water-saving precision drip irrigation optimization method based on a swarm algorithm proposed in the present invention. DETAILED DESCRIPTION

[0073] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0074] refer to Figure 1 A high-water-saving precision drip irrigation optimization method based on a swarm algorithm includes the following steps:

[0075] S1. Collect a digital terrain model dataset and a crop water requirement parameter dataset for the drip irrigation area, integrate micro-topography elevation information, soil physical and chemical property information, and crop evapotranspiration information by stage, and construct a comprehensive input dataset.

[0076] S2. Generate a set of drip irrigation network structure variables and a set of zoned irrigation scheduling variables based on the comprehensive input dataset. Use a multidimensional collaborative vector encoding method to encode the drip irrigation network structure variables into a first subvector and the zoned irrigation scheduling variables into a second subvector, forming a multidimensional collaborative vector encoding set.

[0077] S3. Construct a multi-objective optimization model. This model takes maximizing irrigation uniformity of the drip irrigation system as the primary objective, minimizing irrigation energy consumption per unit area, and satisfying root zone water deficit indicators as auxiliary objectives. Furthermore, constraints on pipe laying economic costs, maximum branch pipe connection length, and irrigation area boundaries are set, forming an optimization problem model with multiple objectives and multiple constraints.

[0078] S4. Initializing the population of the improved shark search algorithm for the multi-dimensional collaborative vector encoding set to generate an initial population of the improved shark search algorithm;

[0079] S5. In the global exploration phase, the multi-objective optimization model is used to calculate the objective function value for each multi-dimensional collaborative vector code. Evaluation results, including irrigation uniformity index, energy consumption estimate, and water deficit response index, are obtained to form an evaluation result dataset.

[0080] S6. During the local development phase, based on the multi-objective optimization model evaluation values ​​in the evaluation result dataset, a Pareto-sort-based fitness evaluation method is used to select dominant individuals. Lévy jump perturbations and soft constraint penalty mechanisms are then implemented on the selected individuals. The shark search algorithm population is then updated and improved, and the current optimal multi-objective solution is recorded.

[0081] S7. Encode and decode the multidimensional collaborative vector corresponding to the current optimal multi-objective solution into a drip irrigation network implementation dataset and an irrigation scheduling implementation dataset. The drip irrigation network implementation dataset includes branch pipe paths, pipe diameter configuration, and layout spacing. The irrigation scheduling implementation dataset includes the irrigation sequence of each zone, irrigation volume setting value, and pump station start and stop duration.

[0082] S8. Generate an irrigation control instruction set based on the drip irrigation network implementation data set and the irrigation scheduling implementation data set, and send the irrigation control instruction set to the valve controller and the variable frequency pump controller through the irrigation control network to perform the zoned irrigation operation.

[0083] In this embodiment, S1 includes the following steps:

[0084] S11. Construct a digital terrain model dataset for the drip irrigation area. The digital terrain model dataset consists of multiple three-dimensional terrain sampling points. Each three-dimensional terrain sampling point consists of plane coordinates and corresponding elevation values. The digital terrain model dataset is used to represent the micro-topography undulation information of the drip irrigation area.

[0085] The digital terrain model dataset is obtained by T The regular grid sampling is obtained and further a two-dimensional elevation interpolation matrix is ​​generated. The two-dimensional elevation interpolation matrix is ​​used to describe the continuous distribution of the entire irrigation area terrain.

[0086] S12. Construct a crop water requirement parameter dataset. The crop water requirement parameter dataset includes the crop evaporation rate, root zone depth, and maximum allowable water deficit rate for each crop unit. The crop water requirement parameter dataset is used to describe the water requirement characteristics of different crop units at specific growth stages.

[0087] S13. Construct a soil physical and chemical property information set. The soil physical and chemical property information set includes the field water holding capacity, wilting point, initial soil moisture content, and effective permeability coefficient for each soil unit. The soil physical and chemical property information set is used to assess the soil's water storage capacity and water conductivity.

[0088] S14. Calculate the daily root zone water requirement for each crop unit based on the crop water requirement parameter dataset and the soil physical and chemical property information dataset. The daily root zone water requirement is calculated by multiplying the crop evaporation rate of the crop unit by the root zone depth, and then multiplying it by a crop stage coefficient. The crop stage coefficient is used to adjust for differences in water consumption at different growth stages. The daily root zone water requirement is used to determine the daily water requirement baseline for each crop unit.

[0089] S15. Integrate the digital terrain model dataset, the crop water requirement parameter dataset, the soil physical and chemical property information dataset, and the daily root zone water demand dataset to form a comprehensive input dataset. The comprehensive input dataset includes all micro-topography information within the irrigation area, crop unit water requirement characteristics, soil physical properties, and irrigation water demand.

[0090] In this embodiment, S2 includes the following steps:

[0091] S21. Based on the digital terrain model dataset and irrigation area boundary information in the comprehensive input dataset, spatially grid the drip irrigation area and establish a set of pipe layout analysis units. Each pipe layout analysis unit in the set corresponds one-to-one to a sampling point in the two-dimensional elevation interpolation matrix in the digital terrain model dataset.

[0092] S22. Construct a set of drip irrigation network structure variables. The set includes layout direction variables, layout spacing variables, branch pipe diameter variables, and branch pipe connection path variables corresponding to each pipe layout analysis unit. The layout direction variables are used to describe the layout azimuth of the branches, the layout spacing variables are used to describe the distance between branches, the branch pipe diameter variables are used to describe the diameter of the branches, and the branch pipe connection path variables are used to describe the length of the branches within the pipe layout analysis unit. All parameters correspond to each pipe layout analysis unit and are used to express the drip irrigation network structure.

[0093] S23. Construct an irrigation scheduling unit set based on the crop water requirement parameter dataset and the daily root zone water requirement set in the comprehensive input dataset, where each irrigation scheduling unit in the irrigation scheduling unit set spatially overlaps with one or more pipe layout analysis units;

[0094] S24. Construct a set of zoned irrigation scheduling variables. The zoned irrigation scheduling variable set includes daily irrigation volume variables, zoned irrigation period variables, and pump station start / stop control variables corresponding to each irrigation scheduling unit. The daily irrigation volume variable is used to describe the target irrigation volume for each irrigation scheduling unit. The zoned irrigation period variable is used to describe the irrigation start and end periods for each irrigation scheduling unit. The pump station start / stop control variable is used to describe the start / stop status of the pump station. All parameters correspond to each irrigation scheduling unit and are used to describe the irrigation scheduling strategy.

[0095] S25. Arrange the drip irrigation network structure variable set in the order of layout direction variable, layout spacing variable, branch pipe diameter variable, and branch pipe connection path variable, and encode it into a first subvector. The first subvector contains the layout structure parameters corresponding to all the layout analysis units.

[0096] S26. Arrange the set of zoned irrigation scheduling variables in the order of daily irrigation amount variables, zoned irrigation period variables, and pump station start / stop control variables, and encode them into a second sub-vector. The second sub-vector contains the irrigation scheduling parameters corresponding to all irrigation scheduling units.

[0097] S27. Concatenate the first sub-vector and the second sub-vector to form a multi-dimensional collaborative vector encoding set.

[0098] In this embodiment, S3 includes the following steps:

[0099] S31. Taking maximizing irrigation uniformity as the main goal, construct the irrigation uniformity objective function F CU , the irrigation uniformity objective function value is used to measure the outflow balance of the entire drip irrigation system:

[0100]

[0101] Among them, q i represents the actual outflow rate of the i-th dripper unit, Indicates the average outflow of all drippers, N e Indicates the total number of drippers; the closer the irrigation uniformity index CU is to 1, the more uniform the irrigation is. To optimize the target in a unified direction, the opposite number - CU is used to minimize the objective function F CU ;

[0102] S31 is the actual flow rate q of the dripper i Deviation from average outflow The average absolute relative difference of the outflow balance is measured and the number of drippers N e Normalization is performed to obtain the irrigation uniformity index CU. To adapt to the multi-objective minimization framework, sign inversion is introduced to transform CU into an irrigation uniformity objective function F that can directly participate in the minimization search.CU ,The objective function of irrigation uniformity keeps CU consistent with the ,traditional industry standard, so that the indicators are consistent with the ,target directions of energy consumption, water deficit, etc. in the heuristic optimization.

[0103] Traditional uniformity measurement mostly uses the coefficient of variation or maximum-minimum ratio for single-point evaluation, which cannot be directly compared with energy consumption and moisture indicators in the same metric space. The present invention naturally integrates uniformity into multi-objective optimization through sign inversion and normalization, reflects indicator coupling, reduces additional engineering steps of target direction conversion or target conflict reconciliation, and improves algorithm integration.

[0104] S32. Taking the minimization of irrigation energy consumption per unit area as the auxiliary goal, construct the irrigation energy consumption objective function F per unit area. E The energy consumption per unit area irrigation objective function value is used to evaluate the energy efficiency level of the entire irrigation process. The energy consumption per unit area irrigation objective function value is obtained by dividing the total amount of electricity consumed during the entire irrigation process by the total area of ​​the irrigation area. It is used to reflect the energy consumption burden required for irrigation per unit land. The smaller the value, the more energy-efficient the system.

[0105] S33. Taking the root zone water deficit index as the auxiliary goal, construct the root zone water deficit objective function F SWDI , the root zone water deficit objective function value is used to measure the deviation between the water content of the crop root zone after irrigation and the target water content state:

[0106]

[0107] in, represents the field water capacity of the j-th crop unit, represents the average water content of the root zone of the crop unit after irrigation, Indicates the moisture content at the wilting point. Indicates the maximum allowable water deficit rate of the crop unit, N C represents the total number of crop units;

[0108] S33 is based on field capacity wilting point Average water content in the root zone after irrigation Constructs a dimensionless deficit ratio, which is used to evaluate the water status of a single crop unit and is related to the allowable deficit rate of each crop unit. After making the difference, take the absolute value and sum it over all crop units to form the root zone water deficit objective function F SWDI The root zone water deficit objective function takes into account the differences in crop types and soils, avoiding local over-irrigation or under-irrigation caused by the traditional single threshold for the entire region.

[0109] Conventional methods usually compare regional average moisture content with a single threshold, ignoring differences in crop zones. The present invention allows the deficit rate to be refined and aggregated to the crop unit dimension, significantly improving the refinement and controllability of moisture indicators.

[0110] S34. Construct a comprehensive objective function vector F by combining the irrigation uniformity objective function, the irrigation energy consumption per unit area objective function, and the root zone water deficit objective function. The comprehensive objective function vector F is used to evaluate the comprehensive adaptability of each set of pipe layout and irrigation strategy parameters.

[0111] S35. Construct a constraint function set G, where the constraint function set includes a pipe laying economic cost constraint, a maximum branch pipe connection length constraint, and an irrigation area boundary constraint;

[0112] S36. The objective function vector F and the constraint function set G constitute a complete multi-objective optimization model M. opt , the optimization variable is the multidimensional collaborative vector encoding set X total .

[0113] In this implementation, the pipe laying economic cost constraint means that the total cost of the current pipe laying plan cannot exceed the preset budget upper limit; the maximum branch pipe connection length constraint means that the total length of the branch pipe path in each pipe laying analysis unit cannot exceed the maximum connection length allowed by the area; and the irrigation area boundary constraint means that the node coordinates of all pipe laying paths must be within the preset irrigation area boundary range, and there must be no crossing of the boundary or invalid laying area.

[0114] In this embodiment, S4 includes the following steps:

[0115] S41. Construct an initial population coding matrix P composed of multiple multi-dimensional cooperative vectors (0) Each multidimensional collaborative vector represents a set of drip irrigation network structural parameters and zone irrigation scheduling parameters to be optimized. The dimension of the multidimensional collaborative vector and the multidimensional collaborative vector encoding set X total Stay consistent;

[0116] S42. Assign an initial velocity vector to each multidimensional cooperative vector in the initial population and define the dynamic inertia factor sequence ω for the improved shark search algorithm. (t) The dynamic inertia factor sequence is used to describe the motion inertia of the search individual in different iteration rounds. Each inertia factor in the dynamic inertia factor sequence is calculated by linearly decreasing from the maximum inertia factor to the minimum inertia factor. The inertia factor value decreases with the increase of the number of iterations. It is used to balance the exploration in the early stage of the search and the convergence in the later stage.

[0117] S43. Introduce an adaptive step size adjustment mechanism, set the basic step size λ0, and update the step size factor corresponding to each variable dimension through the adaptive step size adjustment mechanism in each round of iteration

[0118]

[0119] in, represents the search step length of the d-th dimension in the t-th iteration, represents the objective function value of the best individual in the entire population in round t, ∈ is a positive constant to prevent division by zero;

[0120] S43 takes two consecutive iterations to find the global optimal value and The relative improvement rate is the core adjustment signal, and the basic step size λ0 is dynamically scaled. The larger the improvement rate, the faster the algorithm is progressing. When the improvement rate approaches zero, the step size automatically converges to achieve a smooth transition from global to local search and fractal dimension assignment. Ensure consistency of step size under variable scale differences.

[0121] Traditional shark search or particle swarm search often uses linear decrement or fixed step size, which makes it difficult to respond to the actual convergence trend of the objective function. The adaptive mechanism of the present invention adjusts the search intensity in real time by feeding back the improvement rate of the optimal solution, significantly enhancing the convergence efficiency and stability.

[0122] S44. The initial population encoding matrix P (0) According to the fields to which the pipe structure variables and irrigation scheduling variables belong in the multi-dimensional collaborative vector coding set, a logical division is performed to construct a subgroup set G = {G pipe ,G irri}, each subgroup in the subgroup set corresponds to a different variable subspace in the multidimensional cooperative vector encoding set:

[0123] Pipe structure variable subgroup G pipe Contains the dimensional areas corresponding to the layout direction variables, layout spacing variables, branch pipe diameter variables, and branch pipe connection path variables of all population individuals, representing the pipe layout behavior characteristics of the population in the drip irrigation network structure variable subspace;

[0124] Irrigation scheduling variable subgroup G irri Contains the dimensional areas corresponding to the daily irrigation amount variables, the zoned irrigation period variables, and the pump station start-stop control variables of all population individuals, representing the control strategy characteristics of the population in the zoned irrigation scheduling variable subspace;

[0125] In the pipe structure variable subgroup G pipe and irrigation scheduling variable subgroup G irri According to the reasonable objective function f pipe=F CU And the irrigation uniformity objective function f irri =w1·F SWDI +w2·F E The local guiding individuals with the corresponding fitness values ​​are selected. The local guiding individuals are used to guide the evolutionary direction of each subgroup in its own subspace, and the global optimal individual is selected from all the local guiding individuals as the global guiding individual.

[0126] S45. In all subgroups, local guide individuals are selected according to the complete multi-objective optimization model M. opt Perform multi-objective fitness evaluation on each local guide individual, and select the one with the best comprehensive performance of the objective function vector as the global guide individual:

[0127]

[0128] in, represents the local guide individual of the k-th subgroup, Used to guide the overall search direction of the initial improved shark search algorithm in the multi-objective optimization space.

[0129] In this embodiment, S4 includes the following steps:

[0130] S51. Encode each multidimensional cooperative vector in the current iteration population of the improved shark search algorithm Perform decoding operations. The decoded multidimensional collaborative vector includes layout direction variables, layout spacing variables, branch pipe diameter variables, branch pipe connection path variables, daily irrigation volume variables, zoned irrigation period variables, and pump station start and stop control variables.

[0131] S52. Calculate the irrigation uniformity objective function value The irrigation uniformity objective function value is used to measure the outflow balance of the pipe layout scheme and irrigation strategy corresponding to the multi-dimensional collaborative vector on each dripper unit;

[0132] S53. Calculate the target function value of irrigation energy consumption per unit area The objective function value of irrigation energy consumption per unit area is used to reflect the energy efficiency level of the pipe layout and scheduling strategy represented by the multi-dimensional synergy vector in actual irrigation;

[0133] S54. Calculate the root zone water deficit objective function value The root zone water deficit objective function value is used to measure whether the current irrigation scheduling strategy meets the root zone water requirements of each crop unit;

[0134] S55. Set the irrigation uniformity objective function value Objective function value of irrigation energy consumption per unit area and the root zone water deficit objective function value Combined into the objective function vector of the current individual Each objective function vector is used to characterize the comprehensive performance of an individual in the multi-objective optimization model. The objective function vectors of all individuals in the current population are aggregated to form the evaluation result dataset of the current iteration round.

[0135] In this embodiment, S6 includes the following steps:

[0136] S61. Based on the evaluation result dataset The objective function vector of each individual in Execute the non-dominated sorting method based on Pareto sorting on the current iterative population of the improved shark search algorithm to construct the Pareto non-dominated frontier set And all individuals that are not comprehensively inferior to other individuals are defined as dominant individuals;

[0137] S62. Based on the subgroup set G, perform local guided individual selection within the subgroups in the pipe layout variable subgroup and the irrigation scheduling variable subgroup;

[0138] S63. For each individual in the population Perform disturbance update operations on the variable dimensions of the pipe layout structure variable subgroup and the irrigation scheduling variable subgroup, respectively. Combined with the dynamic inertia factor and the adaptive step size factor, the Lévy jump mechanism is used to calculate the updated subvectors:

[0139] For the pipe structure variable dimension d∈G pipe :

[0140]

[0141] For the irrigation scheduling variable dimension d∈G irri :

[0142]

[0143] Where β is the jump scale factor, L(μ) is a random variable that satisfies the Lévy distribution, and μ is the distribution index. Indicates that in the pipe layout structure variable subgroup, the reasonable pipe layout objective function f pipe is the value of the local guided individual in the dth dimension obtained by screening the fitness evaluation index, Indicates that in the irrigation scheduling variable subgroup, the irrigation uniformity objective function f irri The value of the local guide individual in the dth dimension obtained by screening the fitness evaluation index;

[0144] S63 with the current individual The difference between the local guide value of the subgroup is used as the direction vector, and the heavy-tailed Lévy jump is generated with the random variable L(μ) to achieve the long-range raid feature and integrate the dynamic inertia factor ω (t) Control the global exploration degree and integrate the step size factor To ensure convergence granularity, the jump scale β is uniformly scaled and the dimensions are consistent to avoid numerical imbalance.

[0145] Traditional algorithms use either uniform search or Gaussian perturbation. The present invention couples inertia, step size, and Lévy heavy-tailed distribution to implement a hybrid search mechanism, improving the ability to escape from local extremes.

[0146] S64. Restore the updated subgroup variables to a complete multidimensional synergy vector And for each Constraint checks are performed, including pipe laying economic cost constraints, maximum branch connection length constraints, and irrigation area boundary constraints. If any constraint violations occur, a soft constraint penalty mechanism is introduced to calculate the penalty term:

[0147]

[0148] in, represents the soft constraint penalty term of the i-th individual in the t+1th iteration, which is used to quantify the degree of constraint violation. γ1 represents the penalty weight coefficient of the pipelaying economic cost constraint, which is used to adjust the influence weight of the constraint in the overall penalty. represents the total pipe laying cost of the i-th individual in the t+1th round of iteration, reflecting the economic consumption of the selected pipe laying scheme, C max It represents the budget upper limit of the pipe laying economic cost, which is the decision constraint boundary, γ2 represents the penalty weight coefficient of the maximum branch connection length constraint, represents the actual total connection length of the branch pipes corresponding to the mth pipe layout analysis unit of the i-th individual in the t+1th iteration, represents the maximum branch connection length allowed for the mth pipe layout analysis unit, which serves as the upper limit of the spatial layout. γ3 represents the penalty weight coefficient of the irrigation area boundary constraint. Represents the indicator function. If the coordinates of the j-th pipe node of the i-th individual in the t+1th iteration are If it is outside the irrigation area boundary Ω, the value is 1, otherwise it is 0. represents the plane coordinates of the jth pipe layout node of the i-th individual in the t+1th iteration, Ω represents the spatial boundary of the irrigation area, which is defined as the two-dimensional space set of all allowed pipe layout node coordinates, and M represents the total number of pipe layout analysis units;

[0149] S64 designs corresponding penalty items for three types of engineering hard constraints: economic cost, branch pipe length, and irrigation area boundary. Any violation is calculated by multiplying the excess amount by the penalty weight and adding it to the total penalty value. Adjustable penalty weights γ1, γ2, and γ3 are used to achieve differentiated processing of constraint severity. The indicator function Minimize space out-of-bounds detection overhead.

[0150] Most conventional penalty functions unify the multipliers of all constraints or directly discard infeasible solutions, resulting in convergence oscillation or a sharp contraction of the solution space. The present invention quantifies the degree of violation through item-by-item weights and linear penalties to achieve gradual convergence, maintain diversity in the early stage, and strengthen the feasible domain in the later stage. The layered soft penalty strategy effectively balances feasibility and search activity, and combines dynamic inertia, step size, and Lévy jumps to form a complete closed loop.

[0151] S65. Update the objective function vector of each individual and penalty items Add together to form a comprehensive penalty fitness vector

[0152] S66. Among all individuals that meet the constraints, select the individual with the best overall performance in the penalty fitness vector as the optimal multi-objective solution for the current iteration round.

[0153] In this embodiment, S7 includes the following steps:

[0154] S71. Read the multi-dimensional collaborative vector code corresponding to the optimal multi-objective solution of the current iteration round The decoded multidimensional collaborative vector contains layout direction variables, layout spacing variables, branch pipe diameter variables, branch pipe connection path variables, daily irrigation volume variables, zoned irrigation period variables, and pump station start and stop control variables.

[0155] S72. Generate a drip irrigation network implementation data set based on the layout direction variable, layout spacing variable, branch pipe diameter variable, and branch pipe connection path variable;

[0156] Branch pipe path entries are determined by connecting the branch pipe node sequence point by point; branch pipe diameter entries are determined by extracting the pipe diameter parameters set on each path; and branch pipe spacing entries are determined by extracting the spacing parameters set on each path. The drip irrigation network implementation dataset is used to clarify the spatial layout, physical specifications, and installation parameters of each branch pipe, providing a complete engineering basis for on-site drip irrigation system layout.

[0157] S73. Generate an irrigation scheduling implementation data set based on daily irrigation volume variables, zone irrigation period variables, and pump station start / stop control variables;

[0158] The zone irrigation order entry is obtained by extracting the scheduling priority number of each irrigation zone; the irrigation volume set value entry is obtained by extracting the target daily irrigation volume parameter set for each irrigation zone; the pump station start and stop duration entry is obtained by extracting the pump station operation time parameter set for each irrigation zone. The irrigation scheduling implementation dataset is used to guide the formulation of irrigation execution sequence, daily water supply quota, and pump station start and stop scheduling plans for each irrigation zone.

[0159] S74. Combine and output the drip irrigation network implementation dataset and the irrigation scheduling implementation dataset to generate a final data pair for guiding on-site precision irrigation implementation, which is used to directly deploy pipe layout plans and irrigation plans in actual irrigation district applications.

[0160] Example 1: In the high-standard farmland demonstration area, the smart agriculture technology team conducted a drip irrigation system optimization test on "Plot 8". The total area of ​​the plot is 41.36 hectares, mainly planted with corn, which is currently in the rapid jointing stage. A set of soil sensor nodes, model SoilNode-S7, deployed in the northwest corner of the field uploaded monitoring data, indicating that the surface soil moisture content in the central area of ​​the plot is 17.3%, which is lower than the minimum water requirement threshold of corn at this stage (set at 21.5%). The system recorded the alarm data as follows:

[0161] Node number: S7-192-3; surface soil moisture content: 17.3%; target moisture content during the crop stage: ≥21.5%; root layer depth: 31 cm; abnormal flag: true.

[0162] The drip irrigation optimization management platform activated the algorithm module of the present invention and entered the data collection and analysis phase. First, the system retrieved the latest three-dimensional terrain point cloud from the on-site drone DEM modeling platform on June 1st. The cloud contained 43,280 sampling points with a maximum slope of 8.6°. It also retrieved crop evapotranspiration monitoring data from the previous day for the plot, with an average ET of 5.2 mm / day.

[0163] The project leader manually triggered the collaborative variable coding process through the platform. The system automatically divided the drip irrigation sub-unit into 96 pipe layout units and 72 irrigation scheduling units according to the spatial grid. Subsequently, the platform completed the splicing of the first sub-vector (pipe layout direction, spacing, branch pipe diameter, and connection path) and the second sub-vector (zone irrigation volume, irrigation period, and pump station start and stop), generating an initial population vector matrix with a coding length of 798 bits, totaling 120 groups.

[0164] The system enters the multi-objective optimization stage, and the objective function is set as follows:

[0165] Main goal: FCU maximization (irrigation uniformity); Auxiliary goal 1: FE minimization (energy consumption per unit area); Auxiliary goal 2: FSWDI minimization (root zone water deficit); Budget ceiling setting: 125,000 yuan; Maximum branch pipe length limit: 160 meters.

[0166] After 18 rounds of iterative evolution, the system achieved a breakthrough in the 19th round, with a new non-dominated frontier set update. In the 23rd round, the population converged to a set of Pareto optimal solutions, labeled P84. The corresponding control parameters are as follows:

[0167] P84 pipe layout strategy: Units G22 to G38 are laid in the southeast direction along the slope, with branch pipe spacing of 3.1 meters and pipe diameter of Φ20mm; the longest branch pipe length is 157.6 meters.

[0168] P84 irrigation scheduling: The daily irrigation volume of the central high-water-demand area Z17 is 6.3 mm, and the irrigation time is set to 02:30–04:00 in the morning; the daily irrigation volume of the southern low-lying area Z44 is reduced to 3.2 mm, and the time period is adjusted to 05:30–06:20 in the morning; the start and stop sequence of the pump station is fitted with the pump pressure curve, and the total number of switches is: 5.

[0169] The system completes the generation of irrigation control instructions and automatically pushes them to the PLC irrigation control terminal. The start time of the irrigation operation for the entire plot is set to the early morning of the next day.

[0170] The first irrigation process after the optimization was started on the plot. The main pressure sensor of the southern No. 3 pump station uploaded real-time power curve data. The peak power was reduced by 14.7% compared with the original. The unit irrigation energy consumption dropped from 5.3kWh / ha to 3.8kWh / ha. All 72 dispatching units completed the irrigation. The system recorded a total water volume of 1692.4m 3 .

[0171] The platform compared and analyzed the changes in soil moisture content at 72 nodes before and after irrigation, and found that the average moisture content in the root zone (0-30cm) increased from 20.1% to 24.2%, among which the increase in the Z17 area was the most significant; the root zone water deficit rate decreased from an average of 16.8% the previous day to 5.4%; and the irrigation uniformity FCU reached 0.89 (the previous traditional scheduling was 0.72).

[0172] Table 1 Comparison data between this optimized irrigation system and the traditional irrigation system of the same period last year are as follows

[0173]

[0174]

[0175] The present invention constructs a multi-dimensional collaborative vector encoding method to uniformly encode the drip irrigation network structure variables and the partition irrigation scheduling variables, avoiding the global decoupling failure problem caused by the split modeling of structural design and scheduling strategy, and can simultaneously optimize the rationality of pipe layout and water resource utilization efficiency, showing stronger convergence and generalization capabilities in complex terrain and heterogeneous crop distribution environments.

[0176] The present invention divides the initial group into pipe layout structure variable subgroups and irrigation scheduling variable subgroups, guides each subgroup separately through an independent fitness function, and guides each subgroup to evolve independently in its corresponding subspace, effectively overcoming the problem of inconsistent evolutionary rhythms of different subproblems in the traditional full-variable evolution process. The global optimal body is selected as the global guide body among all local guide individuals of the subgroup, realizing knowledge transfer and cross-guidance between individuals, and significantly improving optimization quality and search efficiency.

[0177] The present invention introduces a Lévy jump distribution perturbation mechanism to enhance the long-range jump capability of individuals in the solution space, thereby avoiding falling into local optimal solutions. At the same time, combined with a soft constraint penalty strategy, penalty terms are dynamically assigned to individuals that violate the pipe laying cost upper limit, maximum branch pipe length, and irrigation area boundary constraints. This allows boundary constraints to provide moderate guidance while ensuring the integrity of the global search space, thereby improving the practical deployability of the solution.

[0178] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A high water-saving precision drip irrigation optimization method based on swarm algorithm, characterized in that: The steps include: S1. Collect the digital terrain model dataset and crop water requirement parameter dataset of the drip irrigation area to construct a comprehensive input dataset; S2. Generate a set of drip irrigation network structure variables and a set of zoned irrigation scheduling variables based on the comprehensive input data set, encode the drip irrigation network structure variables into a first sub-vector, and encode the zoned irrigation scheduling variables into a second sub-vector, to form a multi-dimensional collaborative vector encoding set; S3. Build a multi-objective optimization model and set constraints to form an optimization problem model with multiple objectives and multiple constraint boundaries; S4. Initializing the population of the improved shark search algorithm for the multi-dimensional collaborative vector encoding set to generate an initial population of the improved shark search algorithm; S5. In the global exploration phase, the multi-objective optimization model is used to calculate the objective function value for each multi-dimensional collaborative vector code to form an evaluation result dataset; S6. During the local development phase, based on the multi-objective optimization model evaluation values ​​in the evaluation result dataset, a Pareto-sort-based fitness evaluation method is used to select dominant individuals. Lévy jump perturbations and soft constraint penalty mechanisms are then implemented on the selected individuals. The shark search algorithm population is then updated and improved, and the current optimal multi-objective solution is recorded. S7. Encoding and decoding the multidimensional collaborative vector corresponding to the current optimal multi-objective solution into a drip irrigation network implementation dataset and an irrigation scheduling implementation dataset; S8. Generate an irrigation control instruction set based on the drip irrigation network implementation dataset and the irrigation scheduling implementation dataset, and execute the zoned irrigation operation.

2. The high water-saving precision drip irrigation optimization method based on swarm algorithm according to claim 1 is characterized in that: Said S1 comprises the following steps: S11. Construct a digital terrain model dataset for the drip irrigation area. The digital terrain model dataset consists of multiple three-dimensional terrain sampling points. Each three-dimensional terrain sampling point consists of plane coordinates and corresponding elevation values. The digital terrain model dataset is used to represent the micro-topography undulation information of the drip irrigation area. S12. Constructing a crop water requirement parameter dataset, the crop water requirement parameter dataset includes the crop evapotranspiration rate, root layer depth, and maximum allowable water deficit rate for each crop unit; S13. Constructing a soil physical and chemical property information set, which includes the field water holding capacity, wilting point, initial soil moisture content, and effective permeability coefficient of each soil unit; S14. Calculate the daily root zone water requirement of each crop unit based on the crop water requirement parameter dataset and the soil physical and chemical property information dataset; S15. Integrate the digital terrain model dataset, the crop water requirement parameter dataset, the soil physical and chemical property information dataset, and the daily root zone water demand dataset to form a comprehensive input dataset.

3. The high water-saving precision drip irrigation optimization method based on swarm algorithm according to claim 2 is characterized in that: The S2 comprises the following steps: S21. Based on the digital terrain model dataset and irrigation area boundary information in the comprehensive input dataset, spatial grid division is performed on the drip irrigation area to establish a set of pipe layout analysis units; S22. Construct a set of drip irrigation network structure variables. The set of drip irrigation network structure variables includes layout direction variables, layout spacing variables, branch pipe diameter variables, and branch pipe connection path variables corresponding to each pipe layout analysis unit. All variables correspond to each pipe layout analysis unit and are used to express the drip irrigation network structure. S23. Construct an irrigation scheduling unit set based on the crop water requirement parameter dataset and the daily root zone water requirement set in the comprehensive input dataset, where each irrigation scheduling unit in the irrigation scheduling unit set spatially overlaps with one or more pipe layout analysis units; S24. Construct a set of zoned irrigation scheduling variables. The set of zoned irrigation scheduling variables includes daily irrigation volume variables, zoned irrigation period variables, and pump station start / stop control variables corresponding to each irrigation scheduling unit. All parameters correspond to each irrigation scheduling unit and are used to describe the irrigation scheduling strategy. S25. Arrange the drip irrigation network structure variable set in the order of layout direction variable, layout spacing variable, branch pipe diameter variable, and branch pipe connection path variable, and encode it into a first subvector. The first subvector contains the layout structure parameters corresponding to all the layout analysis units. S26. Arrange the set of zoned irrigation scheduling variables in the order of daily irrigation amount variables, zoned irrigation period variables, and pump station start / stop control variables, and encode them into a second sub-vector. The second sub-vector contains the irrigation scheduling parameters corresponding to all irrigation scheduling units. S27. Concatenate the first sub-vector and the second sub-vector to form a multi-dimensional collaborative vector encoding set.

4. The high water-saving precision drip irrigation optimization method based on swarm algorithm according to claim 3 is characterized in that: The S3 includes the following steps: S31. Taking maximizing irrigation uniformity as the main goal, construct the irrigation uniformity objective function F CU ,The irrigation uniformity objective function value is used to measure the outflow balance of the entire drip irrigation system; S32. Taking the minimization of irrigation energy consumption per unit area as the auxiliary goal, construct the irrigation energy consumption objective function F per unit area. E ,The objective function value of irrigation energy consumption per unit area is used to evaluate the energy efficiency level of the entire irrigation process; S33. Taking the root zone water deficit index as the auxiliary goal, construct the root zone water deficit objective function F SWDI , the root zone water deficit objective function value is used to measure the deviation between the water content of the crop root zone after irrigation and the target water state; S34. Constructing a comprehensive objective function vector F by combining the irrigation uniformity objective function, the irrigation energy consumption per unit area objective function, and the root zone water deficit objective function; S35. Construct a constraint function set G, where the constraint function set includes a pipe laying economic cost constraint, a maximum branch pipe connection length constraint, and an irrigation area boundary constraint; S36. The objective function vector F and the constraint function set G constitute a complete multi-objective optimization model M. opt , the optimization variable is the multidimensional collaborative vector encoding set X total .

5. The high water-saving precision drip irrigation optimization method based on swarm algorithm according to claim 4 is characterized in that: The pipe laying economic cost constraint means that the total cost of the current pipe laying plan cannot exceed the preset budget upper limit; the maximum branch pipe connection length constraint means that the total length of the branch pipe path in each pipe laying analysis unit cannot exceed the maximum connection length allowed by the area; the irrigation area boundary constraint means that the node coordinates of all pipe laying paths must be within the preset irrigation area boundary range, and there must be no crossing of the boundary or invalid laying area.

6. The high water-saving precision drip irrigation optimization method based on swarm algorithm according to claim 4 is characterized in that: The S4 comprises the following steps: S41. Construct an initial population coding matrix P composed of multiple multi-dimensional cooperative vectors (0) ,Each multi-dimensional collaborative vector represents a set of drip irrigation network structural parameters and zoning irrigation scheduling parameters to be optimized; S42. Assign an initial velocity vector to each multidimensional cooperative vector in the initial population and define the dynamic inertia factor sequence ω for the improved shark search algorithm. (t) , each inertia factor of the dynamic inertia factor sequence is calculated by linearly decreasing from the maximum inertia factor to the minimum inertia factor; S43. Introduce an adaptive step size adjustment mechanism, set the basic step size λ0, and update the step size factor corresponding to each variable dimension through the adaptive step size adjustment mechanism in each round of iteration S44. The initial population encoding matrix P (0) According to the fields to which the pipe structure variables and irrigation scheduling variables belong in the multi-dimensional collaborative vector coding set, a logical division is performed to construct a subgroup set G = {G pipe ,G irri }, each subgroup in the subgroup set corresponds to a different variable subspace in the multidimensional cooperative vector encoding set: Pipe structure variable subgroup G pipe represents the pipe layout behavior characteristics of the population in the drip irrigation network structure variable subspace; the irrigation scheduling variable subgroup G irri Represents the control strategy characteristics of the population in the subspace of the zoned irrigation scheduling variable; In the pipe structure variable subgroup G pipe and irrigation scheduling variable subgroup G irri According to the reasonable objective function f pipe =F CU And the irrigation uniformity objective function f irri =w1·F SWDI +w2·F E The local guiding individuals with the corresponding fitness values ​​are selected. The local guiding individuals are used to guide the evolutionary direction of each subgroup in its own subspace, and the global optimal individual is selected from all the local guiding individuals as the global guiding individual. S45. In all subgroups, local guide individuals are selected according to the complete multi-objective optimization model M. opt Perform multi-objective fitness evaluation on each local guide individual, and select the one with the best comprehensive performance of the objective function vector as the global guide individual 7. The high water-saving precision drip irrigation optimization method based on swarm algorithm according to claim 6 is characterized in that: The S4 comprises the following steps: S51. Encode each multidimensional cooperative vector in the current iteration population of the improved shark search algorithm Perform decoding operations; S52. Calculate the irrigation uniformity objective function value Calculate the objective function value of irrigation energy consumption per unit area Calculate the root zone water deficit objective function value S53. Set the irrigation uniformity objective function value Objective function value of irrigation energy consumption per unit area and the root zone water deficit objective function value Combined into the objective function vector of the current individual Each objective function vector is used to characterize the comprehensive performance of an individual in the multi-objective optimization model. The objective function vectors of all individuals in the current population are aggregated to form the evaluation result dataset of the current iteration round.

8. The high water-saving precision drip irrigation optimization method based on swarm algorithm according to claim 7 is characterized in that: The S6 comprises the following steps: S61. Based on the evaluation result dataset The objective function vector of each individual in Execute the non-dominated sorting method based on Pareto sorting on the current iterative population of the improved shark search algorithm to construct the Pareto non-dominated frontier set And all individuals that are not comprehensively inferior to other individuals are defined as dominant individuals; S62. Based on the subgroup set G, perform local guided individual selection within the subgroups in the pipe layout variable subgroup and the irrigation scheduling variable subgroup; S63. For each individual in the population Perform disturbance update operations on the variable dimensions of the pipe layout structure variable subgroup and the irrigation scheduling variable subgroup, respectively. Combined with the dynamic inertia factor and the adaptive step size factor, the Lévy jump mechanism is used to calculate the updated subvectors: For the pipe structure variable dimension d∈G pipe : For the irrigation scheduling variable dimension d∈G irri : Where β is the jump scale factor, L(μ) is a random variable that satisfies the Lévy distribution, and μ is the distribution index. Indicates that in the pipe layout structure variable subgroup, the reasonable pipe layout objective function f pipe is the value of the local guided individual in the dth dimension obtained by screening the fitness evaluation index, Indicates that in the irrigation scheduling variable subgroup, the irrigation uniformity objective function f irri The value of the local guide individual in the dth dimension obtained by screening the fitness evaluation index; S64. Restore the updated subgroup variables to a complete multidimensional synergy vector And for each Execute constraint condition detection, including pipe laying economic cost constraint, maximum branch connection length constraint and irrigation area boundary constraint. If there is a constraint violation, a soft constraint penalty mechanism is introduced to calculate the penalty item. S65. Update the objective function vector of each individual and penalty items Add together to form a comprehensive penalty fitness vector S66. Among all individuals that meet the constraints, select the individual with the best overall performance in the penalty fitness vector as the optimal multi-objective solution for the current iteration round.

9. The high water-saving precision drip irrigation optimization method based on swarm algorithm according to claim 8, characterized in that: The S7 comprises the following steps: S71. Read the multi-dimensional collaborative vector code corresponding to the optimal multi-objective solution of the current iteration round And perform decoding operations on it; S72. Generate a drip irrigation network implementation data set based on the layout direction variable, layout spacing variable, branch pipe diameter variable, and branch pipe connection path variable; S73. Generate an irrigation scheduling implementation data set based on daily irrigation volume variables, zone irrigation period variables, and pump station start / stop control variables; S74. Combine and output the drip irrigation network implementation dataset and the irrigation scheduling implementation dataset to generate a final data pair for guiding on-site precision irrigation implementation, which is used to directly deploy pipe layout plans and irrigation plans in actual irrigation district applications.

Images