A rolling correction intelligent irrigation decision-making method for regional wheat-maize rotation systems

Abstract

This invention discloses a rolling correction intelligent irrigation decision-making method for a regional wheat-maize rotation system, belonging to the field of agricultural irrigation technology. It constructs a regional high-water-efficiency intelligent irrigation decision-making model. In this model, irrigation data within the region no longer solely relies on annual average meteorological and soil data, but instead reads and processes detailed meteorological elements and soil moisture information in real time to dynamically optimize irrigation strategies based on daily changes. By optimizing the irrigation amount of the regional grid daily, it can more accurately reflect the spatial heterogeneity of the crop growth environment within a large-scale region, thereby improving the accuracy and reliability of the crop model in large-scale applications.

Description

Technical Field

[0001] This invention belongs to the field of agricultural irrigation technology, specifically relating to a rolling correction intelligent irrigation decision-making method for a regional wheat-maize rotation system. Background Technology

[0002] Current regional-scale optimization methods for wheat-maize irrigation systems are mostly based on hydrological year divisions, failing to fully consider the seasonal variations in water demand within the wheat-maize rotation system. Winter wheat and summer maize have significantly different water requirements during their growing seasons and exhibit a degree of water complementarity; therefore, rational allocation of water resources is crucial for efficient agricultural water use.

[0003] The water management of the wheat-maize rotation system lacks research on the soil moisture allocation mechanism between the two crops during the growing season, the dynamic regulation of the upper and lower limits of soil moisture is insufficient, the effect of straw return on water infiltration and irrigation demand improvement is not fully quantified, and the irrigation system is not adaptable to regional real-time water supply conditions and meteorological changes, thus failing to adequately meet the growth needs of wheat and maize.

[0004] Precision and intelligent irrigation decision-making in wheat-maize rotation systems requires meticulous management and intelligent decision-making based on the actual needs of wheat and maize growth, including sowing date, irrigation volume, irrigation date, and irrigation frequency. With the development of information technology and intelligent optimization algorithms, intelligent decision-making using data from remote sensing, meteorology, and hydrology has become possible. However, currently, there is no dynamic optimization strategy based on real-time meteorological data and soil moisture changes to achieve a new balance between agricultural irrigation water and wheat-maize yield, thereby maximizing economic benefits. Summary of the Invention

[0005] In order to solve the technical problems existing in the background art, the present invention aims to provide a rolling correction intelligent irrigation decision-making method for regional wheat-maize rotation systems.

[0006] To solve the technical problem, the technical solution of the present invention is as follows:

[0007] A rolling correction intelligent irrigation decision-making method for a regional wheat-maize rotation system, the method comprising:

[0008] S1: Input the acquired driving data into the crop growth model, use the Sobol method to perform sensitivity analysis on the crop growth model parameters, filter out the non-sensitive parameters in the crop growth model and set them as default values, and calibrate the sensitive parameters; input the calibrated sensitive parameters into the localized growth model, and couple it with the multi-objective optimization model, and combine the particle swarm optimization algorithm PSO with the crop growth model to calibrate the model parameters. With the goal of minimizing the error between the measured and simulated values ​​of crop growth indicators, the sensitive parameters in the crop growth model are automatically calibrated to obtain the localized crop parameters and soil parameters of the crop growth model, and establish a localized wheat-maize rotation system growth model;

[0009] S2: Input the crop model-driven data of the planning year into the localized wheat-maize rotation system growth model to construct a distributed localized growth model of the regional wheat-maize rotation system for the planning year. Couple this model with a multi-objective optimization model to determine the impact of water allocation and straw return on soil moisture and irrigation demand. Based on the objectives of maximizing regional yield, minimizing irrigation volume, and maximizing economic benefits, obtain the high-resolution Pareto optimal solution for the irrigation regime. Then, use the TOPSIS method to assign weights to the long-term irrigation regimes of different regions to obtain different long-term irrigation regimes for the regional spatial grid wheat-maize rotation system, thus obtaining the regional spatial long-term irrigation decision optimization model.

[0010] S3: Using long-term irrigation regimes as a reference, real-time crop data, soil data, hydrological data, meteorological data, field management data, and water supply data are collected and input into the regional spatial long-term irrigation decision optimization model. This model determines the impact of climate change on phenological periods and optimizes field management strategies. The irrigation regime is adjusted in real-time with constraints on biomass, water demand, and available water supply for each time period to determine whether to modify the reference irrigation regime. By coupling a multi-objective optimization model, a dynamic irrigation modification method is constructed to maximize biomass, minimize water demand decision errors, and optimize economic benefits, thereby adjusting the irrigation scheme in real time.

[0011] Furthermore, prior to step S1, the method further includes:

[0012] Driven by data acquisition, remote sensing, field soil testing instruments and weather stations are used to collect crop data, meteorological data, field management data and soil data for winter wheat-summer maize; regional division is carried out by dividing the decision area into multiple sub-regions based on crop distribution maps, soil types and water supply conditions, and then performing rasterization processing.

[0013] Furthermore, the specific process of the Sobol method in step S1 and the parameter calibration of the combined PSO algorithm with the crop growth model include:

[0014] S101: Determine the model input parameters and target variables: Determine the input parameters and parameter range of the wheat-maize growth model. The input parameters include: crop parameters, soil parameters, and meteorological parameters; Determine the target variables of the wheat-maize growth model. The target variables are the wheat / maize leaf area index and aboveground dry matter accumulation in the time series output and the wheat / maize yield in the non-time series output.

[0015] S102: Sampling of input variables: Latin hypercube sampling is used to ensure that each variable is evenly distributed within its range, and a set of sample points is generated for each input variable for model calculation.

[0016] S103: Model Evaluation: Output the sample data generated by the Latin hypercube sampling method, run the model to calculate its output results, repeat this process, and perform sensitivity analysis to ensure the sample size and the accuracy of the analysis.

[0017] S104: Calculating Variance Decomposition: The Sobol method evaluates the impact of input variables through variance decomposition; the output variance is represented as the sum of the variance contributions of each input variable.

[0018]

[0019] Among them, V i V represents the contribution of a single input variable to the output variance. ij This represents the contribution of the interaction effect of the two input variables to the output variance, where k is the number of input parameters and Var(Y) is the total output variance of the model.

[0020] S105: Calculate the Sobol sensitivity index: The Sobol index is a quantitative indicator used to measure the contribution of each input parameter to the model output; the first-order sensitivity index, second-order sensitivity index, and full-order sensitivity index are calculated using programming software:

[0021] First-order sensitivity index (S i ): This indicates the contribution of a single input variable to the output variance, representing the degree of influence of that input variable on the output result without considering other interactions. The calculation formula is as follows:

[0022]

[0023] S i It is a first-order sensitivity index, V i Var(Y) represents the contribution of a single input variable to the output variance, and Var(Y) is the total output variance of the model.

[0024] Second-order sensitivity index (S ij ): This represents the contribution of the interaction between two input variables to the output variance. The calculation formula is as follows:

[0025]

[0026] S ij It is a second-order sensitivity index, V ij This represents the contribution of the interaction effect of two input variables to the output variance.

[0027] Global Sensitivity Index (ST) i ): This represents the total contribution of a given input variable to the output, including the first-order effect of that variable and its interaction effects with other input variables. The calculation formula is as follows:

[0028]

[0029] ST i It is the global sensitivity index, V -i This represents the variance of all parameters except the i-th parameter;

[0030] S106: Results Analysis and Interpretation: Based on the calculated sensitivity indices, analyze the sensitivity of each parameter to wheat / maize yield, leaf area index, and aboveground dry matter accumulation, and calibrate the input parameters with higher sensitivity indices.

[0031] S107: Calibration of parameters for the wheat-maize rotation system growth model: Sensitive parameters of the model were screened using the Sobol analysis method, and parameter calibration was performed in conjunction with PSO and the wheat-maize rotation system growth model.

[0032] Furthermore, the specific steps of PSO in step S107 are as follows:

[0033] S1071: Initialization: Set the initial particle swarm size, define the maximum number of algorithm iterations, set the initial values ​​of particle position and velocity. Each particle represents a set of model parameter combinations, with the position being the model parameter combination and the velocity being the step size when the particle is updated.

[0034] S1072: Based on the objective function, evaluate the fitness of each particle, i.e., minimize the error between the measured and simulated values ​​of crop growth indicators. The fitness value is used to measure the quality of the solution of that particle.

[0035] S1073: Update speed and position: Based on the particle's historical best position and the global best position in the population, adjust the particle's speed and position using the particle swarm optimization update formula;

[0036] S1074: Local and Global Optimal Updates: After updating its velocity and position, each particle recalculates its fitness value and compares it with its historical best position. If the current fitness is better, the particle's best position is updated. Through fitness evaluation of the entire population, the particle with the best fitness is selected as the global optimum.

[0037] S1075: Evaluation: The updated particles are evaluated for fitness based on the objective function, and their performance in the optimization objective is calculated. The fitness evaluation is used to determine whether to update the individual optimal position and the global optimal position of the particles.

[0038] S1076: Iterative update: Repeat steps S2073 to S2075 until the maximum number of iterations or the fitness value converges.

[0039] S1077: Termination condition: The algorithm terminates when the maximum number of iterations is reached or the fitness converges;

[0040] S1078: Output result: Output the optimal solution set, which is the optimal combination of model parameters that minimizes the error between the simulated and measured values.

[0041] Furthermore, the objective function for establishing the localized wheat-maize rotation system growth model is as follows:

[0042]

[0043] In the formula, and These are the measured and simulated values ​​of the wheat / jade leaf area index, respectively. and These are the measured and simulated values ​​of aboveground dry matter accumulation in wheat and corn, respectively. and These represent the measured and simulated values ​​of wheat / maize yield, respectively; n represents the number of samplings for wheat / maize leaf area index and aboveground dry matter accumulation in the measured sampling or experiment, and m represents the number of samplings for wheat / maize yield.

[0044] Furthermore, step S2 includes:

[0045] S201: Data Input: Input the data required for the localized distributed growth model of the wheat-maize rotation system under the planning year into the crop model, including meteorological data, soil data, field management data and localized model parameters, to create a localized distributed growth model of the wheat-maize rotation system adapted to the planning year; Determine the objective function: Determine the optimization objective, namely, maximizing crop yield, minimizing irrigation water consumption and maximizing economic effect;

[0046] S202: Set constraints: With the upper and lower limits of soil volumetric moisture content and regional water supply conditions as constraints, consider the soil allocation characteristics and water complementarity in the wheat-maize rotation system: At the end of the winter wheat growing season, the soil moisture should be sufficient to meet the water demand when the maize growing season begins, in order to ensure the availability of resources in actual production and to ensure the normal growth of wheat and maize.

[0047] Formulas relating soil moisture content to wheat and maize:

[0048] SM cron (t0)=SM wheat (t1)+SM(t2,t3)

[0049] In the formula: SM cron (t0) represents the soil moisture content at the initial stage of maize growth; SM wheat (t1) represents the soil moisture content at the time of winter wheat harvest; ΔSM(t2,t3) represents the change in soil moisture between the end of winter wheat growth and the beginning of maize growth, including the effects of precipitation, evaporation and irrigation.

[0050] S203: Water resource allocation and complementarity among crops: Combining the seasonal characteristics of wheat-maize rotation, optimize the water resource requirements of winter wheat and summer maize in the model according to soil moisture at different crop growth stages;

[0051] S204: The impact of straw return to the field on soil moisture infiltration and irrigation demand: In the localized distributed growth model of the wheat-maize rotation system, the improvement effect of straw return to the field on soil porosity and infiltration rate after the maize growing season is considered.

[0052] Formula for the relationship between summer corn straw return to the field and the initial soil volumetric moisture content of winter wheat:

[0053] SM t =SM0+SM s -SM e

[0054] SM s =ΔSOM×S water

[0055] In the formula, SM t SM0 is the initial soil volumetric moisture content of winter wheat; SM is the initial soil volumetric moisture content of winter wheat; SM s It is the change in soil volumetric moisture content caused by returning straw to the field; SM e It is the soil moisture evaporation rate; ΔSOM is the amount of straw returned to the field; S water It is the coefficient of soil organic matter's ability to retain water;

[0056] S205: Model Coupling: Couple the distributed growth localization model of the wheat-maize rotation system with the multi-objective optimization model. The crop input data of the decision variable of the optimization model is transferred to the distributed growth localization model of the wheat-maize rotation system. The yield, irrigation water volume and economic benefits of the distributed growth localization model of the wheat-maize rotation system after the operation is completed are fed back to the optimization model.

[0057] S206: Model Solving: The NSGA-II algorithm is used to solve the optimization model, and the Pareto optimal solution set of irrigation regime under high-resolution raster in the region is output.

[0058] S207: TOPSIS Comprehensive Evaluation Method Selection: Pareto solution set selection is performed, and the TOPSIS comprehensive evaluation method is used to further select the best Pareto solution set.

[0059] Furthermore, step S207 includes:

[0060] S2071: Determine evaluation indicators: Clarify decision-making objectives and select relevant evaluation indicators, namely, the higher the crop yield value, the lower the irrigation amount value, and the higher the economic benefit value, the better.

[0061] S2072: Constructing a normalized decision matrix: The raw data is organized and normalized. The crop yield, irrigation amount, and economic benefit data in the Pareto solution set are organized into a decision matrix Di×j, where i represents the number of Pareto solutions and j represents the number of decision objectives. Each row of this matrix is ​​an alternative scheme. Linear normalization is used to transform the data into a uniform scale. Determining the weights of evaluation indicators: Different weight combinations are made for each evaluation indicator according to the decision-maker's preference perspective. These can be divided into long-term irrigation system schemes for wheat-maize rotation systems in high water efficiency and economic benefit areas, long-term irrigation system schemes for wheat-maize rotation systems in high yield and economic benefit areas, and long-term irrigation system schemes for wheat-maize rotation systems in water-saving and economic benefit areas.

[0062] S2073: Calculate the positive and negative ideal solutions: Calculate the distance from each alternative solution to the positive and negative ideal solutions;

[0063] S2074: Sort: Calculate the relative proximity of each alternative to the positive and negative ideal solutions, sort the alternatives according to the relative proximity, and the one with the largest value is the best long-term irrigation decision scheme for the regional wheat-maize rotation system under that preference angle.

[0064] Furthermore, the objective function of the regional spatial long-term irrigation decision optimization model is as follows:

[0065] Obj1 = maxTWSO

[0066] Obj2=minI

[0067] Obj3 = maxNEB

[0068] In the formula, TWSO is the wheat / maize yield per unit area; I is the irrigation water volume per unit area; NEB is the economic benefit of wheat / maize per unit area.

[0069] The constraints of the distributed growth localization model for the regional wheat-maize rotation system are as follows:

[0070] Soil moisture content constraints:

[0071] SM min ≤SM≤SM max

[0072] In the formula, SM represents the soil moisture content in the root zone. min and SM max These represent the lower and upper limits of soil moisture content in the root zone, respectively.

[0073] Irrigation water constraints:

[0074] I<R

[0075] In the formula, I is the irrigation water volume per unit area; R is the available water volume per unit area.

[0076] Production constraints:

[0077] ΣTWSO·A=TWSO REG

[0078] In the formula, TWSO REG It represents the grain demand in the area where the irrigation district is located.

[0079] Furthermore, step S3 includes:

[0080] S301: The regional long-term irrigation regime is used as a reference long-term irrigation regime in the study area, and the aboveground biomass corresponding to the reference long-term irrigation regime is set as the reference aboveground biomass.

[0081] S302: Remote sensing images of the irrigation decision area are acquired via satellite platforms or drones, and crop data such as aboveground biomass and leaf area index are obtained by combining them with regression models; real-time soil data is acquired by deploying soil moisture monitoring instruments in the irrigation decision area; hydrological data is acquired by a series of runoff monitoring data from hydrological stations; real-time meteorological data is acquired by setting up small weather stations in the irrigation decision area; and field management data is acquired by field managers by inputting agricultural management measures in real time, including: single irrigation amount, irrigation date, fertilizer amount, and fertilizer date.

[0082] S303: After the winter wheat harvest or summer maize harvest, real-time meteorological conditions are read and driven data are read to explore the impact of real-time meteorological conditions on the phenological period of wheat and maize. Future meteorological data are input into the regional wheat-maize rotation system distributed growth localization model. With the goal of maximizing yield, the optimal sowing date of summer maize or winter wheat is explored.

[0083] S304: Input real-time driving data into the regional wheat-maize rotation system distributed growth localization model. Determine whether to modify the reference irrigation regime based on constraints. Grid points that do not meet any of the three constraints need to re-determine the irrigation regime, while grid points that meet the constraints follow the reference irrigation regime. The specific constraints are as follows:

[0084] Time-based biomass constraints:

[0085] K·Bio≥Bio R

[0086] In the formula, K is the correction coefficient; Bio is the biomass for the time period; Bio R For reference irrigation regimes, the biomass of the time period is used;

[0087] Water supply constraints:

[0088] QS ≥Q D

[0089] In the formula, Q S For water supply; Q D This refers to water demand;

[0090] Water demand decision error constraints:

[0091]

[0092] In the formula, I R P represents the total irrigation amount for the current period, P represents the rainfall for the current period, and ET represents the water demand for the current period.

[0093] S305: For grids that require irrigation regime updates, the irrigation regime for the next 15 days is re-determined. To ensure the efficiency of the model operation, the particle swarm optimization algorithm is used to solve the model. With the goal of maximizing biomass during the time period and minimizing the water demand decision error, an adaptive correction decision model is constructed.

[0094] Compared with the prior art, the advantages of the present invention are as follows:

[0095] This invention constructs a regional high-water-efficiency intelligent irrigation decision model. In this model, irrigation data within the region no longer relies solely on annual average meteorological and soil data, but instead reads and processes detailed meteorological elements and soil moisture information in real time to dynamically optimize irrigation strategies based on daily changes. By optimizing the irrigation amount of the regional grid on a daily basis, the spatial heterogeneity of the crop growth environment within a large-scale region can be reflected more precisely, thereby improving the accuracy and reliability of the crop model in large-scale applications. Attached Figure Description

[0096] Figure 1 The present invention provides a technical flowchart of a rolling correction intelligent irrigation decision-making method for a regional wheat-maize rotation system. Detailed Implementation

[0097] The specific implementation of the present invention is described below with reference to embodiments:

[0098] It should be noted that the structures, proportions, sizes, etc. shown in this specification are only used to complement the content disclosed in the specification for those skilled in the art to understand and read, and are not intended to limit the conditions under which the present invention can be implemented. Any modifications to the structure, changes in the proportions, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should still fall within the scope of the technical content disclosed in the present invention.

[0099] Furthermore, the terms such as "upper," "lower," "left," "right," "middle," and "one" used in this specification are merely for clarity of description and are not intended to limit the scope of the invention. Any changes or adjustments to their relative relationships, without substantially altering the technical content, should also be considered within the scope of the invention.

[0100] Example 1:

[0101] like Figure 1 As shown, this invention provides a rolling correction intelligent irrigation decision-making method for a regional wheat-maize rotation system, comprising the following steps:

[0102] S1: Basic Data Collection and Irrigation Decision-Making Area Delineation

[0103] Remote sensing technology, field soil testing instruments, and weather stations were used to collect crop data, meteorological data, field management data, and soil data for the winter wheat-summer maize region. Based on the regional winter wheat-summer maize planting distribution map, soil type, and regional water supply conditions, the decision-making region was divided into four sub-regions. Each sub-region was then rasterized.

[0104] S2: Sensitivity Analysis and Crop Model Localization

[0105] The model-driven data is input into the crop growth model, and the Sobol method is used to perform sensitivity analysis on the crop growth model parameters. The non-sensitive parameters in the crop growth model are screened out and set as default values, while the sensitive parameters are calibrated.

[0106] The Sobol method's flow is as follows:

[0107] 1. Determine the model input parameters and target variables:

[0108] Determine the input parameters and parameter ranges for the wheat-maize growth model. The input parameters include crop parameters, soil parameters, and meteorological parameters. Determine the target variables for the wheat-maize growth model. The target variables are the wheat / maize leaf area index and aboveground dry matter accumulation in the time series output, and the wheat / maize yield in the non-time series output.

[0109] 2. Sampling of input variables:

[0110] Latin hypercube sampling is used to ensure that each variable is uniformly distributed within its range. A set of sample points needs to be generated for each input variable; these sample points will be used in the model's computation.

[0111] 3. Model Evaluation:

[0112] The sample data generated by the Latin hypercube sampling method is output, the model is run to calculate the output results, and this process is repeated until enough sample output points are obtained to perform sensitivity analysis, so as to ensure that the sample size is large enough (several thousand samples) to ensure the accuracy of the analysis.

[0113] 4. Calculate variance decomposition:

[0114] The Sobol method evaluates the impact of input variables through variance decomposition. Specifically, the output variance can be represented as the sum of the variance contributions of each input variable:

[0115]

[0116] Among them, V i V represents the contribution of a single input variable to the output variance. ij This represents the contribution of the interaction effect of the two input variables to the output variance, where k is the number of input parameters and Var(Y) is the total output variance of the model.

[0117] 5. Calculate the Sobol sensitivity index:

[0118] The Sobol index is a quantitative indicator used to measure the contribution of each input parameter to the model output; the first-order sensitivity index, second-order sensitivity index, and full-order sensitivity index are calculated using programming software.

[0119] First-order sensitivity index (S i ): This represents the contribution of a single input variable to the output variance. It can be understood as the degree of influence of that input variable on the output result without considering other interactions. The calculation formula is as follows:

[0120]

[0121] S i It is a first-order sensitivity index, V i Var(Y) represents the contribution of a single input variable to the output variance, and Var(Y) is the total output variance of the model.

[0122] Second-order sensitivity index (S ij ): This represents the contribution of the interaction between two input variables to the output variance. The calculation formula is as follows:

[0123]

[0124] S ij It is a second-order sensitivity index, V ij This represents the contribution of the interaction effect of two input variables to the output variance.

[0125] Global Sensitivity Index (ST) i(): Represents the total contribution of an input variable to the output, including the first-order effect of that variable and the interaction effect with other input variables. The calculation formula is as follows:

[0126]

[0127] ST i It is the global sensitivity index, V -i This represents the variance of all parameters except the i-th parameter;

[0128] 6. Results Analysis and Interpretation:

[0129] Based on the calculated sensitivity indices, the sensitivity of each parameter to wheat / corn yield, leaf area index and aboveground dry matter accumulation is analyzed, and parameter calibration is performed on input parameters with larger sensitivity indices.

[0130] The sensitivity scores that need to be calibrated for each sub-region are determined. The Particle Swarm Optimization (PSO) algorithm is combined with the crop growth model to calibrate the model parameters. With the goal of minimizing the error between the measured and simulated values ​​of crop growth indicators, the sensitivity parameters in the crop growth model are automatically calibrated to obtain the localized crop parameters and soil parameters of the crop growth model, and a localized wheat-maize rotation system growth model is established.

[0131] The specific steps of PSO are as follows:

[0132] 1. Initialization: Set the initial particle swarm size, define the maximum number of algorithm iterations, and set the initial values ​​for particle position and velocity. Each particle represents a set of model parameter combinations; its position is the model parameter combination, and its velocity is the step size for updating that particle.

[0133] 2. Based on the objective function, evaluate the fitness of each particle, i.e., minimize the error between the measured and simulated values ​​of crop growth indicators. The fitness value is used to measure the quality of the solution of that particle.

[0134] 3. Update speed and position: Based on the particle's historical best position (individual best) and the global best position in the swarm, adjust the particle's speed and position using the particle swarm optimization update formula;

[0135] 4. Local and Global Optimal Updates: After updating its velocity and position, each particle recalculates its fitness value and compares it with its historical best position. If the current fitness is better, the particle's best position is updated. The particle with the best fitness across the entire population is selected as the global optimum through a fitness evaluation.

[0136] 5. Evaluation: The updated particles are evaluated for fitness based on the objective function, and their performance in the optimization objective is calculated. Fitness evaluation determines whether to update the individual optimal position and global optimal position of the particles.

[0137] 6. Iterative update: Repeat steps 3 to 5 until the maximum number of iterations or the fitness value converges.

[0138] 7. Termination condition: The algorithm terminates when the maximum number of iterations is reached or the fitness converges.

[0139] 8. Output results: Output the optimal solution set, which is the optimal combination of model parameters that minimizes the error between the simulated and measured values.

[0140] According to the implementation method, the objective function of the localized optimization model of the wheat-corn rotation system growth model in step S27 is as follows:

[0141]

[0142] In the formula, and These are the measured and simulated values ​​of the wheat / jade leaf area index, respectively. and These are the measured and simulated values ​​of aboveground dry matter accumulation in wheat and corn, respectively. and These represent the measured and simulated values ​​of wheat / maize yield, respectively; n represents the number of samplings for wheat / maize leaf area index and aboveground dry matter accumulation in the measured sampling or experiment, and m represents the number of samplings for wheat / maize yield.

[0143] S3: Regional Long-Term Reference Irrigation System Decision-Making Scheme

[0144] The crop model-driven data (crop data, field management data, meteorological data, and soil data) for the planning year are input into the localized wheat-maize rotation system growth model to construct a distributed localized growth model of the regional wheat-maize rotation system for the planning year. The distributed localized growth model of the wheat-maize rotation system is coupled with a multi-objective optimization model to clarify the soil moisture allocation and complementarity relationship of the wheat-maize rotation system, clarify the regulatory role of straw return in the wheat-maize rotation system on soil moisture infiltration and irrigation demand, and construct a regional grid-based refined dynamic irrigation decision-making technology. With the objectives of maximizing regional wheat / maize yield, minimizing irrigation volume, and maximizing wheat / maize economic benefits, and with the upper and lower limits of soil volumetric moisture content and water supply conditions as constraints, the Pareto optimal solution of the irrigation regime of the wheat-maize rotation system under regional high resolution is obtained. The TOPSIS comprehensive evaluation method is used to assign weights to the multi-objectives to obtain different regional spatial grid-based long-term irrigation regimes for the wheat-maize rotation system.

[0145] Developing a long-term regional irrigation decision-making scheme requires coupling a distributed growth localization model and a multi-objective optimization model for the wheat-maize rotation system. The specific steps are as follows:

[0146] 1. Data Input: Input the data required for the localized distributed growth model of the wheat-maize rotation system under the planning year into the crop model, including meteorological data, soil data, field management data and localized model parameters, to create a localized distributed growth model of the wheat-maize rotation system adapted to the planning year; Determine the objective function: Determine the optimization objectives, namely, maximizing crop yield, minimizing irrigation water consumption and maximizing economic benefits;

[0147] 2. Setting Constraints: Based on the upper and lower limits of soil volumetric moisture content and regional water supply conditions, and considering the soil allocation characteristics and water complementarity in the wheat-maize rotation system: At the end of the winter wheat growing season, soil moisture reserves should be sufficient to meet the water requirements of maize when the growing season begins. This is to ensure the availability of resources in actual production while guaranteeing the normal growth of wheat and maize;

[0148] Formulas relating soil moisture content to wheat and maize:

[0149] SM cron (t0)=SM wheat (t1)+SM(t2,t3)

[0150] In the formula: SM cron (t0) represents the soil moisture content at the beginning of the maize growing season; SMwheat(t1) represents the soil moisture content at the time of winter wheat harvest; ΔSM(t2,t3) represents the change in soil moisture (including the effects of precipitation, evaporation and irrigation) between the end of winter wheat growth and the beginning of maize growth.

[0151] 3. Water resource allocation and complementarity among crops: Combining the seasonal characteristics of wheat-maize rotation, the water resource requirements of winter wheat and summer maize are allocated in the optimization model according to soil moisture at different crop growth stages.

[0152] 4. Impact of straw return to the field on soil moisture infiltration and irrigation demand: In the localized distributed growth model of the wheat-maize rotation system, the improvement effect of straw return to the field on soil porosity and infiltration rate after the maize growing season is considered.

[0153] Formula for the relationship between summer corn straw return to the field and the initial soil volumetric moisture content of winter wheat:

[0154] SM t =SM0+SM s -SM e

[0155] SM s =ΔSOM×S water

[0156] In the formula, SM t SM0 is the initial soil volumetric moisture content of winter wheat; SM is the initial soil volumetric moisture content of winter wheat; SM s It is the change in soil volumetric moisture content caused by returning straw to the field; SM e It is the soil moisture evaporation rate; ΔSOM is the amount of straw returned to the field; S water It is the coefficient of soil organic matter's ability to retain water;

[0157] 5. Model Coupling: The distributed growth localization model of the wheat-maize rotation system is coupled with the multi-objective optimization model. The crop input data of the decision variable of the optimization model is transferred to the distributed growth localization model of the wheat-maize rotation system. After the distributed growth localization model of the wheat-maize rotation system is completed, the yield, irrigation water volume and economic benefits are fed back to the optimization model.

[0158] 6. Model Solving: The NSGA-II algorithm is used to solve the optimization model, and the Pareto optimal solution set of the irrigation regime under the high-resolution raster is output.

[0159] Furthermore, the specific steps for NSGA-II are as follows:

[0160] Initialization: Set the initial population size, define the maximum number of algorithm iterations, set the crossover probability and mutation probability, and randomly generate the initial population. Each individual is a combination of irrigation regimes for each grid.

[0161] Evaluation: Based on the objective function, evaluate the fitness of each individual, namely, maximizing regional crop yield, minimizing irrigation, and maximizing economic benefits;

[0162] Selection: Individuals are ranked by non-dominated ordering and crowding distance is calculated to maintain diversity, with priority given to individuals with low non-dominated rank and large crowding distance.

[0163] Crossover: The selected parent individuals are crossovered to generate new individuals; the crossover operation is determined by the crossover probability, and a new individual is generated by combining the genetic information of two parent individuals.

[0164] Mutation: Mutation is performed on the crossover individuals to increase population diversity and avoid getting trapped in local optima. The mutation operation determines whether to perform mutation based on the mutation probability, and generates new individuals by randomly changing the genetic information of the individuals.

[0165] Evaluation: The fitness of newly generated individuals is evaluated based on the objective function, and their performance in the optimization objective is calculated for ranking and selection.

[0166] Replacement: Parents and offspring are selected using non-dominated sorting and crowding distance to optimize and retain high-quality individuals, generating the next generation of the population;

[0167] Termination condition: The algorithm terminates when the maximum number of iterations is reached or the fitness converges;

[0168] Output: Output the optimal solution set, which is the optimal combination of grid irrigation systems that maximizes regional crop yield, minimizes irrigation water consumption, and maximizes economic benefits.

[0169] The TOPSIS comprehensive evaluation method is used to select the best Pareto solution set, and the TOPSIS comprehensive evaluation method is used to further select the best Pareto solution set.

[0170] The specific steps of the TOPSIS comprehensive evaluation method are as follows:

[0171] 1. Determine evaluation indicators: Clarify decision-making objectives and select relevant evaluation indicators, namely, the higher the crop yield value, the lower the irrigation amount value, and the higher the economic benefit value, the better;

[0172] 2. Constructing a normalized decision matrix: The raw data is organized and normalized. The crop yield, irrigation amount, and economic benefit data in the Pareto solution set are organized into a decision matrix Di×j, where i represents the number of Pareto solutions and j represents the number of decision objectives. Each row of this matrix represents an alternative solution. Linear normalization is used to transform the data into a uniform scale. Determining the weights of evaluation indicators: Different weight combinations are made for each evaluation indicator based on the decision-maker's preference perspective. These can be divided into long-term irrigation system schemes for wheat-maize rotation systems in high water efficiency and economic benefit areas, long-term irrigation system schemes for wheat-maize rotation systems in high-yield and economic benefit areas, and long-term irrigation system schemes for wheat-maize rotation systems in water-saving and economic benefit areas.

[0173] 3. Calculate the positive and negative ideal solutions: Calculate the distance from each alternative solution to the positive and negative ideal solutions;

[0174] 4. Sort: Calculate the relative proximity of each alternative to the positive and negative ideal solutions, and sort the alternatives according to the relative proximity. The one with the largest value is the best long-term irrigation decision scheme for the wheat-maize rotation system in the region from the perspective of preference.

[0175] According to some implementation methods, the objective function of the regional spatial long-term irrigation decision optimization model is as follows:

[0176] Obj1 = maxTWSO

[0177] Obj2=minI

[0178] Obj3 = maxNEB

[0179] In the formula, TWSO is the wheat / maize yield per unit area; I is the irrigation water volume per unit area; NEB is the economic benefit of wheat / maize per unit area.

[0180] The constraints of the distributed growth localization model for the regional wheat-maize rotation system are as follows:

[0181] Soil moisture content constraints:

[0182] SM min ≤SM≤SM max

[0183] In the formula, SM represents the soil moisture content in the root zone. min and SM max These represent the lower and upper limits of soil moisture content in the root zone, respectively.

[0184] Irrigation water constraints:

[0185] I<R

[0186] In the formula, I is the irrigation water volume per unit area; R is the available water volume per unit area.

[0187] Production constraints:

[0188] ΣTWSO·A=TWSO REG

[0189] In the formula, TWSO REG It represents the grain demand in the area where the irrigation district is located;

[0190] S4: Real-time Dynamic Correction Scheme for Regional Irrigation Systems

[0191] The long-term irrigation regime of the regional spatial grid-based wheat-maize rotation system is used as a reference irrigation regime and applied to the decision-making area. Real-time crop data, soil data, hydrological data, meteorological data, field management data, and water supply data are input into the regional wheat-maize rotation system distributed growth localization model to explore the impact of climate change on the phenological period of the wheat-maize rotation system and optimize field management strategies. The irrigation regime is adjusted in real time with constraints of biomass, water demand, and available water for each time period to determine whether the reference irrigation regime should be modified. The regional grid-based refined dynamic irrigation correction technology is constructed by coupling the wheat-maize rotation system distributed growth localization model with a multi-objective optimization model. With the objectives of maximizing biomass for each time period, minimizing water demand decision error, and maximizing economic benefits, and with soil volumetric moisture content and water supply conditions as constraints, the reference irrigation regime is modified in real time.

[0192] The specific steps for the real-time dynamic correction scheme of regional irrigation systems are as follows:

[0193] 1. The regional long-term irrigation regime was used as a reference long-term irrigation regime in the study area, and the aboveground biomass corresponding to the reference long-term irrigation regime was set as the reference aboveground biomass.

[0194] 2. Obtain remote sensing images of the irrigation decision area through satellite platforms or drones, and combine them with regression models to obtain crop data on aboveground biomass and leaf area index; obtain real-time soil data by deploying soil moisture monitoring instruments in the irrigation decision area; obtain hydrological data by using a series of runoff monitoring data from hydrological stations; obtain real-time meteorological data by setting up small weather stations in the irrigation decision area; obtain field management data by having field managers input agricultural management measures in real time, including but not limited to: single irrigation amount, irrigation date, fertilizer amount, and fertilizer date;

[0195] 3. After the winter wheat harvest or summer maize harvest, the impact of real-time weather changes on the phenological period of wheat and maize will be explored by reading driving data in real time. Future weather data will be input into the regional wheat-maize rotation system distributed growth localization model to explore the optimal sowing date of summer maize or winter wheat with the goal of maximizing yield.

[0196] 4. Input real-time driving data into the regional wheat-maize rotation system distributed growth localization model. Use constraints to determine whether to modify the reference irrigation regime. Grid points that do not meet any of the three constraints need to re-determine the irrigation regime; grid points that meet the constraints follow the reference irrigation regime. Specific constraints are as follows:

[0197] Time-based biomass constraints:

[0198] K·Bio≥Bio R

[0199] In the formula, K is the correction coefficient; Bio is the biomass for the time period; Bio R For reference irrigation regimes, the biomass of the time period is used;

[0200] Water supply constraints:

[0201] Q S ≥Q D

[0202] In the formula, Q S For water supply; Q D This refers to water demand;

[0203] Water demand decision error constraints:

[0204]

[0205] In the formula, I R P represents the total irrigation amount for the current period, P represents the rainfall for the current period, and ET represents the water demand for the current period.

[0206] 5. For grids that require irrigation regime updates, the irrigation regime for the next 15 days is re-determined. To ensure the efficiency of the model operation, the particle swarm optimization algorithm is used to solve the model. An adaptive correction decision model is constructed with the objectives of maximizing biomass during the time period and minimizing water demand decision error.

[0207] The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

[0208] Many other changes and modifications can be made without departing from the concept and scope of this invention. It should be understood that this invention is not limited to the specific embodiments, and the scope of this invention is defined by the appended claims.

Claims

1. A rolling correction intelligent irrigation decision-making method for a regional wheat-maize rotation system, characterized in that, The method includes: S1: Input the acquired driving data into the crop growth model, use the Sobol method to perform sensitivity analysis on the crop growth model parameters, screen out the non-sensitive parameters in the crop growth model and set them as default values, calibrate the sensitive parameters, and combine the particle swarm optimization algorithm (PSO) with the crop growth model to calibrate the model parameters. With the goal of minimizing the error between the measured and simulated values ​​of crop growth indicators, the sensitive parameters in the crop growth model are automatically calibrated to obtain the localized crop parameters and soil parameters of the crop growth model, and establish a localized wheat-maize rotation system growth model. S2: Input the crop model-driven data of the planning year into the localized wheat-maize rotation system growth model to construct a distributed localized growth model of the regional wheat-maize rotation system for the planning year. Couple this model with a multi-objective optimization model to determine the impact of water allocation and straw return on soil moisture and irrigation demand, thus obtaining a regional spatial long-term irrigation decision optimization model. Based on the objectives of maximizing regional yield, minimizing irrigation volume, and maximizing economic benefits, obtain high-resolution Pareto optimal solutions for irrigation regimes. Then, use the TOPSIS method to assign weights to long-term irrigation regimes in different regions to obtain different regional spatial grid-based wheat-maize rotation system long-term irrigation regimes. Step S2 includes: Set constraints: The upper and lower limits of soil volumetric moisture content and regional water supply conditions are used as constraints. Consider the soil allocation characteristics and water complementarity in the wheat-maize rotation system: At the end of the winter wheat growing season, the soil moisture should be sufficient to meet the water demand of maize when it enters the growing season. This is to ensure the availability of resources in actual production and to ensure the normal growth of wheat and maize. Formulas relating soil moisture content to wheat and maize: ； In the formula: This refers to the soil moisture content during the early stages of corn growth. This refers to the soil moisture content at the time of winter wheat harvest. The change in soil moisture between the end of winter wheat growth and the beginning of maize growth includes the effects of precipitation, evaporation, and irrigation. Water resource allocation and complementarity among crops: Combining the seasonal characteristics of wheat-maize rotation, the water resource requirements of winter wheat and summer maize are allocated in the optimization model according to soil moisture at different crop growth stages. The impact of straw return to the field on soil moisture infiltration and irrigation demand: In the localized distributed growth model of the wheat-maize rotation system, the improvement effect of straw return to the field after the maize growing season on soil porosity and infiltration rate is considered. Formula for the relationship between summer corn straw return to the field and the initial soil volumetric moisture content of winter wheat: ； ； In the formula, SM t SM0 is the initial soil volumetric moisture content of winter wheat; SM is the initial soil volumetric moisture content of winter wheat; SM s It is the change in soil volumetric moisture content caused by returning straw to the field; SM e It is the amount of soil moisture evaporation; This refers to the amount of straw returned to the field; S water It is the coefficient of soil organic matter's ability to retain water; Model Coupling: A distributed localized growth model of the wheat-maize rotation system is coupled with a multi-objective optimization model. Crop input data, the decision variables of the optimization model, are transferred to the distributed localized growth model of the wheat-maize rotation system. Yield, irrigation water volume, and economic benefits after the distributed localized growth model of the wheat-maize rotation system are fed back to the optimization model. S3: Using a long-term irrigation regime as a reference, real-time crop data, soil data, hydrological data, meteorological data, field management data, and water supply data are collected and input into the regional spatial long-term irrigation decision optimization model to determine the impact of climate change on phenological periods and optimize field management. The strategy involves adjusting the irrigation regime in real time, using time-bound biomass, water demand, and available water as constraints, to determine whether to modify the reference irrigation regime. By coupling a multi-objective optimization model, a dynamic irrigation correction method is constructed to maximize biomass, minimize water demand decision-making errors, and optimize economic benefits, thereby adjusting the irrigation scheme in real time. Step S3 includes: at the end of winter wheat or summer maize harvest, real-time reading of driving data is used to explore the impact of real-time meteorological changes on wheat-maize phenology. Future meteorological data is then input into the regional wheat-maize rotation system distributed growth localization model to explore the optimal sowing date for summer maize or winter wheat.

2. The rolling correction intelligent irrigation decision-making method for a regional wheat-maize rotation system according to claim 1, characterized in that, Prior to step S1, the method further includes: Driven by data acquisition, remote sensing, field soil testing instruments and weather stations are used to collect crop data, meteorological data, field management data and soil data for winter wheat-summer maize; regional division is carried out by dividing the decision area into multiple sub-regions based on crop distribution maps, soil types and water supply conditions, and then performing rasterization processing.

3. The rolling correction intelligent irrigation decision-making method for a regional wheat-maize rotation system according to claim 1, characterized in that, The specific process of the Sobol method in step S1 and the parameter calibration of the combined PSO algorithm with the crop growth model include: S101: Determine the model input parameters and target variables: Determine the input parameters and parameter range of the wheat-maize growth model. The input parameters include: crop parameters, soil parameters, and meteorological parameters; Determine the target variables of the wheat-maize growth model. The target variables are the wheat / maize leaf area index and aboveground dry matter accumulation in the time series output and the wheat / maize yield in the non-time series output. S102: Sampling of input variables: Latin hypercube sampling is used to ensure that each variable is evenly distributed within its range, generating a set of sample points for each input variable for model calculation; S103: Model Evaluation: Output the sample data generated by the Latin hypercube sampling method, run the model to calculate its output results, repeat this process, and perform sensitivity analysis to ensure the sample size and the accuracy of the analysis. S104: Calculating Variance Decomposition: The Sobol method evaluates the impact of input variables through variance decomposition; the output variance is represented as the sum of the variance contributions of each input variable. ； Among them, V i V represents the contribution of a single input variable to the output variance. ij This represents the contribution of the interaction effect of the two input variables to the output variance, where k is the number of input parameters and Var(Y) is the total output variance of the model. S105: Calculate the Sobol sensitivity index: The Sobol index is a quantitative indicator used to measure the contribution of each input parameter to the model output; the first-order sensitivity index, second-order sensitivity index, and full-order sensitivity index are calculated using programming software: First-order sensitivity index (S i The contribution of a single input variable to the output variance represents the degree of influence of that input variable on the output result without considering other interactions. The calculation formula is as follows: ； S i It is a first-order sensitivity index, V i Var(Y) represents the contribution of a single input variable to the output variance, and Var(Y) is the total output variance of the model. Second-order sensitivity index (S ij The variance denoted by represents the contribution of the interaction between two input variables to the output variance, and is calculated using the following formula: ； S ij It is a second-order sensitivity index, V ij This represents the contribution of the interaction effect of two input variables to the output variance. Global Sensitivity Index (ST) i ): This represents the total contribution of a given input variable to the output, including the first-order effect of that variable and its interaction with other input variables. The calculation formula is as follows: ； ST i It is the global sensitivity index, V -i This represents the variance of all parameters except the i-th parameter; S106: Results Analysis and Interpretation: Based on the calculated sensitivity indices, analyze the sensitivity of each parameter to wheat / maize yield, wheat / maize leaf area index and aboveground dry matter accumulation, and perform parameter calibration for input parameters with higher sensitivity indices. S107: Calibration of parameters for the wheat-maize rotation system growth model: Sensitive parameters of the model were screened using the Sobol analysis method, and parameter calibration was performed in conjunction with PSO and the wheat-maize rotation system growth model.

4. The rolling correction intelligent irrigation decision-making method for a regional wheat-maize rotation system according to claim 3, characterized in that, The specific steps of PSO in step S107 are as follows: S1071: Initialization: Set the initial particle swarm size, define the maximum number of algorithm iterations, set the initial values ​​of particle position and velocity. Each particle represents a set of model parameter combinations, with the position being the model parameter combination and the velocity being the step size when the particle is updated. S1072: Based on the objective function, evaluate the fitness of each particle, i.e., minimize the error between the measured and simulated values ​​of crop growth indicators. The fitness value is used to measure the quality of the solution of that particle. S1073: Update speed and position: Based on the particle's historical best position and the global best position in the population, adjust the particle's speed and position using the particle swarm optimization update formula; S1074: Local and Global Optimal Updates: After updating its velocity and position, each particle recalculates its fitness value and compares it with its historical best position. If the current fitness is better, the particle's best position is updated. Through fitness evaluation of the entire population, the particle with the best fitness is selected as the global optimum. S1075: Evaluation: The updated particles are evaluated for fitness based on the objective function, and their performance in the optimization objective is calculated. The fitness evaluation is used to determine whether to update the individual optimal position and the global optimal position of the particles. S1076: Iterative update: Repeat steps S1073 to S1075 until the maximum number of iterations or the fitness value converges; S1077: Termination condition: The algorithm terminates when the maximum number of iterations is reached or the fitness converges; S1078: Output result: Output the optimal solution set, which is the optimal combination of model parameters that minimizes the error between the simulated and measured values.

5. The rolling correction intelligent irrigation decision-making method for a regional wheat-maize rotation system according to claim 1, characterized in that, The objective function for establishing the localized wheat-maize rotation system growth model is as follows: In the formula, LAI i m and LAI i s These are the measured and simulated values ​​of the wheat / jade leaf area index, respectively; TAGP i m and TAGP i s These are the measured and simulated values ​​of aboveground dry matter accumulation in wheat / maize, respectively; TWSO i m and TWSO i s These represent the measured and simulated values ​​of wheat / maize yield, respectively; n represents the number of samplings for wheat / maize leaf area index and aboveground dry matter accumulation in the measured sampling or experiment, and m represents the number of samplings for wheat / maize yield.

6. The rolling correction intelligent irrigation decision-making method for a regional wheat-maize rotation system according to claim 1, characterized in that, The objective function of the regional spatial long-term irrigation decision optimization model is as follows: ； ； ； In the formula, TWSO is the wheat / corn yield per unit area; I represents the irrigation water volume per unit area; NEB represents the economic benefits of wheat / maize per unit area. The constraints of the distributed growth localization model for the regional wheat-maize rotation system are as follows: Soil moisture content constraints: ； In the formula, SM represents the soil moisture content in the root zone. min and SM max These represent the lower and upper limits of soil moisture content in the root zone, respectively. Irrigation water constraints: ； In the formula, I is the irrigation water volume per unit area; R is the available water volume per unit area. Production constraints: ； In the formula, TWSO REG It represents the grain demand in the area where the irrigation district is located.

7. The rolling correction intelligent irrigation decision-making method for a regional wheat-maize rotation system according to claim 1, characterized in that, Step S3 includes: S301: The regional long-term irrigation regime is used as a reference long-term irrigation regime in the study area, and the aboveground biomass corresponding to the reference long-term irrigation regime is set as the reference aboveground biomass. S302: Remote sensing images of the irrigation decision area are acquired via satellite platforms or drones, and crop data such as aboveground biomass and leaf area index are obtained by combining them with regression models; real-time soil data is acquired by deploying soil moisture monitoring instruments in the irrigation decision area; hydrological data is acquired by a series of runoff monitoring data from hydrological stations; real-time meteorological data is acquired by setting up small weather stations in the irrigation decision area; and field management data is acquired by field managers by inputting agricultural management measures in real time, including: single irrigation amount, irrigation date, fertilizer amount, and fertilizer date. S303: After the winter wheat harvest or summer maize harvest, real-time meteorological conditions are read and driven data are read to explore the impact of real-time meteorological conditions on the phenological period of wheat and maize. Future meteorological data are input into the regional wheat-maize rotation system distributed growth localization model. With the goal of maximizing yield, the optimal sowing date of summer maize or winter wheat is explored. S304: Input real-time driving data into the regional wheat-maize rotation system distributed growth localization model. Determine whether to modify the reference irrigation regime based on constraints. Grid points that do not meet any of the three constraints need to re-determine the irrigation regime, while grid points that meet the constraints follow the reference irrigation regime. The specific constraints are as follows: Time-based biomass constraints: ； In the formula, K is the correction coefficient; Bio is the biomass for the time period; Bio R For reference irrigation regimes, the biomass of the time period is used; Water supply constraints: ； In the formula, Q S For water supply; Q D This refers to water demand; Water demand decision error constraints: ； In the formula, I R P represents the total irrigation amount for the current period, P represents the rainfall for the current period, and ET represents the water demand for the current period. S305: For grids that require irrigation regime updates, the irrigation regime for the next 15 days is re-determined. To ensure the efficiency of the model operation, the particle swarm optimization algorithm is used to solve the model. With the goal of maximizing biomass during the time period and minimizing the water demand decision error, an adaptive correction decision model is constructed.

Images