An irrigation system operation optimization method and system based on digital twinning

Abstract

This invention relates to the field of data processing, specifically to a method and system for optimizing the operation of an irrigation system based on digital twins. The method includes: defining decision variables and optimization objectives based on collected data from a horticultural area; applying the NSGA-II algorithm for multi-objective optimization to obtain a non-dominated solution set; calculating the improvement potential of each non-dominated solution in the non-dominated solution set; selecting and performing corresponding genetic operations for each non-dominated solution based on the calculated improvement potential to generate a new generation population; using the new generation population as the current population, repeatedly performing the operations of obtaining non-dominated solutions, calculating improvement potential, and generating a new generation population until the iteration termination condition is met, outputting the final non-dominated solution set; and adjusting the operating parameters of the horticultural irrigation system based on the final non-dominated solution set. This invention transforms traditional "indiscriminate" random evolution into "differentiated" precise evolution, resulting in higher overall search efficiency.

Description

Technical Field

[0001] This invention relates to the field of data processing. More specifically, this invention relates to a method and system for optimizing the operation of an irrigation system based on digital twins. Background Technology

[0002] In smart agriculture, such as horticultural irrigation, achieving efficient water resource utilization and precise crop irrigation often requires comprehensive optimization of multiple operational objectives of the irrigation system (such as water conservation and moisture retention). Digital twin technology provides an effective platform for constructing virtual irrigation systems and conducting optimization simulations, while multi-objective optimization algorithms such as NSGA-II (Non-dominated Sorting Genetic Algorithm II) are often used on such platforms to solve Pareto optimal solutions, providing diverse irrigation strategy options.

[0003] However, when applying the NSGA-II algorithm for irrigation optimization in practice, its inherent genetic operation strategy has significant limitations: in each generation of evolution, the algorithm typically uses preset, indiscriminate crossover and mutation probabilities for solutions in the population. This approach does not consider the quality of the solutions themselves, leading to frequent random crossover and mutation of high-quality solutions that are close to the Pareto front. This not only easily generates duplicate or invalid solutions, wasting computational resources, but also significantly slows down the algorithm's convergence speed, making it difficult to meet the efficiency requirements of irrigation systems for real-time optimization decision-making. Summary of the Invention

[0004] To address the aforementioned technical problems, the present invention provides solutions in the following aspects.

[0005] In the first aspect, a method for optimizing the operation of an irrigation system based on digital twins includes:

[0006] Collect environmental and meteorological data of the horticultural area and preprocess the collected data;

[0007] Define the decision variables and optimization objectives for the optimization problem based on the collected data;

[0008] Based on the optimization objective, the NSGA-II algorithm is applied to perform multi-objective optimization on the decision variables to obtain a non-dominated solution set.

[0009] For each nondominated solution in the nondominated solution set, its improvement potential is calculated. The improvement potential is obtained by comprehensively calculating three indicators: the water distribution potential, the time efficiency potential, and the energy efficiency index of the corresponding nondominated solution.

[0010] Based on the calculated improvement potential, select and perform corresponding genetic operations for each non-dominated solution to generate a new generation of population;

[0011] Using the new generation population as the current population, repeatedly execute the operations of obtaining non-dominated solutions, calculating improvement potential, and generating a new generation population until the iteration termination condition is met, and output the final non-dominated solution set.

[0012] Adjust the operating parameters of the horticultural irrigation system based on the final non-dominated solution set.

[0013] Preferably, the collected data also includes facility operation data, the environmental data including soil moisture and soil pH value, the meteorological data including rainfall, air temperature, wind speed and humidity, and the facility operation data including real-time pipeline flow rate and pipeline pressure;

[0014] The decision variables include irrigation amount, irrigation duration, irrigation frequency, and irrigation time point, and the optimization objectives include water saving rate and soil moisture.

[0015] Preferably, the moisture distribution potential, time efficiency potential, and energy efficiency index are multiplied together to obtain a first product; the first product is then subjected to an exponential function operation to obtain a first exponent value.

[0016] The difference between 1 and the first exponent value is taken as the improvement potential.

[0017] Preferably, the acquisition of the moisture distribution potential includes:

[0018] Calculate the mean of irrigation amounts for all non-dominated solutions, and use the difference between the current irrigation amount for a non-dominated solution and the mean of irrigation amounts for all non-dominated solutions as the first factor. Divide the first factor by the current irrigation amount for a non-dominated solution to obtain the relative deviation. Use the difference between 1 and the normalized value of the relative deviation as the water distribution potential.

[0019] Preferably, acquiring the time efficiency potential includes:

[0020] Divide the current irrigation amount of the non-dominated solution by the pipeline flow rate to obtain the theoretical irrigation duration;

[0021] Calculate the absolute value of the difference between the current non-dominated solution's irrigation duration and the theoretical irrigation duration, and divide the absolute value by the theoretical irrigation duration to obtain the deviation ratio;

[0022] The time efficiency potential is obtained by subtracting the deviation ratio from 1.

[0023] Preferably, the energy efficiency index is obtained by:

[0024] Find the highest value of irrigation amount among all non-dominated solutions in the non-dominated solution set, and use the pipeline pressure corresponding to the highest value of irrigation amount as the reference pressure;

[0025] The energy efficiency index is obtained by dividing the irrigation amount of the current non-dominated solution by the difference between the pipeline pressure corresponding to the irrigation amount of the current non-dominated solution and the reference pressure.

[0026] Preferably, the types of genetic operations include performing crossover and mutation simultaneously, performing only crossover, or performing no genetic operations.

[0027] Preferably, performing the corresponding genetic operation includes:

[0028] If the improvement potential is greater than or equal to a preset first threshold, then perform crossover and mutation operations simultaneously.

[0029] If the improvement potential is greater than or equal to a preset second threshold and less than the first threshold, then perform a crossover operation only;

[0030] If the improvement potential is less than the second threshold, then no genetic operation is performed.

[0031] Preferably, the iteration termination condition is that the Jaccard similarity of two consecutive generations of non-dominated solution sets is greater than 0 or the preset maximum number of iterations is reached.

[0032] Secondly, an irrigation system operation optimization system based on digital twins includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the irrigation system operation optimization method based on digital twins described in any one of the claims is implemented.

[0033] The beneficial effects of this invention are:

[0034] 1. This invention transforms traditional "indiscriminate" random evolution into "differentiated" precise evolution by calculating the improvement potential of each non-dominated solution and intelligently selecting the genetic type. This mechanism concentrates computational resources on the solution with the greatest improvement potential while protecting solutions that are close to optimal, thereby significantly reducing ineffective iterations and wasted computational resources. This enables multi-objective optimization algorithms to converge to a high-quality Pareto optimal front at a faster speed, meeting the real-time decision-making needs of irrigation systems.

[0035] 2. This invention introduces a comprehensive evaluation system consisting of water distribution potential, time efficiency potential, and energy efficiency indicators. The algorithm can accurately identify the optimization state of each solution from multiple dimensions. Based on the evaluation results, dynamic decisions are made (e.g., performing crossover and mutation on high-potential solutions, only performing crossover on medium-potential solutions, and retaining high-quality solutions), achieving on-demand allocation of evolutionary computing power. This strategy avoids the destructive perturbation of high-quality solutions and insufficient exploration of inferior solutions caused by traditional fixed-probability operations, resulting in higher overall search efficiency.

[0036] 3. This invention achieves a closed loop from data acquisition and simulation optimization to control execution by embedding an improved optimization algorithm into the digital twin environment of a horticultural irrigation system. The final output non-dominated solution set provides a variety of irrigation strategies that achieve the optimal trade-off between water conservation and moisture retention, allowing users to flexibly choose according to actual conditions. This transforms irrigation decision-making from experience-driven to data and model-driven, improving the scientific rigor, accuracy, and system adaptability of the decisions. Attached Figure Description

[0037] Figure 1 This is a flowchart of steps S1-S4 in an irrigation system operation optimization method based on digital twins according to an embodiment of the present invention.

[0038] Figure 2 This is a structural block diagram of an irrigation system operation optimization system based on digital twins according to an embodiment of the present invention. Detailed Implementation

[0039] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0040] Reference Figure 1 A method for optimizing the operation of an irrigation system based on digital twins includes steps S1-S4, as detailed below:

[0041] S1: Collect multi-source data on horticulture and define the optimization problem.

[0042] Precise optimization requires establishing a digital model that accurately maps the state of the physical system and its external environment. This necessitates acquiring comprehensive, high-quality input data and formally defining the engineering optimization problem to be solved, providing reliable input and a clear search direction for subsequent algorithms.

[0043] First, collect data from multiple sources:

[0044] By deploying soil moisture sensors in the crop root zone, the soil volumetric water content U is acquired in real time. This parameter is the core basis for directly reflecting the available water for crops, calculating irrigation needs, and evaluating irrigation effects.

[0045] Soil pH values ​​are obtained by using a soil pH sensor. This parameter serves as an auxiliary environmental factor input, used to incorporate soil suitability assessments into the model and provide a reference for whether the irrigation-fertilization combined strategy needs to be adjusted.

[0046] By connecting to a meteorological data service interface, forecast data for the next irrigation decision cycle (e.g., the next 24 hours) is obtained, including rainfall, air temperature, wind speed, and relative humidity. This data is the key input driving the crop transpiration Z-prediction model, used to estimate crop water consumption in future periods.

[0047] The irrigation system's monitoring and data acquisition system reads facility operation data in real time, including pipeline flow rate (GL) and pipeline pressure (P). Pipeline flow rate is used to accurately calculate irrigation efficiency and theoretical operating time, while pipeline pressure is a core variable for evaluating system energy consumption and calculating energy efficiency.

[0048] The raw data collected above is preprocessed as follows: statistical filtering methods (such as sliding window and Z-score method) are used to remove sensor outliers; for missing data caused by communication interruption, linear interpolation of time series or previous value preservation method is used to fill in the missing data; finally, the timestamps and physical quantity units of all data are unified to form a high-quality, time-synchronized standardized dataset.

[0049] Then, based on the preprocessed data described above, the variables and objectives involved in the optimization are clarified, that is, the optimization problem is defined:

[0050] I. Decision Variables: These are defined as the directly controllable operating parameters of the irrigation system, including irrigation volume ( ), irrigation duration ( ), irrigation frequency ( ) and irrigation time ( Optimization is the process of finding the optimal combination of these variables.

[0051] II. Optimization Objectives: Establish two mutually restrictive core performance indicators, including water saving rate ( ) and soil moisture ( The priority of objectives can be adjusted according to actual needs.

[0052] Among them, water saving rate The specific calculation formula is as follows:

[0053]

[0054] In the formula, Theoretical water demand This represents the current soil volumetric water content. The upper limit of suitable humidity for crops, For soil volume, This is an estimate of crop transpiration based on meteorological data. Irrigation volume is a decision variable. This water-saving rate index measures the degree of water saving in actual irrigation relative to the theoretical crop demand.

[0055] Soil moisture The specific calculation formula is as follows:

[0056]

[0057] In the formula, This represents the current soil volumetric water content. For irrigation amount in the decision variables, For soil volume, This is the soil water holding capacity index (usually taken as 0.8). This soil moisture index predicts the amount of irrigation to be applied. After that, the soil will reach a certain level of moisture.

[0058] The above steps provide a precise data foundation and a clear optimization framework for the entire optimization process. By fusing environmental, meteorological, and facility operation data, the system can accurately perceive the status; by defining decision variables and optimization objectives, it provides clear and quantifiable inputs and evaluation criteria for subsequent algorithms, ensuring the mathematical rigor and correct direction of the optimization process.

[0059] S2: Perform NSGA-II optimization based on the defined optimization problem to generate a non-dominated solution set.

[0060] After defining the decision variables and optimization objectives in S1, an effective multi-objective optimization algorithm is needed to search the solution space for a set of high-quality solutions that can simultaneously balance multiple objectives. In this invention, the NSGA-II algorithm is used for solving the problem, and the solution process is as follows:

[0061] Initialize the population: Randomly generate a population containing Individuals (e.g.) The initial population. Each individual is encoded as a specific set of decision variable values, such as individual... .

[0062] Fast non-dominated sorting: For each individual in the population, according to the definition in S1 above... and Calculate its target value. Perform non-dominated ranking through pairwise comparisons. If individuals Not inferior to individuals in all objectives And at least in one objective, it is strictly superior to Then it is called Dominate The first non-dominated frontier is formed by selecting all individuals in the population that are not dominated by other individuals.

[0063] It should be noted that this invention focuses only on the first non-dominated frontier, taking all individuals in the first non-dominated frontier as the output non-dominated solution set. Each solution in this non-dominated solution set represents an irrigation strategy that, at the current search stage, cannot be simultaneously surpassed by other solutions in terms of both water conservation and moisture retention objectives, and possesses Pareto optimal characteristics.

[0064] The above steps utilize the NSGA-II algorithm to perform an efficient global search on the defined optimization problem. Its output non-dominated solution set provides a group of candidate strategies that are mutually non-dominant and each possesses its own advantages under the given objective. This solution set combines superior performance with diverse solutions, providing a high-quality starting point and a clear optimization target for subsequent intelligent and refined improvements.

[0065] S3: Evaluate the improvement potential of each non-dominated solution in the non-dominated solution set and drive intelligent genetic evolution accordingly.

[0066] The traditional NSGA-II algorithm applies fixed crossover and mutation probabilities to all individuals in the population. This "indiscriminate" operation wastes computational resources by perturbing solutions that are already close to optimal, while potentially failing to explore more promising solutions. To improve convergence efficiency, a mechanism is needed to quantify and evaluate the "optimization value" of each solution and implement differentiated operations accordingly.

[0067] This invention introduces "improvement potential" as an evaluation indicator and uses it to dynamically guide genetic evolution.

[0068] For the k-th non-dominated solution in the non-dominated solution set obtained from S2 above, its improvement potential is calculated. This value is obtained by comprehensively calculating the indices of three dimensions:

[0069] Calculate the mean irrigation amount of all non-dominated solutions. Calculate the difference between the irrigation amount of the k-th non-dominated solution and the mean irrigation amount of all non-dominated solutions as the first factor. Divide the first factor by the irrigation amount of the k-th non-dominated solution to obtain the relative deviation. The difference between 1 and the normalized value of the relative deviation is taken as the water distribution potential of the k-th non-dominated solution, which can be expressed as:

[0070]

[0071] In the formula, For the first The water distribution potential of a non-dominated solution For the first The relative deviation of a nondominated solution. This indicates normalization processing. Through calculation... , measuring the first The degree of deviation of the irrigation amount for a non-dominated solution from the population average is set. The higher the value, the better. The more "conventional" the irrigation amount setting of a non-dominated solution, the less potential problems there are in terms of the uniformity of water spatial distribution, and the lower the expected benefits of making significant changes to it; conversely, it suggests that the irrigation amount may be abnormal and there is clear room for optimization.

[0072] The first Dividing the irrigation amount of the first non-dominated solution by the real-time pipeline flow rate yields the theoretical irrigation duration; the calculation of the... The absolute value of the difference between the irrigation duration and the theoretical irrigation duration of the first non-dominated solution is used to divide the calculated absolute value by the theoretical irrigation duration to obtain the deviation ratio; subtracting the deviation ratio from 1 yields the result of the first solution. The time efficiency potential of the non-dominated solution is measured by calculating the time efficiency potential. The degree to which the irrigation duration set in a non-dominated solution matches the theoretically optimal duration indicates that the time efficiency potential is high, meaning that the time efficiency of the irrigation operation is close to the theoretical limit and there is little room for improvement in the time dimension; the lower the duration, the more likely there is significant redundancy or insufficiency in the duration setting, which is a key optimization entry point.

[0073] Find the highest irrigation amount among all non-dominated solutions in the non-dominated solution set, and use the pipeline pressure corresponding to the highest irrigation amount as the reference pressure; [The text abruptly ends here, likely due to an incomplete sentence or a formatting error.] The irrigation amount of the non-dominated solution divided by the first The difference between the pipeline pressure corresponding to the irrigation volume of the non-dominated solution and the aforementioned reference pressure is used to obtain the first... The energy efficiency index of the non-dominated solution. Among them, the first... The irrigation amount of the non-dominated solution represents the "irrigation output" of that solution. The difference between the pipeline pressure corresponding to the irrigation volume of a non-dominated solution and the aforementioned reference pressure represents the "extra pressure cost" incurred by that solution to achieve the output, exceeding the efficiency benchmark. Therefore, the... The energy efficiency index of the non-dominated solution, i.e., the irrigation gain obtained per unit of additional pressure consumption, is the... The higher the energy efficiency index of the first nondominated solution, the greater the irrigation effect achieved with relatively less additional pressure; its pressure utilization efficiency is higher, and its energy efficiency is better. Conversely, the lower the energy efficiency index of the second nondominated solution, the better the energy efficiency. The lower the energy efficiency index of a non-dominated solution, the higher its energy consumption cost and the lower its energy efficiency, making it a potential target for energy-saving optimization.

[0074] After calculating the number After determining the water distribution potential, time efficiency potential, and energy efficiency index of each non-dominated solution, these three positive indices, each ranging from 0 to 1, are further integrated into a comprehensive evaluation value between 0 and 1 – the improvement potential, which can be expressed as:

[0075]

[0076] In the formula, For the first The potential for improvement of a non-dominated solution. For the first The water distribution potential of a non-dominated solution For the first The time efficiency potential of a non-dominated solution. For the first Energy efficiency index of a non-dominated solution It is an exponential function with the natural number e as its base.

[0077] When all three indicators perform well Approaching 1 means that the first The non-dominated solutions exhibit excellent overall performance, but have limited potential for further improvement; any low value in any one of these indicators will lead to... A significant decrease indicates that the first Each non-dominated solution has a weakness in a certain aspect, but has high potential for overall improvement.

[0078] Furthermore, according to the above-mentioned... The process of calculating the improvement potential of a non-dominated solution is similar to calculating the improvement potential of all other non-dominated solutions in the non-dominated solution set.

[0079] Furthermore, based on the quantification results of the calculated improvement potential, a genetic operation is adaptively selected and performed for each non-dominated solution in the non-dominated solution set:

[0080] like If the value is greater than or equal to a preset first threshold (e.g., 0.8), then the first value is determined. The non-dominated solution has great potential for improvement (i.e., its current quality is relatively poor). We should carry out high-intensity exploration, that is, crossover and mutation at the same time, in order to make the solution jump out of the current poor quality region and find a performance breakthrough through drastic gene recombination and perturbation.

[0081] like If the value is greater than or equal to a preset second threshold (e.g., 0.5) and less than a first threshold, then the condition is determined to be... The non-dominated solution has moderate improvement potential; performing only crossover operations, such as simulated binary crossover, can improve the performance of the first solution. The non-dominated solution exchanges genes with other excellent solutions, aiming to merge and generate new characteristics with greater advantages, while avoiding the performance degradation risk that may be caused by strong random mutation.

[0082] like If it is less than the second threshold, determine the first The non-dominated solution is already very close to Pareto optimality and has low improvement potential. Implementing protective preservation, that is, without performing any genetic operations, directly preserves it completely to the next generation population, protecting its excellent genetic structure from being destroyed by random perturbations, avoiding invalid computation, and consolidating the optimal frontier already obtained by the algorithm.

[0083] A new set of individuals is generated through the above intelligent genetic operations.

[0084] In summary, the traditional NSGA-II algorithm employs pre-set fixed probabilities for crossover and mutation, which is essentially an indiscriminate and blind random perturbation strategy. This strategy has obvious drawbacks: First, it performs unnecessary crossover and mutation on high-quality solutions that are already close to the Pareto front, which easily generates a large number of duplicate solutions or inferior nearest neighbor solutions, resulting in a significant waste of computational resources and ineffective iterations; Second, it does not explore low-potential inferior solutions sufficiently, making it difficult to guide the population out of local optima.

[0085] This invention introduces "improvement potential" as a dynamic quantitative evaluation index and constructs an "evaluation-decision" mechanism based on this index, achieving a fundamental shift from "blind random operation" to "intelligent and precise operation." Specifically, this mechanism accurately diagnoses the state of each non-dominated solution through water distribution potential, time efficiency potential, and energy efficiency indicators, and performs differentiated genetic operations based on the diagnosis results: high-quality solutions are protectively shelved, effectively avoiding the waste of valuable computing resources and the destruction of excellent gene structures; medium-potential solutions are mildly improved to promote the fusion of excellent traits; and high-potential inferior solutions are aggressively explored, concentrating computing power to open up new regions. This intelligent strategy of allocating resources on demand and adapting operations to solutions fundamentally overcomes the inherent defects of traditional algorithms; and greatly reduces redundant computation and ineffective exploration in the population evolution process, significantly improving the search efficiency and directionality of the solution space, thereby enabling the algorithm to converge to a wider and higher-quality global Pareto optimal front in fewer iterations.

[0086] S4: Control the optimization iteration until convergence, output and execute the optimal irrigation strategy.

[0087] The optimization process is a search process that approximates the optimal solution through repeated iterations (S2 / S3 loops). Therefore, there is a need for clear criteria to determine when the search can terminate (converge) and to reliably map and apply the optimal strategy found in the digital twin world to control the physical system, thereby completing the closed loop of "perception-optimization-control".

[0088] The new set of individuals generated by the intelligent genetic operation in S3 is taken as the new generation population. Then, the system automatically repeats the steps of S2 (performing non-dominated sorting based on the new generation population and outputting a new solution set) and S3 (evaluating the new solution set and intelligently driving evolution), forming a continuous iterative optimization loop.

[0089] After each iteration, it automatically checks whether any of the following preset termination conditions are met:

[0090] Condition 1: Calculate the Jaccard similarity between two consecutive generations of non-dominated solution sets (the first non-dominated front). The Jaccard similarity is defined as the ratio of the size of the intersection to the size of the union of the two generations of non-dominated solution sets. If the Jaccard similarity is greater than 0 (in practice, it can be set to be greater than a very small threshold, such as 0.05), then the optimal solution set is considered to have stabilized and the algorithm has converged.

[0091] Condition 2: The number of iterations reaches the preset maximum limit (e.g., 100 generations).

[0092] When the iteration termination condition is triggered, the algorithm immediately stops and outputs the final non-dominated solution set to the strategy knowledge base of the digital twin platform. Users can interactively select the current optimal irrigation strategy from the final solution set based on real-time operating conditions and preferences (e.g., prioritizing water conservation during periods of water scarcity and moisture conservation during droughts). The system also supports fully automated decision-making based on preset rules (e.g., always selecting the solution with the highest water-saving rate).

[0093] The digital twin system will select the decision variables of the strategy ( , , , The control commands are transformed into specific control instruction sets and sent to the programmable logic controller (PLC) of the physical irrigation system via the Industrial Internet of Things (IIoT) protocol. The PLC then precisely drives the pumps, solenoid valves, and other actuators according to the instructions to complete the irrigation operation.

[0094] The above steps achieve automated management and closed-loop decision execution throughout the entire optimization process. A scientific convergence judgment mechanism ensures optimization efficiency and result quality. Ultimately, by seamlessly and reliably distributing the optimal strategy obtained from virtual space optimization to the physical irrigation system for execution, a complete closed-loop digital twin application is realized, encompassing environmental perception, data fusion, intelligent optimization simulation, and precise control of the physical system. This enables the irrigation system to operate independently of experience, continuously and dynamically maintaining its optimal state of high efficiency, water conservation, and precision.

[0095] This invention also provides an irrigation system operation optimization system based on digital twins. For example... Figure 2As shown, the system includes a processor and a memory, the memory storing computer program instructions, which, when executed by the processor, implement the digital twin-based irrigation system operation optimization method according to the first aspect of the present invention.

[0096] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0097] It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of this invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for optimizing the operation of an irrigation system based on digital twins, characterized in that, include: Collect environmental and meteorological data of the horticultural area and preprocess the collected data; Define the decision variables and optimization objectives for the optimization problem based on the collected data; Based on the optimization objective, the NSGA-II algorithm is applied to perform multi-objective optimization on the decision variables to obtain a non-dominated solution set. For each nondominated solution in the nondominated solution set, its improvement potential is calculated. The improvement potential is obtained by comprehensively calculating three indicators: the water distribution potential, the time efficiency potential, and the energy efficiency index of the corresponding nondominated solution. Calculate the mean of irrigation amounts for all non-dominated solutions, and use the difference between the current irrigation amount for a non-dominated solution and the mean of irrigation amounts for all non-dominated solutions as the first factor; divide the first factor by the current irrigation amount for a non-dominated solution to obtain the relative deviation, and use the difference between 1 and the normalized value of the relative deviation as the water distribution potential. Divide the current irrigation amount of the non-dominated solution by the pipeline flow rate to obtain the theoretical irrigation duration; Calculate the absolute value of the difference between the current non-dominated solution's irrigation duration and the theoretical irrigation duration, and divide the absolute value by the theoretical irrigation duration to obtain the deviation ratio; The time efficiency potential is obtained by subtracting the deviation ratio from 1. Find the highest value of irrigation amount among all non-dominated solutions in the non-dominated solution set, and use the pipeline pressure corresponding to the highest value of irrigation amount as the reference pressure; The energy efficiency index is obtained by dividing the irrigation amount of the current non-dominated solution by the difference between the pipeline pressure corresponding to the irrigation amount of the current non-dominated solution and the reference pressure. Based on the calculated improvement potential, select and perform corresponding genetic operations for each non-dominated solution to generate a new generation of population; Using the new generation population as the current population, repeatedly execute the operations of obtaining non-dominated solutions, calculating improvement potential, and generating a new generation population until the iteration termination condition is met, and output the final non-dominated solution set. Adjust the operating parameters of the horticultural irrigation system based on the final non-dominated solution set.

2. The method for optimizing the operation of an irrigation system based on digital twins according to claim 1, characterized in that, The collected data also includes facility operation data, the environmental data including soil moisture and soil pH, the meteorological data including rainfall, air temperature, wind speed and humidity, and the facility operation data including real-time pipeline flow and pipeline pressure; The decision variables include irrigation amount, irrigation duration, irrigation frequency, and irrigation time point, and the optimization objectives include water saving rate and soil moisture.

3. The method for optimizing the operation of an irrigation system based on digital twins according to claim 2, characterized in that, Multiply the water distribution potential, time efficiency potential, and energy efficiency index to obtain the first product; then perform an exponential function operation on the first product to obtain the first exponent value. The difference between 1 and the first exponent value is taken as the improvement potential.

4. The method for optimizing the operation of an irrigation system based on digital twins according to claim 1, characterized in that, The types of genetic operations include performing crossover and mutation simultaneously, performing crossover only, or performing no genetic operations.

5. The method for optimizing the operation of an irrigation system based on digital twins according to claim 1, characterized in that, The iteration termination condition is that the Jaccard similarity of two consecutive generations of non-dominated solution sets is greater than 0 or the preset maximum number of iterations is reached.

6. A digital twin-based irrigation system operation optimization system, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement the irrigation system operation optimization method based on digital twins according to any one of claims 1-5.

Images