Optimal method for water distribution in three-level canal system based on the third generation non-dominated sorting genetic algorithm
Through the third generation of non-dominant sorting genetic algorithm, multi-objective optimization functions and reference points are constructed, which solves the problem that water distribution optimization in the irrigation zone canal system is difficult to achieve the allocation of water resources in three-level canal system in the existing technology, and realizes efficient water resources management and Irrigation efficiency is improved.
Patent Information
- Application Number
- CN202310952388.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-08-01
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2043-08-01
AI Technical Summary
The existing water distribution optimization method for irrigation zone canal system is difficult to effectively solve the problem of water resource optimization in the third-level canal system, especially when optimizing high-dimensional targets, the convergence speed is slow, the accuracy is low, and the water resource allocation at multiple angles cannot be fully considered.
The third-generation non-dominant sorting genetic algorithm is adopted to construct multi-objective optimization functions, and the population diversity is maintained using well-distributed reference points, and the water distribution scheme of the three-level canal system is optimized, including the water resource allocation model of the stem, branch, and bucket canal to the fields. Set constraints such as flow, time, and water volume balance, and use non-dominant sorting genetic algorithm to solve them.
The algorithm convergence speed and optimization efficiency have been improved, the optimization allocation of water resources in the third-level canal system has been achieved, water conservation and energy saving, irrigation efficiency has been improved, and the Changgang irrigation area has been provided with optimized water distribution decision support.
Smart Images

Figure CN116796997B_ABST
Abstract
Description
Technical Field
[0001] The invention provides a three-level canal system water distribution optimization method based on a third-generation non-dominated sorting genetic algorithm, belonging to the technical field of agricultural irrigation. Background Art
[0002] At present, there have been some research results at home and abroad in the study of irrigation canal water distribution schemes. For example, genetic algorithms or particle swarm algorithms are used to solve the problem of optimizing water distribution in two-level channels. However, in previous studies, most multi-objective evolutionary algorithms are more effective when solving problems with lower target dimensions. Most canal water distribution optimization models are generalized models, which are theoretically feasible but difficult to use in practice, and do not reflect the optimal allocation of water resources in the three-level canal system from trunk, branch, and ditches to the field.
[0003] For example, CN116108942A provides an irrigation area water resources optimization allocation method based on an artificial bee colony algorithm, which uses an artificial bee colony algorithm to optimize the allocation of irrigation area water resources, takes the maximum economic benefit of the irrigation area as the goal, establishes an optimal water distribution model under multiple constraints and solves the model. This method has the following disadvantages: 1. The artificial bee colony algorithm used has a slow convergence speed, low accuracy, and is too sensitive to parameters, and needs to be adjusted according to different problems. 2. The objective function in the model is only the maximum economic benefit, which is relatively simple and cannot comprehensively consider the irrigation area water resources allocation from multiple angles.
[0004] The canal water distribution optimization model based on multi-objective genetic algorithm (Journal of China Agricultural University, 2017, No. 07, Guo Shanshan) uses vector evaluation genetic algorithm to solve the objective function of the most stable water flow in the upper channel and the minimum water loss in the two-level canals of the irrigation area. This method has the following disadvantages: 1. The solution searched by the vector evaluation genetic algorithm only tends to the optimal individual of a single objective function; 2. It can solve the water distribution optimization of the two-level canal system of the main canal and branch canal, and cannot reach the ditches or fields. Summary of the invention
[0005] The purpose of the present invention is to use the third generation non-dominated sorting genetic algorithm to maintain the diversity of the population and accelerate the convergence and optimization efficiency of the algorithm when facing the high-dimensional target optimization problem of irrigation canal water distribution. At the same time, it embodies the three-level canal water resource optimization configuration model of the main, branch, and ditches to the field, saves water and energy, improves irrigation efficiency, provides an optimized water distribution plan, and actually applies the model to the Changgang irrigation area to provide decision support for the water supply and distribution of the Changgang irrigation area.
[0006] The specific technical solution provided by the present invention is:
[0007] The three-level canal water distribution optimization method based on the third-generation non-dominated sorting genetic algorithm includes the following steps:
[0008] S1. Collect necessary irrigation district data;
[0009] S2. Construct a paddy field water balance model and calculate the target irrigation time based on the rice growth period division and paddy field depth table in Changgang irrigation area;
[0010] S3, construct a multi-objective optimization function to meet the timeliness requirements of irrigation in different channels and achieve high efficiency in the water distribution process;
[0011] S4. Setting the model boundary. The model boundary constraints include flow constraint, time constraint, water balance constraint, and integer 0-1 constraint;
[0012] S5. Based on the above-mentioned model, the third-generation non-dominated sorting genetic algorithm is used to solve the model, and the optimized water distribution plan provides a reference for the actual water distribution in Changgang Irrigation District.
[0013] The necessary irrigation district data collected in S1 include;
[0014] (1) Design flow of 1 main canal, 18 branch canals and 159 ditches, m 3 / s;
[0015] (2) Length of 1 main canal, 18 branch canals and 159 ditches, km;
[0016] (3) Collect the effective rainfall of the control field of the jth lower channel in the tth time unit, mm;
[0017] (4) Collect the paddy field leakage of the jth lower-level channel control field in the tth time unit, mm;
[0018] (5) Collect the crop coefficient Kc, soil water stress factor Ks, and crop evapotranspiration ET0 at each growth stage of rice.
[0019] In S2, the target irrigation time is calculated;
[0020] H jt' =P jt +IR jt +H jt -ETa jt -SP jt -D jt
[0021] In the formula, H jt' represents the water depth of paddy fields in the area controlled by the jth lower channel at the end of the calculation period, mm; the calculation time unit in this model is 4h, the same below;
[0022] IR jt It represents the irrigation amount of the field controlled by the jth subordinate channel in the tth time unit, mm;
[0023] P jt represents the effective rainfall in the field controlled by the jth subordinate channel in the tth time unit, mm;
[0024] ETa represents the actual evapotranspiration of the field controlled by the jth subordinate channel in the tth time unit, mm;
[0025] SP jt represents the paddy field leakage of the jth subordinate channel control field in the tth time unit, mm;
[0026] D jt It represents the drainage volume of the field controlled by the jth subordinate channel in the tth time unit, mm.
[0027] S3 specifically includes:
[0028] (1) Calculate the target irrigation time of the ditches based on the above paddy field water balance model, and construct an objective function that minimizes the difference between the target irrigation time and the actual irrigation time, so as to determine the supply according to demand and improve irrigation efficiency;
[0029]
[0030] Where j is the channel number, J is the total number of channels, It j * and It j The target irrigation time and the actual irrigation time of the j-th canal are respectively;
[0031] (2) Construct a function that minimizes the fluctuation of water flow in the upper channel to ensure the stability of water flow in the upper channel;
[0032]
[0033] In the formula, Q tu is the water flow rate of the upper channel during time period t, m 3 / s; is the average water flow rate of the upper channel during the entire T period, m 3 / s; T is the number of time periods;
[0034]
[0035] In the formula, q' tj is the water flow at the inlet of the j-th water distribution channel; q tj is the net water distribution flow of the jth water distribution channel;
[0036] f tj (x) is a state variable expressed as:
[0037]
[0038] Indicates whether the jth subordinate channel is filled with water in the kth time period;
[0039] (3) Establish an objective function to minimize water loss during the entire water distribution process, control seepage loss and improve the water utilization coefficient of the canal system;
[0040]
[0041] Where S is the total amount of water seepage in the water distribution channel, m 3 ; A is the soil permeability coefficient of the canal bed, which is generally 0.7-3.4, determined according to soil characteristics, and set to 1.9; m is the soil permeability index of the canal bed, which is 0.3-0.5, determined according to soil characteristics, and can be set to 0.4; L u is the water conveyance length of the upper channel, km, determined by searching data; l j is the water delivery length of the lower channel, km, determined by searching data; Q tu is the net water distribution flow of the upstream channel in t time periods, m 3 / s, calculated by adding formula and data; q tj is the net water flow of the lower channel during time period t, m 3 / s; β is the reduction coefficient of the leakage volume after anti-seepage measures are taken, refer to "Agricultural Water Conservancy" or other references, and take the value within the range; t u ,t i ,t j are the water distribution durations for the main, branch and ditch channels respectively;
[0042] (4) Use the Theil Index to determine fairness based on the amount of water in each channel to ensure fairness in water distribution;
[0043]
[0044] (5) Construct an objective function to minimize the energy loss of irrigation gates during water distribution and transmission to save energy and improve efficiency;
[0045]
[0046] Where P i , P j are the gate powers of branch and ditch channels respectively;
[0047] (6) Construct an objective function with the minimum drainage volume in the irrigation area, consider the impact of rainfall on paddy field water depth, make full use of rainfall, and improve agricultural water productivity;
[0048]
[0049] Where D is the total discharge during the cycle;
[0050] The constraints of S4 are;
[0051] α d q j ≤q tj ≤α u q j
[0052] γ d Q u ≤Q tu ≤γ u Q u
[0053] 0≤S j <S j +I tj ≤T
[0054]
[0055] In the formula, q j is the design water flow of the jth water distribution channel; Q u is the designed water flow of the water conveyance channel; S j is the water distribution start time of the jth water distribution channel; Sum water is the allowable water supply in the irrigation area; α d and α u is the minimum and maximum coefficient of the flow fluctuation range of the jth lower channel, which is set to 0.6 and 1.2; γ d and γ u The minimum and maximum coefficients of the main channel flow fluctuation range are set to 0.6 and 1.2; f tj (x) indicates whether the jth subordinate channel is filled with water in the kth time period.
[0056] S5 specifically includes:
[0057] (1) Setting of reference points
[0058] in accordance with Calculate the number of reference points, where M is the number of targets and p is the number of divisions in each direction;
[0059] (2) Adaptive standardization of population
[0060] ① Find the ideal point: that is, find the minimum value of all objectives of this generation of population;
[0061] a. Select the minimum value of each dimension target of individuals in the current population St The ideal point that constitutes the current population
[0062] b. Translate the population St Make the ideal point become the origin;
[0063] ②Calculate extreme points:
[0064] Take the point corresponding to the minimum value of the above scalar function; where w is the unit direction vector of the coordinate axis;
[0065] ③Construct the hyperplane, find the intercept, and normalize the target;
[0066] The three lines formed by the extreme points and the origin, i.e. the ideal point, form a surface. The intersection of this surface and the three coordinate axes is the intercept a1, a2 and a3 that need to be solved in the end. After finding the intercept, use the intercept to normalize according to the following equation;
[0067]
[0068] (3) Association operation
[0069] To associate the individuals in the population with the corresponding reference points, define a reference line, which is the line connecting the origin and the reference point in the target space; calculate the vertical distance from each individual in the population to the reference line, and associate the individual with the reference line closest to it;
[0070] (4) Individual retention operation
[0071] ①First select the reference point J with the least number of associations min If there are multiple such reference points, one is randomly selected from them. As reference point the individual being contacted; is empty, reselect the reference point; if If it is not empty, then look at Is it 0?
[0072] ②If from Select distance The smallest individual enters the next generation; if from Randomly select an individual to enter the next generation; ρ j :Population P t+1 =S t / F l , that is, F l The number of individuals associated with reference point j in all non-dominated layers before the layer;
[0073] ③ Repeat the operation until the size of the next generation is equal to N.
[0074] When facing the high-dimensional target optimization problem of irrigation canal water distribution, the present invention uses the third-generation non-dominated sorting genetic algorithm, uses well-distributed reference points to maintain the diversity of the population, and accelerates the convergence of the algorithm and the optimization efficiency. At the same time, it embodies the three-level canal water resource optimization configuration model from the main, branch, and ditches to the fields, saves water and energy, improves irrigation efficiency, provides an optimized water distribution plan, and actually applies the model to the Changgang irrigation area to provide decision support for the water distribution in the Changgang irrigation area. BRIEF DESCRIPTION OF THE DRAWINGS
[0075] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION
[0076] The specific technical solution of the present invention is described in conjunction with the accompanying drawings.
[0077] like Figure 1 As shown, the three-level canal water distribution optimization method based on the third-generation non-dominated sorting genetic algorithm includes the following steps:
[0078] S1. Collect necessary irrigation district data including;
[0079] (1) Design flow of 1 main canal, 18 branch canals and 159 ditches, m 3 / s;
[0080] (2) Length of 1 main canal, 18 branch canals and 159 ditches, km;
[0081] (3) Collect the effective rainfall of the control field of the jth lower channel in the tth time unit, mm;
[0082] (4) Collect the paddy field leakage of the jth lower-level channel control field in the tth time unit, mm;
[0083] (5) Collect the crop coefficient Kc, soil water stress factor Ks, and crop evapotranspiration ET0 at each growth stage of rice.
[0084] S2. Construct a paddy field water balance model and calculate the target irrigation time based on the rice growth period division and paddy field depth table in Changgang irrigation area;
[0085] In S2, the target irrigation time is calculated;
[0086] H jt' =P jt +IR jt +H jt -ETa jt -SP jt -D jt
[0087] In the formula, H jt'represents the water depth of paddy fields in the area controlled by the jth lower channel at the end of the calculation period, mm; the calculation time unit in this model is 4h, the same below;
[0088] IR jt It represents the irrigation amount of the field controlled by the jth subordinate channel in the tth time unit, mm;
[0089] P jt represents the effective rainfall in the field controlled by the jth subordinate channel in the tth time unit, mm;
[0090] ETa represents the actual evapotranspiration of the field controlled by the jth subordinate channel in the tth time unit, mm;
[0091] SP jt represents the paddy field leakage of the jth subordinate channel control field in the tth time unit, mm;
[0092] D jt It represents the drainage volume of the field controlled by the jth subordinate channel in the tth time unit, mm.
[0093] S3, construct a multi-objective optimization function to meet the timeliness requirements of irrigation in different channels and achieve high efficiency in the water distribution process;
[0094] S3 specifically includes:
[0095] (1) Calculate the target irrigation time of the ditches based on the above paddy field water balance model, and construct an objective function that minimizes the difference between the target irrigation time and the actual irrigation time, so as to determine the supply according to demand and improve irrigation efficiency;
[0096]
[0097] Where j is the channel number, J is the total number of channels, It j * and It j The target irrigation time and the actual irrigation time of the j-th canal are respectively;
[0098] (2) Construct a function that minimizes the fluctuation of water flow in the upper channel to ensure the stability of water flow in the upper channel;
[0099]
[0100] In the formula, Q tu is the water flow rate of the upper channel during time period t, m 3 / s; is the average water flow rate of the upper channel during the entire T period, m 3 / s; T is the number of time periods;
[0101]
[0102] In the formula, q' tj is the water flow at the inlet of the j-th water distribution channel; q tj is the net water distribution flow of the jth water distribution channel;
[0103] f tj (x) is a state variable expressed as:
[0104]
[0105] Indicates whether the jth subordinate channel is filled with water in the kth time period;
[0106] (3) Establish an objective function to minimize water loss during the entire water distribution process, control seepage loss and improve the water utilization coefficient of the canal system;
[0107]
[0108] Where S is the total amount of water seepage in the water distribution channel, m 3 ; A is the soil permeability coefficient of the canal bed, which is generally 0.7-3.4, determined according to soil characteristics, and set to 1.9; m is the soil permeability index of the canal bed, which is 0.3-0.5, determined according to soil characteristics, and can be set to 0.4; L u is the water conveyance length of the upper channel, km, determined by searching data; l j is the water delivery length of the lower channel, km, determined by searching data; Q tu is the net water distribution flow of the upstream channel in t time periods, m 3 / s, calculated by adding formula and data; q tj is the net water flow of the lower channel during time period t, m 3 / s; β is the reduction coefficient of the leakage volume after anti-seepage measures are taken, refer to "Agricultural Water Conservancy" or other references, and take the value within the range; t u ,t i ,t j are the water distribution durations for the main, branch and ditch channels respectively;
[0109] (4) Use the Theil Index to determine fairness based on the amount of water in each channel to ensure fairness in water distribution;
[0110]
[0111] (5) Construct an objective function to minimize the energy loss of irrigation gates during water distribution and transmission to save energy and improve efficiency;
[0112]
[0113] Where P i, P j are the gate powers of branch and ditch channels respectively;
[0114] (6) Construct an objective function with the minimum drainage volume in the irrigation area, consider the impact of rainfall on paddy field water depth, make full use of rainfall, and improve agricultural water productivity;
[0115]
[0116] Where D is the total discharge during the cycle;
[0117] S4. Setting model boundaries: Model boundary constraints include flow constraints, time constraints, water balance constraints, and integer 0-1 constraints;
[0118] The constraints of S4 are;
[0119] α d q j ≤q tj ≤α u q j
[0120] γ d Q u ≤Q tu ≤γ u Q u
[0121] 0≤S j <S j +I tj ≤T
[0122]
[0123]
[0124] In the formula, q j is the design water flow of the jth water distribution channel; Q u is the designed water flow of the water conveyance channel; S j is the water distribution start time of the jth water distribution channel; Sum water is the allowable water supply in the irrigation area; α d and α u is the minimum and maximum coefficient of the flow fluctuation range of the jth lower channel, which is set to 0.6 and 1.2; γ d and γ u The minimum and maximum coefficients of the main channel flow fluctuation range are set to 0.6 and 1.2; f tj (x) indicates whether the jth subordinate channel is filled with water in the kth time period.
[0125] S5. Based on the above-mentioned model, the third-generation non-dominated sorting genetic algorithm is used to solve the model, and the optimized water distribution plan provides a reference for the actual water distribution in Changgang Irrigation District.
[0126] S5 specifically includes:
[0127] (1) Setting of reference points
[0128] in accordance with Calculate the number of reference points, where M is the number of targets and p is the number of divisions in each direction;
[0129] (2) Adaptive standardization of population
[0130] ① Find the ideal point: that is, find the minimum value of all objectives of this generation of population;
[0131] a. Select the minimum value of each dimension target of individuals in the current population St The ideal point that constitutes the current population
[0132] b. Translate the population St Make the ideal point become the origin;
[0133] ②Calculate extreme points:
[0134] Take the point corresponding to the minimum value of the above scalar function; where w is the unit direction vector of the coordinate axis;
[0135] ③Construct the hyperplane, find the intercept, and normalize the target;
[0136] The three lines formed by the extreme points and the origin, i.e. the ideal point, form a surface. The intersection of this surface and the three coordinate axes is the intercept a1, a2 and a3 that need to be solved in the end. After finding the intercept, use the intercept to normalize according to the following equation;
[0137]
[0138] (3) Association operation
[0139] To associate the individuals in the population with the corresponding reference points, define a reference line, which is the line connecting the origin and the reference point in the target space; calculate the vertical distance from each individual in the population to the reference line, and associate the individual with the reference line closest to it;
[0140] (4) Individual retention operation
[0141] ①First select the reference point J with the least number of associations min If there are multiple such reference points, one is randomly selected from them. As reference point the individual being contacted; is empty, reselect the reference point; if If it is not empty, then look at Is it 0?
[0142] ②If from Select distance The smallest individual enters the next generation; if from Randomly select an individual to enter the next generation; ρ j :Population P t+1 =S t / F l , that is, F l The number of individuals associated with reference point j in all non-dominated layers before the layer;
[0143] ③ Repeat the operation until the size of the next generation is equal to N.
[0144] When facing the high-dimensional target optimization problem of irrigation canal water distribution, the present invention uses the third-generation non-dominated sorting genetic algorithm, uses well-distributed reference points to maintain the diversity of the population, and accelerates the convergence of the algorithm and the optimization efficiency. At the same time, it comprehensively considers the actual situation of the irrigation area from the aspects of irrigation time, water transmission loss, water flow conditions of the upper channel, gate loss energy, water distribution fairness, and drainage volume, and embodies the three-level canal water resource optimization configuration model from the main, branch, and ditches to the field, saving water and energy, improving irrigation efficiency, and providing an optimized water distribution plan.
Claims
1. A three-level canal water distribution optimization method based on the third-generation non-dominated sorting genetic algorithm, characterized in that: The following steps are involved: S1. Collect irrigation district data; S2. Construct a paddy field water balance model and calculate the target irrigation time based on the rice growth period division and paddy field depth table in Changgang irrigation area; S3, construct a multi-objective optimization function to meet the timeliness requirements of irrigation in different channels and achieve high efficiency in the water distribution process; S4. Setting the model boundary. The model boundary constraints include flow constraint, time constraint, water balance constraint, and integer 0-1 constraint; S5. A three-level canal water distribution optimization method based on the third-generation non-dominated sorting genetic algorithm. The model is solved using the third-generation non-dominated sorting genetic algorithm. The optimized water distribution scheme provides a reference for the actual water distribution in Changgang Irrigation District. Among them, the irrigation area data collected in S1 include: (1) Design flow of 1 main canal, 18 branch canals and 159 ditches; (2) the length of 1 main canal, 18 branch canals, and 159 ditches; (3) Collect the effective rainfall of the field controlled by the jth canal in the tth time unit; (4) Collect the paddy field leakage of the jth canal control field in the tth time unit; (5) Collect the crop coefficient Kc, soil water stress factor Ks, and crop evapotranspiration ET0 at each growth stage of rice; Calculation of paddy field water depth in paddy field depth table in S2: H jt' =P jt +IR jt +H jt -ETa jt -SP jt -D jt In the formula, H jt' It represents the water depth of the paddy field controlled by the j-th canal at the end of the calculation time unit; the calculation time unit is 4h; IR jt It represents the irrigation volume of the field controlled by the j-th canal in the t-th time unit; P jt It represents the effective rainfall of the field controlled by the j-th canal in the t-th time unit; ETa jt It represents the actual evapotranspiration of the field controlled by the j-th canal in the t-th time unit; SP jt It represents the paddy field leakage of the j-th canal-controlled field in the t-th time unit; D jt It represents the drainage volume of the field controlled by the j-th canal in the t-th time unit; S3 specifically includes: (1) Calculate the target irrigation time of the ditches based on the paddy field water balance model in S2, and construct an objective function Z that minimizes the difference between the target irrigation time and the actual irrigation time, so as to determine the supply according to demand and improve irrigation efficiency; Where j is the channel number, J is the total number of channels, It j * and It j The target irrigation time and the actual irrigation time of the j-th canal are respectively; (2) Construct a function f with the goal of minimizing the fluctuation of water flow in the main canal to ensure the stability of water flow in the main canal: In the formula, Q tu is the water flow rate of the main canal in the tth time unit; is the average water flow rate of the main canal over the entire T time units; T is the number of time units: In the formula, q' tj is the water flow at the inlet of the j-th ditch in the t-th time unit; q tj is the net water distribution flow of the j-th ditch in the t-th time unit; f tj (x) is a state variable, expressed as: Indicates whether the j-th canal is irrigated in the x-th time unit; (3) Construct the objective function S with the minimum water loss in the entire water distribution process to control the seepage loss and improve the water utilization coefficient of the canal system: Where S is the total amount of seepage in the canal; A is the soil permeability coefficient of the canal bed, which is set to 1.9; m is the soil permeability coefficient index of the canal bed, which is set to 0.4; L u is the water delivery length of the main canal, which can be determined by searching the data; l j is the water conveyance length of the ditch, which is determined by searching data; β is the reduction coefficient of the leakage volume after anti-seepage measures are taken; t u ,t i ,t j are the water distribution durations for the main, branch and ditch channels respectively; (4) Use the Theil Index to determine fairness based on the amount of water in each channel to ensure fairness in water distribution: (5) Construct an objective function W with the minimum energy loss of the irrigation gate during water distribution to save energy and improve efficiency: Where P i , P j are the gate powers of branch and ditch channels respectively; (6) Construct the objective function D with the minimum drainage volume in the irrigation area, consider the impact of rainfall on the water depth of paddy fields, make full use of rainfall, and improve agricultural water productivity: Where D is the total discharge during the cycle.
2. The three-level canal water distribution optimization method based on the third-generation non-dominated sorting genetic algorithm according to claim 1 is characterized in that: The constraints of S4 are: α d q j ≤q tj ≤α u q j γ d Q u ≤Q tu ≤γ u Q u 0≤S j <S j +I tj ≤T In the formula, q j is the design water flow of the jth ditch; Q u is the design water flow of the main canal; S j is the start time of water distribution for the jth canal; Sum water is the allowable water supply in the irrigation area; α d and α u are the minimum and maximum coefficients of the flow fluctuation range of the jth ditch, which are set to 0.6 and 1.2 respectively; d and γ u are the minimum and maximum coefficients of the main canal flow fluctuation range, which are set to 0.6 and 1.2 respectively; f tj (x) indicates whether the jth canal is irrigated in the xth time unit.
3. The three-level canal water distribution optimization method based on the third-generation non-dominated sorting genetic algorithm according to claim 2 is characterized in that: S5 specifically includes: (1) Setting of reference points in accordance with Calculate the number of reference points, where M is the number of targets and p is the number of divisions in each direction; (2) Adaptive Standardization of Population ① Find the ideal point: that is, find the minimum value of all objectives of this generation of population; a. Select the minimum value of each dimension target of individuals in the current population St The ideal point that constitutes the current population ; b. Translate the population St Make the ideal point become the origin; ②Calculate extreme points: Take the point corresponding to the minimum value of the objective function; where w is the unit direction vector of the coordinate axis; ③Construct the hyperplane, find the intercept, and normalize the target; The three lines formed by the extreme points and the origin, i.e. the ideal point, form a surface. The intersection of this surface and the three coordinate axes is the intercept a1, a2 and a3 that need to be solved in the end. After finding the intercept, use the intercept to normalize according to the following equation: (3) Association Operation To associate the individuals in the population with the corresponding reference points, define a reference line, which is the line connecting the origin and the reference point in the target space; calculate the vertical distance from each individual in the population to the reference line, and associate the individual with the reference line closest to it; (4) Individual retention operation ①First select the reference point K with the least number of associations min If there are multiple such reference points, one is randomly selected from them. As reference point the individual being contacted; is empty, reselect the reference point; if If it is not empty, then look at Is it 0? ②If from Select distance The smallest individual enters the next generation; if from An individual is randomly selected to enter the next generation; Population P n+1 =S n / F l , that is, F l All non-dominated layers before the layer and the reference point The number of individuals associated; ③ Repeat the operation until the size of the next generation is equal to N.
