High-dimensional multi-objective production scheduling optimization method, terminal and medium for creep forming production line of aluminum alloy components

Abstract

The present invention discloses a high-dimensional multi-objective production scheduling optimization method, terminal and medium for creep forming production line of aluminum alloy components. First, a multi-dimensional objective function for optimizing the scheduling of creep forming production line of aluminum alloy components is constructed, and then three co-evolving populations are constructed. The first parent population uses an improved VAEA method, integrates a neighborhood search strategy, and expands the solution set distribution space; the second parent population uses an improved MOEA / D method, which can coordinate the convergence and diversity of the population and obtain a uniformly distributed solution set; the third parent population uses SPEA based on SDE, and its dominance relationship reduces the difficulty of zone decomposition, increases environmental selection pressure, and improves convergence. Multiple algorithms can effectively improve performance by integrating their respective advantages to search collaboratively, avoid premature convergence of a single population due to falling into a local minimum, keep the algorithm at a high search efficiency, improve the quality of non-dominated solution sets, and reasonably allocate production resources.

Description

Technical Field

[0001] The present invention relates to the technical field of aluminum alloy component production, and in particular to a high-dimensional multi-objective production scheduling optimization method, a terminal and a medium for an aluminum alloy component creep forming production line. Background Art

[0002] With the continuous development of intelligent manufacturing, production scheduling plays an increasingly important role in manufacturing enterprises. The quality of production scheduling schemes will directly affect the production efficiency, cost, and resource consumption of enterprises. Intelligent production scheduling of creep forming production lines for aluminum alloy components is of great significance for improving the production efficiency of production lines and reducing production costs. Therefore, it is necessary to study high-dimensional multi-objective intelligent scheduling methods for creep forming production lines for aluminum alloy components. The scheduling optimization problem of creep forming production lines for aluminum alloy components has the characteristics of high-dimensional multi-objective. In order to achieve high quality, flexibility, high efficiency and low consumption in the component manufacturing process, it is necessary to comprehensively consider and optimize multiple scheduling objectives in the scheduling process. With the increase of the target dimension, the conflict between the convergence and distribution of the population intensifies. The selection operator in the traditional multi-objective evolutionary algorithm can no longer effectively maintain the balance between the convergence and distribution of the population. Therefore, the optimization performance of high-dimensional multi-objective production scheduling decision optimization methods needs to be strengthened.

[0003] At present, for the production scheduling optimization methods of creep forming production lines of aluminum alloy aerospace components, traditional scheduling optimization methods are single-species evolutionary algorithms. Single-species evolutionary algorithms can easily fall into local optimal solutions and have limited search space. At the same time, there are many shortcomings such as the ability to distinguish problem solutions decreases with the increase in the number of targets, the strategy for maintaining the diversity of solutions is poor, and the computational cost is high. The level of reasonable allocation of production resources needs to be further improved. Summary of the invention

[0004] In view of the deficiencies in the prior art, the purpose of the present invention is to provide a high-dimensional multi-objective production scheduling optimization method, terminal and medium for a creep forming production line of aluminum alloy components, so as to rationally allocate production resources, improve the production efficiency of manufacturing enterprises and reduce production costs.

[0005] In a first aspect, a high-dimensional multi-objective production scheduling optimization method for a creep forming production line of an aluminum alloy component is provided, comprising the following steps:

[0006] Construct a multi-dimensional objective function for optimizing the scheduling of creep forming production lines for aluminum alloy components;

[0007] Initialize the parent population and split it into the first parent population, the second parent population and the third parent population; each individual in the population uses a double-layer encoding, the individual length is the number of components to be processed, the first-layer encoding chromosome is the order of components entering the tank, and the second-layer encoding chromosome is the placement of each component;

[0008] For the first parent population, the genetic operator and the local search strategy are used to obtain the first child population, the first parent population and the first child population are merged to obtain the first merged population, the vector angle-based environmental selection method is used to select the next generation first parent population from the first merged population, and this process is iterated;

[0009] For the second parent population, a decomposition-based evolutionary algorithm is used to update the population, and this process is iterated;

[0010] For the third parent population, a genetic operator is used to obtain a third child population, the third parent population and the third child population are merged to obtain a third merged population, a mobile-based density estimation strategy method is used to select the next generation of the third parent population from the third merged population, and this process is iterated;

[0011] When the iteration termination condition is met, the three populations evolved based on the first parent population, the second parent population, and the third parent population are merged to obtain the final optimal population;

[0012] The fuzzy decision-making method is used to find the optimal compromise solution of the multi-dimensional objective function from the optimal population, and the optimal compromise solution is decoded to obtain the production scheduling optimization plan.

[0013] Furthermore, the multidimensional objective function includes at least three of a volume utilization function of the autoclave, a completion time function, a delivery delay time function, an equipment utilization function, and a number of autoclave brackets function.

[0014] Furthermore, the first parent population evolution process specifically includes:

[0015] S11: Apply genetic operator operation to the first parent population to obtain the initial first child population;

[0016] S12: Use a large-area search algorithm to perform local search and update on the individuals of the initial first-generation population to obtain an updated first-generation population;

[0017] S13: merging the first parent population and the first child population to obtain a first merged population;

[0018] S14: Selecting the first parent population of the next generation from the first merged population using a vector angle-based environmental selection method;

[0019] S15: Determine whether the iteration termination condition is met, if not, return to step S11; if yes, end.

[0020] Furthermore, the second parent population evolution process specifically includes:

[0021] S21: The uniform random method and WS transformation method are used to generate a weight vector set, in which the weight vectors in the weight vector set correspond one-to-one to the individuals in the second parent population;

[0022] S22: Calculate the minimum value of each objective function value in the current second parent population as the initial reference point;

[0023] S23: Randomly select two individuals from the neighborhood of the individual to perform genetic operations to generate offspring individuals, and update the reference point;

[0024] S24: Based on the offspring individuals, the improved Chebyshev aggregation function is used to update the individuals in the neighborhood to obtain a new second parent population;

[0025] S25: Determine whether the iteration termination condition is met, if not, return to step S21; if so, end.

[0026] Furthermore, the uniform random method and the WS conversion method are used to generate the weight vector set, specifically including:

[0027] A uniform random method is used to generate X weight vectors to form a weight set, where X is greater than the size of the second parent population;

[0028] Generate an m-dimensional identity matrix and split it into m unit vectors to form a vector set, where m is the number of objective functions;

[0029] Calculate the Euclidean distance between each weight vector in the weight set and each vector in the vector set, and take the maximum value of the Euclidean distance as the eigenvalue of the weight vector;

[0030] Select the N2-m weight vector with the largest eigenvalue in the weight set and add it to the vector set to obtain the initial weight vector set; where N2 represents the size of the second parent population;

[0031] The WS transformation method is used to transform the initial weight vector set to obtain the transformed weight vector set. The transformation formula is as follows:

[0032]

[0033] In the formula, λ q and λ q' They represent the qth weight vector before and after WS transformation respectively; Represents the i-th element of the q-th weight vector before transformation.

[0034] Furthermore, the third parent population evolution process specifically includes:

[0035] S31: Calculate the fitness of the individuals in the third parent population, use the genetic operator to operate the third parent population, and obtain the third child population;

[0036] S32: merging the third parent population and the third child population to obtain a third merged population;

[0037] S33: Select the third parent population of the next generation from the third merged population using a mobile-based density estimation strategy method;

[0038] S34: Determine whether the iteration termination condition is met, if not, return to step S31; if so, end.

[0039] Further, the method of selecting a third parent population of the next generation from the third merged population using a mobile-based density estimation strategy method specifically includes:

[0040] The fitness of each individual in the third combined population is calculated by the following formula;

[0041] F(i)=R(i)+D(i,P)

[0042]

[0043] S(i)=|{j|j∈P+Q∧i>j}|

[0044]

[0045] In the formula, F(i) represents the fitness of individual i, R(i) represents the original fitness of individual i, D(i,P) represents the density of individual i in the population; S(j) represents the strength value of individual j, Pop represents the set of all individuals that dominate individual i; i>j means that individual i dominates individual j; dist(i,q′ N-1 ) represents individual i and individual q′ N-1 , P and Q represent the third parent population and the third child population respectively, and N3 is the size of the third parent population;

[0046] If the total number of individuals with fitness less than 1 is less than N3, all individuals with fitness less than 1 are added to the third parent population of the next generation, and the individuals with the smallest fitness are selected from the remaining individuals in the third merged population and added to the third parent population of the next generation, until the population size reaches N3;

[0047] If the total number of individuals with fitness less than 1 is greater than N3, all individuals with fitness less than 1 are added to the third parent population of the next generation, and the Euclidean distances between individuals in the third parent population of the next generation and other individuals are calculated in turn, and the individuals with the smallest Euclidean distance are deleted until the population size is N3.

[0048] Furthermore, the method further includes: for each iteration, a preset number of iterations are performed, and the current first parent population and the second parent population are interacted, and the interaction process includes:

[0049] S41: Calculate the sparsity level of each individual in the current second parent population;

[0050] S42: Delete A individuals with the lowest sparsity level in the second parent population and the corresponding weight vector; where A is the lower value of the number B of non-dominated solutions in the current first parent population and the preset number C, and C is a positive integer less than half of the size of the second parent population;

[0051] S43: merge the non-dominated solution individuals in the current first parent population with the current second parent population to form a new population;

[0052] S44: Calculate the sparsity level of each individual in the new population, retain the individual with the largest sparsity level, and then repeat the process until the size of the retained individuals reaches N2, and use all the retained individuals as the new second parent population; and supplement the missing weight vector according to the following formula to achieve the update of the weight vector set;

[0053]

[0054] In the formula, λ sp represents the weight vector corresponding to the non-dominated solution individual sp in the first subpopulation added to the second parent population, represents the kth objective function value of the non-dominated solution individual sp in the first subpopulation added to the second parent population, Represents the minimum value of the kth objective function in the current second parent population;

[0055] S45: Select D non-dominated solutions in the current second parent population to replace the D individuals with the lowest sparsity level in the current first parent population, where D is the lower value of the number E of non-dominated solutions in the current second parent population and the preset number F, and F is a positive integer less than half the size of the first parent population.

[0056] In a second aspect, an electronic terminal is provided, comprising:

[0057] a memory having a computer program stored thereon;

[0058] A processor is used to load and execute the computer program to implement the high-dimensional multi-objective production scheduling optimization method for the creep forming production line of aluminum alloy components as described above.

[0059] In a third aspect, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the high-dimensional multi-objective production scheduling optimization method for the creep forming production line of aluminum alloy components as described above is implemented.

[0060] The present invention proposes a high-dimensional multi-objective production scheduling optimization method, terminal and medium for creep forming production lines of aluminum alloy components, which includes three co-evolving populations. The three populations use different optimization methods respectively. Multiple algorithms can effectively improve performance by integrating their respective advantages for collaborative search. Different sub-populations can search different areas. Multi-population collaborative search is conducive to maintaining the diversity of populations, and the sharing and integration of search information can be achieved through interaction between populations; it avoids premature convergence of a single population due to falling into a local minimum, so that the algorithm can maintain a high search efficiency. Under the premise of ensuring the diversity of the solution space, the global search and local search of the solution are co-evolved, the quality of the non-dominated solution set is improved, the production resources can be reasonably allocated, the production efficiency of manufacturing enterprises can be improved, the production cost can be reduced, the production management and operation level can be improved, and the intelligent manufacturing level of enterprises can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0062] Figure 1 It is a flow chart of a high-dimensional multi-objective production scheduling optimization method for an aluminum alloy component creep forming production line provided by an embodiment of the present invention;

[0063] Figure 2 is an xy plane diagram of an autoclave support provided by an embodiment of the present invention;

[0064] Figure 3 is a schematic diagram of chromosome encoding provided by an embodiment of the present invention;

[0065] Figure 4 It is a schematic diagram of updating the remaining rectangles provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0066] To make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be described in detail below. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other implementation methods obtained by ordinary technicians in this field without creative work belong to the scope of protection of the present invention.

[0067] At present, for the production scheduling optimization methods of creep forming production lines of aluminum alloy aerospace components, traditional scheduling optimization methods are single-population evolutionary algorithms, which are easy to fall into local optimal solutions and have limited search space. At the same time, there are many shortcomings such as the ability to distinguish problem solutions decreases with the increase in the number of targets, the diversity of solutions is poor, and the calculation cost is high. The level of reasonable allocation of production resources needs to be further improved. Based on this, the present invention proposes a high-dimensional multi-objective production scheduling optimization method for creep forming production lines of aluminum alloy components, which includes three co-evolving populations. The three populations use different optimization methods respectively, and each type of optimization method is improved to better solve the production line scheduling model. The first sub-population uses the improved VAEA method, which integrates the large neighborhood search strategy to expand the distribution space of the solution set; the second sub-population uses the improved MOEA / D method, which can coordinate the convergence and diversity of the population and obtain a uniformly distributed solution set; the third sub-population uses SPEA (strength Pareto evolutionary algorithm) based on SDE, and its dominance relationship reduces the difficulty of zone decomposition, increases the pressure of environmental selection, and improves convergence. Collaborative search by integrating the advantages of multiple algorithms can effectively improve performance. Different sub-populations can search different areas. Collaborative search of multiple populations is conducive to maintaining the diversity of the population, and the sharing and integration of search information can be achieved through interaction between populations; it avoids premature convergence of a single population due to falling into a local minimum, so that the algorithm can maintain a high search efficiency. Under the premise of ensuring the diversity of the solution space, the coordinated evolution of global search and local search for solutions is achieved. The technical solution of the present invention is described in detail below in conjunction with specific embodiments.

[0068] like Figure 1 As shown, an embodiment of the present invention provides a high-dimensional multi-objective production scheduling optimization method for an aluminum alloy component creep forming production line, comprising the following steps:

[0069] Step S1: Construct a multi-dimensional objective function for optimizing the scheduling of creep forming production lines for aluminum alloy components.

[0070] The multidimensional objective function includes at least three of the volume utilization function of the autoclave, the completion time function, the delivery delay time function, the equipment utilization function, and the number of autoclave brackets function. In this embodiment, all five objective functions are taken as an example for explanation. Let f1~f5 represent the five optimization objective functions respectively. Therefore, a full-process production scheduling model F of the creep aging forming production line containing five optimization objectives is established, and its specific definition is as follows.

[0071] min F=(-f1,f2,f3,f4,f5) (1)

[0072] The volume utilization rate of the autoclave is f1. Improving the volume utilization rate of the autoclave is helpful to improve the production capacity of the autoclave equipment, improve the production efficiency of the creep forming production line and reduce production costs. The volume utilization rate of the autoclave is f1 as follows:

[0073]

[0074] Where M is the total number of supports in the autoclave preparation plan, v i The volume utilization of each bracket.

[0075] Completion time f2. The completion time of the creep aging forming production line refers to the processing time to complete all pending orders. Completion time is one of the important scheduling goals. Improving production efficiency is an important manifestation of the full-process intelligent scheduling of the creep aging forming production line. The completion time f2 is expressed as follows:

[0076] f2=max(C i |i=1,2,3…P) (3)

[0077] In the formula, C i It represents the time when the last process of the i-th component is completed, and P represents the total number of components.

[0078] Delay time f3. The scheduling goal based on the delay time is to deliver on time according to the delivery date and reduce the delay time. The delay time f3 is expressed as follows:

[0079]

[0080] In the formula, DD i represents the delivery date of the i-th component.

[0081] Equipment idle time f4. Reducing the equipment idle time can indirectly optimize the utilization rate of production line equipment and reduce the production cost of the creep forming production line of aluminum alloy aerospace components. The equipment idle time f4 is expressed as follows:

[0082]

[0083] Where, T k is the stop time after the kth device completes the order processing, The value is 0 or 1, indicating the process Is it on machine M k On processing, Indicates the process On machine M k The processing time, represents the jth process O of the i-th component ij Choose the kth machine M k processing, D represents the total number of machines, Q iRepresents the total number of processes for the i-th component.

[0084] The number of times (number of racks) of the autoclave is filled is expressed as follows:

[0085] f5=M (6)

[0086] Where M is the total number of supports in the autoclave preparation plan.

[0087] The constraints of the full-process production scheduling model of the creep aging forming production line of aluminum alloy aerospace components are as follows:

[0088] max(x i1 -x j3 ,x j1 -x i3 ,y i2 -y j4 ,y j2 -y i4 )≥0 (7)

[0089] 0≤x qh ≤W (8)

[0090] 0≤y qh ≤L (9)

[0091] 0≤z q ≤H (10)

[0092] Figure 2 The placement of the component fixture in the xy plane of the autoclave support is shown. i1 、x i3 They represent the minimum and maximum values ​​of a component on the x-axis in the xy plane of the autoclave support, and y i2 ,y i4 They represent the minimum and maximum values ​​of a component on the x-axis in the xy plane of the autoclave support; j1 、x j3 They represent the minimum and maximum values ​​of another component on the x-axis in the xy plane of the autoclave support, and y j2 ,y j4 They represent the minimum and maximum values ​​of another component on the x-axis in the xy plane of the autoclave support. q is i or j, h is 1, 2, 3, 4, and z is qrepresents the vertical height of the component, and W, L, and H represent the width, length, and height of the autoclave support. Formula (7) indicates that there is no overlap between the positions of the two component fixtures placed on the autoclave support; Formula (8) indicates that the position of the component fixture does not exceed the width of the autoclave support; Formula (9) indicates that the longitudinal position of the component fixture does not exceed the length of the autoclave support; Formula (10) indicates that the vertical position of the component fixture does not exceed the height of the autoclave support; At the same time, the stacking station can only process one component fixture at a time; The creep aging forming process parameters of all components to be processed are known; Only components with the same or similar creep aging forming process parameters can be placed in the same autoclave for creep aging forming process; The number of component fixtures in the autoclave is less than the required number of vacuum nozzles and thermocouples.

[0093] Step S2: Initialize the parent population and split it into the first parent population, the second parent population and the third parent population; each individual in the population adopts double-layer coding, the individual length is the number of components to be processed, the first-layer coding chromosome is the order of components entering the tank, and the second-layer coding chromosome is the placement of each component.

[0094] In order to effectively solve the whole process production scheduling problem of creep forming production line of aluminum alloy aerospace components, this embodiment adopts a double-layer coding method, such as Figure 3 The length of the chromosome is the number of components to be processed, the first layer of coded chromosomes is the order of components entering the tank, and the second layer of coded chromosomes is the placement of each component. The placement of the component determines whether the component needs to be rotated 90° when placed on the autoclave support.

[0095] Decoding is to convert the coding scheme into a feasible production scheduling scheme. In the decoding process, the component processing sequence and the placement of each component on the autoclave bracket that have been specified in the coding chromosome are first determined. The creep forming process is the bottleneck process of the production line. The decoding operation first needs to compile the autoclave tank arrangement plan for all components to be processed. According to the order of all components entering the tank and the placement of each component, the autoclave tank arrangement plan is compiled using the remaining rectangle method. According to the autoclave tank arrangement plan, the number of tanks, the number of bracket layers, and the position coordinates of each component can be determined. After completing the autoclave tank arrangement plan, start from the first process to arrange the production scheduling of each component in each process. In the process of arranging production, arrange the production of each process according to the first-come-first-processed rule (FCFS) and the first idle equipment allocation rule (FAM). Based on the above decoding method, the coding chromosome can be converted into a production scheduling plan for the creep aging forming production line of aluminum alloy aerospace components.

[0096] Step S3: Population evolution. It includes: the first parent population uses the improved VAEA method, which integrates the large neighborhood search strategy to expand the distribution space of the solution set; the second parent population uses the improved MOEA / D method, which can coordinate the convergence and diversity of the population and obtain a uniformly distributed solution set; the third parent population uses SPEA (strength Paretoevolutionary algorithm) based on SDE, and its dominance relationship reduces the difficulty of zone decomposition, increases the pressure of environmental selection, and improves convergence. When the iteration termination condition is met, the three populations evolved based on the first parent population, the second parent population, and the third parent population are merged to obtain the final optimal population. The details are explained below.

[0097] S31: For the first parent population, use genetic operators and local search strategies to obtain the first child population, merge the first parent population and the first child population to obtain the first merged population, use the vector angle-based environmental selection method to select the next generation of the first parent population from the first merged population, and iterate this process. The specific process is as follows:

[0098] Furthermore, the first parent population evolution process specifically includes:

[0099] S311: Genetic operators (selection, crossover, mutation) are applied to the first parent population to obtain an initial first child population.

[0100] In the specific implementation, the genetic selection operator adopts the tournament selection method. The crossover operator of the two layers of chromosomes selects the two-point crossover method. First, the two parent chromosomes are paired. If the random number is less than the crossover probability, the two paired parent chromosomes are crossovered. Two crossover points are randomly selected on the coding chromosome, and then some gene segments are exchanged. For the first layer of chromosomes, the crossover fragment of the second parent chromosome is placed in front of the coding string of the first parent chromosome, and then the repeated component number is checked from the starting gene, and the second repeated component number is deleted. In the same way, the crossover fragment of the first parent chromosome is placed in front of the coding string of the second parent chromosome, and then the genes of the second parent chromosome are deleted in turn; for the two-point crossover method of the second layer of chromosomes, two crossover points are randomly set in the individual coding chromosome, and then some gene segments are exchanged, and finally two offspring individuals are formed. The mutation operator uses the mutation method of two-point exchange on the chromosome, and arbitrarily selects two genes on the coding chromosome, and exchanges the positions of these two genes to generate a new coding chromosome.

[0101] S312: Using a large-area search algorithm, locally search and update the individuals of the initial first-generation population to obtain an updated first-generation population.

[0102] The local search strategy achieved in this embodiment is a large neighborhood search. The fusion of the vector angle based evolutionary algorithm (VAEA) and the large neighborhood search can accelerate the convergence of the population and avoid local optimality. A large neighborhood search is performed on the offspring individuals to find a better solution by searching the neighborhood of the current solution.

[0103] The implementation process of large neighborhood search is as follows:

[0104] In large neighborhood search methods, the neighborhood is implicitly defined by the destroy and repair methods. The destroy method destroys part of the current solution, and the repair method then reconstructs the destroyed solution. The destroy method usually includes an element of randomness so that a different part of the solution is destroyed each time the destroy method is called.

[0105] The specific process of large neighborhood search is as follows: first, set the number of genes p that destroy the current chromosome and record the original target value of individual i. In the destroy function, randomly generate a sequence number containing all genes, record the first p genes in the sequence number, and then find these p genes in the current solution chromosome and remove them. The remaining chromosome genes constitute the remaining chromosome. Secondly, the remaining chromosome is repaired, that is, the removed genes are inserted into the remaining chromosomes one by one; during the repair process, the removed genes are placed on the remaining chromosomes, and all individuals generated in this process are decoded. According to the concept of dominance relationship (that is, each time a removed gene is inserted, the gene is inserted into different positions of the remaining chromosomes, and the objective function values ​​of all chromosomes are calculated, the dominance relationship is determined, and the gene insertion position corresponding to the optimal chromosome is selected as the optimal insertion position), the optimal insertion position of the first removed gene in the remaining chromosome is determined, and then the remaining chromosome is updated. All removed genes are inserted into the remaining chromosomes in this way, and finally a new solution is formed. According to the concept of dominance relationship, if the new solution can dominate the current solution, the new solution chromosome replaces the current chromosome, otherwise, the current offspring chromosome remains unchanged.

[0106] S313: Merge the first parent population and the first child population to obtain a first merged population.

[0107] S314: Decoding the first merged population to obtain the scheduling optimization objective function value of each individual, and then using the vector angle-based environment selection method to select the next generation first parent population from the first merged population.

[0108] S315: Determine whether the iteration termination condition is met, if not, return to step S311; if yes, end.

[0109] S32: For the second parent population, a decomposition-based evolutionary algorithm is used to update the population, and this process is iterated.

[0110] In this embodiment, the second parent population evolution process specifically includes:

[0111] S321: A uniformly random method (UR) and a WS-Transformation method are used to generate a weight vector set, in which the weight vectors in the weight vector set correspond one-to-one to the individuals in the second parent population.

[0112] More specifically, the process includes:

[0113] 1. Generate X weight vectors using uniform random method to form a weight set, where X is greater than the size of the second parent population and the dimension of the weight vector is the number of objective functions;

[0114] 2. Generate an m-dimensional identity matrix and split it into m unit vectors to form a vector set ((1 0...0 0), (0 1...00),..., (0 0...0 1)), where m is the number of objective functions;

[0115] 3. Calculate the Euclidean distance between each weight vector in the weight set and each vector in the vector set, and take the maximum value of the Euclidean distance as the eigenvalue of the weight vector;

[0116] 4. Select the N2-m weight vector with the largest eigenvalue in the weight set and add it to the vector set to obtain the initial weight vector set; where N2 represents the size of the second parent population.

[0117] 5. Use the WS transformation method to transform the initial weight vector set to obtain the transformed weight vector set. The transformation formula is as follows:

[0118]

[0119] In the formula, λ q and λ q' They represent the qth weight vector before and after WS transformation respectively; Represents the i-th element of the q-th weight vector before transformation.

[0120] In general, the uniformity of weight vectors in MOEA / D can ensure the diversity of Pareto optimal solutions, but it is not effective when the high-dimensional multi-objective optimization problem has a complex Pareto front (PF), especially a discontinuous PF. To solve this problem, we incorporate an improved adaptive weight vector adjustment method into the MOEA / D algorithm. That is, after obtaining an initial set of weight vectors, a new transformation method (WS) is used to transform the initial set of weight vectors. The transformed weight vector set can better optimize the case of discontinuous PF and can ensure that a relatively uniform non-dominated solution set can be obtained in the end.

[0121] S322: Calculate the minimum value of each objective function value in the current second parent population as the initial reference point.

[0122] S323: Randomly select two individuals from the neighborhood of the individual to perform genetic operations to generate offspring individuals, and update the reference point.

[0123] S324: Based on the offspring individuals, the improved Chebyshev aggregation function is used to update the individuals in the neighborhood to obtain a new second parent population. The updating process is that when the Chebyshev aggregation function value of the offspring individual is less than the individual in the current area, the individual in the current area is replaced, otherwise, no replacement is performed.

[0124] S325: Determine whether the iteration termination condition is met, if not, return to step S21; if so, end.

[0125] The multi-objective evolutionary algorithm based on decomposition aggregates the various objectives of the original multi-objective problem in a linear or nonlinear manner to obtain a single-objective optimization problem, and uses the single-objective optimization method to obtain the Pareto optimal solution. When dealing with high-dimensional multi-objective optimization problems, the Chebyshev method is used as a decomposition optimization method for the aggregation function. The Chebyshev method limits the receiving area of ​​convergence, so it can better ensure the convergence of the population. However, under uniformly distributed weight vectors, the uniformity of the optimization results obtained by the Chebyshev aggregation function (Tchebycheff) decomposition method is poor. This embodiment uses the improved Chebyshev aggregation function formula as follows:

[0126]

[0127] In the formula, gte(xλ,z*) represents the Chebyshev aggregation function value of the individual, fj(x) represents the jth objective function value of individual x, represents the minimum value of the jth objective function in the population, λ j represents the weight of the j-th objective function.

[0128] The improved Tchebycheff decomposition method can not only solve non-convex problems, but also obtain uniformly distributed solutions.

[0129] S33: For the third parent population, a genetic operator operation is used to obtain a third child population, the third parent population and the third child population are merged to obtain a third merged population, a mobile-based density estimation strategy method is used to select the next generation of the third parent population from the third merged population, and this process is iterated.

[0130] In this embodiment, the third parent population evolution process specifically includes:

[0131] S331: Calculate the fitness of individuals in the third parent population, use genetic operators to operate the third parent population, and obtain the third child population;

[0132] S332: merging the third parent population and the third child population to obtain a third merged population;

[0133] S333: selecting a third parent population of the next generation from the third merged population using a mobile-based density estimation strategy method;

[0134] S334: Determine whether the iteration termination condition is met, if not, return to step S31; if so, end.

[0135] In this embodiment, the method of using the mobile-based density estimation strategy to select the next generation third parent population from the third merged population specifically includes:

[0136] The fitness of each individual in the third combined population is calculated by the following formula;

[0137] F(i)=R(i)+D(i,P) (13)

[0138]

[0139] S(i)=|{j|j∈P+Q∧i>j}| (15)

[0140]

[0141] In the formula, F(i) represents the fitness of individual i, R(i) represents the original fitness of individual i, D(i,P) represents the density of individual i in the population; S(j) represents the strength value of individual j, Pop represents the set of all individuals that dominate individual i; i>j means that individual i dominates individual j; dist(i,q′ N-1 ) represents individual i and individual q′ N-1 , P and Q represent the third parent population and the third child population respectively, and N3 is the size of the third parent population;

[0142] If the total number of individuals with fitness less than 1 is less than N3, all individuals with fitness less than 1 are added to the third parent population of the next generation, and the individuals with the smallest fitness are selected from the remaining individuals in the third merged population and added to the third parent population of the next generation, until the population size reaches N3;

[0143] If the total number of individuals with fitness less than 1 is greater than N3, all individuals with fitness less than 1 are added to the third parent population of the next generation, and the Euclidean distances between individuals in the third parent population of the next generation and other individuals are calculated in turn, and the individuals with the smallest Euclidean distance are deleted until the population size is N3.

[0144] It should be noted that in some preferred embodiments, the method further includes: each iteration is performed for a preset number of iterations Z, and the current first parent population interacts with the second parent population. During the evolution of the second parent population, the weight vector is updated once for each iteration Z times, and a population interaction is performed, and during the Z iterations, the weight vector set is kept unchanged. The interaction process specifically includes:

[0145] S341: Calculate the sparsity level of each individual in the current second parent population; the calculation formula of the sparsity level is as follows:

[0146]

[0147] In the formula, SL j represents the sparsity level of the jth individual; is the Euclidean distance from the jth individual to the i-th individual closest to it; m takes the number of objective functions, that is, the sparsity level of the jth individual is calculated based on the m individuals closest to its Euclidean distance.

[0148] S342: Delete A individuals with the lowest sparsity level in the second parent population and the corresponding weight vector; A is the lower value of the number of non-dominated solutions B in the current first parent population and the preset number C, C is a positive integer less than half of the size of the second parent population, and C is preferably one-third of the size of the second parent population;

[0149] S343: merging the non-dominated solution individuals in the current first parent population with the current second parent population to form a new population;

[0150] S344: Calculate the sparsity level of each individual in the new population, retain the individual with the largest sparsity level, and then repeat the process (calculate the sparsity level of each individual in the new population, and retain the individual with the largest sparsity level) until the size of the retained individuals reaches N2, and use all the retained individuals as the new second parent population, and the remaining individuals and their weight vectors are deleted; and supplement the missing weight vector according to the following formula to achieve the update of the weight vector set;

[0151]

[0152] In the formula, λ sp represents the weight vector corresponding to the non-dominated solution individual sp in the first subpopulation added to the second parent population, represents the kth objective function value of the non-dominated solution individual sp in the first subpopulation added to the second parent population, Represents the minimum value of the kth objective function in the current second parent population;

[0153] S45: Select D non-dominated solutions in the current second parent population to replace the D individuals with the lowest sparsity level in the current first parent population, where D is the lower value of the number E of non-dominated solutions in the current second parent population and the preset number F, and F is a positive integer less than half the size of the first parent population, and F is preferably one-third the size of the first parent population.

[0154] The weights are adjusted periodically so that the weights of the subproblems can be adaptively redistributed, thereby obtaining better uniformity of solutions. At the same time, the computational effort of subproblems with repeated optimal solutions is saved. In addition, the introduction of an external elite population helps to add new subproblems to the real sparse region of the complex PF, rather than the pseudo-sparse region, i.e., the discontinuous region of the PF. The second subpopulation uses an improved MOEA / D method, which can coordinate the convergence and diversity of the population and obtain a uniformly distributed solution set. At the same time, the global search performance of the multi-population coevolution strategy is good, which improves the quality of the non-dominated solution set.

[0155] S4: Use fuzzy decision-making methods to find the optimal compromise solution of the multi-dimensional objective function from the optimal population, decode the optimal compromise solution, and obtain the production scheduling optimization plan.

[0156] The optimal compromise solution is obtained by the following method:

[0157] Each objective function is processed using the fuzzy membership function. The membership of the i-th objective function is u i , its mathematical expression is as follows:

[0158]

[0159] In the formula, F i max and F i min They represent the maximum and minimum values ​​of the i-th objective function of the N individuals in the population obtained at the end;

[0160] The membership of the nth individual is standardized using the following formula:

[0161]

[0162] In the formula, u opt represents the membership degree of the nth individual after standardized processing, represents the membership degree of the nth individual, and m is the number of all objective functions;

[0163] The individual with the largest membership degree after standardized processing in the final population is selected as the optimal compromise solution.

[0164] Decoding the optimal compromise solution involves:

[0165] According to the order of all the processed components entering the tank and the placement of each component in the optimal compromise solution, the remaining rectangle method is used to prepare the autoclave arrangement plan;

[0166] According to the autoclave scheduling plan, the production scheduling of each component in each process is formulated starting from the first process. During the scheduling process, the production of each process is arranged according to the first-come-first-served rule and the first idle equipment allocation rule.

[0167] In order to effectively solve the full-process production scheduling model of the creep forming production line of aluminum alloy aerospace components, the problem of autoclave canning must be solved first. The tooling in the autoclave must be placed on the bracket. The autoclave bracket is a double-layer structure. The autoclave canning plan is determined by the component tooling placed on the double-layer bracket. According to the known sequence of component tooling entering the can, the three-dimensional double-layer bracket canning method needs to determine the number of bracket layers and their corresponding positions for each component tooling entering the can, and also determines the number of autoclaves required to process the current component order.

[0168] In this embodiment, the remaining rectangle method is used to compile the autoclave tank arrangement plan. Before the arrangement, the order of the components to be arranged and the orientation of each component (whether it is rotated 90°) are already determined. B1 = [0 0 L1 H1] is the size information of the first layer bracket, which is represented as a rectangle and represented by the two diagonal points of the rectangle, (0,0) represents the origin, (L1, H1) represents the diagonal point of the origin, where L1 and H1 are the length and width of the first layer bracket of the autoclave; similarly, B2 = [0 0 L2H2] is the size information of the second layer bracket, and the matrix A stores the coordinates and size of each component placement.

[0169] The steps of the remaining rectangle method are as follows:

[0170] 1. First, determine whether the component needs to be rotated 90° according to the orientation of the component. Components that do not need to be rotated should maintain their original length and width. Components that need to be rotated should have their length and width changed. Determine the overall dimensions of each component based on the above rules.

[0171] 2. Place the current component i in the remaining rectangle. Find the rectangle set re that can contain the current component i in the remaining rectangle set rest, and determine which rectangle the component i is placed in according to the BL (bottom-up left-justified) condition. At the same time, the information in the matrix A needs to be updated.

[0172] 3. Update the number of bracket layers. If there is no rectangle in the remaining rectangle set rest that can be used to place the current component tooling i, then the number of brackets needs to be updated. If the first layer of bracket placement tooling is completed, it is necessary to start compiling the placement plan for the second layer of brackets; if the second layer of bracket placement tooling is completed, it is necessary to start a new tank and start compiling the placement plan for the first layer of brackets at the same time. When starting the placement plan for a new layer of brackets, it is necessary to update the information in the remaining rectangle set rest and matrix A in a timely manner.

[0173] 4. Update the remaining rectangle set rest. When placing the current component tooling i in the remaining rectangle j, it is necessary to determine whether other remaining rectangles will be generated and delete the remaining rectangle j. Figure 4 As shown, when component tooling i is placed in the remaining rectangle j = [x0, y0, x1, y1], remaining rectangles a1 = [x2, y0, x1, y1] and a2 = [x0, y2, x1, y1] will be generated.

[0174] The above embodiment proposes a high-dimensional multi-objective production scheduling optimization method for creep forming production lines of aluminum alloy components, which includes three co-evolving populations. The three populations use different optimization methods respectively. Multiple algorithms can effectively improve performance by integrating their respective advantages for collaborative search. Different sub-populations can search different areas. Multi-population collaborative search is conducive to maintaining the diversity of populations, and sharing and fusion of search information can be achieved through interaction between populations; it avoids premature convergence of a single population due to falling into a local minimum, so that the algorithm can maintain a high search efficiency. Under the premise of ensuring the diversity of the solution space, the global search and local search of the solution are co-evolved, the quality of the non-dominated solution set is improved, the production resources can be reasonably allocated, the production efficiency of manufacturing enterprises can be improved, the production cost can be reduced, the production management and operation level can be improved, and the intelligent manufacturing level of enterprises can be improved.

[0175] An embodiment of the present invention further provides an electronic terminal, including:

[0176] a memory having a computer program stored thereon;

[0177] A processor is used to load and execute the computer program to implement the high-dimensional multi-objective production scheduling optimization method for the creep forming production line of aluminum alloy components as described in the above-mentioned embodiment.

[0178] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the high-dimensional multi-objective production scheduling optimization method for the creep forming production line of aluminum alloy components as described in the aforementioned embodiment is implemented.

[0179] It can be understood that the same or similar parts of the above embodiments can be referenced to each other, and the contents not described in detail in some embodiments can refer to the same or similar contents in other embodiments.

[0180] For further explanation, the technical solution of the present invention is further described below in conjunction with specific experiments.

[0181] In order to verify the efficiency and feasibility of the proposed high-dimensional multi-objective production scheduling optimization method for creep forming production lines of aluminum alloy components, 10 benchmark examples (ACF01~ACF10) were constructed. The information of the benchmark examples is shown in Table 1. The examples are generated with reference to the actual process parameters of each aluminum alloy component.

[0182] Table 1

[0183]

[0184] In order to prove the efficiency of the optimization method (MPCOA optimization method) proposed in this invention in solving the full-process production scheduling model of the aluminum alloy creep forming component production line. We use RPD-NSGA-2, spea2+SDE, EFR_RR and PICEAg as comparison optimization methods. In order to ensure the fairness of the comparison, the population size, crossover mutation method, and crossover mutation probability are all kept consistent. The compared methods use the mainstream BL (bottom-up left-justified) based canning method. At the same time, in the comparison process, we use a fixed time as the iteration stop criterion. The MPCOA optimization method calculates each benchmark example 5 times, stops after 50 iterations each time, and uses the average calculation time of each example as the iteration stop time. The average calculation time of each example is shown in Table 2.

[0185] Table 2

[0186]

[0187] The parameters in this experiment are set as follows.

[0188] Table 3

[0189]

[0190] This experiment uses two performance indicators, IGD (Inverted Generational Distance) and HV (Hypervolume), to verify the superiority of the proposed HMOMA algorithm. Both indicators are comprehensive performance evaluation indicators. The smaller the IGD value, the better the diversity and convergence of the algorithm. The larger the HV value, the better the diversity and convergence of the algorithm.

[0191] IGD calculates the individuals in the true Pareto optimal solution set P to the non-dominated solution set P obtained by the algorithm * The average distance.

[0192]

[0193] Where min dis(x,P) represents P * The minimum Euclidean distance from individual x in P to individuals in P.

[0194] Hypervolume allows evaluating the quality of a solution set by a single value of its hypervolume. Hypervolume is a comprehensive metric for evaluating the convergence and diversity of an approximate solution set.

[0195]

[0196] In the formula, δ represents the Lebesgue measure, |S| represents the number of non-dominated solution sets, and v i represents the hypervolume formed by the reference point and the i-th solution in the solution set.

[0197] In order to more accurately judge the performance of each multi-objective scheduling decision optimization method in solving different benchmark cases, each decision method was calculated repeatedly 30 times when solving each benchmark case, and the 30 performance measurement index values ​​were subjected to the Wilcoxon rank sum test with a significance level of 0.05.

[0198] The average values ​​of the IGD index and the HV index and the P value of the rank sum test are shown in Tables 4 and 5. In the experimental results, the index values ​​that are significantly better than other algorithms are indicated in bold. The MPCOA optimization method outperforms other algorithms in the IGD index and HV index of most test instances. Based on the above results, the efficiency and feasibility of the full-process production scheduling model of the aluminum alloy creep forming component production line and the MPCOA optimization decision-making method are verified.

[0199] Table 4

[0200]

[0201] Table 5

[0202]

[0203] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0204] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0205] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0206] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0207] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A high-dimensional multi-objective production scheduling optimization method for creep forming production lines of aluminum alloy components, characterized in that: The steps include: Construct a multi-dimensional objective function for optimizing the scheduling of creep forming production lines for aluminum alloy components; Initialize the parent population and split it into the first parent population, the second parent population and the third parent population; each individual in the population uses a double-layer encoding, the individual length is the number of components to be processed, the first-layer encoding chromosome is the order of components entering the tank, and the second-layer encoding chromosome is the placement of each component; For the first parent population, the genetic operator and the local search strategy are used to obtain the first child population, the first parent population and the first child population are merged to obtain the first merged population, the vector angle-based environmental selection method is used to select the next generation first parent population from the first merged population, and this process is iterated; For the second parent population, a decomposition-based evolutionary algorithm is used to update the population, and this process is iterated; For the third parent population, a genetic operator is used to obtain a third child population, the third parent population and the third child population are merged to obtain a third merged population, a mobile-based density estimation strategy method is used to select the next generation of the third parent population from the third merged population, and this process is iterated; When the iteration termination condition is met, the three populations evolved based on the first parent population, the second parent population, and the third parent population are merged to obtain the final optimal population; The fuzzy decision-making method is used to find the optimal compromise solution of the multi-dimensional objective function from the optimal population, and the optimal compromise solution is decoded to obtain the production scheduling optimization plan; It also includes: each iteration presets the number of iterations, and interacts the current first parent population with the second parent population, and the interaction process includes: S41: Calculate the sparsity level of each individual in the current second parent population; S42: Delete A individuals with the lowest sparsity level in the second parent population and the corresponding weight vector; where A is the lower value of the number B of non-dominated solutions in the current first parent population and the preset number C, and C is a positive integer less than half of the size of the second parent population; S43: merge the non-dominated solution individuals in the current first parent population with the current second parent population to form a new population; S44: Calculate the sparsity level of each individual in the new population, retain the individual with the largest sparsity level, and then repeat the process until the size of the retained individuals reaches N2, and use all the retained individuals as the new second parent population; and supplement the missing weight vector according to the following formula to achieve the update of the weight vector set; In the formula, λ sp represents the weight vector corresponding to the non-dominated solution individual sp in the first subpopulation added to the second parent population, represents the kth objective function value of the non-dominated solution individual sp in the first subpopulation added to the second parent population, Represents the minimum value of the kth objective function in the current second parent population; S45: Select D non-dominated solutions in the current second parent population to replace the D individuals with the lowest sparsity level in the current first parent population, where D is the lower value of the number E of non-dominated solutions in the current second parent population and the preset number F, and F is a positive integer less than half the size of the first parent population.

2. The high-dimensional multi-objective production scheduling optimization method for the creep forming production line of aluminum alloy components according to claim 1 is characterized in that: The multidimensional objective function includes at least three of the autoclave volume utilization function, the completion time function, the delivery delay time function, the equipment utilization function and the autoclave support number function.

3. The high-dimensional multi-objective production scheduling optimization method for the creep forming production line of aluminum alloy components according to claim 1 is characterized in that: The first parent population evolution process specifically includes: S11: Apply genetic operator operation to the first parent population to obtain the initial first child population; S12: Use a large-area search algorithm to perform local search and update on the individuals of the initial first-generation population to obtain an updated first-generation population; S13: merging the first parent population and the updated first child population to obtain a first merged population; S14: Selecting the first parent population of the next generation from the first merged population using a vector angle-based environmental selection method; S15: Determine whether the iteration termination condition is met, if not, return to step S11; if yes, end.

4. The high-dimensional multi-objective production scheduling optimization method for the creep forming production line of aluminum alloy components according to claim 1 is characterized in that: The second parent population evolution process specifically includes: S21: The uniform random method and WS transformation method are used to generate a weight vector set, in which the weight vectors in the weight vector set correspond one-to-one to the individuals in the second parent population; S22: Calculate the minimum value of each objective function value in the current second parent population as the initial reference point; S23: Randomly select two individuals from the neighborhood of the individual to perform genetic operations to generate offspring individuals, and update the reference point; S24: Based on the offspring individuals, the improved Chebyshev aggregation function is used to update the individuals in the neighborhood to obtain a new second parent population; S25: Determine whether the iteration termination condition is met, if not, return to step S21; if so, end.

5. The high-dimensional multi-objective production scheduling optimization method for the creep forming production line of aluminum alloy components according to claim 4 is characterized in that: The uniform random method and WS transformation method are used to generate the weight vector set, including: A uniform random method is used to generate X weight vectors to form a weight set, where X is greater than the size of the second parent population; Generate an m-dimensional identity matrix and split it into m unit vectors to form a vector set, where m is the number of objective functions; Calculate the Euclidean distance between each weight vector in the weight set and each vector in the vector set, and take the maximum value of the Euclidean distance as the eigenvalue of the weight vector; Select the N2-m weight vector with the largest eigenvalue in the weight set and add it to the vector set to obtain the initial weight vector set; where N2 represents the size of the second parent population; The WS transformation method is used to transform the initial weight vector set to obtain the transformed weight vector set. The transformation formula is as follows: In the formula, λ q and λ q' They represent the qth weight vector before and after WS transformation respectively; Represents the i-th element of the q-th weight vector before transformation.

6. The high-dimensional multi-objective production scheduling optimization method for the creep forming production line of aluminum alloy components according to claim 1 is characterized in that: The evolution process of the third parent population specifically includes: S31: Calculate the fitness of the individuals in the third parent population, use the genetic operator to operate the third parent population, and obtain the third child population; S32: merging the third parent population and the third child population to obtain a third merged population; S33: Select the third parent population of the next generation from the third merged population using a mobile-based density estimation strategy method; S34: Determine whether the iteration termination condition is met, if not, return to step S31; if so, end.

7. The high-dimensional multi-objective production scheduling optimization method for the creep forming production line of aluminum alloy components according to claim 6 is characterized in that: The method of using the mobile-based density estimation strategy to select the next generation third parent population from the third merged population specifically includes: The fitness of each individual in the third combined population is calculated by the following formula; F(i)=R(i)+D(i,P) S(i)=|{j|j∈P+Q∧i>j}| In the formula, F(i) represents the fitness of individual i, R(i) represents the original fitness of individual i, D(i,P) represents the density of individual i in the population; S(j) represents the strength value of individual j, Pop represents the set of all individuals that dominate individual i; i>j means that individual i dominates individual j; Represents individual i and individual , P and Q represent the third parent population and the third child population respectively, and N3 is the size of the third parent population; If the total number of individuals with fitness less than 1 is less than N3, all individuals with fitness less than 1 are added to the third parent population of the next generation, and the individuals with the smallest fitness are selected from the remaining individuals in the third merged population and added to the third parent population of the next generation, until the population size reaches N3; If the total number of individuals with fitness less than 1 is greater than N3, all individuals with fitness less than 1 are added to the third parent population of the next generation, and the Euclidean distances between individuals in the third parent population of the next generation and other individuals are calculated in turn, and the individuals with the smallest Euclidean distance are deleted until the population size is N3.

8. An electronic terminal, characterized in that: include: a memory having a computer program stored thereon; A processor is used to load and execute the computer program to implement the high-dimensional multi-objective production scheduling optimization method for the creep forming production line of aluminum alloy components as described in any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the high-dimensional multi-objective production scheduling optimization method for the creep forming production line of an aluminum alloy component as described in any one of claims 1 to 7 is implemented.

Images