A multi-objective optimization allocation method for manufacturing tasks

Abstract

The multi-objective manufacturing task optimization allocation method of this invention is implemented through the following steps: a) establishing a set of tasks and equipment; b) setting conditions; c) constructing an optimization model, implemented through steps c-1) to c-2): c-1) establishing a set of decision variables; c-2) establishing known boundary conditions; d) optimizing the objective function, using a combination of genetic algorithm and simulated annealing algorithm to optimize, allocating n independent manufacturing tasks to m manufacturing equipment according to the optimal set of decision variables matrix x. This multi-objective manufacturing task optimization allocation method of the present invention comprehensively considers time, quality, and load factor factors, scientifically allocating multiple manufacturing tasks among limited manufacturing equipment. While ensuring that the load factor of the manufacturing equipment meets the predetermined requirements, it achieves multiple objectives: minimizing the total processing time of manufacturing tasks, accelerating task completion, and maximizing task completion quality.

Description

Technical Field

[0001] The present invention relates to a method for optimizing the allocation of manufacturing tasks, and more specifically, to a method for optimizing the allocation of multi-objective manufacturing tasks. Background Art

[0002] At present, the traditional discrete single-variety mass production mode is gradually changing to a multi-variety small and medium-batch production mode. At the same time, with the promotion of intelligent manufacturing equipment, manufacturing equipment is becoming more and more flexible, and the same manufacturing equipment can usually process multiple manufacturing tasks. Under the dual effects of market pull and technology push, the production planning and scheduling of modern manufacturing enterprises have encountered new challenges. How to scientifically allocate multiple manufacturing tasks among limited manufacturing equipment to achieve the multiple goals of minimizing the total processing time of manufacturing tasks, completing tasks early, and completing tasks with the highest quality, while trying to ensure that the load rate of manufacturing equipment meets the established requirements, is a complex decision-making problem.

[0003] It is crucial to establish a multi-objective optimization allocation model for manufacturing tasks by comprehensively considering factors such as time, quality, and load rate, and to propose a solution algorithm for the model. Summary of the invention

[0004] In order to overcome the disadvantages of the above technical problems, the present invention provides a multi-objective manufacturing task optimization allocation method.

[0005] The multi-objective manufacturing task optimization allocation method of the present invention is characterized by being implemented by the following steps:

[0006] a). Establishment of task and equipment set. Suppose there are n independent manufacturing tasks and m manufacturing equipment in total. The n independent manufacturing tasks are represented by the set Task = {T1, T2, ..., T n} indicates that T i is the i-th task; m manufacturing equipments use the set P = {P1, P2, ..., P m} indicates that P i is the i-th manufacturing equipment; the tasks in the set Task must be allocated to all m manufacturing equipment, and one task is processed independently by only one manufacturing equipment;

[0007] b). Condition setting: Assume that the decision maker allocates tasks based on the CT cycle, allocates different task sets in different cycles, and the manufacturing equipment executes the assigned processing tasks without time delay; the manufacturing equipment P k Once the manufacturing task begins, i , cannot be interrupted to process another manufacturing task T j , the same manufacturing equipment can only process one manufacturing task at the same time and cannot process multiple tasks in parallel;

[0008] In this step, when the status of a manufacturing equipment is "idle" or changes from "busy" to "idle", if there are still tasks waiting to be processed in its task queue, task processing will start immediately without time delay.

[0009] c) The construction of the optimization model is achieved through steps c-1) to c-2):

[0010] c-1). Establish a decision variable set, and use an n×m 0-1 matrix X to represent the decision variable set. The value range of the elements in the matrix X is 0 or 1. ij =1 indicates task T i Assigned to manufacturing equipment P j , X ij =0 indicates task T i Not allocated to manufacturing equipment P j ;

[0011] c-2). Establish known boundary conditions and build an n×m processing time matrix T. The element t in the processing time matrix T is ij Represents task T i By manufacturing equipment P j The processing time required for processing; establish an n×m processing quality matrix Q, and the element q in the processing quality matrix Q ij Represents task T i By manufacturing equipment P j Quality level during processing; establish an m×3 equipment load parameter matrix E, the element e in the equipment load parameter matrix E ia represents the effective utilization time of the i-th manufacturing equipment in the time period CT, a=1, 2, 3, is the lower limit of the load rate constraint, and is the upper limit of the load rate constraint;

[0012] Establish the mathematical model shown in formula (1) and formula (2):

[0013] OPT f(x)=OPT(f1(X),f2(X),f3(X))=(minT total ,minT end ,maxQ total ) (1)

[0014]

[0015] The constraints are:

[0016]

[0017] OPT f(x) is the objective function, T total is the total processing time of the task, T end Q is the earliest completion time of all tasks to be assigned within CT time.total For the quality of all tasks, T max Indicates the upper limit of the total processing time of the task, T emax Q is the earliest completion time limit of the task to be assigned within the CT time. min is the lower limit of the total quality level of the task;

[0018] d) To optimize the objective function, first unify the three objectives of the objective function in formula (1) into a single objective: i (x) The exponential method in the power factor method is used to convert it into a power factor d with a value between 0 and 1. i , and take its geometric mean as the new objective function; the converted single objective function is shown in formula (3):

[0019]

[0020] T 1 is the qualified threshold of the total processing time of the task, and the corresponding efficiency coefficient is e -1 =0.3679; T 0 is the qualified threshold of the total processing time of the task, and its efficiency coefficient is e -e =0.06598; is the task completion time qualification threshold, is the unqualified threshold for task completion time, Q 1 is the total quality threshold for task completion, Q 0 T is the total quality failure threshold for task completion; 1 、T 0 , Q 1 , Q 0 The value of reflects the importance that decision makers attach to each goal;

[0021] Then, a method combining genetic algorithm and simulated annealing algorithm is used to search for the optimal solution: first, the approximate area of ​​the optimal solution is found through the genetic algorithm, and then the simulated annealing algorithm is used to further search for the optimal solution near the approximate area until the optimal decision variable set matrix x, x∈X is obtained, and the n independent manufacturing tasks are assigned to m manufacturing equipment for processing according to the optimal decision variable set matrix x.

[0022] In the multi-objective manufacturing task optimization allocation method of the present invention, the approximate area of ​​finding the optimal solution by genetic algorithm in step d) is achieved by the following steps:

[0023] d-1-1). Chromosome encoding, using integer encoding rules to encode the variables as chromosomes;

[0024] d-1-2). Genetic manipulation, using a genetic algorithm that combines roulette and elite retention for genetic manipulation;

[0025] d-1-3). Interactive operation, using two-point crossover method, select two parent individuals, randomly generate two integers a and b between 1 and n, assuming a < b, and exchange the gene values ​​between the ath position and the bth position of the two parent individuals as a whole, while the other gene values ​​remain unchanged;

[0026] d-1-4). Adaptive crossover rate. To ensure the stability of good solutions and increase the possibility of transforming bad solutions into good solutions, different crossover rates are used for different individuals. That is, when the fitness of an individual is lower than the average fitness of the generation population, a larger crossover rate should be used. On the contrary, a smaller crossover rate should be given. The adaptive crossover rate is shown in formula (4):

[0027]

[0028] In the formula, f max is the maximum fitness value of individuals in the group, f avg is the average fitness value of individuals in the group, f is the larger fitness value of the two individuals to be crossed, P cmax is the upper limit of the adaptive crossover rate, P cmin is the lower limit of the adaptive crossover rate;

[0029] d-1-5). Mutation operation, using the double-point mutation method, select an individual, randomly generate two integers a and b between 1 and n, set a<b, randomly generate an integer c and d between 1 and m, and replace the gene values ​​corresponding to points a and b with c and d;

[0030] d-1-6). Adaptive mutation rate: Adaptive mutation rate as shown in formula (5):

[0031]

[0032] P m max is the upper limit of the adaptive mutation rate, P c min is the lower limit of the adaptive mutation rate;

[0033] d-1-7). Parent update strategy: In order to make the best solution as close to the optimal solution as possible and improve the computational efficiency of the subsequent simulated annealing algorithm, a multi-stage parent update strategy is adopted; inter is used to control the number of stages, and the condition for exiting the iteration at each stage is f max <f min That is, when this condition is met, exit the current generation process and regenerate a new population; at the same time, replace the worst individual in the new population with the current best individual.

[0034] In the multi-objective manufacturing task optimization allocation method of the present invention, the further optimization using the simulated annealing algorithm near the approximate area of ​​the optimal solution in step d) is achieved by the following steps:

[0035] d-2-1). Assign the optimal individual found by genetic algorithm and its fitness to R a 、f a , and let the current temperature t = t0, t0 is the set value;

[0036] d-2-2). Determine if t>t f Is it true? If so, the current individual is the optimal solution and the optimal solution is output; if not, execute step d-2-3); t f is to set the temperature threshold;

[0037] d-2-3). Let i = 1;

[0038] d-2-4). Determine whether i≤L is established. If so, execute step d-2-5); if not, execute the cooling operation of t=t×alpha, alpha<1, and jump to step d-2-2);

[0039] d-2-5). By R a Mutation produces R b ; Execute step d-2-6);

[0040] d-2-6). Calculate R b The corresponding fitness f b ;

[0041] d-2-7). Determine whether f is true b >f a , if it holds, then accept this new solution and let R a =R b 、f a =f b ; If not, generate a random real number rand from 0 to 1 and judge rand <exp((f a -f b ) / t) is true, if true, accept the new solution, otherwise discard the new solution; execute step d-2-8);

[0042] d-2-8). Execute step i=i+1 and go to step d-2-4).

[0043] The multi-objective manufacturing task optimization allocation method of the present invention, in step d-1-2), the genetic algorithm combining roulette and elite retention is specifically implemented by the following steps:

[0044] d-1-2-1). Set a variable A to store the best individual at the current moment;

[0045] d-1-2-2). Calculate the individual cumulative selection probability and obtain the selection probability vector P = [P1 P2 … P popsize ] T ;

[0046] In this step, the genetic algorithm combining roulette and elite retention is implemented by the following steps:

[0047] d-1-2-1). Set a variable A to store the best individual at the current moment;

[0048] d-1-2-2). Calculate the individual cumulative selection probability and obtain the selection probability vector P = [P1 P2 … P popsize ] T ;

[0049] d-1-2-3). Select two individuals at a time by roulette;

[0050] d-1-2-4). After the two individuals are subjected to adaptive crossover and adaptive mutation operations, two new individuals are obtained. Whether the new individuals are feasible is determined, and the feasible individuals are added to the offspring;

[0051] d-1-2-5). Repeat steps d-1-2-3) to d-1-2-4) until the size of the offspring reaches popsize;

[0052] d-1-2-6). Find the best individual B and the worst individual C from the offspring;

[0053] d-1-2-7). Compare A and B. If B is better than A, replace A with B.

[0054] d-1-2-8). Replace C with A;

[0055] d-1-2-9). Replace the parent with the child.

[0056] d-1-2-3). Select two individuals at a time by roulette;

[0057] d-1-2-4). After the two individuals are subjected to adaptive crossover and adaptive mutation operations, two new individuals are obtained. Whether the new individuals are feasible is determined, and the feasible individuals are added to the offspring;

[0058] d-1-2-5). Repeat steps d-1-2-3) to d-1-2-4) until the size of the offspring reaches popsize;

[0059] d-1-2-6). Find the best individual B and the worst individual C from the offspring;

[0060] d-1-2-7). Compare A and B. If B is better than A, replace A with B.

[0061] d-1-2-8). Replace C with A;

[0062] d-1-2-9). Replace the parent with the child.

[0063] The beneficial effects of the present invention are as follows: the multi-objective manufacturing task optimization allocation method of the present invention first establishes a set of independent manufacturing tasks and manufacturing equipment, and gives set conditions, then establishes an optimization model, and then uses a method combining a genetic algorithm with a simulated annealing algorithm to search for the best, until the optimal decision variable set matrix is ​​obtained to execute the processing work of the manufacturing equipment. It can be seen that the multi-objective manufacturing task optimization allocation method of the present invention comprehensively considers the factors of time, quality, and load rate, and scientifically allocates multiple manufacturing tasks among limited manufacturing equipment. Under the premise of ensuring that the load rate of the manufacturing equipment meets the established requirements, it achieves the multiple goals of minimizing the total processing time of the manufacturing task, completing the task early, and achieving the highest quality of task completion. DETAILED DESCRIPTION

[0064] As a type of random search optimization algorithm that simulates the natural evolution and genetic process of the biological world, genetic algorithms have the ability to solve approximate optimal solutions in large problem spaces, and show strong advantages in solving large combinatorial optimization problems. However, genetic algorithms have defects such as easy premature maturity and slow convergence speed when approaching the optimal solution. Experiments show that its defects are more prominent for the model in this section. The simulated annealing method has a strong local search ability, which can enable the search process to jump out of the local optimal solution and then search for the global optimal solution, but it has little knowledge of the entire search process, and the quality of the initial solution selection greatly affects its computational efficiency. If the genetic algorithm is combined with the simulated annealing algorithm, the approximate area of ​​the optimal solution is found by the genetic algorithm, and then the simulated annealing algorithm is used near the area for further optimization, it can play to its strengths and avoid its weaknesses, improve the effectiveness of the algorithm and improve its computational efficiency. Based on the above ideas, the present invention comprehensively proposes and designs a multi-stage parent-updating adaptive genetic-simulated annealing algorithm.

[0065] The multi-objective manufacturing task optimization allocation method of the present invention is implemented by the following steps:

[0066] a). Establishment of task and equipment set. Suppose there are n independent manufacturing tasks and m manufacturing equipment in total. The n independent manufacturing tasks are represented by the set Task = {T1, T2, ..., T n} indicates that T i is the i-th task; m manufacturing equipments use the set P = {P1, P2, ..., P m} indicates that P iis the i-th manufacturing equipment; the tasks in the set Task must be allocated to all m manufacturing equipment, and one task is processed independently by only one manufacturing equipment;

[0067] b). Condition setting: Assume that the decision maker allocates tasks based on the CT cycle, allocates different task sets in different cycles, and the manufacturing equipment executes the assigned processing tasks without time delay; the manufacturing equipment P k Once the manufacturing task begins, i , cannot be interrupted to process another manufacturing task T j , the same manufacturing equipment can only process one manufacturing task at the same time and cannot process multiple tasks in parallel;

[0068] c) The construction of the optimization model is achieved through steps c-1) to c-2):

[0069] c-1). Establish a decision variable set, and use an n×m 0-1 matrix X to represent the decision variable set. The value range of the elements in the matrix X is 0 or 1. ij =1 indicates task T i Assigned to manufacturing equipment P j , X ij =0 indicates task T i Not allocated to manufacturing equipment P j ;

[0070] c-2). Establish known boundary conditions and build an n×m processing time matrix T. The element t in the processing time matrix T is ij Represents task T i By manufacturing equipment P j The processing time required for processing; establish an n×m processing quality matrix Q, and the element q in the processing quality matrix Q ij Represents task T i By manufacturing equipment P j Quality level during processing; establish an m×3 equipment load parameter matrix E, the element e in the equipment load parameter matrix E ia represents the effective utilization time of the i-th manufacturing equipment in the time period CT, a=1, 2, 3, is the lower limit of the load rate constraint, and is the upper limit of the load rate constraint;

[0071] Establish the mathematical model shown in formula (1) and formula (2):

[0072] OPT f(x)=OPT(f1(X),f2(X),f3(X))=(minT total ,minT end ,maxQ total ) (1)

[0073]

[0074] The constraints are:

[0075]

[0076] OPT f(x) is the objective function, T total is the total processing time of the task, T end Q is the earliest completion time of all tasks to be assigned within CT time. total For the quality of all tasks, T max Indicates the upper limit of the total processing time of the task, T emax Q is the earliest completion time limit of the task to be assigned within the CT time. min is the lower limit of the total quality level of the task;

[0077] d) To optimize the objective function, first unify the three objectives of the objective function in formula (1) into a single objective: i (x) The exponential method in the power factor method is used to convert it into a power factor d with a value between 0 and 1. i , and take its geometric mean as the new objective function; the converted single objective function is shown in formula (3):

[0078]

[0079] T 1 is the qualified threshold of the total processing time of the task, and the corresponding efficiency coefficient is e -1 =0.3679; T 0 is the qualified threshold of the total processing time of the task, and its efficiency coefficient is e -e =0.06598; is the task completion time qualification threshold, is the unqualified threshold for task completion time, Q 1 is the total quality threshold for task completion, Q 0 T is the total quality failure threshold for task completion; 1 、T 0 , Q 1 , Q 0 The value of reflects the importance that decision makers attach to each goal;

[0080] Then, a method combining genetic algorithm and simulated annealing algorithm is used to search for the optimal solution: first, the approximate area of ​​the optimal solution is found through the genetic algorithm, and then the simulated annealing algorithm is used to further search for the optimal solution near the approximate area until the optimal decision variable set matrix x, x∈X is obtained, and the n independent manufacturing tasks are assigned to m manufacturing equipment for processing according to the optimal decision variable set matrix x.

[0081] In this step, the approximate area of ​​the optimal solution found by the genetic algorithm is achieved by the following steps:

[0082] d-1-1). Chromosome encoding, using integer encoding rules to encode the variables as chromosomes;

[0083] d-1-2). Genetic manipulation, using a genetic algorithm that combines roulette and elite retention for genetic manipulation;

[0084] d-1-3). Interactive operation, using two-point crossover method, select two parent individuals, randomly generate two integers a and b between 1 and n, assuming a < b, and exchange the gene values ​​between the ath position and the bth position of the two parent individuals as a whole, while the other gene values ​​remain unchanged;

[0085] d-1-4). Adaptive crossover rate. To ensure the stability of good solutions and increase the possibility of transforming bad solutions into good solutions, different crossover rates are used for different individuals. That is, when the fitness of an individual is lower than the average fitness of the generation population, a larger crossover rate should be used. On the contrary, a smaller crossover rate should be given. The adaptive crossover rate is shown in formula (4):

[0086]

[0087] In the formula, f max is the maximum fitness value of individuals in the group, f avg is the average fitness value of individuals in the group, f is the larger fitness value of the two individuals to be crossed, P cmax is the upper limit of the adaptive crossover rate, P cmin is the lower limit of the adaptive crossover rate;

[0088] d-1-5). Mutation operation, using the double-point mutation method, select an individual, randomly generate two integers a and b between 1 and n, set a<b, randomly generate an integer c and d between 1 and m, and replace the gene values ​​corresponding to points a and b with c and d;

[0089] d-1-6). Adaptive mutation rate: Adaptive mutation rate as shown in formula (5):

[0090]

[0091] P m max is the upper limit of the adaptive mutation rate, P c min is the lower limit of the adaptive mutation rate;

[0092] d-1-7). Parent update strategy: In order to make the best solution as close to the optimal solution as possible and improve the computational efficiency of the subsequent simulated annealing algorithm, a multi-stage parent update strategy is adopted; inter is used to control the number of stages, and the condition for exiting the iteration at each stage is fmax <f min That is, when this condition is met, exit the current generation process and regenerate a new population; at the same time, replace the worst individual in the new population with the current best individual.

[0093] In this step,

[0094] The further optimization using the simulated annealing algorithm near the approximate area of ​​the optimal solution is achieved by the following steps:

[0095] d-2-1). Assign the optimal individual found by genetic algorithm and its fitness to R a 、f a , and let the current temperature t = t0, t0 is the set value;

[0096] d-2-2). Determine if t>t f Is it true? If so, the current individual is the optimal solution and the optimal solution is output; if not, execute step d-2-3); t f is to set the temperature threshold;

[0097] d-2-3). Let i = 1;

[0098] d-2-4). Determine whether i≤L is established. If so, execute step d-2-5); if not, execute the cooling operation of t=t×alpha, alpha<1, and jump to step d-2-2);

[0099] d-2-5). By R a Mutation produces R b ; Execute step d-2-6);

[0100] d-2-6). Calculate R b The corresponding fitness f b ;

[0101] d-2-7). Determine whether f is true b >f a , if it holds, then accept this new solution and let R a =R b 、f a =f b ; If not, generate a random real number rand from 0 to 1 and judge rand <exp((f a -f b ) / t) is true, if true, accept the new solution, otherwise discard the new solution; execute step d-2-8);

[0102] d-2-8). Execute step i=i+1 and go to step d-2-4).

[0103] In summary, the multi-objective manufacturing task optimization allocation method of the present invention, aimed at the task allocation problem of the manufacturing workshop, proposes a multi-objective optimization allocation model that comprehensively considers the three objectives of the total processing time of the task, the task completion time and the total quality of the task completion, and takes the manufacturing equipment load rate as a constraint, and on this basis, combines the respective advantages of the genetic algorithm and the simulated annealing algorithm, and proposes and designs a multi-stage parent generation update adaptive genetic-simulated annealing algorithm. The application example shows that the model established by the present invention is correct, and the proposed multi-stage parent generation update genetic-simulated annealing algorithm has the advantages of strong convergence and high computational efficiency. The optimization model and its algorithm can provide auxiliary decision-making information for workshop production management and multi-objective optimization allocation of manufacturing tasks.

Claims

1. A multi-objective manufacturing task optimization allocation method, characterized in that: It is achieved through the following steps: a). Establishment of task and equipment set. Suppose there are n independent manufacturing tasks and m manufacturing equipment in total. The n independent manufacturing tasks are represented by the set Task = {T1, T2, ..., T n } indicates that T i is the i-th task; The set P for m manufacturing equipment is P = {P1, P2, ..., P m } indicates that P j is the jth manufacturing equipment; the tasks in the set Task must be allocated to all m manufacturing equipment, and one task can only be processed independently by one manufacturing equipment; b). Condition setting: Assume that the decision maker assigns tasks in cycles of CT, and different task sets Task are assigned in different cycles. The manufacturing equipment executes the assigned processing tasks without delay. Manufacturing Equipment k Once the manufacturing task begins, i , cannot be interrupted to process another manufacturing task T i′ , the same manufacturing equipment can only process one manufacturing task at the same time and cannot process multiple tasks in parallel; c). Construction of the optimization model: It is achieved through steps c-1) to c-2): c-1). Establish a decision variable set, and use an n×m 0-1 matrix X to represent the decision variable set. The value range of the elements in the matrix X is 0 or 1. ij =1 indicates task T i Assigned to manufacturing equipment P j , X ij =0 indicates task T i Not allocated to manufacturing equipment P j ; c-2). Establish known boundary conditions and build an n×m processing time matrix T. The element t in the processing time matrix T is ij Represents task T i By manufacturing equipment P j The processing time required for processing; establish an n×m processing quality matrix Q, and the element q in the processing quality matrix Q ij Represents task T i By manufacturing equipment P j The quality level during processing; Establish an m×3 equipment load parameter matrix E, and the element T in the equipment load parameter matrix E aj represents the effective utilization time of the jth manufacturing equipment in the time period CT, r j1 is the lower limit of load factor constraint, r j2 is the upper limit of load constraint rate; Establish the mathematical models shown in formulas (1) and (2): OPTf(X)=OPT(f1(X),f2(X),f3(X))=(minT total ,minT end ,maxQ total ) (1) In formula (1), OPTf(X) is the objective function of the optimization model, f1(X), f2(X), and f3(X) are the corresponding sub-objective functions, which are the shortest total processing time of the task, the earliest completion time of all tasks to be assigned within the CT time, and the optimal sub-objective function of the completion quality of all tasks, that is, f1(X) = min T total f2(X) = min T end f3(X) = max Q total ; The constraint conditions are: OPT f(x) is the objective function, T total is the total processing time of the task, T end Q is the earliest completion time of all tasks to be assigned within CT time. total For the quality of all tasks, T max Indicates the upper limit of the total processing time of the task, T emax Q is the earliest completion time limit of the task to be assigned within the CT time. min is the lower limit of the total quality level of the task, T aj is the effective utilization time of the jth device; d) To optimize the objective function, first unify the three objectives of the objective function in formula (1) into a single objective: i (x) The exponential method in the power factor method is used to convert it into a power factor d with a value between 0 and 1. i , i = 1, 2, 3; and take its geometric mean as the new objective function; the converted single objective function is shown in formula (3): T 1 is the qualified threshold of the total processing time of the task, and the corresponding efficiency coefficient is e -1 =0.3679; T 0 is the qualified threshold of the total processing time of the task, and its efficiency coefficient is e -e =0.06598; is the task completion time qualification threshold, is the unqualified threshold for task completion time, Q 1 is the total quality threshold for task completion, Q 0 T is the total quality failure threshold for task completion; 1 , T 0 , Q 1 , Q 0 The value of reflects the importance that decision makers attach to each goal; Then, a method combining the genetic algorithm and the simulated annealing algorithm is used for optimization: First, the approximate region of the optimal solution is found through the genetic algorithm, and then the simulated annealing algorithm is used for further optimization near the approximate region of the optimal solution until the optimal decision variable set matrix x is obtained, where x ∈ X. The n independent manufacturing tasks are assigned to m manufacturing equipment for processing according to the optimal decision variable set matrix x.

2. The multi-objective manufacturing task optimization allocation method according to claim 1 is characterized in that: The step of finding the approximate region of the optimal solution through the genetic algorithm described in step d) is achieved through the following steps: d-1-1). Chromosome encoding: The variables used as chromosomes are encoded using the integer encoding rule. d-1-2). Genetic operation: A genetic algorithm combining roulette wheel selection and elitist retention is used for genetic manipulation. d-1-3). Crossover operation: The two-point crossover method is adopted. Select two parent individuals, randomly generate two integers a and b between 1 and R, assuming a < b. Exchange the gene values between the a-th and b-th positions of the two parent individuals as a whole, and keep the other gene values unchanged. d-1-4). Adaptive crossover rate: To ensure the stability of good solutions and increase the possibility of inferior solutions transforming into good solutions, different crossover rates are used for different individuals. That is, when the fitness of an individual is lower than the average fitness of the generation population, a larger crossover rate should be used. Conversely, a smaller crossover rate should be given. The adaptive crossover rate is shown in formula (4): In the formula, f max is the maximum fitness value of individuals in the group, f avg is the average fitness value of individuals in the group, f is the larger fitness value of the two individuals to be crossed, P cmax is the upper limit of the adaptive crossover rate, P cmin is the lower limit of the adaptive crossover rate; d-1-5). Mutation operation: The two-point mutation method is adopted. Select an individual, randomly generate two integers a and b between 1 and S, assuming a < b, and randomly generate two integers c and d between 1 and S. Replace the gene values corresponding to points a and b with c and d. d-1-6). Adaptive mutation rate: The adaptive mutation rate shown in formula (5) is adopted: P mmax is the upper limit of the adaptive mutation rate, P mmin is the lower limit of the adaptive mutation rate; d-1-7). Parent update strategy: In order to make the best solution as close to the optimal solution as possible and improve the computational efficiency of the subsequent simulated annealing algorithm, a multi-stage parent update strategy is adopted; inter is used to control the number of stages, and the condition for exiting the iteration at each stage is f max <f min That is, when this condition is met, exit the current generation process and regenerate a new population; at the same time, replace the worst individual in the new population with the current best individual.

Images