Method and device for optimizing production line configuration based on collaboration of digital twin and double-population genetic algorithm

Abstract

The present invention discloses a method and device for optimizing production line configuration based on the collaboration of digital twin and double-population genetic algorithm. The steps of the method include: constructing a twin model of the PCB production line; constructing a single-line layout model of the PCB production line, and constructing a multi-objective programming model with the goals of maximum production efficiency, maximum equipment utilization rate, and minimum logistics cost; obtaining twin data in real time, obtaining the priority order of each process unit of the PCB production line according to the twin data, using the double-population genetic algorithm to encode the priority order of each process unit and the interval distance between each process unit respectively, and using the fitness function constructed based on the multi-objective programming model to perform iterative solution to obtain an optimized configuration plan; mapping the optimized configuration plan obtained by iterative solution to the twin model to evaluate the optimization effect. The present invention can efficiently generate accurate and reliable optimization plans for the resource configuration of the PCB production line, and improve the production line efficiency and equipment utilization rate.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent manufacturing of PCB (Printed Circuit Board), and particularly to a method and device for optimizing production line configuration based on the collaboration of digital twin and double-population genetic algorithm. Background Art

[0002] The manufacturing process of PCB includes a series of complex technological processes, usually including processes such as blanking, pretreatment, inner layer processing, inner layer lamination, drilling, chemical / electroplating copper deposition, outer layer image transfer, outer layer circuit, silk screen process, electrical testing, forming inspection and packaging, etc. Each process usually also includes multiple sub-processes. For example, the inner layer processing specifically includes sub-processes such as cleaning and baking, exposure, development and etching, stripping and inspection, etc. And in the whole production process, each process needs to be precisely controlled to ensure the quality and performance of the PCB. Therefore, the entire PCB production line is relatively complex, and there will be large-scale resource scheduling. The resource allocation method will directly affect the efficiency of PCB manufacturing and the utilization rate of production equipment.

[0003] For the resource scheduling of complex production systems, in the prior art, methods based on operations research, heuristic scheduling rules, intelligent algorithms, simulation methods, etc. are usually adopted, but each method has certain limitations. Among them, the method based on operations research establishes a mathematical programming model and selects the optimal scheduling scheme from all possible combinations by using methods such as branch and bound method and dynamic programming method. However, since the optimization solution of production line resource scheduling belongs to the category of "NP (NP-hard) difficult", this type of method based on operations research will be subject to certain applicable limitations. The simulation-based method verifies the effect of resource scheduling through simulation. Compared with the traditional manual scheduling method, it can improve the scheduling efficiency. However, since it needs to rely on the establishment and design of the simulation model, the accuracy will be limited by the simulation model. The heuristic method based on intelligent algorithms guides the search process by using heuristic information to find an approximate solution or a feasible solution. However, the solution obtained by this type of method is only a local optimal solution and cannot obtain the global optimal solution. When applied to solve the resource scheduling problem of complex production systems, it is difficult to directly obtain the globally optimal resource scheduling scheme. Summary of the Invention

[0004] The technical problem to be solved by the present invention lies in: aiming at the technical problems existing in the prior art, the present invention provides a method and device for optimizing production line configuration based on the collaboration of digital twin and double-population genetic algorithm, which can efficiently generate accurate and reliable optimization schemes for PCB production line resource configuration, thereby improving the production line efficiency and equipment utilization rate, and at the same time reducing the overall production cost.

[0005] To solve the above technical problems, the technical solution proposed by the present invention is:

[0006] An optimization method for production line configuration based on the collaboration of digital twin and double-population genetic algorithm, the steps include:

[0007] Construct a digital twin model of the PCB production line. In the model, each process unit is arranged at intervals in a single-row layout manner, and each process unit corresponds to a process.

[0008] Construct a single-row layout model of the PCB production line. Equivalent each process unit in the PCB production line to multiple rectangles in a single-row layout, and there is a specified interval distance between each process unit. A multi-objective programming model is constructed with the goals of maximum production efficiency, maximum equipment utilization rate, and minimum logistics cost.

[0009] Obtain the twin data generated by the digital twin model of the PCB production line in real time. According to the twin data, obtain the priority order of each process unit in the PCB production line. Use the double-population genetic algorithm to encode the priority order of each process unit and the interval distance between each process unit respectively, and use the fitness function constructed based on the multi-objective programming model for iterative solution to obtain the priority order of each process unit in the PCB production line and the optimized configuration plan of the interval between each process unit. When using the double-population genetic algorithm, the first population is the process sequence population, which is used to encode the priority order of each process unit, and the second population is the net spacing population, which is used to encode the interval distance between each process unit.

[0010] Map the optimized configuration plan obtained by iterative solution using the double-population genetic algorithm to the digital twin model of the PCB production line, and evaluate the optimization effect according to the twin data generated by the digital twin model of the PCB production line.

[0011] Furthermore, when constructing the multi-objective programming model, use the sequential algorithm to determine the priority order of each goal. The constructed multi-objective programming model is:

[0012]

[0013] Among them, z represents the objective function, P k is the priority factor corresponding to the kth goal, and are the positive and negative deviation amounts of the kth goal deviating from the target value respectively, represent the positive and negative deviation amounts of the first goal deviating from the target value respectively, represent the positive and negative deviation amounts of the second goal deviating from the target value respectively, represent the positive and negative deviation amounts of the third goal deviating from the target value respectively, and They are the weight values of the positive and negative deviation amounts corresponding to the kth target. α1, α2, and α3 correspond to the expected values of the production efficiency target, the equipment utilization rate target, and the total transportation distance target respectively. N is the total output of PCBs within the specified duration, T is the total production time per unit time, and d ij is the net interval distance between the center points of the ith process unit and the jth process unit. C is the fixed transportation cost per unit distance, n represents the number of process units, and T i represents the processing time of the ith process unit.

[0014] Furthermore, the total output of PCBs N within the specified duration is calculated as follows:

[0015] N = T / (T p + T on )

[0016]

[0017] where T p is the time required to process one PCB, T on is the feeding interval time, t i is the processing time of the ith process unit, v is the material transfer speed per unit distance, and △x ij represents the net interval distance between the boundaries of the ith process unit and the jth process unit.

[0018] Furthermore, when constructing the multi-objective programming model, multi-objective constraint conditions are also included. The multi-objective constraint conditions include spatial constraints, interval constraints, and boundary constraints. The spatial constraints are the constraint conditions that each process unit needs to satisfy within the spatial range. The interval constraints are the conditions that the interval distances between each process unit need to satisfy. The boundary constraints are the conditions that the boundaries between each process unit need to satisfy. The calculation expressions of the multi-objective constraint conditions are:

[0019]

[0020] where x i , x j are the center coordinates of the ith and jth process units respectively, D x , D y are the minimum distances between the ith and jth process units and other process units or the wall in the x-axis and y-axis directions respectively, L i and L j are the lengths of the ith process unit respectively, W i is the width of the ith process unit, L and W are the length and width of the production workshop respectively, and △x ijIt represents the net interval distance between the boundary of the i-th process unit and the boundary of the j-th process unit.

[0021] Further, when using the double-population genetic algorithm, the partially matched crossover method is selected for the crossover operator to determine the crossover point and exchange the middle part of the genes; the mutation operator adopts the random mutation method, randomly generates a mutation point with a specified probability, divides the chromosome into two front and back segments, the front segment of the chromosome remains unchanged, the back segment of the chromosome is gene-recombined according to the priority matrix, and the recombined segment is spliced with the front segment of the chromosome to generate the offspring chromosome, and gene recombination is performed according to the priority matrix to make the mutated chromosome satisfy the relative sequence of processes.

[0022] Further, when using the double-population genetic algorithm, the fitness function is constructed based on the multi-objective programming model using the penalty function, and the expression of the fitness function is:

[0023] Fit = 1 / (z + P)

[0024]

[0025] Where, Fit represents the fitness function, z represents the objective function of the multi-objective programming model, P is the penalty function, and Q is the penalty value.

[0026] Further, when using the double-population genetic algorithm, the roulette wheel selection method is adopted for the selection operator to select the next population, and the probability C that an individual is selected and inherited to the next generation i is:

[0027]

[0028] Where, Fit(i) represents the value of the fitness function of the i-th individual, and groupsize represents the number of all individuals in the population.

[0029] Further, when using the double-population genetic algorithm, the partially matched crossover method is selected for the crossover operator to determine the crossover point and exchange the middle part of the genes, and the mutation operator adopts the random mutation method, randomly generates a mutation point with a certain probability, divides the chromosome into two front and back segments, the front segment of the chromosome remains unchanged, the back segment of the chromosome is gene-recombined according to the priority matrix, and the recombined segment is spliced with the front segment of the chromosome to generate the offspring chromosome.

[0030] Further, when using the double-population genetic algorithm, a co-evolution strategy based on similarity and dynamic mapping is adopted for regular information exchange between the two populations, and the steps include:

[0031] Determine the mapping relationship: Search for individuals in the net spacing population and the process sequence population whose fitness values differ by no more than a preset threshold. Count the corresponding gene positions of each individual pair among all individuals in the two populations, and obtain the mapping relationship pairs of multiple gene values of the two populations based on the gene value pairs with the most corresponding relationships between the chromosomes of the two populations at the corresponding positions.

[0032] Information exchange based on similarity: Exchange the information of the net spacing population and the process sequence population according to the similarity between the two populations. The steps include:

[0033] Take the optimal individual with the highest fitness in the net spacing population as the current individual of the net spacing population;

[0034] Calculate the similarity index between the current individual of the net spacing population and all individuals in the process sequence population respectively, and find the individual with the highest similarity, that is, the individual of the process sequence population with the highest similarity to the current individual of the net spacing population;

[0035] Exchange the gene values of the current individual of the net spacing population and the found individual of the process sequence population with the highest similarity according to the mapping relationship of gene value pairs. Specifically, replace the first N bits of the chromosome of the individual of the process sequence population with the highest similarity with the corresponding gene values of the first N bits in the current individual of the net spacing population, and at the same time replace the central coordinate genes of the last N bits of the chromosome of the individual of the process sequence population with the highest similarity with the central coordinate genes of the last N bits in the current individual of the net spacing population. Then judge whether the exchanged gene values meet the priority order. If not, the exchange fails, abandon the current exchange, take the individual with the next highest fitness in the net spacing population as the current individual of the net spacing population, and continue the exchange. If the exchange fails for all individuals until the last one with the highest fitness, directly enter the next iteration. If the exchanged gene values meet the priority order, update the population with the current exchanged result and enter the next iteration. N represents the number of PCB processes;

[0036] After completing the iteration, finally obtain the relatively optimal solutions of the net spacing population and the process sequence population.

[0037] A production line configuration optimization device based on the cooperation of digital twin and double-population genetic algorithm, including a processor and a memory. The memory is used to store computer programs, and the processor is used to execute the computer programs to execute the method as described above.

[0038] A computer-readable storage medium storing a computer program, where the computer program, when executed by a processor, implements the method as described above.

[0039] Compared with the prior art, the advantages of the present invention are:

[0040] 1. The present invention synergistically interacts the digital twin technology with the double-population genetic algorithm. The digital twin technology is used to simulate the PCB production line resource allocation scheme, and the double-population genetic algorithm is used to rapidly and iteratively solve the optimal PCB production line resource allocation scheme based on multiple objectives of maximum production efficiency, maximum equipment utilization rate, and minimum logistics cost. It can effectively improve the efficiency, accuracy, and flexibility of resource allocation optimization. When using the double-population genetic algorithm to solve the optimal allocation scheme, the priority order of each process unit and the interval distance between each process unit correspond to different populations respectively, and it can also comprehensively consider the priority order of process units and the interval distance between process units to further improve the reliability of resource optimization.

[0041] 2. The present invention further takes into account the specific priorities and relative orders between production processes. Based on the digital twin technology, an improved double-population genetic algorithm is adopted. On the basis of using the random mutation method for mutation operation, gene recombination is carried out according to the priority matrix to ensure that the mutated chromosomes also meet the relative sequence of processes, thereby further ensuring the reliability of production line configuration optimization.

[0042] 3. When the present invention adopts the improved double-population genetic algorithm to determine the priority order of process units and the interval distance between process units, further regular information exchange between the two populations of process sequence population and net spacing population is carried out through a co-evolution strategy based on similarity and dynamic mapping, so as to quickly and accurately solve the optimal solution of the double-population that meets the priority order by matching different chromosome coding methods and gene representation ranges, thereby determining the optimal priority order of process units and the optimal interval distance between process units. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 is a schematic flowchart of the implementation of the production line configuration optimization method based on the collaboration of digital twin and double-population genetic algorithm in this embodiment.

[0044] Figure 2 is a schematic diagram of the implementation architecture principle of the PCB production line configuration optimization based on the collaborative interaction of the double-population genetic algorithm and the digital twin technology in this embodiment.

[0045] Figure 3 is a schematic diagram of the principle of the single-line layout mathematical model constructed in this embodiment.

[0046] Figure 4 is a schematic diagram of the effect of the PCB production line twin model constructed in this embodiment.

[0047] Figure 5 is a directed graph obtained by converting the process distribution table into a process flow in the specific application embodiment of the present invention.

[0048] Figure 6 It is a schematic diagram of the detailed implementation process of using the digital twin-based double-population genetic algorithm to optimize the resource allocation of the PCB production line in this embodiment.

[0049] Figure 7 It is a schematic diagram of the layout results of each working unit of the PCB production line obtained in the specific application embodiment of the present invention.

[0050] Figure 8 It is a schematic diagram of the sub-processes and their sequences on each process unit obtained in the specific application embodiment of the present invention. Specific Embodiments

[0051] The present invention will be further described below in conjunction with the accompanying drawings of the specification and specific preferred embodiments, but the protection scope of the present invention is not limited thereby.

[0052] Digital twin, as a virtual simulation technology, creates a twin of a physical entity, constructs a two-way connection between the virtual and the real, enables the virtual model to simulate the behavior of the physical entity, and realizes the integration of the physical world and the information world. Applying digital twin technology can achieve precise simulation and optimization of the production process, so the resource allocation of large-scale production lines can be optimized using digital twin technology. However, the manufacturing process of PCBs includes a series of complex technological processes (such as blanking, pretreatment, inner layer processing, inner layer lamination, drilling, chemical / electroplating copper deposition, outer layer image transfer, outer layer circuit, silk screen process, electrical testing, forming inspection and packaging, etc.). Each process includes multiple sub-processes, and each process needs to be precisely controlled. Different process sequences and different intervals between process units will affect the overall production efficiency, equipment utilization rate, production cost, etc. Digital twin technology can only simulate specific resource allocation schemes. Different process sequences and different gaps between process units can form a large number of resource allocation schemes, and it is impossible to quickly simulate the optimal resource allocation scheme directly using digital twin technology.

[0053] The present invention synergistically interacts the digital twin technology with a heuristic intelligent algorithm (double-population genetic algorithm), uses the digital twin technology to simulate the PCB production line resource allocation scheme, and uses the double-population genetic algorithm to quickly iteratively solve the optimal resource allocation scheme for the PCB production line based on multiple objectives of maximum production efficiency, maximum equipment utilization rate, and minimum logistics cost. The interaction of the digital twin technology and the double-population genetic algorithm can effectively improve the efficiency, accuracy, and flexibility of resource allocation optimization. When using the double-population genetic algorithm to solve the optimal allocation scheme, the priority order of each process unit and the interval distance between each process unit correspond to different populations, and the priority order of the process unit and the interval distance between the process units can be comprehensively considered to further improve the reliability of resource optimization, so as to obtain a high-precision and high-reliability production line resource optimal allocation scheme.

[0054] As Figures 1 - 2 shown, the steps of the production line configuration optimization method based on the collaboration of digital twin and double-population genetic algorithm in this embodiment include:

[0055] Step S01. Construct a twin model of the PCB production line, in which each process unit is arranged at intervals in a single-row layout manner, and each process unit corresponds to a process.

[0056] In this embodiment, the twin model of the PCB production line process structure is first analyzed, including constructing a device geometric model, rendering the device model, building a production line scene and then rendering the scene. Further, a corresponding specification model and a corresponding setting rule model can be set for the device behavior specification and the personnel behavior specification.

[0057] Optionally, the MIoT.VC Meiqin simulation software can be used to construct a twin model of the PCB production line. Through this software, the modeling (importing the model), model layout parameter setting, and model dynamic simulation of the production line simulation system can be realized. Among them, the processing equipment Shrink Tunnel can be used to build equipment such as a laminator, an exposure machine, a solder mask printer, a developer, an acid-alkali etching machine, and a stripping machine. The ST-30 model can be used as a hot air leveling workstation and a drying oven required for the baking step. The EffiMat Box model can be used as a cleaning machine and an electrolytic electroplating tank. The Automatic Shrink Wrap Machine model can be used as a packaging device. In addition, the Conveyor model (3D model of the conveyor) in the model library is selected as the conveyor belt, and the KR 10R1420 model is used as the robotic arm, and at the same time, a smart car (Robot TransportController) model is equipped.

[0058] Since the steps of the PCB production process flow form a directed acyclic graph, and to simplify the spatial layout, facilitate management and material flow, reduce the handling time between processes, and improve production efficiency, in this embodiment, the PCB production line is arranged in a single-line layout. At the same time, the production workshop is equivalent to a rectangle W with length and width, and each process unit is equivalent to a rectangle with length and width. Each rectangle corresponds to a process unit, and the floor areas of each process unit are the same and do not overlap. Each process unit corresponds to one process, that is, all work units are arranged in a straight line, and there is a specified net interval distance between the boundaries of adjacent process units. The single-line layout mathematical model constructed in this embodiment is as Figure 3 shown, where L and W are the length and width of the production workshop respectively, L i and W i are the length and width of the i-th process unit respectively, M i is the number of the i-th work unit, x i is the central coordinate of the i-th work unit, D x and D y are the minimum distances between the i-th and j-th process units and other process units or walls in the x-axis and y-axis directions respectively, and △x ij represents the net interval distance between the boundary of the i-th process unit and the boundary of the j-th process unit.

[0059] For ease of analysis, in this embodiment, some processes in the PCB production line are simplified, and the simplified process flow is shown in Table 1.

[0060] Table 1: Sub-process table and its labels

[0061]

[0062]

[0063] Based on the above division of each process of the PCB production line, a single-line layout of the PCB production line is carried out, that is, all processes are arranged in a straight line. The production line twin body is constructed mainly considering the order of sub-processes and the center distance between processes. In a specific application embodiment, the PCB production line twin body model constructed in an equidistant manner is as Figure 4 shown. By adjusting the relative order of the production process and the net distance between each process, a new model can be re-constructed, and the PCB production system twin data set is composed of the operation data of the PCB production line twin body model. To intuitively reflect the sequence and parallel relationship between PCB processes, based on the ECRS principle, the process distribution table is converted into a process flow directed graph, as Figure 5 shown.

[0064] Step S02. Construct a single-line layout model for the PCB production line. Equivalent each process unit in the PCB production line to multiple rectangles in a single-line layout, and there is a specified interval distance between each process unit. A multi-objective programming model is constructed with the goals of maximizing production efficiency, maximizing equipment utilization rate, and minimizing logistics cost.

[0065] Since unreasonable layout order and spacing of the production line work units will lead to problems such as high material handling frequency, low production efficiency, and low equipment utilization rate, in this embodiment, the multi-objective function is to maximize production efficiency, high equipment utilization rate, and minimize logistics cost. Assume that each process unit can be arbitrarily arranged within the workshop space limit, the facility placement positions in each process unit in the workshop are parallel to the workshop walls, the geometric centers of each production equipment are on a straight line, the conveyor belt flow between each production equipment can only move in the direction of the axis parallel line, the materials are transferred in a straight line during the handling process of the intelligent trolley robotic arm or manually, the PCB production sequence follows the layout order of each production equipment on the production line, and there are no faults or outages of each equipment on the PCB production line during the total production time of a day. In this embodiment, a multi-objective function design is carried out for the PCB production line. The first objective is to maximize the production efficiency f1, that is, the number of PCB printed circuit boards produced per unit time is the largest, and it can be calculated by the following formula:

[0066]

[0067] where N is the total output of PCBs within a specified time, T is the total production time per unit time (such as per day), T p is the time required to process one PCB printed circuit board, T on is the feeding interval time, T p +T on is the total production time of one PCB product, T i is the processing time of the i-th process unit (i.e., the processing time corresponding to the i-th process), v is the material transfer speed per unit distance, △x ij represents the net interval distance between the boundaries of the i-th and j-th work units.

[0068] In this embodiment, the second objective is to maximize the equipment utilization rate f2, that is, the proportion of the actual production time of the production equipment per unit working time is the largest. The more operation tasks are completed per unit time, the higher the corresponding equipment utilization rate. Maximizing the equipment utilization rate f2 can be calculated by the following expression:

[0069]

[0070] In this embodiment, the third objective is to minimize the logistics cost f3, that is, the total transportation distance multiplied by the logistics cost per unit distance is the lowest, and it can be calculated by the following expression:

[0071]

[0072] Among them, C is the fixed transportation cost per unit distance, and d ij is the net interval distance between the center points of the i-th process unit and the j-th process unit. Further, the constraint conditions of the spatial constraint can be set. For example, the placement of the work unit in the x-axis direction and the width in the y-axis cannot exceed the spatial range of the workshop.

[0073] To construct the final multi-objective programming model, in this embodiment, the sequential algorithm (Sequential Algorithm) is used to determine the priority order of each objective. The sequential algorithm processes only one decision or calculation at each step and then continues with the next step based on the current state. The optimal solutions of the high-priority objectives obtained are used as fixed constraint conditions in the optimization process of the subsequent low-priority objectives, and the problem can be solved step by step in order.

[0074] Taking the construction of each objective according to the above formulas (1), (2), and (3) as an example, the multi-objective programming model finally constructed using the sequential algorithm can be expressed as:

[0075]

[0076] Among them, z represents the objective function, and P k is the priority factor corresponding to the k-th objective, and are the positive and negative deviation amounts of the k-th objective from the target value respectively, represent the positive and negative deviation amounts of the first objective from the target value respectively, represent the positive and negative deviation amounts of the second objective from the target value respectively, represent the positive and negative deviation amounts of the third objective from the target value respectively, and are the weight values of the positive and negative deviation amounts corresponding to the k-th objective respectively. α1, α2, and α3 correspond to the expected values of the production efficiency objective, the equipment utilization rate objective, and the total transportation distance objective respectively. N is the total output of PCBs within the specified duration, T is the total production time per unit time, and d ij is the net interval distance between the center points of the i-th process unit and the j-th process unit, C is the fixed transportation cost per unit distance, n represents the number of process units, and T i represents the processing time of the i-th process unit.

[0077] Optionally, to solve for the maximum production efficiency and the maximum equipment utilization rate, the expected values of production efficiency and equipment utilization rate can be set as: α1 = 0, α2 = 0. To solve for the minimum logistics cost, the expected value of logistics cost is set as α3 = R (R is a constant).

[0078] Furthermore, when constructing the multi-objective programming model in this embodiment, it also includes setting multi-objective constraint conditions. The multi-objective constraint conditions include space constraint, interval constraint, and boundary constraint. The space constraint is the constraint condition that each process unit needs to satisfy within the spatial range. The interval constraint is the condition that the interval distance between each process unit needs to satisfy. The boundary constraint is the condition that needs to be satisfied between the boundaries of each process unit. Optionally, the space constraint can be configured such that the placement of the process unit in the x-axis direction and the width in the y-axis direction do not exceed the spatial range of the workshop. The interval constraint can be configured such that the process units do not overlap pairwise. Specifically, the calculation expression of the multi-objective constraint conditions can be expressed as:

[0079]

[0080] where x i and x j are the central coordinates of the i-th and j-th process units respectively, D x and D y are the minimum distances between the i-th and j-th process units and other process units or the wall in the x-axis and y-axis directions respectively, L i and L j are the lengths of the i-th process unit respectively, W i is the width of the i-th process unit, L and W are the length and width of the production workshop respectively, and △x ij represents the net interval distance between the boundary of the i-th process unit and the boundary of the j-th process unit.

[0081] Step S03. Obtain the twin data generated by the twin model of the PCB production line in real time, obtain the priority order of each process unit of the PCB production line according to the twin data, encode the priority order of each process unit and the interval distance between each process unit using the double-population genetic algorithm, and perform iterative solution using the fitness function constructed based on the multi-objective programming model to obtain the priority order of each process unit in the PCB production line and the optimized configuration scheme of the interval between each process unit. When using the double-population genetic algorithm, the first population is used to encode the priority order of each process unit, and the second population is used to encode the interval distance between each process unit.

[0082] In this embodiment, the twin data acquisition platform collects and stores in real time the twin data generated during the operation of the twin body model, provides the real-time generated twin data to the double-population genetic algorithm as input data. The double-population genetic algorithm encodes the priority order of each process unit and the interval distance between each process unit according to the twin data. After iterative solution, an optimized configuration plan is obtained, and the solved optimized configuration plan is provided to the twin body model for collaborative interaction, that is, the iterative solution and verification of the PCB production line configuration plan can be realized, and finally an optimized configuration plan of the PCB production line that meets the requirements is determined.

[0083] In the genetic algorithm, through encoding, complex actual production problems can be transformed into mathematical sequence logic. In this embodiment, when optimizing the resource scheduling of the PCB production line, two factors, namely the layout of the production process sequence and the net interval distance of the boundaries of each working unit, are considered simultaneously, corresponding to the double-population encoding method. The first population is the process sequence population, which is used to encode the priority order of each process unit (the process production sub-process order), and the second population is the net interval population, which is used to encode the interval distance between each process unit.

[0084] For example, when using the double-population genetic algorithm, the first population (process sequence population) is encoded according to F1 = [M1, M2, …, M i , …, M n , x1, x2, …, x i , …, x n , that is, the priority order of the process unit and the center distance are combined for encoding, where M i represents the i-th process unit, and x i represents the center coordinates of the i-th process unit. The specific encoding method of the first population is designed as: F1 = [M1, M2, …, M i , …, M 16 , x1, x2, …, x i , …, x 16 . M i is the i-th working unit, M i = 1, 2, 3, …, 16; if M i = p, p ∈ [1, 16], it means that the sub-process p of the PCB production line is located in the i-th working unit. x i is the center coordinate position of the i-th working unit. The sub-process order included in each process can be expressed as: [1, 2, 3, [4, 5], 6, [7, 8], 9, 10, [11, 12, 13], 14, 15, 16]. Taking M as an example, the chromosome [M1, M2, …, M i , …, M 16It is [2, 1, 3, 5, 4, 6, 7, 8, 9, 11, 10, 12, 13, 14, 15, 16]. Through the encoding method of this chromosome, the priority assignment order of sub-process operations can be intuitively understood. That is, the first operation is sub-process p02, the second operation is sub-process p01, the third operation is p03, and so on. The chromosome of the encoding method adopted in this embodiment contains 32 gene positions. Since the order of sub-processes in the process can be adjusted arbitrarily, the initial population can be generated:

[0085]

[0086] The second population (net spacing population) can be encoded according to F2 = [△x 0,1 , △x 1,2 , …, △x i,j , …, △x 15,16 , x1, x2, …, x i , …, x 16 , that is, the equipment spacing and the center distance are combined and encoded. △x ij represents the net spacing distance between the boundary of the i-th process unit and the boundary of the j-th process unit, and the initial population of the second population is iteratively generated according to the maximum and minimum values of D x , D y . D x , D y are respectively the minimum spacing between the i-th and j-th process units and other process units or the wall in the x-axis and y-axis directions.

[0087] For example, the specific encoding method for the net spacing population to encode using the net spacing distance between the boundaries of work units can be: F2 = [△x 0,1 , △x 1,2 , …, △x i,j , …, △x 15,16 , x1, x2, …, x i , …, x 16 . The chromosome of this encoding method also contains 32 gene positions. The minimum spacing D x can be set to 0.5m (configurable), the maximum spacing is 2m (configurable), and the minimum difference in spacing is 0.1m (configurable). That is, an initial population can be iteratively generated at intervals of 0.1, that is, the range of the generated initial population is:

[0088]

[0089] Among them, △x 0,1 represents the net spacing between the boundary of the first process unit and the wall.

[0090] The fitness function is the core criterion for transforming the objective function of a problem into a numerical form to evaluate the quality of individuals. Individuals with higher fitness values are more likely to be selected and passed on to the next generation, thereby driving the algorithm to search in a better solution space. The fitness function directly determines the performance of the genetic algorithm and the quality of the final solution. In the face of constraints, the fitness function can be made unconstrained by introducing a penalty function to ensure that the algorithm avoids solutions that violate the constraints during the search process. In this embodiment, the double-population genetic algorithm is used to solve the problem with the minimum value as the optimization objective. Since the higher the fitness value of the genetic algorithm, the better the individual, this embodiment can further improve the accuracy and reliability of the solution by using a penalty function in combination with the objective function to establish the fitness function.

[0091] Optionally, based on the multi-objective programming model, the penalty function can be used to construct the fitness function according to the following formula:

[0092] Fit = 1 / (z + P) (7)

[0093]

[0094] Where, Fit represents the fitness function, z represents the objective function of the multi-objective programming model, P is the penalty function, Q is the penalty value, usually a relatively large positive number, and as the value exceeding the constraint condition becomes larger, the Q value becomes correspondingly larger. The penalty function P can be set according to the required rigid constraint conditions. For example, it can be configured as:

[0095]

[0096] During the evolution process of the genetic algorithm, the selection operator selects individuals with higher fitness from the current population to reproduce to the next generation. Optionally, for the single-line layout multi-objective optimization model of the PCB production line, considering the complexity and multi-objective characteristics of the problem, when using the double-population genetic algorithm in this embodiment, the selection operator can use the roulette wheel selection method for the selection of the next population. Roulette wheel selection, as a proportional selection strategy based on the individual fitness value, can ensure that individuals with higher fitness have a greater probability of being selected, thereby driving the population to evolve towards a better solution. At the same time, by proportionally distributing the selection probability according to the fitness value, the selection pressure and population diversity can be effectively balanced.

[0097] Specifically, the probability C that an individual is selected and passed on to the next generation i is:

[0098]

[0099] Where, Fit(i) represents the value of the fitness function of the i-th individual, and groupsize represents the number of all individuals in the population. It represents the sum of the fitness values of all individuals in the population. It can be seen from the above formula that the higher the individual fitness, the higher the probability of being selected and inherited to the next generation, and it can take precedence over individuals with lower fitness, that is, the selection probability is positively correlated with the fitness.

[0100] The crossover operator determines how genetic information is transmitted and recombined among populations. In this embodiment, to achieve the multi-objective optimization of the single-line layout of the PCB production line, considering the complexity of layout optimization and the need to balance multiple performance indicators, the partially matched crossover (PMX) method is selected as the crossover operator when using the double-population genetic algorithm. By determining the crossover points and exchanging the middle part of the genes while maintaining the inheritance of the non-crossed part of the genes, it helps to maintain the diversity of individuals and the search space, thus avoiding premature convergence. By adopting the above method of finely adjusting the crossover probability, the search space can be guaranteed while improving the ability of the algorithm to search for excellent solutions locally.

[0101] For example:

[0102] The genes of the parent generation are: A = [1 2 3 4 5 6 7], B = [2 1 3 5 7 6 4];

[0103] Randomly select the crossover point positions as: A1 = [1 2|3 4 5|6 7], B1 = [2 1|3 5 7|6 4];

[0104] Exchange some genes in the middle segment to get: A2 = [1 2|3 5 7|6 7], B2 = [2 1|3 4 5|6 4];

[0105] Retain the non-repeated genes inherited from the parent generation genes, and use x to replace the repeated genes in the middle segment to form: A3 = [1 2|3 5x|6 7], B3 = [2 1|3x 5|6 4];

[0106] Determine the genes of the offspring according to the relative order of the parent generation genes and the middle segment, and obtain the non-repeated segments: A4 = [1 2|3 5 4|6 7], B4 = [2 1|3 7 5|6 4];

[0107] Attention should be paid to the process priority in the production line layout. In this embodiment, by judging whether the generated offspring chromosomes are qualified after each crossover operation, the offspring chromosomes that meet the sub-process priority matrix can be retained, so that the offspring chromosomes after the crossover operation still meet the process priority relationship, further ensuring the reliability of the production line configuration.

[0108] The mutation operator is a mechanism that simulates the phenomenon of gene mutation in biological evolution. It can simulate the randomness in biological evolution, helping the genetic algorithm search for the optimal solution in a broader solution space and enhancing the global search ability of the algorithm. In the genetic algorithm, the mutation operator randomly selects individuals in the population with a certain probability and changes some gene values in their gene coding strings, thereby introducing new gene combinations. Such changes can be operations such as replacement, insertion, or deletion, which can increase the diversity of the population and prevent the algorithm from prematurely converging to a local optimal solution. In this embodiment, the mutation operator adopts the random mutation method, randomly generates a mutation point with a certain probability, divides the chromosome into two front and back segments, keeps the front segment of the chromosome unchanged, and performs gene recombination on the back segment of the chromosome according to the priority matrix, and then stitches the recombined segment with the front segment of the chromosome to generate the offspring chromosome. Considering that there is a specific priority and relative order between production processes, in this embodiment, by improving the double-population genetic algorithm, on the basis of using the random mutation method for mutation operations, gene recombination is carried out according to the priority matrix to ensure that the mutated chromosome also satisfies the relative order of the processes, thereby further ensuring the reliability of the production line configuration optimization.

[0109] Since the chromosome coding methods and gene representation ranges of the two populations are different, it is not feasible to simply select the optimal individuals in the populations and exchange them. In this embodiment, the double-population genetic algorithm is further improved, and a co-evolution strategy based on similarity and dynamic mapping is adopted for regular information exchange between the two populations. The specific steps are as follows:

[0110] Step ① Determine the mapping relationship: Search for individuals whose fitness values differ by no more than a preset threshold (for example, 0.8), and count the corresponding relationships of the corresponding gene positions of the chromosomes of the individuals, that is, count the corresponding gene positions of each individual pair in all individuals in the two populations, and obtain the mapping relationship pairs of multiple gene values of the two populations according to the gene value pairs with the most corresponding relationships of the gene values at the corresponding positions of the two population chromosomes.

[0111] For example, the PCB production line has 16 processes. Then, the gene coding range of the first half of the chromosome of the process sequence population is [1, 16], the difference between genes is 1, there are 16 gene values and 16 gene positions. The gene coding range of the first half of the chromosome of the net spacing population is [0.5, 2], the difference is 0.1, and there are also 16 gene values and 16 gene positions. Among them, 0.5 is the distance D between the first process unit and the wall x , and 2 represents the distance D between the last process unit and the wall x , so the mapping relationship of the first 16 gene values of the two populations can be obtained one by one, that is, the mapping relationship pairs between the gene values of the chromosome of the net spacing population and the gene values of the chromosome of the process sequence population, which can be represented by (gene value of the chromosome of the net spacing population, gene value of the chromosome of the process sequence population).

[0112] For example, the gene value pairs at the same position of all chromosome genes in the process sequence population and the net spacing population are statistically analyzed. The gene value pair with the most statistics at the 5th index of the chromosome gene list is (0.9, 6), where 0.9 represents the process position and 6 represents the process number. The gene value pair (0.9, 6) is recorded in the mapping list as a gene value pair. This gene value pair only represents a mapping relationship and is independent of the position. If the gene value pair with the most statistics at the 7th index in the chromosome gene list is retrieved later as (0.9, 8), since the 0.9 in this gene value pair is the same as the value in the mapping relationship obtained from the previous index 5, and the mapping relationship must be one-to-one (the process position is unique), it is necessary to continue to select the gene value pair ranked second for judgment until the value in this gene value pair is not the same as the value in the previously obtained mapping relationship, that is, all the obtained mapping relationships must satisfy that the x value in each gene value pair (x, y) ranges from 0.5 to 2 (interval of 0.1), the y value ranges from 1 to 16 (interval of 1), and is one-to-one.

[0113] Step ② Information exchange based on similarity:

[0114] Step 2.1. Take the optimal individual with the highest fitness in the net spacing population as the current individual in the net spacing population;

[0115] Step 2.2. Calculate the similarity index (such as pearson or spearman coefficient) between the current individual in the net spacing population and all individuals in the process sequence population respectively, and find the individual in the process sequence population with the highest similarity to the current individual in the net spacing population;

[0116] Step S2.3 exchanges the gene values of the current net spacing population individuals and the process sequence population individuals with the highest similarity found in Step 2.2 according to the mapping relationship of the first 16 gene values obtained in Step ①. Specifically, the first 16 bits of the chromosome of the process sequence population individual with the highest similarity are replaced with the corresponding gene values of the first 16 bits in the current net spacing population individual, that is, the gene positions of the individual remain unchanged, only the gene values of the first 16 bits are changed. At the same time, the 16-bit central coordinate genes at the back of the chromosome of the process sequence population individual with the highest similarity are replaced with the 16-bit central coordinate genes at the back of the current net spacing population individual, that is, to ensure that the second half of the optimal solutions of the two populations finally obtained, namely the 16-bit genes of the central coordinates, are consistent. Then, it is judged whether the exchanged gene values meet the priority order. If not, the exchange fails, the current exchange is abandoned, the individual with the next fitness in the net spacing population is taken as the current net spacing population individual, and Step S2.2 is returned to continue the exchange. If the exchange fails for all individuals with the last fitness, it directly enters the next iteration. If the exchanged gene values meet the priority order, the population is updated with the current exchanged result and enters the next iteration;

[0117] Step S2.4. After completing the iteration, the relative optimal solutions of the net spacing population and the process sequence population are obtained.

[0118] For example, suppose the optimal individual with the highest fitness in the net spacing population is [1.1, 0.5, 2, 3.6, 9.1, 16.1], and suppose the individual of the process sequence population with the highest similarity calculated according to the similarity index is [1, 2, 3, 3, 9, 15, 3]. Also suppose the gene value pairs related to the first 3 genes of the net spacing are (1.1, 3), (0.5, 1), (2, 2). Then the individual of the process sequence population with the highest similarity is replaced with [3, 1, 2, 3.6, 9.1, 16.1]. However, due to the restriction of the sub-process priority in the coding rule of the process sequence population, the priority is [[1, 2], 3]. The position with subscript 0 in the list is a sub-list, and the values in the sub-list can be swapped arbitrarily, but the relative order in the priority list is fixed, that is, processes 1 and 2 take precedence over process 3, and the value "3" has subscript 1 and will be after the two values "1" and "2" in the sub-list with subscript 0. If the first 3 genes of the individual of the process sequence population after information exchange and replacement are 3, 1, 2, it will violate the coding rule of the process sequence population. Therefore, this individual is invalid, and then it continues to exchange information with the individual with the second highest similarity in the process sequence population. If it conforms to the coding rule of the process sequence population, the population can be updated and enter the next iteration to generate the next generation. If it does not conform to the rule, it continues to exchange information with the individual with the next fitness until the information exchange fails for the last fitness individual, ending the current information exchange and entering the next iteration.

[0119] In this embodiment, specifically, after each generation of genetic operator operations, information exchange is carried out between the two populations to ensure the co-evolution between the two populations, and finally the relative optimal solutions of the two populations are obtained, which can quickly and accurately solve the optimal solutions of the two populations that meet the priority order by matching different chromosome coding methods and gene representation ranges, so as to determine the priority order of the optimal process units and the optimal interval distance between the process units.

[0120] Step S04. Map the optimized configuration solution obtained by iterative solution using the double-population genetic algorithm to the twin model of the PCB production line, and evaluate the optimization effect according to the twin data generated by the twin model of the PCB production line.

[0121] After using the double-population genetic algorithm to iteratively solve the optimized configuration solution, the obtained optimized configuration solution is feedback-mapped to the twin model of the PCB production line. The twin model of the PCB production line then operates according to the optimized configuration solution solved in real time, and evaluates whether the production line resource configuration solution is optimal based on the real-time operation results of the production line twin data. If it is, it is added to the candidate solution set, and finally the optimal production line resource configuration solution is obtained through simulation verification and evaluation in the candidate solution set. The evaluation process aims to maximize productivity, maximize equipment utilization, and minimize logistics costs.

[0122] As Figure 6 shown, in this embodiment, based on the twin data, the process equipment configuration is analyzed, and the single population of the genetic algorithm is divided into two independent but related sub-populations. The first sub-population considers the order of equipment sub-processes and the equipment center coordinates, and the second sub-population considers the interval distance between sub-processes and the equipment center coordinates. The two sub-populations have different genetic operation parameters, such as crossover rate, mutation rate, etc., and execute different evolutionary strategies, and exchange and share information under certain conditions, so as to realize the collaborative interaction between digital twin and double-population genetic algorithm. At the same time, the multi-objective function is used as the fitness function to broaden the search range and avoid premature convergence. During the iterative solution process of the double-population genetic algorithm, the survival of the fittest is continuously carried out through the competition mechanism to maintain the vitality and progressiveness of the population, and a candidate solution set is obtained. The production line balance loss rate is evaluated through the twin body, and finally the optimal resource configuration solution is determined.

[0123] To verify the effectiveness of the present invention, in a specific application embodiment, the above method of the present invention is used to optimize the resource configuration of a certain PCB production line. First, in the step of initializing the population, the first input parameter is the sub-process priority matrices F1 and F2, and F1 and F2 come from the twin data; the second parameter is the initial population size popualr_size, and the parameter settings are specifically shown in Table 2. The parameter defined in the genetic operator operation part is the mutation probability mutation_rate = 0.4.

[0124] Table 2 Parameter Settings of Double-Population Genetic Algorithm

[0125]

[0126] According to the above configuration, after iterative solution, the optimal PCB production line layout plan obtained is as follows: the sub-process sequence is [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16], the net spacing is [1.3, 1.3, 1.7, 1.2, 1.5, 2.0, 1.0, 0.5, 1.0, 1.3, 1.3, 1.6, 0.5, 0.5, 1.9, 1.6], and the central coordinates are [6.9, 13.2, 19.9, 26.1, 32.6, 39.6, 45.6, 51.1, 56.8, 63.5, 69.8, 76.4, 83.0, 88.5, 95.4, 102.0]. The layout of each working unit of the obtained PCB production line is as Figure 7 shown, and the sub-processes and their sequences on each process unit are as Figure 8 shown.

[0127] Further verify the above scheme. After simulation operation on the twin platform, set the simulation time to 8 hours of a day's working production time, and run to obtain the equipment utilization rate of each sub-process. From the experimental results, it can be seen that the equipment utilization rate of each optimized production line has been improved. Compared with before optimization, the average utilization rate has increased by nearly twice. Further evaluate the results of the twin system and verify the optimized resource allocation scheme. The balance loss rate of the production line can evaluate the results of the production line resource allocation, and its formula is as follows:

[0128]

[0129] Among them: n is the total number of production line processes; ω i is the equipment utilization rate of the i-th station; ω Max is the highest utilization rate, and η1 is the production line balance loss rate. The comparison results of various data of the twin system and genetic algorithm obtained in this embodiment are shown in Table 3.

[0130] Table 3 Comparison Results

[0131]

[0132] From the above results, it can be seen that the output of the initial layout of the twin system is 636, the production efficiency is about 36%, the average equipment utilization rate is about 29.80%, and the production line balance loss rate is about 58%; the output of the optimized layout scheme is 1147, the production efficiency is about 59%, the average equipment utilization rate is about 45.98%, and the production line balance loss rate is about 36%. Among them, the production efficiency has increased by 64%, the output has increased by 80%, and the maximum equipment utilization rate has increased by about 24%, verifying the reliability of the current optimal solution, that is, the present invention can accurately and reliably optimize the resource allocation of the PCB production line by using the method of collaborative interaction between the double-population genetic algorithm and digital twin. Further feedback the optimization results to the physical system, and real-time adjustment and optimization can also be achieved.

[0133] This embodiment further provides a production line configuration optimization device based on the collaboration of digital twin and double-population genetic algorithm, including a processor and a memory. The memory is used to store computer programs, and the processor is used to execute the computer programs to execute the above method.

[0134] It can be understood that the above method of this embodiment can be executed by a single device, such as a computer or a server, etc., or can also be applied to a distributed scenario where multiple devices cooperate with each other to complete. In the case of a distributed scenario, one of the multiple devices can only execute one or more steps of the above method of this embodiment, and the multiple devices interact with each other to complete the above method. The processor can be implemented in the form of a general-purpose CPU, a microprocessor, an application-specific integrated circuit, or one or more integrated circuits, etc., and is used to execute relevant programs to implement the above method of this embodiment. The memory can be implemented in the form of a read-only memory ROM, a random access memory RAM, a static storage device, and a dynamic storage device, etc. The memory can store an operating system and other application programs. When implementing the above method of this embodiment through software or firmware, the relevant program codes are stored in the memory and are called and executed by the processor.

[0135] This embodiment further provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the above method is implemented.

[0136] Those skilled in the art should understand that the above embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) containing computer-usable program code. The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in the process Figure 1 one process or multiple processes and / or blocks Figure 1 or multiple blocks. These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means realizes the functions specified in the process Figure 1 one process or multiple processes and / or blocks Figure 1 or multiple blocks. These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for realizing the functions specified in the process Figure 1 one process or multiple processes and / or blocks Figure 1 or multiple blocks.

[0137] The above are only the preferred embodiments of the present invention, and do not impose any formal limitations on the present invention. Although the present invention has been disclosed above with preferred embodiments, it is not intended to limit the present invention. Therefore, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention shall fall within the scope of protection of the technical solution of the present invention.

Claims

1. A production line configuration optimization method based on the collaboration of digital twins and dual-population genetic algorithm is applied to optimize the resource configuration of PCB production lines, characterized in that: The steps of the method include: Construct a twin model of the PCB production line. In the model, each process unit is arranged in sequence in a single row layout, and each process unit corresponds to a process. Construct a single-row layout model for PCB production lines, equate each process unit in the PCB production line to multiple rectangles in a single-row layout with a specified spacing between each process unit, and construct a multi-objective planning model with the goals of maximizing production efficiency, maximizing equipment utilization, and minimizing logistics costs; Acquire the twin data generated by the twin model of the PCB production line in real time, obtain the priority order of each process unit of the PCB production line according to the twin data, use the dual-population genetic algorithm to encode the priority order of each process unit and the interval distance between each process unit, and use the fitness function constructed based on the multi-objective programming model to iteratively solve to obtain the priority order of each process unit in the PCB production line and the optimal configuration scheme of the interval between each process unit, wherein when using the dual-population genetic algorithm, the first population is the process order population, which is used to encode the priority order of each process unit, and the second population is the net spacing population, which is used to encode the interval distance between each process unit; The optimized configuration solution obtained by iterative solution using the dual-population genetic algorithm is mapped to the PCB production line twin model, and the optimization effect is evaluated based on the twin data generated by the PCB production line twin model; When using the dual-population genetic algorithm, a co-evolution strategy based on similarity and dynamic mapping is adopted to exchange information between the two populations regularly. The steps include: Determine the mapping relationship: find individuals in the net spacing population and the process sequence population whose fitness values ​​do not differ by more than the preset threshold, count the corresponding gene positions of each pair of individuals in all individuals in the two populations, and obtain the mapping relationship pairs of multiple gene values ​​of the two populations based on the gene value pairs with the most corresponding relationships of the gene values ​​at the corresponding positions of the chromosomes of the two populations counted; Information exchange based on similarity: Based on the similarity between the two populations, the information of the two populations, the clear spacing population and the process sequence population, is exchanged. The steps include: Take the best individual with the highest fitness in the net spacing population as the current net spacing population individual; Calculate the similarity index of the individuals in the current clear spacing population and all the individuals in the process sequence population, find the individuals with the highest similarity, and the individuals in the process sequence population with the highest similarity to the individuals in the current clear spacing population; According to the mapping relationship of gene value pairs, the gene values ​​of the current net spacing population individual and the process sequence population individual with the highest similarity found are exchanged, wherein, according to the mapping relationship, the first N bits of the chromosome of the process sequence population individual with the highest similarity are replaced with the gene values ​​corresponding to the first N bits in the current net spacing population individual, and at the same time, the chromosome center coordinate genes of the process sequence population individual with the highest similarity are replaced with the center coordinate genes of the last N bits of the current net spacing population individual, and it is judged whether the exchanged gene values ​​meet the priority order. If not, the exchange fails, the current exchange is abandoned, and the individual with the next fitness in the net spacing population is taken as the current net spacing population individual, and the exchange is continued. If the exchange fails until the last fitness individual, the next round of iteration is directly entered. If the exchanged gene values ​​meet the priority order, the population is updated by the current exchange result and the next round of iteration is entered, and N represents the number of PCB processes; After completing the iteration, the relative optimal solutions of the two populations, the net spacing population and the process sequence population, are finally obtained.

2. The production line configuration optimization method based on the collaboration of digital twin and dual population genetic algorithm according to claim 1 is characterized in that: When constructing the multi-objective planning model, a sequential algorithm is used to determine the priority order of each objective, and the constructed multi-objective planning model is: Among them, z represents the objective function, P k is the priority factor corresponding to the kth target, and are the positive and negative deviations of the kth target from the target value, Respectively represent the positive and negative deviation of the first target from the target value. Respectively represent the positive and negative deviation of the second target from the target value. Respectively represent the positive and negative deviation of the third target from the target value. and are the weight values ​​of the positive and negative deviations corresponding to the kth target, α1, α2 and α3 correspond to the expected values ​​of the production efficiency target, equipment utilization target and total transportation distance target, N is the total PCB output within the specified time, T is the total production time per unit time, d ij is the net distance between the center point of the i-th process unit and the center point of the j-th process unit, C is the fixed transportation cost per unit distance, n represents the number of process units, T i Represents the processing time of the i-th process unit.

3. The production line configuration optimization method based on the collaboration of digital twin and dual population genetic algorithm according to claim 2 is characterized in that: The total PCB production N within a specified period of time is calculated as follows: N=T / (T p +T on ) Among them, T p The time required to process a PCB, T on is the feeding interval time, t i is the processing time of the i-th process unit, v is the material conveying speed per unit distance, Δx ij Represents the net distance between the boundary of the i-th process unit and the boundary of the j-th process unit.

4. The production line configuration optimization method based on the collaboration of digital twin and dual population genetic algorithm according to claim 2 is characterized in that: When constructing the multi-objective planning model, it also includes setting multi-objective constraints, which include space constraints, interval constraints and boundary constraints. The space constraints are constraints that each process unit needs to meet in the spatial range, the interval constraints are conditions that the interval distance between each process unit needs to meet, and the boundary constraints are conditions that need to be met between the boundaries of each process unit. The calculation expression of the multi-objective constraints is: Among them, x i 、x j are the center coordinates of the i-th and j-th process units, respectively, x , D y are the minimum distances between the i-th and j-th process units and other process units or walls in the x-axis and y-axis directions, respectively. i and L j are the lengths of the i-th and j-th process units, respectively, W i is the width of the i-th process unit, L and W are the length and width of the production workshop respectively, △x ij Represents the net distance between the boundary of the i-th process unit and the boundary of the j-th process unit.

5. The production line configuration optimization method based on the collaboration of digital twin and dual population genetic algorithm according to claim 1 is characterized in that: When using the dual population genetic algorithm, the first population is based on F1 = [M1, M2, ..., M i , ..., M n , x1, x2, ..., x i , ..., x n ] is encoded, where M i represents the i-th process unit, x i represents the center coordinates of the i-th process unit; the second population is calculated according to F2 = [△x 0，1 , △x 1，2 , ..., △x i，j , ..., △x 15，16 , x1, x2, ..., x i , ..., x 16 ] is encoded, △x ij represents the net distance between the boundary of the i-th process unit and the boundary of the j-th process unit, and according to D x , D y The maximum and minimum values ​​of iteratively generate the initial population of the second population, D x , D y The minimum distances between the i-th and j-th process units and other process units or walls in the x-axis and y-axis directions respectively.

6. The production line configuration optimization method based on the collaboration of digital twin and dual population genetic algorithm according to any one of claims 1 to 5, characterized in that: When using the dual population genetic algorithm, the fitness function is constructed based on the multi-objective programming model using a penalty function. The fitness function is expressed as: Fit=1 / (z+P) Among them, Fit represents the fitness function, z represents the objective function of the multi-objective planning model, P is the penalty function, and Q is the penalty value.

7. The production line configuration optimization method based on the collaboration of digital twin and dual population genetic algorithm according to any one of claims 1 to 5, characterized in that: When using the dual population genetic algorithm, the selection operator uses the roulette wheel selection method to select the next population. The probability C of an individual being selected and inherited to the next generation is i for: Among them, Fit(i) represents the value of the fitness function of the i-th individual, and groupsize represents the number of all individuals in the population.

8. The production line configuration optimization method based on the collaboration of digital twin and dual population genetic algorithm according to any one of claims 1 to 5, characterized in that: When using the dual-population genetic algorithm, the crossover operator selects the partial matching crossover method to determine the crossover point and exchange the middle part of the genes; the mutation operator uses the random mutation method to randomly generate a mutation point with a specified probability, dividing the chromosome into two fragments, the front chromosome remains unchanged, and the rear chromosome undergoes genetic recombination according to the priority matrix. The recombined fragment is spliced ​​with the front chromosome to generate a daughter chromosome, and genetic recombination is performed according to the priority matrix so that the mutated chromosome meets the relative sequence of the process.

9. A production line configuration optimization device based on the collaboration of digital twins and dual-population genetic algorithm, comprising a processor and a memory, wherein the memory is used to store a computer program, characterized in that: The processor is configured to execute the computer program to perform the method according to any one of claims 1 to 8.

Images