Production line configuration optimization method and device based on digital twinborn and double-population genetic algorithm collaboration

Abstract

The invention discloses a production line configuration optimization method and device based on digital twinborn and dual-population genetic algorithm collaboration. The method comprises the following steps: constructing a PCB production line twinborn model; constructing a single-row layout model of the PCB production line, and constructing and forming a multi-target planning model by taking the maximum production efficiency, the maximum equipment utilization rate and the minimum logistics cost as targets; the method comprises the following steps: acquiring twinborn data in real time, acquiring a priority sequence of each process unit of a PCB production line according to the twinborn data, and encoding the priority sequence of each process unit and spacing distances among the process units by using a double-population genetic algorithm, carrying out iterative solution by using a fitness function constructed based on a multi-objective planning model to obtain an optimal configuration scheme; and mapping an optimal configuration scheme obtained by iterative solution to the twin model, and evaluating an optimization effect. According to the invention, an accurate and reliable PCB production line resource configuration optimization scheme can be efficiently generated, and the production line efficiency and the equipment utilization rate are improved.

Description

Technical Field

[0001] The present invention relates to the field of PCB (Printed Circuit Board) intelligent manufacturing technology, and in particular to a production line configuration optimization method and device based on the collaboration of digital twins and dual-population genetic algorithms. Background Art

[0002] The manufacturing process of PCB includes a series of complex process flows, usually including cutting, pre-treatment, inner layer treatment, inner layer lamination, drilling, chemical / electroplating copper, outer layer image transfer, outer layer circuit, silk screen process, electrical testing, molding inspection and packaging, etc. Each process usually also includes multiple sub-processes. For example, the inner layer treatment specifically includes cleaning and baking, exposure, development and etching, film stripping and maintenance, etc. In the entire 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 resource scheduling of complex production systems, the existing technology usually adopts methods based on operations research, heuristic scheduling rules, intelligent algorithms and simulation methods, but each method has certain limitations. Among them, the method based on operations research establishes a mathematical programming model and uses methods such as branch and bound method and dynamic programming method to select the optimal scheduling solution from all possible combinations. However, since the optimization solution of production line resource scheduling belongs to the category of "NP (NP-hard) difficulty", this type of method based on operations research will be subject to certain applicability restrictions. 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, but because it depends on the establishment and design of the simulation model, the accuracy will be limited by the limitations of the simulation model. The heuristic method based on intelligent algorithm is to use heuristic information to guide the search process to find an approximate solution or a feasible solution, but this type of method only obtains the local optimal solution, and cannot obtain the overall optimal solution. When applied to solve the resource scheduling problem of complex production systems, it is difficult to directly obtain the overall optimal resource scheduling solution. Summary of the invention

[0004] The technical problem to be solved by the present invention is: In response to the technical problems existing in the prior art, the present invention provides a production line configuration optimization method and device based on the collaboration of digital twins and dual-population genetic algorithms, which can efficiently generate accurate and reliable PCB production line resource configuration optimization plans, thereby improving production line efficiency and equipment utilization, while reducing overall production costs.

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

[0006] A production line configuration optimization method based on the collaboration of digital twins and dual-population genetic algorithm, the steps comprising:

[0007] 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.

[0008] 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;

[0009] 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;

[0010] The optimized configuration scheme obtained by iterative solution using a dual-population genetic algorithm is mapped to the twin model of the PCB production line, and the optimization effect is evaluated based on the twin data generated by the twin model of the PCB production line.

[0011] Furthermore, when constructing the multi-objective programming model, a sequential algorithm is used to determine the priority order of each objective, and 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 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.

[0014] Furthermore, the total PCB production N within a specified period of time is calculated as follows:

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

[0016]

[0017] 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, △x ij Represents the net distance between the boundary of the i-th process unit and the boundary of the j-th process unit.

[0018] Furthermore, when constructing the multi-objective planning model, it also includes setting multi-objective constraints, and the multi-objective constraints 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:

[0019]

[0020] 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 length of the i-th process unit, 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 ijRepresents the net distance between the boundary of the i-th process unit and the boundary of the j-th process unit.

[0021] Furthermore, 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 adopts the random mutation method to randomly generate a mutation point with a specified probability, and divides the chromosome into two fragments, the front chromosome remains unchanged, and the rear chromosome undergoes genetic recombination according to the priority matrix, and 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.

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

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

[0024]

[0025] 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.

[0026] Furthermore, when using the dual population genetic algorithm, the selection operator adopts the roulette wheel selection method to select the next population, and the probability C of an individual being selected and inherited to the next generation is i for:

[0027]

[0028] 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.

[0029] Furthermore, 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 genes. The mutation operator adopts the random mutation method to randomly generate a mutation point with a certain probability, and divides the chromosome into two fragments, the front chromosome remains unchanged, and the rear chromosome undergoes genetic recombination according to the priority matrix, and the recombined fragment is spliced ​​with the front chromosome to generate the offspring chromosome.

[0030] Furthermore, when using the dual population genetic algorithm, a co-evolution strategy based on similarity and dynamic mapping is adopted to perform regular information exchange between the dual populations, and the steps include:

[0031] 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;

[0032] 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:

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

[0034] 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;

[0035] 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;

[0036] After completing the iteration, the relative optimal solutions of the two populations, the net spacing population and the process sequence population, are finally obtained.

[0037] A production line configuration optimization device based on the collaboration of digital twins and dual-population genetic algorithms, comprising a processor and a memory, wherein the memory is used to store a computer program, and the processor is used to execute the computer program to perform the above method.

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

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

[0040] 1. The present invention uses digital twin technology to interact with a dual-population genetic algorithm, uses digital twin technology to simulate the PCB production line resource configuration plan, and uses a dual-population genetic algorithm to quickly iterate and solve the PCB production line resource optimization configuration plan based on multiple objectives such as maximum production efficiency, maximum equipment utilization, and minimum logistics cost. The efficiency, accuracy and flexibility of resource configuration optimization can be effectively improved. When using the dual-population genetic algorithm to solve the optimization configuration plan, the priority order of each process unit and the interval distance between each process unit are respectively corresponded to different populations. The priority order of the process units and the interval distance between the process units can also be comprehensively considered 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 dual-population genetic algorithm is adopted. On the basis of the random mutation method for mutation operation, gene recombination is performed according to the priority matrix to ensure that the mutated chromosomes also meet the relative order of the processes, thereby further ensuring the reliability of the production line configuration optimization.

[0042] 3. The present invention adopts an improved dual-population genetic algorithm to determine the priority order of process units and the interval distance between process units. It further uses a collaborative evolution strategy based on similarity and dynamic mapping to conduct regular information exchange between the process order population and the net spacing population, so that different chromosome encoding methods and gene representation ranges can be matched to quickly and accurately solve the dual-population optimal solution that meets the priority order, 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 It is a schematic diagram of the implementation flow of the production line configuration optimization method based on the collaboration of digital twins and dual-population genetic algorithm in this embodiment.

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

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

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

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

[0048] Figure 6 It is a detailed implementation flow chart of optimizing PCB production line resource configuration using a dual-population genetic algorithm based on digital twins in this embodiment.

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

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

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

[0052] As a virtual simulation technology, digital twin creates a twin of a physical entity, builds a two-way connection between the virtual and the real, and allows the virtual model to simulate the behavior of the physical entity, realizing the integration of the physical world and the information world. The application of digital twin technology can realize the accurate simulation and optimization of the production process, so the use of digital twin technology can realize the optimization of resource allocation for large-scale production lines. However, the manufacturing process of PCB will include a series of complex process flows (cutting, pre-treatment, inner layer treatment, inner layer lamination, drilling, chemical / electroplating copper, outer layer image transfer, outer layer circuit, silk screen process, electrical testing, molding detection and packaging, etc.), each of which will include multiple sub-processes, each of which needs to be precisely controlled. Different process sequences and different intervals between process units will affect the overall production efficiency, equipment utilization, production cost, etc., while digital twin technology can only simulate specific resource allocation schemes. Different process sequences and different intervals between process units can be combined to form a large number of resource allocation schemes. Directly using digital twin technology cannot quickly simulate the optimal resource allocation scheme.

[0053] The present invention realizes collaborative interaction between digital twin technology and heuristic intelligent algorithm (dual-population genetic algorithm), utilizes digital twin technology to simulate PCB production line resource configuration scheme, and utilizes dual-population genetic algorithm to rapidly iterate and solve the optimal configuration scheme of PCB production line resources based on multiple objectives of maximum production efficiency, maximum equipment utilization and minimum logistics cost. The comprehensive interaction between digital twin technology and dual-population genetic algorithm can effectively improve the efficiency, accuracy and flexibility of resource configuration optimization. When the dual-population genetic algorithm is used to solve the optimal configuration scheme, the priority order of each process unit and the interval distance between each process unit are respectively corresponded to different populations. The priority order of the process units and the interval distance between the process units can also be comprehensively considered to further improve the reliability of resource optimization, thereby obtaining a high-precision and high-reliability production line resource optimization configuration scheme.

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

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

[0056] This embodiment first analyzes the PCB production line process to construct a twin model, including building a device geometry model, rendering the device model, and rendering the scene after building the production line scene. It can also set corresponding specification models for device behavior specifications and personnel behavior specifications and configure corresponding setting rule models.

[0057] Optionally, you can use MIoT.VC simulation software to build a twin model of the PCB production line. Through this software, you can achieve modeling (importing models), model layout parameter settings, and model dynamic simulation of the production line simulation system. The processing equipment Shrink Tunnel can be used to build laminators, exposure machines, solder mask printers, developers, acid and alkali etching machines, film stripping machines, etc. 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 electrolytic plating tank, and the Automatic Shrink Wrap Machine model can be used as a packaging device. In addition, the Conveyor model (a three-dimensional model of a conveyor) in the model library is selected as the conveyor belt, the KR 10R1420 model is used as the robotic arm, and the intelligent car (Robot TransportController) model is also equipped.

[0058] Since the steps of the PCB production process are a one-way acyclic graph, and in order to simplify the spatial configuration, facilitate management and material flow, reduce the handling time between processes, and improve production efficiency, this embodiment arranges the PCB production line in a single-row layout, and at the same time, the production workshop is equivalent to a rectangle W with a length and a width, and each process unit is equivalent to a rectangle with a length and a width. Each rectangle corresponds to a process unit, and the floor space of each process unit is consistent and non-overlapping. Each process unit corresponds to a process, that is, all work units are arranged along a straight line, and there is a specified net spacing distance between the boundaries of adjacent process units. The single-row layout mathematical model constructed in this embodiment is as follows: Figure 3 As shown, L and W are the length and width of the production workshop respectively. i and W i are the length and width of the i-th process unit, M i is the number of the i-th work unit, x i is the center coordinate of the i-th working unit, D 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. ij Represents the net distance between the boundary of the i-th process unit and the boundary of the j-th process unit.

[0059] For the convenience of analysis, this embodiment simplifies some processes in the PCB production line to form a simplified process flow as shown in Table 1.

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

[0061]

[0062]

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

[0064] Step S02. Construct a single-row layout model for the PCB production line, and equate each process unit in the PCB production line to multiple rectangles in a single-row layout with a specified spacing distance between each process unit. A multi-objective planning model is constructed with the goals of maximizing production efficiency, maximizing equipment utilization, and minimizing logistics costs.

[0065] Since unreasonable layout order and spacing of work units in the production line will lead to problems such as high material handling frequency, low production efficiency, and low equipment utilization, this embodiment uses maximum production efficiency, high equipment utilization, and minimum logistics cost as multi-objective functions. Assume that each process unit can be arbitrarily laid out within the space limitations of the workshop, the facilities in each process unit in the workshop are placed parallel to the workshop wall, the geometric center of each production equipment is located in a straight line, and the conveyor belt flow between each production equipment can only move in the direction of the axis parallel line. Materials are transferred in a straight line during the intelligent car mechanical arm or manual handling process. The PCB production sequence follows the layout order of each production equipment on the production line. During the total production time of one day, each device on the PCB production line is free of faults and outages. This embodiment designs a multi-objective function for the PCB production line. The first goal is to maximize the production efficiency f1, that is, the maximum number of PCB printed circuit boards produced per unit time, which can be calculated using the following formula:

[0066]

[0067] Where N is the total PCB output within a specified time, T is the total production time per unit time (for example, per day), and T p The time required to process a PCB printed circuit board, T on is the feeding interval time, T p +T on is the total production time of a 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 conveying speed per unit distance, △x ij The element represents the net separation distance between the boundaries of the i-th and j-th work units.

[0068] The second goal in this embodiment is to maximize the equipment utilization rate f2, that is, the actual production time per unit working time of the production equipment is maximized. The more tasks completed per unit time, the higher the corresponding equipment utilization rate. The maximum equipment utilization rate f2 can be calculated using the following expression:

[0069]

[0070] The third goal in this embodiment is to minimize the logistics cost f3, that is, the total transportation distance multiplied by the logistics cost per unit distance is the lowest, which can be calculated using the following expression:

[0071]

[0072] Where C is the fixed transportation cost per unit distance, 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. Further, space constraints can be set, such as the placement of the work unit in the x-axis direction and the y-axis width cannot exceed the spatial range of the workshop.

[0073] In order to construct the final multi-objective planning model, this embodiment uses a sequential algorithm to determine the priority order of each goal. The sequential algorithm processes only one decision or calculation at each step, and then proceeds to the next step based on the current state. The optimized solution of the high-priority goal that has been obtained will be used as a fixed constraint in the subsequent optimization process of the low-priority goal, and the problem can be solved step by step in order.

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

[0075]

[0076] 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.

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

[0078] Furthermore, this embodiment also includes setting multi-objective constraints when constructing a multi-objective planning model. The multi-objective constraints include space constraints, spacing constraints, and boundary constraints. Space constraints are constraints that each process unit needs to meet in the spatial range, spacing constraints are conditions that the spacing distance between each process unit needs to meet, and boundary constraints are conditions that need to be met between the boundaries of each process unit. Optionally, the space constraint can be configured so that the placement of the process unit in the x-axis direction and the y-axis width cannot exceed the spatial range of the workshop, and the spacing constraint can be configured so that the process units do not overlap with each other. Specifically, the calculation expression of the multi-objective constraint can be expressed as:

[0079]

[0080] 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 length of the i-th process unit, 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.

[0081] Step S03. 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 a 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 planning 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, where when using the dual-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 the twin data generated by the operation of the twin model in real time, and provides the twin data generated in real time to the dual-population genetic algorithm as input data. The dual-population genetic algorithm encodes the priority order of each process unit and the interval distance between each process unit according to the twin data, and obtains the optimal configuration plan after iterative solution. The solved optimal configuration plan is then provided to the twin model for collaborative interaction, that is, the continuous iterative solution and verification of the PCB production line configuration plan can be realized, and finally the optimal configuration plan for the PCB production line that meets the needs is determined.

[0083] In genetic algorithms, complex actual production problems can be converted into mathematical sequence logic through coding. In this embodiment, two factors, the production process sequence layout and the net spacing distance between the boundaries of each work unit, are considered simultaneously when optimizing the resource scheduling of the PCB production line, corresponding to the dual population coding method, where the first population is the process sequence population, which is used to encode the priority sequence of each process unit (process production sub-process sequence), and the second population is the net spacing population, which is used to encode the spacing distance between each process unit.

[0084] For example, when using a dual population genetic algorithm, the first population (process sequence population) is calculated according to F1 = [M1, M2, ..., M i ,…,M n ,x1,x2,…,x i ,…,x n ] is encoded, that is, the process unit priority order and the center distance are combined to encode, where M i represents the i-th process unit, x i represents the center coordinates of the i-th process unit. The specific encoding method of the first group is designed as: F1 = [M1, M2, ..., M i ,…,M 16 ,x1,x2,…,x i ,…,x 16 ]. i is the i-th working unit, M i =1,2,3,…,16; if M i =p, p∈[1,16], it means that the PCB production line sub-process p is located in the i-th work unit. i is the center coordinate of the i-th work unit. The order of sub-processes in each process can be expressed as follows: [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 16] is [2,1,3,5,4,6,7,8,9,11,10,12,13,14,15,16]. The encoding method of the chromosome can intuitively understand the priority allocation order of the sub-process operations, 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 used in this embodiment contains 32 gene positions. Since the order of sub-processes in the process can be adjusted at will, the initial population can be generated:

[0085]

[0086] The second population (clear spacing population) can be calculated according to F2 = [△x 0,1 ,△x 1,2 ,…,△x i,j ,…,△x 15,16 ,x1,x2,…,x i ,…,x 16 ] is encoded, that is, the device spacing and center distance are combined and 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.

[0087] For example, the specific encoding method of using the net spacing distance between the boundaries of the working units to encode the net spacing population 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 loci, which can reduce the minimum spacing D x Set to 0.5m (configurable), the maximum spacing is 2m (configurable), and the minimum difference in spacing is 0.1m (configurable), that is, the initial population can be iteratively generated at intervals of 0.1, that is, the range of generating the initial population is:

[0088]

[0089] Among them, △x 0,1 Indicates the clear distance between the first process unit and the boundary of the wall.

[0090] The fitness function is used to convert the objective function of the problem into a numerical form to evaluate the core standard of the quality of individuals. Individuals with high fitness values ​​are more likely to be selected and inherited to the next generation, thereby driving the algorithm to search for 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 unconstrained by introducing a penalty function to ensure that the algorithm avoids solutions that violate the constraints during the search process. This embodiment uses a dual-population genetic algorithm to solve the problem with the minimum value as the optimization goal. Since the higher the fitness value of the genetic algorithm, the better the individual will be, this embodiment uses a penalty function in combination with the objective function to establish a fitness function, which can further improve the accuracy and reliability of the solution.

[0091] Optionally, the penalty function can be used to construct a fitness function based on the multi-objective programming model as follows:

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

[0093]

[0094] Among them, Fit represents the fitness function, z represents the objective function of the multi-objective programming model, P represents the penalty function, and Q represents the penalty value, which is usually a large positive number. As the value exceeding the constraint condition increases, the Q value increases accordingly. The penalty function P can be set according to the required rigid constraint condition, for example, it can be configured as:

[0095]

[0096] In the evolution process of the genetic algorithm, individuals with higher fitness are selected from the current population through the selection operator to reproduce to the next generation. Optionally, for the multi-objective optimization model of the single-line layout of the PCB production line, considering the complexity and multi-objective characteristics of the problem, the selection operator can use the roulette selection method to select the next population when using the dual-population genetic algorithm in this embodiment. As a proportional selection strategy based on individual fitness values, roulette selection can ensure that individuals with higher fitness have a greater probability of being selected, thereby promoting the population to evolve towards a better solution. At the same time, by allocating the selection probability in proportion to the fitness value, it can effectively balance the selection pressure and population diversity.

[0097] Specifically, the probability C of an individual being selected and inherited to the next generation i for:

[0098]

[0099] Among them, Fit(i) represents the value of the fitness function of the i-th individual, groupsize represents the number of all individuals in the population, Represents the sum of the fitness values ​​of all individuals in the population. From the above formula, we can see that the higher the individual fitness, the higher the probability of being selected and inherited to the next generation, and it can be given priority over individuals with low fitness, that is, the selection probability is positively correlated with fitness.

[0100] The crossover operator determines how genetic information is transmitted and recombined between populations. In this embodiment, in order to achieve multi-objective optimization of the single-line layout of the PCB production line, considering the complexity of layout optimization and the need to take into account multiple performance indicators, the crossover operator selects the partial matching crossover (PMX) method when using the dual-population genetic algorithm. By determining the crossover point and exchanging the middle part of the gene, while maintaining the inheritance of the uncrossed part of the gene, it helps to maintain the diversity of individuals and the search space, thereby avoiding premature convergence. By adopting the above-mentioned method of finely adjusting the crossover probability, the ability of the algorithm to search for excellent solutions locally can be improved while ensuring the search space.

[0101] For example:

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

[0103] The intersection positions are randomly selected as follows: A1 = [1 2 | 3 4 5 | 6 7], B1 = [2 1 | 3 5 7 | 6 4];

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

[0105] The non-repeated genes inherited from the parental genes are retained, and the repeated genes in the middle segment are replaced by x to form: A3 = [12|3 5x|6 7], B3 = [2 1|3x 5|6 4];

[0106] The offspring genes were determined according to the relative order of the parental gene loci and the middle segment, and the non-repeating segments were obtained: A4 = [1 2|35 4|6 7], B4 = [2 1|3 7 5|6 4];

[0107] In the production line layout, it is necessary to pay attention to the process priority. In this embodiment, after each crossover operation, it is determined whether the generated offspring chromosomes are qualified. 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, help the genetic algorithm to search for the optimal solution in a wider solution space, and improve the global search ability of the algorithm. In the genetic algorithm, the mutation operator introduces new gene combinations by randomly selecting individuals in the population with a certain probability and changing certain gene values ​​in their gene encoding strings. This change can be operations such as replacement, insertion or deletion, which can increase the diversity of the population and prevent the algorithm from converging to the local optimal solution too early. In this embodiment, the mutation operator uses a random mutation method to randomly generate a mutation point with a certain probability, divide the chromosome into two fragments, the front chromosome remains unchanged, and the rear chromosome is genetically recombined according to the priority matrix, and the recombined fragments are spliced ​​with the front chromosome to generate offspring chromosomes. Considering that there are specific priorities and relative orders between production processes, this embodiment improves the dual-population genetic algorithm. On the basis of using the random mutation method for mutation operation, genetic recombination is performed according to the priority matrix to ensure that the mutated chromosomes also meet the relative order of the process, thereby further ensuring the reliability of the production line configuration optimization.

[0109] Since the chromosome encoding methods and gene representation ranges of the two populations are different, it is not feasible to simply select the best individuals in the population and exchange them. This embodiment further improves the dual-population genetic algorithm and adopts a co-evolution strategy based on similarity and dynamic mapping to perform regular information exchange between the two populations. The specific steps are as follows:

[0110] Step ① Determine the mapping relationship: Find individuals whose fitness values ​​do not differ by more than a preset threshold (e.g. 0.8), and count the corresponding relationships between the individual pairs and the corresponding gene positions of the chromosomes, that is, 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 between the gene values ​​at the corresponding positions of the chromosomes of the two populations.

[0111] For example, the PCB production line has 16 processes, and 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, 16 gene positions, and 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, there are also 16 gene values, 16 gene positions, where 0.5 is the distance D between the first process unit and the wall x , 2 represents the distance D between the last process unit and the wall x Therefore, we can get a one-to-one mapping relationship between the first 16 gene values ​​of the two populations, that is, the mapping relationship between the chromosome gene value of the net spacing population and the chromosome gene value of the process sequence population, which can be expressed as (chromosome gene value of the net spacing population, chromosome gene value 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 counted. The gene value pair with the most statistics at index 5 of the chromosome gene list is (0.9, 6), where 0.9 represents the process position and 6 represents the process number. (0.9, 6) is recorded as a gene value pair in the mapping list. This gene value pair only represents a mapping relationship and has nothing to do with the position. If the gene value pair with the most statistics at index 7 in the chromosome gene list is subsequently retrieved as (0.9, 8), since 0.9 in this gene value pair is the same as the value in the mapping relationship obtained by the aforementioned index 5, and the mapping relationship must be one-to-one (the process position is unique), it is necessary to continue to select the second-ranked gene value pair for further judgment until the value in this gene value pair is not the same as the value in the mapping relationship obtained previously, that is, all the mapping relationships obtained must satisfy that the x value in each gene value pair (x, y) ranges from 0.5 to 2 (interval is 0.1), and the y value ranges from 1 to 16 (interval is 1) and is one-to-one corresponding.

[0113] Step ② Information exchange based on similarity:

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

[0115] Step 2.2. Calculate the similarity index (such as Pearson or Spearman coefficient) of the individuals in the current clear spacing population and all the individuals in the process sequence population, and find the individual in the process sequence population with the highest similarity to the individuals in the current clear spacing population;

[0116] Step S2.3 exchanges the gene values ​​of the current clear spacing population individual and the process sequence population individual with the highest similarity found in step 2.2 according to the mapping relationship of the first 16 gene values ​​obtained in step ①, wherein the first 16 chromosome bits of the process sequence population individual with the highest similarity are replaced with the gene values ​​corresponding to the first 16 bits in the current clear spacing population individual according to the mapping relationship, that is, the gene bits of the individual remain unchanged, and only the gene values ​​of the first 16 bits are changed; at the same time, the center coordinate genes of the last 16 chromosome bits of the process sequence population individual with the highest similarity are replaced with the last 16 chromosome bits of the current clear spacing population individual. 6-bit center coordinate gene, that is, to ensure that the second half of the optimal solution of the two populations obtained at the end, that is, the 16-bit center coordinate gene, is consistent, and to determine whether the gene value after the exchange meets 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 return to step S2.2 to continue the exchange. If the exchange fails until the last fitness individual, the next round of iteration is directly entered. If the gene value after the exchange meets the priority order, the population is updated by the result of the current exchange and the next round of iteration is entered;

[0117] Step S2.4. After the iteration is completed, the relative optimal solution of the net spacing population and the process sequence population is obtained.

[0118] For example, if the best individual with the highest fitness in the net spacing population is [1.1, 0.5, 2, 3.6, 9.1, 16.1], and the individual in the process sequence population with the highest similarity calculated based on the similarity index is [1, 2, 3, 3, 9, 15, 3], and the gene value pairs related to the first three genes of the net spacing are (1.1, 3), (0.5, 1), and (2, 2), respectively, then the individual in 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 sub-process priority in the encoding rule of the process sequence population, the priority is [[1, 2], 3]. According to the list order, the position with the subscript 0 is a sublist, and the values ​​in the sublist can be The positions 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 subscript with the value "3" is 1, which will be after the two values ​​"1" and "2" in the sublist with the subscript 0. If the first three genes of the individual in the process sequence population replaced after information exchange are 3, 1, and 2, it will violate the encoding rules of the process sequence population. Therefore, the individual is invalid, and then continue to exchange information with the individual with the second highest similarity in the process sequence population. If it meets the encoding rules of the process sequence population, the population can be updated to enter the next round of iteration to generate the next generation. If it does not meet the rules, continue to exchange information with the individual with the next fitness until the exchange of the last fitness individual fails, ending the current information exchange and entering the next round of iteration.

[0119] This embodiment specifically exchanges information between the two populations after each generation of genetic operator operations is completed to ensure the co-evolution between the two populations, and ultimately obtains a relatively optimal solution for the two populations. It is capable of matching different chromosome encoding methods and gene representation ranges to quickly and accurately solve the optimal solution for the two populations that satisfies the priority order, thereby determining the optimal priority order of the process units and the optimal spacing distance between the process units.

[0120] Step S04. Map the optimized configuration solution obtained by iteratively solving the dual-population genetic algorithm to the PCB production line twin model, and evaluate the optimization effect based on the twin data generated by the PCB production line twin model.

[0121] After using the dual-population genetic algorithm to iteratively solve the optimal configuration plan, the solved optimal configuration plan is fed back and mapped to the PCB production line twin model. The PCB production line twin model then runs according to the optimal configuration plan solved in real time. The real-time operation results of the production line twin data are used to evaluate whether the production line resource configuration plan is optimal. If so, it is added to the candidate solution set. Finally, the optimal production line resource configuration plan is obtained from the candidate solution set after simulation verification and evaluation. The goals of the evaluation process are to maximize productivity, maximize equipment utilization, and minimize logistics costs.

[0122] like Figure 6 As shown, this embodiment analyzes the process equipment configuration based on the twin data, and divides the single population of the genetic algorithm into two independent but interconnected sub-populations. The first sub-population considers the order of the equipment sub-processes and the equipment center coordinates, and the second sub-population considers the interval distance of the sub-processes and the equipment center coordinates. The two sub-populations have different genetic operation parameters, such as crossover rate, mutation rate, etc., and respectively implement different evolutionary strategies, and exchange and share information under certain conditions, thereby realizing the collaborative interaction between the digital twin and the dual-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. In the iterative solution process of the dual-population genetic algorithm, the competition mechanism is continuously used to eliminate the fittest, maintain the vitality and progress of the population, obtain the candidate solution set, and evaluate the production line balance loss rate through the twin, and finally determine the optimal resource allocation plan.

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

[0124] Table 2 Parameter settings of dual population genetic algorithm

[0125]

[0126] According to the above configuration, after iterative solution, the optimal PCB production line layout solution is obtained: sub-process sequence [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], the center 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], and the layout of each working unit of the PCB production line is as follows Figure 7 As shown, the sub-processes and their sequences on each process unit are as follows Figure 8 shown.

[0127] The above scheme was further verified by running the twin platform simulation, setting the simulation time to 8 hours of working production time per day, and running to obtain the equipment utilization rate of each sub-process. The experimental results show that the utilization rate of each equipment in the production line has increased after optimization. Compared with before optimization, the average utilization rate has increased by nearly two times. The results of the twin system were further evaluated, and the optimized resource allocation scheme was verified. The balance loss rate of the production line can be used to evaluate the results of the production line resource allocation, and the formula is as follows:

[0128]

[0129] Where: n is the total number of production line processes; ω i is the equipment utilization rate of the i-th workstation; ω Max is the maximum utilization rate, and η1 is the balance loss rate of the production line. The data comparison results of the twin system and the 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 is about 29.80%, and the production line balance loss rate is about 58%; the output of the optimized layout solution is 1147, the production efficiency is about 59%, the average equipment utilization is about 45.98%, and the production line balance loss rate is about 36%, among which the production efficiency is increased by 64%, the output is increased by 80%, and the maximum equipment utilization is increased by about 24%, which verifies the reliability of the current optimal solution, that is, the present invention adopts the dual-population genetic algorithm and the digital twin collaborative interaction method to accurately and reliably realize the optimization of PCB production line resource configuration. Further feedback of the optimization results to the physical system can also realize real-time adjustment and optimization.

[0133] This embodiment further provides a production line configuration optimization device based on the collaboration of digital twins and dual-population genetic algorithms, 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 perform the above method.

[0134] It is understandable that the above method of this embodiment can be executed by a single device, such as a computer or server, etc., and can also be applied to a distributed scenario and completed by multiple devices in cooperation with each other. In the case of a distributed scenario, one of the multiple devices can only execute one or more steps in the above method of this embodiment, and multiple devices interact 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., for executing related 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. The memory can store an operating system and other applications. When the above method of this embodiment is implemented by software or firmware, the relevant program code is stored in the memory and 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 will appreciate that the above-mentioned embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, may be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the functions in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including an instruction device, which implements the functions specified in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 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 produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0137] The above is only a preferred embodiment of the present invention, and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment, it is not intended to limit the present invention. Therefore, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the content of 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 scheme obtained by iterative solution using a dual-population genetic algorithm is mapped to the twin model of the PCB production line, and the optimization effect is evaluated based on the twin data generated by the twin model of the PCB production line.

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 length of the i-th process unit, W i is the width of the i-th process unit, L and W are the length and width of the production workshop respectively, Δx 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 based on F2 = [Δx 0，1 , Δx 1，2 , ..., Δx i，j , ..., Δx 15，16 , x1, x2, ..., x i , ..., x 16 ] is encoded, Δx 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. 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, 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 individual of the current net spacing population; 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.

10. 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