Optimized substrate splitting method, system and storage medium for multi-chip placement machine CT
In the substrate splitting method optimized by patch machine CT, the substrate allocation is optimized by genetic algorithm and chromosome gene mapping index, and the problems of uneven production load and inflexible distribution strategies in the existing technology are solved, and the cycle time is achieved Optimization and productivity improvement.
Patent Information
- Application Number
- CN202411217291.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-02
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-09-02
AI Technical Summary
The existing substrate splitting method fails to fully consider the actual production capacity and configuration differences of the patch machine, and cannot effectively respond to the miniaturization, high density and personalized requirements of components, resulting in uneven distribution of production loads and lack of flexibility in distribution strategies. and adaptability, the optimization of cycle time (CT) cannot be achieved.
The substrate splitting method optimized by multi-patch machine CT is adopted. By loading machine configuration parameters and original substrate data, combining genetic algorithms and chromosome gene mapping indexes, parallel computing tasks are generated to optimize substrate allocation, and the production load balance and CT are achieved. Optimized.
It significantly improves production efficiency, shortens production cycles, improves the optimization quality of the production process, can flexibly adapt to diversified production needs, supports users to customize allocation rules, and shows strong adaptability and customization.
Smart Images

Figure CN118760526B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of chip placement machine mounting optimization, and in particular to a substrate splitting method and system and a storage medium for CT optimization of multiple chip placement machines. Background Art
[0002] With the rapid development of the electronic assembly industry, the electronic manufacturing industry has increasingly higher requirements for production efficiency and quality. As one of the core technologies of modern electronic assembly, surface mount technology (SMT) directly determines the speed and quality of electronic product manufacturing. SMT technology automatically mounts electronic components on printed circuit boards (PCBs) and achieves good mechanical and electrical connections between components and PCBs through welding.
[0003] In SMT production lines, configuring multiple placement machines for the production line has become a routine operation to improve production efficiency. The multi-patch machine system can process multiple PCB substrates in parallel, significantly shortening the production cycle and increasing production capacity. However, with the miniaturization and high density of electronic components, and the increasing demand of consumers for personalized and diversified electronic products, the complexity and challenges of SMT production lines have also increased.
[0004] In SMT production lines, CT (Cycle Time) is an important indicator for measuring production efficiency. The shorter the CT, the higher the production capacity of the production line and the lower the production cost. In order to optimize CT, an effective method is to optimize the substrate splitting strategy, that is, to reasonably allocate PCB substrates to multiple placement machines to achieve balanced production load and maximize overall production efficiency.
[0005] Existing methods for splitting substrates are mostly based on experience or simple rules, such as allocation by PCB size, component type or quantity. These methods have the following shortcomings:
[0006] (1) Failure to fully consider the actual production capacity and configuration differences of each placement machine.
[0007] (2) Failure to effectively respond to the challenges brought about by the miniaturization, high density and personalized needs of components.
[0008] (3) The allocation strategy lacks flexibility and adaptability and cannot adapt to the complex and changing production environment.
[0009] Therefore, it is particularly important to develop a substrate splitting method based on comprehensive consideration of multiple factors to optimize the CT of multiple placement machines in SMT production lines. This method needs to comprehensively consider the number and type of placement machines in the production line, the machine configuration and efficiency of each placement machine, the layout, size, type and other data of substrate components, and optimize the substrate splitting strategy through intelligent algorithms to achieve balanced production load and optimized CT. Summary of the invention
[0010] In order to solve the problems in the prior art, the present invention provides a substrate splitting method and system, and a storage medium for optimizing the CT of multiple placement machines, which can efficiently split a complete PCB substrate file into multiple sub-substrate files so that these sub-substrate files can be mounted in parallel by multiple placement machines on an SMT production line, thereby achieving optimization of the production line CT.
[0011] In order to achieve the above object, the first aspect of the present invention provides a substrate splitting method for optimizing a multi-mounting machine CT, comprising the following steps:
[0012] Load the production line information including the configuration parameters of all machines, the substrate to be split and the manually set parameters to restore the original substrate data;
[0013] Analyze the original substrate data, combine the configuration parameters of each machine and the manually set component allocation strategy, and calculate the set A of machines to which each type of component can be allocated;
[0014] Based on the original substrate data and component allocation strategy, the chromosome gene mapping index is initialized, and the information stored in the initialized chromosome gene mapping index is the mounting point number corresponding to each component type number;
[0015] Obtain the current number of CPU threads, generate a corresponding number of computing tasks, execute the genetic algorithm separately for each computing task, take the longest CT in the split sub-matrix as the fitness of the individual, and calculate the fitness according to the initialized chromosome gene mapping index and set A;
[0016] Obtain the results of all calculation tasks, take the one with the lowest fitness, use the corresponding split substrate data as the final split result, generate all sub-substrate files, and complete the splitting.
[0017] A second aspect of the present invention provides a substrate splitting system for optimizing CT of multiple placement machines, comprising:
[0018] The substrate splitting module is used to load the production line information including the configuration parameters of all machines, the substrate to be split and the manually set parameters, and restore the original substrate data;
[0019] The component allocation analysis module is used to analyze the original substrate data, combine the configuration parameters of each machine and the manually set component allocation strategy, and calculate the set A of machines to which each type of component can be allocated;
[0020] A chromosome gene mapping index generation module is used to initialize the chromosome gene mapping index based on the original substrate data and the component allocation strategy. The information stored in the initialized chromosome gene mapping index is the mounting point number corresponding to each component type number.
[0021] A genetic algorithm execution module is used to obtain the current number of CPU threads, generate a corresponding number of computing tasks, and execute a genetic algorithm for each computing task separately to calculate the fitness according to the initialized chromosome gene mapping index and set A;
[0022] The split result generation module is used to obtain the results of all calculation tasks, take the one with the lowest fitness, take the corresponding split substrate data as the final split result, generate all sub-substrate files, and complete the splitting.
[0023] A third aspect of the present invention provides a machine-readable storage medium having instructions stored thereon, the instructions being used to enable a machine to execute the steps of the substrate splitting method for optimizing the multi-mount machine CT described above in the present application.
[0024] Through the above technical solution, by integrating the parallel design of genetic algorithms, complex substrate data processing and intelligent splitting logic, innovative chromosome gene mapping indexing strategy, and a comprehensive optimization target achievement mechanism, synergistic efficiency is achieved to achieve the following significant advantages: Efficiency Leap: The innovative parallel processing architecture greatly accelerates the algorithm operation speed, effectively shortens the production cycle, and improves overall work efficiency; Optimization accuracy and efficiency: With flexible component allocation strategies and original gene mapping indexing technology, this method ensures that the optimization results are both accurate and efficient, significantly improving the optimization quality in the production process; High adaptability: This solution is designed to be flexible and changeable, and can seamlessly connect to diverse production needs. At the same time, it supports user-defined allocation rules and strategies, showing strong adaptability and customizability.
[0025] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings:
[0027] Figure 1It is a schematic diagram of the optimized substrate splitting process of the multi-chip mounter CT of the present invention;
[0028] Figure 2 is a schematic diagram of hybrid substrate information according to an embodiment of the present invention;
[0029] Figure 3 Schematic diagram of chromosome decoding process in an embodiment of the present invention. DETAILED DESCRIPTION
[0030] The specific implementation of the embodiment of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present invention, and is not used to limit the embodiment of the present invention.
[0031] Existing substrate splitting methods often have the following technical disadvantages:
[0032] (1) Uneven distribution of production load: resulting in a decrease in the overall efficiency of the production line.
[0033] (2) Single allocation strategy: It cannot guarantee high-quality allocation results in different environments.
[0034] (3) Low degree of strategy customization: Humans cannot freely intervene in the algorithm’s allocation strategy, and the algorithm has poor adaptability to actual production conditions.
[0035] (4) Incomplete consideration of constraints: Incomplete consideration of the constraints brought by machine configuration and substrate data may lead to infeasible allocation results.
[0036] (5) Low computational efficiency: It is impossible to fully utilize computer performance to obtain better allocation results within a certain period of time.
[0037] The solution described in the embodiment of the present invention is an algorithm submodule of the mounting optimization software. In order to solve the above problems, the first aspect of the embodiment of the present invention provides a substrate splitting method for optimizing the CT of multiple placement machines, such as Figure 1 As shown, the following steps are included:
[0038] S1. Load the production line information including the configuration parameters of all machines, the substrate to be split and the manually set parameters, and restore the original substrate data;
[0039] S2, analyzing the original substrate data, combining the configuration parameters of each machine and the manually set component allocation strategy, and calculating the set A of machines to which each type of component can be allocated;
[0040] S3, based on the original substrate data and component allocation strategy, the chromosome gene mapping index is initialized, and the information stored in the initialized chromosome gene mapping index is the mounting point number corresponding to each component type number;
[0041] S4, obtaining the current number of CPU threads, generating a corresponding number of computing tasks, executing a genetic algorithm for each computing task separately, taking the longest CT in the split sub-substrate as the fitness of the individual, and calculating the fitness according to the initialized chromosome gene mapping index and set A;
[0042] S5. Obtain the results of all calculation tasks, take the one with the lowest fitness, use the corresponding split substrate data as the final split result, generate all sub-substrate files, and complete the splitting.
[0043] The core part of the algorithm of the present invention is implemented based on a genetic algorithm, which takes advantage of the parallel characteristics of the genetic algorithm and uses multi-threading technology to fully mobilize the hardware performance of the computer and improve the algorithm's computational efficiency. As a load balancing distribution algorithm at the production line level, the algorithm frequently calls the placement optimization algorithm of a single machine during operation to obtain the optimal CT of each machine under the current distribution result, which is then used as an evaluation indicator to measure the quality of this distribution result and affect the subsequent iterative distribution process. It is worth noting that the placement optimization algorithm of a single machine is a calling module of this algorithm and is not the focus of discussion of the present invention.
[0044] In another embodiment of the present invention, step S1 loads the production line information including the configuration parameters of all machines, the substrate to be split and the manually set parameters, and restores the original substrate data, which specifically includes the following process:
[0045] S11, the placement optimization software loads the production line information (including the configuration parameters of all machines), the substrates to be split (multiple substrates with the same name are regarded as sub-substrates after splitting, and the algorithm will merge all substrate data before splitting), and manually sets parameters (sets substrate splitting parameters and single substrate optimization parameters);
[0046] S12. The substrate splitting module first performs a data check on all substrates to be split, mainly to check whether there are fatal errors or optimizable errors in the substrates. After a fatal error is found, the algorithm will directly report an error and will not execute the next steps. If there is only an optimizable error, a warning message will be prompted, but the next steps will still be executed. Exemplarily, fatal errors include but are not limited to: no suction nozzles are configured for the components, which means that the suction and placement process cannot be completed and cannot be solved by optimization; optimizable errors include but are not limited to: the position of the feeder is illegal, for which the algorithm will reassign the optimal position for the feeder, which will not affect the substrate splitting and optimization;
[0047] S13, determine whether the substrate to be split is composed of multiple sub-substrates (multiple sub-substrates means that the original substrate has been split once and now needs to be split again). If so, first perform a mixed calculation of the substrate data to restore the original substrate data. When restoring the original substrate data, first load and generate all sub-substrate data, then merge the same information data, reorder, name, and merge different information data, and finally process to obtain the data of the original substrate. For example, Figure 2 As shown, there are substrates with the same name in two machines on the production line. The substrate of machine 1 contains information 1, information 2, and information 3, and the substrate of machine 2 contains information 3, information 4, and information 5. Then the mixed substrate will contain information 1, information 2, information 3, information 4, and information 5.
[0048] Before splitting, the algorithm will check the data of all substrates to be split and handle possible errors. For substrates to be split that are composed of multiple sub-substrates, the algorithm can first merge the sub-substrate data and restore the original substrate data to implement complex substrate data processing and splitting logic. In addition, by detecting fatal errors and optimizable errors in substrate data, directly reporting errors for fatal errors that cannot be resolved, and providing warning information and continuing execution for optimizable errors, the algorithm is enhanced in terms of robustness and user-friendliness.
[0049] In another embodiment of the present invention, step S2, analyzing the original substrate data, combining the configuration parameters of each machine and the manually set component allocation strategy, and calculating the set A of machines to which each type of component can be allocated, specifically includes the following process:
[0050] S21. Traverse all component data and calculate key attributes for component allocation, including: component feeding type (feeder feeding, automatic tray feeding), component recognition camera type (flying camera, bottom mirror camera), whether the component is a large-size component, whether the component allows multi-machine allocation, and the set of manually restricted allocable machines, etc.;
[0051] S22. Traverse all production line machines and calculate the key configuration attributes of the machines, including: whether they support flying camera recognition, whether they support bottom mirror camera recognition, whether they support feeder feeding, whether they support automatic tray feeding, whether they support mounting large-size components, etc.;
[0052] S23. Combine the key attributes of component allocation calculated in step S21 and the key configuration attributes of the machine calculated in step S22, and cooperate with the manually set component allocation strategy to determine the set of machines to which each type of component can be allocated, that is, set A. The set is stored in a mapping table (map), named component machine allocation mapping table. The key represents the type number of the component, and the value is a variable-length array, and each bit of the array represents the number of a machine that is allowed to be allocated. Exemplarily, the set of machines to which each type of component can be allocated shown in Table 1 generates the component machine allocation mapping table shown in Table 2.
[0053] Table 1 The set of machines to which each type of component can be assigned
[0054]
[0055] Table 2 Component machine allocation mapping table
[0056]
[0057] S24. Combined with the result calculated in step S23, if the length of the variable-length array is 0, it is determined that there are elements that cannot be allocated to any machine, and the method steps are terminated and an error is reported, otherwise, the method proceeds to the next step.
[0058] The technical solution of the present invention supports the splitting of any number of machines, and the lengths of the variable-length arrays in the component machine allocation mapping table for processing the multiple machines are different.
[0059] In this embodiment, based on the various attributes of the components (such as feeding type, recognition camera type, size, etc.) and the configuration parameters of the machine, combined with the manually set allocation strategy, the set of machines to which each type of component can be allocated is accurately calculated to ensure the rationality and optimality of the allocation. The flexible component allocation strategy supports allowing or not allowing a certain type of component to be allocated to multiple machines at the same time, and can control the specific machines to which a certain type of component can be allocated, adapting to different allocation requirements and enhancing the flexibility of the algorithm. It effectively solves the problem that the single allocation strategy cannot guarantee high-quality allocation results in different environments, and the low degree of customization of the strategy causes the inability of humans to freely intervene in the algorithm's allocation strategy and poor adaptability to actual production conditions.
[0060] Further, step S3, based on the original substrate data and the component allocation strategy, the chromosome gene mapping index is initialized, and the information stored in the initialized chromosome gene mapping index is the mounting point number corresponding to each component type number. The chromosome is further processed by the initialized chromosome gene mapping index to obtain a specific component allocation result. The mapping index is expressed by a matrix, and the matrix is named the gene mapping index matrix. Among them, the first row of the matrix is the type number of the component, and the second row is the mounting point number of the corresponding type of component on the PCB board. If the component allocation strategy does not allow the type of component to be allocated to multiple machines at the same time, the type of component has only one column in the matrix, and the mounting point number is -1, indicating that all mounting points are allocated together during allocation. If the component allocation strategy allows the type of component to be allocated to multiple machines at the same time, the type of component has N columns in the matrix (N=the number of mounting points of the type of component), each column represents a mounting point of the type, and the second row represents the number of the mounting point on the PCB.
[0061] It is worth noting that the corresponding relationship between each component type number and the mounting point number means that for components with "single machine allocation" set, the mounting point number is -1, indicating all mounting points of this type of component, which is equivalent to one-to-many; for components with "multiple machine allocation", there will be multiple columns with the same component number and different mounting point numbers, which are one-to-one corresponding. The mounting point number "-1" refers to the type of component in only one column in the matrix, which does not have the meaning of numerical size, but belongs to the symbolic meaning, that is, "-1" can also be replaced by -2, -3, M, etc.
[0062] An example of a gene mapping index matrix is shown in Table 3:
[0063] Table 3 Gene mapping index matrix
[0064]
[0065] By designing a gene mapping index matrix, the component type, mounting point number and chromosome gene code are matched one by one, thus realizing the mapping from chromosome gene to specific component allocation results, which provides a basis for subsequent genetic algorithm operations.
[0066] Specifically, the chromosome is encoded and decoded by the gene mapping index matrix and the component machine allocation mapping table, and the substrate is split according to the meaning of the chromosome after decoding. Exemplarily, the chromosome encoding and decoding process is as follows:
[0067] The chromosome consists of a series of genes. The number of genes is K, which is the number of columns in the gene mapping index matrix. The genes in the chromosome correspond to the columns in the gene mapping index matrix. Each gene is represented by a number, which represents the number of the machine assigned to it. Then the entire chromosome is an array of length K.
[0068] Coding process: Coding is to convert a solution to the problem (how to allocate mounting points to specific machines to ensure compliance with constraints and optimal production line CT) into a chromosome according to the designed coding rules. The specific process is as follows:
[0069] First, generate an array of length K to represent a chromosome. At this time, the chromosome gene has not been initialized;
[0070] Traverse each bit of the array (or chromosome) (each bit represents a gene) and find the first row number of the corresponding column of the corresponding gene mapping index matrix. This number is the type of element operated by the current gene;
[0071] According to the component type currently being operated, the component machine allocation mapping table is searched for a set of machine numbers that can be allocated to the corresponding component type;
[0072] Choose any one of the optional machine numbers and use it as the value of the gene to fill in the corresponding position of the chromosome;
[0073] After traversing the entire chromosome according to the above steps, all genes are initialized, and the process of randomly decoding the problem into a chromosome is completed.
[0074] Decoding process: Decoding is the reverse process of encoding, which solves the problem of how to convert meaningless chromosomes into a solution according to the designed decoding rules, such as Figure 3 As shown, the specific process is as follows:
[0075] First, traverse every gene of the chromosome and decode them one by one;
[0076] Based on the column of the current gene in the chromosome, find the corresponding column in the gene mapping index matrix to obtain the component type number and mounting point number operated by the gene;
[0077] If the mounting point number is -1, it means that the gene operates the mounting points of all corresponding types of components, otherwise it means that the gene only operates the mounting points with the corresponding number;
[0078] The number of the gene itself indicates the number of the machine to which it is assigned. Combining the above steps, we know which machine a certain type of component or a mounting point of this type of component should be assigned to.
[0079] Based on the above operations, the decoding of all genes is completed, and finally the process of decoding a chromosome into a solution to the problem is realized.
[0080] In another embodiment of the present invention, step S4, obtains the current number of CPU threads, generates a corresponding number of computing tasks, executes a genetic algorithm for each computing task separately, takes the longest CT in the split sub-substrate as the fitness of the individual, and calculates the fitness according to the initialized chromosome gene mapping index and set A, and the specific process is:
[0081] S41, initialize the individual chromosome, traverse each bit of the gene array, find the element type number of the corresponding position in the gene mapping index matrix, and combine the information of the set A to obtain the set B of machines to which the element type number corresponding to the type can be assigned; wherein the chromosome gene of the individual is represented by an array of length K, K is the number of columns of the gene mapping index matrix, so that the chromosome gene of the individual corresponds to the gene mapping index matrix one by one;
[0082] S42, generating a random number, randomly selecting a machine number from the set B, filling the machine number into the corresponding position of the gene array, and completing the chromosome initialization of the individual after the traversal is completed;
[0083] S43, complete the chromosome initialization of all individuals, then calculate the fitness of each individual, and complete the population initialization;
[0084] S44, iterative evolution of the population needs to go through four processes of selection, crossover, mutation, and fitness calculation in sequence, so as to return the optimal substrate splitting data after reaching a specified number of iterations or time.
[0085] The four processes are as follows:
[0086] Selection: Use roulette wheel to select the next generation of individuals for survival of the fittest, and use the longest CT in the split sub-matrix as the fitness of the individual. The final splitting goal is to minimize the longest CT. The higher the fitness, the lower the probability of selection.
[0087] Crossover: The individuals in the population are grouped in pairs and their chromosomes are crossed. During the crossover, a chromosome segment is randomly selected and the corresponding segment of the parent chromosome is exchanged to generate a new individual after the crossover.
[0088] Mutation: traverse the chromosome genes of individuals in the population, generate random numbers within the specified range and combine the mutation probability to determine whether the gene has mutated. If a mutation occurs, randomly select a different machine number from the allocated machines allowed by the gene to replace the gene.
[0089] Calculate fitness: First, generate empty substrate data for the split sub-substrates. Traverse the genes of the chromosome and determine all assigned components based on the gene mapping index. If the mounting number corresponding to the index is -1, extract all mounting points of this type of component from the original substrate data and add them to the sub-substrate data corresponding to the machine represented by the gene. If the index is a certain mounting point, only find the mounting point data from the original substrate data and add it to the sub-substrate data corresponding to the machine represented by the gene. Finally, the corresponding mounting data is assigned to each sub-substrate, and the chromosome analysis is completed. Finally, call the single substrate optimization module to optimize each sub-substrate separately. After the optimization is completed, calculate the CT of each sub-substrate, and take the longest CT as the fitness of the individual.
[0090] By generating a corresponding number of computing tasks according to the number of CPU threads in the computer, each task executes the genetic algorithm in an independent thread, realizing the parallelization of the algorithm and significantly accelerating the execution speed of the algorithm. In addition, as a load balancing allocation algorithm at the production line level, this algorithm frequently calls the placement optimization algorithm of a single machine to obtain the optimal CT (cycle time) of each machine under the current allocation result, and uses it as an evaluation indicator to measure the quality of the allocation result, and minimizes the overall CT through iterative optimization.
[0091] Based on the same inventive concept, a second aspect of an embodiment of the present invention provides a substrate splitting system for optimizing CT of multiple placement machines, comprising:
[0092] The substrate splitting module is used to load the production line information including the configuration parameters of all machines, the substrate to be split and the manually set parameters, and restore the original substrate data;
[0093] The component allocation analysis module is used to analyze the original substrate data, combine the configuration parameters of each machine and the manually set component allocation strategy, and calculate the set A of machines to which each type of component can be allocated;
[0094] A chromosome gene mapping index generation module is used to initialize the chromosome gene mapping index based on the original substrate data and the component allocation strategy. The information stored in the initialized chromosome gene mapping index is the mounting point number corresponding to each component type number.
[0095] A genetic algorithm execution module is used to obtain the current number of CPU threads, generate a corresponding number of computing tasks, and execute a genetic algorithm for each computing task separately to calculate the fitness according to the initialized chromosome gene mapping index and set A;
[0096] The split result generation module is used to obtain the results of all calculation tasks, take the one with the lowest fitness, take the corresponding split substrate data as the final split result, generate all sub-substrate files, and complete the splitting.
[0097] Based on the same inventive concept, a third aspect of an embodiment of the present invention provides a machine-readable storage medium, on which instructions are stored, and the instructions are used to enable a machine to execute the steps of the above-mentioned method of the present application.
[0098] In summary, the technical solution of the present invention achieves the following beneficial effects:
[0099] (1) The gene mapping index matrix provides an innovative solution for mapping chromosome genes to specific element allocation results. This design enables the genetic algorithm to easily generate specific allocation plans based on gene coding during the iteration process, and also facilitates the evaluation and optimization of the allocation results. In contrast, existing technologies may use more complex coding methods or evaluation mechanisms, which increases the difficulty of algorithm implementation or limits the optimization effect.
[0100] (2) Flexible component allocation strategy: Through complex substrate data processing and splitting logic, combined with the various attributes of the components and the configuration parameters of the machine, a flexible component allocation strategy is formulated. This strategy not only takes into account the physical characteristics of the components and the compatibility of the machine, but also allows users to further refine the allocation rules through manual settings, thereby achieving more accurate and efficient component allocation. In contrast, existing technologies may only consider some factors, or the allocation strategy is relatively fixed and difficult to adapt to diverse production needs.
[0101] (3) Parallel processing and efficiency improvement: Compared with the prior art, the present invention introduces a parallel processing mechanism. By generating a corresponding number of computing tasks according to the number of CPU threads of the computer, each task executes the genetic algorithm module in an independent thread. This parallel processing method significantly improves the computing efficiency. Traditional serial genetic algorithms often require huge amounts of calculations and take a long time to solve large-scale problems. The parallel design of the present invention can make full use of the computing resources of multi-core CPUs, accelerate the execution speed of the algorithm, and shorten the optimization cycle.
[0102] (4) By detecting fatal errors and optimizable errors in the substrate data, the algorithm directly reports errors for fatal errors that cannot be resolved, and provides warning information and continues execution for optimizable errors, thereby enhancing the robustness and user-friendliness of the algorithm.
[0103] Finally, the technical solution of the present invention is based on the parallel algorithm design of genetic algorithm, complex substrate data processing and splitting logic, innovative chromosome gene mapping index design, perfect error handling and feedback mechanism and comprehensive optimization goal realization, and the synergistic effect achieves: efficiency improvement: the parallel processing mechanism significantly improves the execution speed of the algorithm and shortens the production cycle; optimization effect: flexible component allocation strategy and innovative gene mapping index design make the optimization results more accurate and efficient; strong adaptability: the algorithm of the present invention can adapt to diverse production needs and support user-defined allocation rules and strategies; good stability: through strict data processing and error checking mechanism, the robustness and stability of the algorithm are guaranteed, and the efficiency is improved.
[0104] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity, or apparatus.
[0105] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.
Claims
1. A substrate splitting method for optimizing CT of multiple placement machines, characterized in that: The following steps are involved: Load the production line information including the configuration parameters of all machines, the substrate to be split and the manually set parameters to restore the original substrate data; Analyze the original substrate data, combine the configuration parameters of each machine and the manually set component allocation strategy, and calculate the set A of machines to which each type of component can be allocated; Based on the original substrate data and the component allocation strategy, the chromosome gene mapping index is initialized. The information stored in the initialized chromosome gene mapping index is the mounting point number corresponding to each component type number. The initialized chromosome gene mapping index is expressed through a gene mapping index matrix, and the first row of the matrix is the type number of the component; wherein, if the component allocation strategy does not allow a certain type of component to be allocated to multiple machines at the same time, then the type of component has only one column in the matrix, and the mounting point number is -1, indicating that all mounting points are allocated together during allocation; if the component allocation strategy allows a certain type of component to be allocated to multiple machines at the same time, then the type of component has N columns in the matrix, N is the number of mounting points of the type of component, each column represents a mounting point of the type, and the second row represents the mounting point number corresponding to the type of component; Obtain the current number of CPU threads and generate a corresponding number of computing tasks. Each computing task executes a genetic algorithm separately, and uses the longest CT in the split sub-substrate as the fitness of the individual. The fitness is calculated according to the initialized chromosome gene mapping index and set A. Each computing task executes a genetic algorithm separately, and the fitness is calculated according to the initialized chromosome gene mapping index and set A, including the following steps: (1) Initialize the individual chromosome, traverse each bit of the gene array, find the component type number at the corresponding position in the gene mapping index matrix, and combine the information of set A to obtain the set B of machines to which the type of component corresponding to the component type number can be assigned; wherein the chromosome gene of the individual is represented by an array of length K, K is the number of columns of the gene mapping index matrix, so that the chromosome gene of the individual corresponds to the gene mapping index matrix one by one; (2) Generate a random number, arbitrarily select a machine number in the set B, and fill the machine number into the base. The corresponding position of the factor array is completed after the traversal is completed. (3) The chromosome initialization of all individuals is completed, and then the fitness of each individual is calculated to complete the population initialization. (4) The iterative evolution of the population needs to go through four processes of selection, crossover, mutation, and fitness calculation in sequence to return the optimal substrate splitting data after reaching the specified number of iterations. The fitness calculation process is specifically as follows: generating empty substrate data for the split sub-substrate, traversing the genes of the chromosome, and determining all assigned components based on the gene mapping index matrix. If the mounting number corresponding to the index is -1, then extracting the mounting points of all corresponding types of components from the original substrate data and adding them to the sub-substrate data corresponding to the machine represented by the gene; if the index is a determined mounting point, then only the mounting point data is found from the original substrate data and added to the sub-substrate data corresponding to the machine represented by the gene; assigning corresponding mounting data to each sub-substrate to complete the chromosome analysis. Each sub-base is optimized separately. After the optimization is completed, the CT of each sub-base is calculated, and the longest CT is used as the fitness of the corresponding individual; Obtain the results of all calculation tasks, take the one with the lowest fitness, use the corresponding split substrate data as the final split result, generate all sub-substrate files, and complete the splitting.
2. The method according to claim 1, characterized in that The process of analyzing the original substrate data, combining the configuration parameters of each machine and the manually set component allocation strategy, and calculating the set A of machines to which each type of component can be allocated includes the following process: Traverse all component data to calculate key attributes when allocating components; Traverse all production line machines to calculate the key configuration attributes of the machines; In combination with the key attributes, key configuration attributes and component allocation strategy, a set A of machines to which each type of component can be allocated is determined. The set A is stored in the form of a component-machine allocation mapping table. The key in the component-machine allocation mapping table represents the type number of the component, and the value is a variable-length array. Each bit of the array represents the number of a machine that is allowed to be allocated.
3. The method according to claim 2, characterized in that If the length of the variable-length array is 0, it is determined that there are elements that cannot be allocated to any machine, and the step is terminated and an error is reported, otherwise, the next step is entered.
4. The method according to claim 1, characterized in that: The selection process is specifically as follows: using a roulette wheel method to select the next generation of individuals for survival of the fittest, taking the longest CT in the split sub-substrate as the fitness of the individual, and the final splitting goal is to minimize the longest CT, the higher the fitness, the lower the probability of selection; The mutation process is specifically as follows: traversing the chromosome genes of individuals in the population, generating random numbers within a specified range and combining the mutation probability to determine whether the gene has mutated; if a mutation has occurred, randomly selecting a different machine number from the allocated machines allowed by the mutated gene to replace the mutated gene.
5. The method according to any one of claims 1 to 4, characterized in that: Before restoring the original substrate data, all substrates to be split are first checked for fatal errors or optimizable errors. If a fatal error exists, an error is reported and the steps are paused; if an optimizable error exists, a warning message is prompted and the subsequent steps are continued.
6. The method according to any one of claims 1 to 4, characterized in that: Before restoring the original substrate data, it is first determined whether the substrate to be split is composed of multiple sub-substrates. If so, a mixed calculation of the substrate data is performed to restore the original substrate data.
7. The method according to claim 6, characterized in that The hybrid calculation of substrate data to restore the original substrate data specifically includes: firstly loading and generating all sub-substrate data, then merging the same information data, and reordering, naming, and merging the different information data.
8. A substrate splitting system for optimizing CT of multiple placement machines, characterized in that: include: The substrate splitting module is used to load the production line information including the configuration parameters of all machines, the substrate to be split and the manually set parameters, and restore the original substrate data; The component allocation analysis module is used to analyze the original substrate data, combine the configuration parameters of each machine and the manually set component allocation strategy, and calculate the set A of machines to which each type of component can be allocated; A chromosome gene mapping index generation module is used to initialize the chromosome gene mapping index based on the original substrate data and the component allocation strategy. The information stored in the initialized chromosome gene mapping index is the mounting point number corresponding to each component type number. The initialized chromosome gene mapping index is expressed through a gene mapping index matrix, and the first row of the matrix is the type number of the component; wherein, if the component allocation strategy does not allow a certain type of component to be allocated to multiple machines at the same time, then the type of component has only one column in the matrix, and the mounting point number is -1, indicating that all mounting points are allocated together during allocation; if the component allocation strategy allows a certain type of component to be allocated to multiple machines at the same time, then the type of component has N columns in the matrix, N is the number of mounting points of the type of component, each column represents a mounting point of the type, and the second row represents the mounting point number corresponding to the type of component; A genetic algorithm execution module is used to obtain the current number of CPU threads and generate a corresponding number of computing tasks. Each computing task executes the genetic algorithm separately and calculates the fitness according to the initialized chromosome gene mapping index and set A. Each computing task executes the genetic algorithm separately and calculates the fitness according to the initialized chromosome gene mapping index and set A, including the following steps: (1) Initializing the individual chromosome, traversing each bit of the gene array, finding the element type number of the corresponding position in the gene mapping index matrix, and combining the information of the set A to obtain the set B of machines to which the element type number corresponding to the type of element can be assigned; wherein the individual chromosome gene is represented by an array of length K, K is the number of columns of the gene mapping index matrix, so that the individual chromosome gene corresponds to the gene mapping index matrix one by one; (2) Generating a random number, arbitrarily selecting a machine number in the set B, and filling the machine number into the corresponding position of the gene array After the traversal is completed, the chromosome initialization of the individual is completed; (3) the chromosome initialization of all individuals is completed, and then the fitness of each individual is calculated to complete the population initialization; (4) the iterative evolution of the population needs to go through four processes of selection, crossover, mutation, and fitness calculation in sequence, so as to return the optimal substrate splitting data after reaching the specified number of iterations; the fitness calculation process is specifically as follows: generating empty substrate data for the split sub-substrates, traversing the genes of the chromosomes, and determining all the assigned components based on the gene mapping index matrix. If the mounting number corresponding to the index is -1, then the mounting points of all corresponding types of components are extracted from the original substrate data and added to the sub-substrate data corresponding to the machine represented by the gene; if the index is a determined mounting point, then only the mounting point data is found from the original substrate data and added to the sub-substrate data corresponding to the machine represented by the gene; the corresponding mounting data is assigned to each sub-substrate to complete the chromosome analysis; Each sub-base is optimized separately. After the optimization is completed, the CT of each sub-base is calculated, and the longest CT is used as the fitness of the corresponding individual; The split result generation module is used to obtain the results of all calculation tasks, take the one with the lowest fitness, take the corresponding split substrate data as the final split result, generate all sub-substrate files, and complete the splitting.
9. A machine-readable storage medium having instructions stored thereon, the instructions being used to enable a machine to execute the steps of the method of any one of claims 1 to 7 of the present application.
