Aviation composite material workshop production scheduling method, program product, computer readable storage medium and production scheduling system
Through recombining the process and using hybrid genetic algorithm optimization, the production scheduling problem of aviation composite materials workshops is solved, global optimization and efficient production are achieved, and the problem of global workshop scheduling that cannot be achieved in the existing technology is solved.
Patent Information
- Application Number
- CN202410877437.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-02
- Publication Date
- 2024-11-15
AI Technical Summary
The existing technology cannot effectively solve the production scheduling problem of aviation composite materials workshops, especially the inability to achieve global optimization of the entire workshop, resulting in low production efficiency and waste of resources.
By recombining the processes, establishing mathematical models and using mixed genetic algorithm optimization, combining taboo search and heuristic rules, optimizing the process sequence and batch processing method, meeting the constraints of raw material life and workpiece type, and achieving global optimization.
The global optimization scheduling of the aviation composite workshop is realized, which reduces the complexity of the problem, improves production efficiency, reduces idle time and unnecessary arrangements, and meets the requirements of high-quality production.
Smart Images

Figure CN118966400A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of workshop scheduling, and more specifically, relates to an aviation composite material workshop production scheduling method, a program product, a computer-readable storage medium and a production scheduling system. Background Art
[0002] Shop scheduling technology is one of the core technologies in the manufacturing field. Good shop scheduling capabilities can significantly improve overall efficiency. Classic shop scheduling problems include single-machine scheduling problems, mixed flow shop scheduling problems, and job shop scheduling problems. These problems originate from actual production and have been fully studied. However, highly simplified models are difficult to directly guide production. With the popularization of digital workshops, more and more production scenarios require scheduling methods that are more in line with actual problems.
[0003] Composite materials are a new type of material with good performance and are widely used in the chemical, pharmaceutical, aviation and other industries. Aviation composite materials have the characteristics of high specific strength, corrosion resistance, high and low temperature resistance, etc., and have gradually attracted attention in recent years. With the continuous development of my country's aviation industry, the demand for aviation composite material production scheduling is increasing.
[0004] The main process of aviation composite material production technology is as follows: When the company receives an order, the material cutter will take out the raw materials (prepreg) from the freezer in advance to thaw them according to the operation plan, and the countdown of the raw material life will begin from this point. After thawing, the automatic material cutter will cut the sheets according to the needs.
[0005] Then comes the parallel operation stage, with three production lines running simultaneously: two processing honeycombs and raw materials respectively, and one preparing tooling. The core of this is the raw material laying process: operators usually work in groups of multiple people, laying and pre-compacting the sheets layer by layer to ensure that the layers are tightly attached, and then checking for any debris left. After the laying process is completed, vacuuming is also required. Only after the vacuum degree is tested to be qualified can the workpiece enter the next process.
[0006] The honeycomb and raw materials are then assembled before the curing step, but the assembled workpieces cannot be placed directly on the bottom plate of the autoclave, but must be packaged in vacuum bags and then batched into the prepared tooling (similar to a frame). The tooling batch is then placed in a batch processor (autoclave), connected to a thermocouple (used to detect the real-time temperature), the autoclave door is closed, and the temperature, duration and other parameters of the autoclave are set before curing can begin. Different types of workpieces require different curing conditions. Under the same curing parameters such as part temperature, pressure, and processing time, the autoclave has the ability to process multiple parts at the same time. During the production process, it is necessary to consider producing parts with the same curing parameters (generally the same type of workpieces) in the same batch. The maximum loading capacity of the autoclave must also be considered, so the process of formulating the autoclave discharge plan is more complicated.
[0007] After the cans are taken out and the bags are removed and demoulded, the workpieces are ground and processed. At this time, some workpieces need to repeat the previous process for general inspection, protection, and storage, while some workpieces can be directly inspected, protected, and stored.
[0008] Among them, except for the curing process which has only one batch processing machine, the other processes have multiple equipments, so it is a mixed flow workshop. All processes can be divided into 5 major parts: honeycomb processing, raw material and tooling processing, curing, grinding, and general inspection and protection. Only the raw material and tooling processing, curing, and general inspection and protection are the processes that all types of workpieces must go through, and the other three parts are only required for some processes. Due to the extremely high product quality requirements, most processes are followed by an additional inspection process. These inspection processes are completed manually, and there is sufficient inspection human resources, so there is no need to consider them in scheduling. It is only necessary to inform when the corresponding process position should be reached after the scheduling Gantt chart is obtained.
[0009] In general, the aviation composite material production workshop is a special hybrid flow workshop, which is special in that it contains three special constraints: parallel operation process, batch processing machine, and limited life of raw materials. At present, the existing algorithms are only for the scheduling of a certain stage and cannot solve the scheduling problem of the entire workshop. Summary of the invention
[0010] In view of the above defects or improvement needs of the prior art, the present invention provides a production scheduling method, program product, computer-readable storage medium and production scheduling system for an aviation composite material workshop, the purpose of which is to solve the technical problem that the existing methods are unable to solve the production scheduling problem of an aviation composite material workshop.
[0011] To achieve the above object, according to one aspect of the present invention, a method for scheduling production of aviation composite materials workshop is provided, comprising the following steps:
[0012] Step 1: Recombine the original process, simplify the process, and establish the corresponding mathematical model as follows:
[0013] Objective function:
[0014] min C max , C max The latest completion time of the last process of all workpieces;
[0015] Constraints:
[0016]
[0017] C max ≥C im ,i=1,2,…,n
[0018] Where i represents the workpiece index number, n represents the number of workpiece index numbers; v represents the workpiece class index number, f represents the number of workpiece class index numbers; l iv Indicates whether workpiece i belongs to category v. If l iv =1, if l iv =0, it does not belong; j represents the stage index number, m is the number of stage index numbers; k j represents the parallel machine index of stage j, k j ∈{1,2,…,p j}, p j is the number of parallel machine indexes in stage j; t ij represents the processing time of workpiece i in stage j; u jv C represents the capacity of the machine at stage j for type v workpieces; ij represents the completion time of job i in stage j; C b represents the completion time of batch b, b max is the batch quantity; C max Indicates the latest completion time of the last process of all workpieces; x jki Indicates whether workpiece i is processed on machine k in stage j. If it is 1, it is processed; if it is 0, it is not processed; y ji1i2 Indicates whether workpiece i1 is processed before workpiece i2 at stage j, if it is 1, it is, if it is 0, it is not; z jib Indicates whether workpiece i is processed in batch b at stage j, if it is 1, it is, if it is 0, it is not; γ jvb Indicates whether workpiece type v is processed in batch b at stage j, if it is 1, it is, if it is 0, it is not;
[0019] Step 2: Set algorithm parameters and initialize the population for the problem in step 1. The initialized population is the initial coding population. Use a hybrid genetic algorithm to optimize the initial coding population to obtain the optimal production plan.
[0020] Furthermore, the method for recombining the original process and simplifying the process in step 1 is as follows:
[0021] (1) For the parallel processes of honeycomb processing, raw material and tooling processing, and curing, the tooling preparation start time is calculated based on the curing time;
[0022] (2) For the less frequently used honeycomb processing parts, merge them with the raw material and tooling processing parts and use the same team. When they are needed at the same time, they are processed serially to avoid the honeycomb processing team being idle;
[0023] (3) The inspection process after each process is incorporated into the process, and the duration of the process is increased accordingly;
[0024] (4) Combine consecutive processes processed by the same team at the same stage into one large process;
[0025] (5) As for workpiece batching and tooling batching, since in actual production an autoclave can hold multiple toolings, and each tooling can hold multiple workpieces depending on the type of workpiece, the two batchings are simplified to one batching based on the actual number of toolings and workpieces.
[0026] Furthermore, the mathematical model of step 1 also includes encoding and decoding, and the basic encoding method is two-dimensional encoding, including discrete processing stage encoding and batch processing stage encoding.
[0027] Furthermore, for each individual decoding, the process is as follows:
[0028] First, in the first discrete processing stage, the discrete processing stage encoding is used, and the process is decoded step by step. The machine is selected using the FCFS and FAM rules, and the order of the subsequent stages is determined by the end time of the previous stage. The first stage to end is the first stage to be processed.
[0029] Then the batch processing stage is decoded, and the order is arranged according to the end time of the discrete processing stage according to the type of workpiece, and the batches are processed according to the batch processing stage code;
[0030] The order of different batches is determined by the order of the last workpiece in each batch; in machine selection, the FCFS and FAM rules are also used;
[0031] In the second discrete processing stage, the order in which the workpieces are put on line is determined according to the end time of the batch processing stage; for the same batch, the order in which the workpieces are put on line is determined according to the end time of the first discrete processing stage, and the machine selection is determined according to FCFS and FAM.
[0032] Furthermore, the mathematical model of step 1 also includes raw material life constraints. Specifically, the optimization target is set to obj=M-makespan-k*n, where M is a preset sufficiently large positive integer, makespan is the maximum completion time, k is the penalty function coefficient, and n is the number of scrapped raw material workpieces.
[0033] Furthermore, in the hybrid genetic algorithm of step 2, multiple populations are used for global optimization, and taboo search is used for local optimization; after optimization to a certain stage, the second discrete processing stage sequence is optimized by taboo search; the neighborhood structure of the taboo search should be selected from the already set neighborhood structures, and the neighborhood structures of the discrete processing stage encoding and the batch processing stage encoding should both be reflected, or the neighborhood structure should be redesigned according to the optimization goal.
[0034] Further, the steps of the hybrid genetic algorithm are as follows:
[0035] First, set the algorithm parameters and initialize the population. After initializing the coding group, a population of size p is obtained, in which each individual has two codes, namely, the discrete processing stage code and the batch processing stage code; then check whether the number of iterations has reached the preset value, if not, perform iterative search;
[0036] Iterative search uses a genetic algorithm framework, including tabu search, local search, selection, crossover, and mutation;
[0037] In the taboo search, the discrete processing stage code and the batch processing stage code are slightly perturbed in turn, and at the same time, it is continuously checked whether both have failed to improve the historical optimal solution of the current taboo search for n consecutive times; if these two conditions are not met, the taboo search continues, and if they are met, it ends.
[0038] According to another aspect of the present invention, there is provided an aviation composite material workshop production scheduling program product, which, when called and executed, implements the aviation composite material workshop production scheduling method as described in any of the preceding items.
[0039] According to another aspect of the present invention, a computer-readable storage medium is provided, storing an aviation composite material workshop production scheduling program product as described above.
[0040] According to another aspect of the present invention, there is provided an aviation composite material workshop production scheduling system, comprising an aviation composite material workshop production scheduling program product as described above.
[0041] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects compared with the prior art:
[0042] (1) The present invention designs a complete set of solutions that can meet the complex process constraints of aviation composite material production workshops. Life check and local search are used to meet the raw material life constraints, and heuristic rules and batch stage encoding and decoding are used to meet the batch constraints.
[0043] (2) The present invention proposes a reconstruction method for the production scheduling problem of aviation composite material workshops. Compared with the original problem, the merged process reduces idle time, the merged process eliminates unnecessary arrangements between multiple processes within the same team, and the batching method is considered more comprehensively to achieve global optimization. In general, this simplified method can reduce the complexity of the problem, highlight the key points of the problem, and improve production efficiency.
[0044] (3) The encoding and decoding of the present invention can ensure that a larger solution space is covered while also focusing on a certain aspect. The discrete processing stage encoding determines the order in which the workpieces are put online, and the batch processing stage encoding determines the batching method, but does not determine the processing order of the batches. The order of batch processing is naturally determined by the arrival time of the last workpiece in each batch that arrives at the curing process. For the second discrete processing stage sequence, since its order adjustment space is small and its impact on the final solution is also small, no encoding is set, and it is only adjusted in the algorithm in the later stage of iteration, saving computing power.
[0045] (4) The present invention provides a workshop scheduling method, which obtains a model that meets practical requirements and is convenient for calculation by reconstructing the problem, and facilitates calculation by encoding abstract problems; uses a hybrid evolutionary algorithm to optimize the encoding, and integrates heuristic rules in decoding, thereby narrowing the solution space and quickly obtaining high-quality solutions; thereby solving the problem that the existing aviation composite material workshop production scheduling algorithm cannot achieve global optimization of the entire workshop.
[0046] (5) The present invention has good guiding significance for actual production. For the calculation examples generated using actual production data, the present invention can obtain solutions close to the lower bound in a relatively short time. BRIEF DESCRIPTION OF THE DRAWINGS
[0047] Figure 1 is a flow chart of a production scheduling method for an aviation composite material workshop in a preferred embodiment of the present invention;
[0048] Figure 2 is a schematic diagram of the original production process in a preferred embodiment of the present invention;
[0049] Figure 3 is a simplified schematic diagram of a production workshop in a preferred embodiment of the present invention;
[0050] Figure 4 is a schematic diagram of a two-dimensional code constructed according to a preferred embodiment of the present invention;
[0051] Figure 5 yes Figure 4 Schematic diagram of the changes in the coding according to the preferred embodiment of the present invention;
[0052] Figure 6 is a flow chart of a hybrid algorithm in a preferred embodiment of the present invention;
[0053] Figure 7 is a schematic diagram of the perturbation of discrete processing stages encoded in a tabu search constructed according to a preferred embodiment of the present invention;
[0054] Figure 8 is a schematic diagram of perturbations of batch stage encoding in tabu search constructed according to a preferred embodiment of the present invention;
[0055] Fig. 9 is a schematic diagram of the disturbance of the second discrete processing stage encoding in the local search constructed according to the preferred embodiment of the present invention;
[0056] Fig.10 is a crossover schematic diagram of discrete processing stages encoded in a genetic algorithm framework constructed in accordance with a preferred embodiment of the present invention;
[0057] Fig.11 is a crossover schematic diagram of batch processing stage encoding in a genetic algorithm framework constructed according to a preferred embodiment of the present invention;
[0058] Fig.12 is a schematic diagram of variations of discrete processing stages encoded in a genetic algorithm framework constructed according to a preferred embodiment of the present invention;
[0059] Fig.13 It is a schematic diagram of a mutation method 1 encoded in a genetic algorithm framework in a batch processing stage constructed according to a preferred embodiment of the present invention;
[0060] Fig.14 It is a schematic diagram of mutation method 2 encoded in a genetic algorithm framework in a batch processing stage constructed according to a preferred embodiment of the present invention;
[0061] Fig.15 is a Gantt chart of the optimal solution of Example 100(1) in the preferred embodiment of the present invention;
[0062] Fig.16 It is a diagram of a batch solution file of the optimal solution of example 100(1) in the preferred embodiment of the present invention. DETAILED DESCRIPTION
[0063] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.
[0064] like Figure 1 As shown, the present invention provides a production scheduling method for an aviation composite material workshop, comprising:
[0065] Step 1: Recombine the original process, simplify the process, and establish the corresponding mathematical model
[0066] S1: Problem reconstruction: Recombining the original processes and simplifying the process to facilitate subsequent calculations. The basic standards are: the original processes completed by the same skill group should be merged into the same process; the auxiliary original processes that are frequently repeated and have a very short time should be merged with the corresponding main processes.
[0067] Figure 2 This is a schematic diagram of the original workshop. Each box represents a process, the Chinese characters are the process name, and the number below is the processing time. Different color blocks represent the processes required for different types of workpieces. The blue part is the process that all workpieces flow through together. The honeycomb processing in the orange part and the machining (grinding) in the green part are only required for some types of workpieces. For this workshop, the specific reconstruction method is:
[0068] (1) For the parallel processes of honeycomb processing, raw material and tooling processing, and curing, since the tooling preparation time is much shorter than that of other branches, it can be ignored. The tooling preparation start time can be reversed based on the curing time.
[0069] (2) Since only a few workpieces need to use the honeycomb processing part, it can be combined with the raw material and tooling processing parts and use the same team. When they are needed at the same time, they can be processed serially. This will improve the team operation efficiency and avoid the honeycomb processing team being idle.
[0070] (3) The inspection process after each process is incorporated into the process, and the duration is increased accordingly. This modification is reasonable because the corresponding inspection will not be separated from the current workstation.
[0071] (4) Combine the continuous processes processed by the same team at the same stage into one large process. In this way, seven major processes are obtained: unloading, drying, laying, tooling, curing, demoulding, machining, and protection. Figure 3 As shown. Unloading, drying, and laying constitute the first discrete processing stage, tooling and curing constitute the batch processing stage, and demoulding, machining, and protection constitute the second discrete processing stage. Among them, the batch processing stage is the bottleneck.
[0072] (5) For workpiece batching and tooling batching, since in actual production the autoclave can hold four toolings, and each tooling can hold 2-4 workpieces depending on the type of workpiece, the autoclave capacity is directly set to 8, 12, or 16 workpieces, simplifying two batchings into one batching.
[0073] S2: Modeling and encoding: Establish a mathematical model for the reconstructed problem, determine the basic constraints, and use the metaheuristic basic framework to encode the problem. The basic encoding method is two-dimensional encoding: discrete processing stage encoding and batch processing stage encoding. Then randomly initialize the encoding group, and the number of encodings in the group is determined according to the scale of the problem, generally between 20 and 100.
[0074] The mathematical model is as follows:
[0075] Index and parameter notation:
[0076] i: Workpiece index number, i∈{1,2,…,n}
[0077] v: artifact class index, v∈{1,2,…,f}
[0078] l iv :Whether workpiece i belongs to category v, if l iv =1, if l iv =0 does not belong to
[0079] j: stage index, j∈{1,2,…,m}
[0080] k j : The parallel machine index of stage j, kj∈{1,2,…,p j}
[0081] t ij : Processing time of workpiece i in stage j
[0082] u jv : The capacity of the machine in stage j for type v workpieces
[0083] Decision variables:
[0084] C ij : Completion time of job i in stage j
[0085] C b : Completion time of batch b
[0086] C max : The latest completion time of the last process of all workpieces
[0087] x jki :Whether workpiece i is processed on machine k in stage j, if it is 1, it is processed, if it is 0, it is not processed
[0088] y ji1i2 : At stage j, whether workpiece i1 is processed before workpiece i2, if it is 1, it is, if it is 0, it is not
[0089] z jib : At stage j, whether workpiece i is processed in batch b, if it is 1, it is, if it is 0, it is not γ jvb : In stage j, whether workpiece type v is processed in batch b, if it is 1, it is, if it is 0, it is not
[0090] Objective function:
[0091] min C max
[0092] Constraints:
[0093] The workpiece can be processed on only one machine at each stage:
[0094]
[0095] The workpiece can only be processed in one batch at each stage:
[0096]
[0097] Only one type of workpiece can be processed in one batch:
[0098]
[0099] The number of workpieces in a batch does not exceed the capacity of the batch processor for that class:
[0100]
[0101] The relationship between the completion time of the previous process and the completion time of the next process of each workpiece:
[0102]
[0103] The processing time of a batch is determined by the type of workpiece:
[0104]
[0105] The relationship between the completion time of the previous batch and the completion time of the next batch on each batch processing machine:
[0106]
[0107] The relationship between the completion time of the previous workpiece and the completion time of the next workpiece on each machine:
[0108]
[0109] The maximum completion time is the maximum value of the completion time of the last process of all workpieces:
[0110] C max ≥C im ,i=1,2,…,n
[0111] According to this mathematical model, a Figure 4 The two codes are discrete processing stage code and batch processing stage code. Take two workpieces A and B as examples, A has 6 workpieces and B has 4 workpieces. The discrete processing stage code represents the order of going online, and its length is the total number of workpieces. Figure 4As shown, the discrete processing stage code represents that workpieces A, A, A, B, B, B, A, A, B, A enter the assembly line in sequence and start the unloading process. The batch processing stage code represents the number of batches composed of each workpiece and the actual loading capacity. Different types of workpieces cannot be mixed in the same batch, so each workpiece can be loaded with a maximum of several batches of workpieces - 1 workpiece per batch. Therefore, the batch processing stage code length is the same as the discrete processing stage code, and each workpiece occupies a length equal to the number of its own types. As shown in the figure, the batch processing stage code represents that type A workpieces can be loaded with a maximum of 6 batches, and 3 batches are actually loaded, with loading capacities of 2, 2, and 2, respectively. Type B workpieces can be loaded with a maximum of 4 batches, and 1 batch is actually loaded, with a loading capacity of 4.
[0112] S3: Model decoding: For the coding group of S2, use each individual to generate a complete production plan while satisfying the basic constraints. Calculate the quality of each individual.
[0113] Specifically, for each individual decoding, the process is as follows: First, in the first discrete processing stage, the discrete processing stage code is used to decode each process step by step, and the machine is selected using the FCFS (first come first serve) and FAM (first available machine) rules. The order of the subsequent stages is determined by the end time of the previous stage, and the first to end is processed first. Then the batch processing stage is decoded, and the order is arranged according to the end time of the discrete processing stage according to the type of workpiece, and processed in batches according to the batch processing stage code; the order of different batches is determined by the order of the last workpiece in each batch; in terms of machine selection, the FCFS and FAM rules are also used. In the second discrete processing stage, the order of workpieces going online is determined according to the end time of the batch processing stage; for the same batch, the order of workpieces going online is determined according to the end time of the first discrete processing stage, and the machine selection is determined according to FCFS and FAM.
[0114] Figure 5 yes Figure 4 Schematic diagram of the decoding process of the hybrid algorithm. In the discrete processing stage, the workpieces are first put online in the order of the discrete processing stage coding, and then scheduled according to the first come first service (FCFS) rule, and the sequence order changes. In the batch processing stage, batches are grouped according to the batch processing stage coding. If the same type of workpiece has not arrived, other workpieces in the same batch will enter the waiting state. Then check the raw material life constraint. If it is not satisfied, all processes of the corresponding workpiece are inserted to the end, and then taboo search is performed. After solidifying the process, a new second discrete processing stage sequence is obtained. Then the order is fine-tuned through local search, and the second discrete processing stage is entered. The final sequence is obtained by heuristic decoding according to the FCFS rule.
[0115] Calculate individual quality based on the optimization goal. The optimization goal is the actual needs of the workshop, such as minimizing the maximum completion time, meeting the most delivery deadlines, minimizing energy consumption, minimizing raw material loss, etc. In the production of aviation composite materials, minimizing raw material loss is an inevitable goal. The calculation method is that the time from the start of material unloading to the moment of entering the batch processing stage cannot exceed a certain limit, otherwise a larger penalty function needs to be added.
[0116] The way to take the raw material life constraint into consideration is as follows: In one embodiment, the optimization objective obj = M-makespan-k*n. M is a preset positive integer that is large enough, makespan is the maximum completion time, that is, the time when all workpieces complete all processes, k is the penalty function coefficient, and n is the number of scrapped raw material workpieces.
[0117] Step 2: Set algorithm parameters and initialize the population for the problem in step 1. The initialized population is the initial coding population. Use a hybrid genetic algorithm to optimize the initial coding population to obtain the optimal production plan.
[0118] S4: Algorithm optimization: The initial coding group in S2 is optimized using a hybrid genetic algorithm framework to ultimately obtain the optimal production solution.
[0119] In the hybrid genetic algorithm, multiple populations are used for global optimization, and taboo search is used for local optimization; after the optimization reaches a certain stage, the taboo search optimization is performed on the second discrete processing stage sequence. The neighborhood structure of the taboo search should be selected from the already set neighborhood structures, and the neighborhood structures of the discrete processing stage encoding and the batch processing stage encoding should be reflected, or the neighborhood structure should be redesigned according to the optimization goal.
[0120] The specific steps of the hybrid genetic algorithm are as follows: Figure 6 The specific steps include:
[0121] First, set the algorithm parameters and initialize the population. After initialization, a population of size p is obtained, in which each individual has two codes: discrete processing stage code and batch processing stage code. Then check whether the number of iterations has reached the preset value. If not, perform iterative search.
[0122] Iterative search uses a genetic algorithm framework, which includes five parts: taboo search, local search, selection, crossover, and mutation.
[0123] In the taboo search, the discrete processing stage code and the batch processing stage code are perturbed slightly in turn, and the two are constantly checked to see if they have failed to improve the historical optimal solution of the current taboo search for n consecutive times. If these two conditions are not met, the taboo search continues, and if they are met, it ends.
[0124] Figure 7 It is a schematic diagram of the perturbation of the discrete processing stage code in the taboo search constructed according to the preferred embodiment of the present invention. P is the original code, and O is the new code. The discrete processing stage code mutation is to randomly exchange the positions of two different types of workpieces. However, the probability of changing a chaotic sequence to a continuous sequence of the same type of workpiece should be set greater than the probability of changing a continuous sequence of the same type of workpiece to a chaotic sequence. Because the continuous sequence can reach the batch processing stage at a similar time, it is more likely to make full use of the autoclave and avoid waiting for the bottleneck process.
[0125] Figure 8 It is a schematic diagram of the perturbation of the batch processing stage code in the taboo search constructed according to the preferred embodiment of the present invention. P is the original code, and O is the new code. The specific method is to add +1 workpiece (only for batches with less than the capacity) and -1 workpiece (only for batches with more than 0 workpieces) to different batches of the same workpiece.
[0126] At the end of the iterative search, local search adjustments are made to the second discrete processing stage. Since the completion times of different batches vary greatly and the time overlaps of workpieces in the second discrete processing stage are rare, the adjustment space here is small and only a small amount of adjustment is required at the end of the iterative search.
[0127] Fig. 9 It is a schematic diagram of the disturbance of the second discrete processing sequence of the workpiece constructed according to the preferred embodiment of the present invention in the local search. P' is the original sequence, and O' is the new sequence. In the local search, only the workpieces that will overlap with the curing end time of the subsequent batches of workpieces participate in the mutation, and only workpieces of different types of batches are found, and only inserted later but not forward. This is because if there is no overlap in the time of the previous and subsequent workpieces, it is impossible to optimize the whole by changing the order. If the types of workpieces in the subsequent batches are the same, local exchange is meaningless. The workpiece cannot be inserted forward either, because the release time is determined by the batch and cannot be changed.
[0128] Use the selection operation to obtain the next generation population from the current population. The selection operation uses a combination of elite and roulette strategies. First, the top two elite solutions are retained, and the remaining individuals are randomly retained to the next generation according to probability.
[0129] Perform a crossover operation on the current population. The crossover operation is to crossover the discrete processing stage encoding and the batch processing stage encoding at the same time, with the purpose of retaining the better encoding fragments of the parent generation.
[0130] Fig.10It is a schematic diagram of the crossover of discrete processing stage codes constructed according to the preferred embodiment of the present invention. P1 and P2 are parent codes, and O1 and O2 are child codes. When crossing, firstly, some fragments are randomly selected, and the corresponding genes of P1 and P2 are directly saved in O1 and O2 respectively. The remaining positions are first arranged in the order of the remaining genes of the other parent code, and finally the missing genes are filled.
[0131] Fig.11 It is a schematic diagram of coding crossover in the batch processing stage constructed according to the preferred embodiment of the present invention. P1 and P2 are parent codes, and O1 and O2 are child codes. When crossing, first randomly select some fragments, and directly save the corresponding genes of P1 and P2 into O1 and O2 respectively, and the remaining positions are arranged in the order of the remaining genes of the other parent code. Finally, check whether the number of each workpiece has changed. If it increases a, randomly select a times, and select a batch -1 with a load greater than 0 each time; if it decreases b, randomly select b times, and select a batch +1 with a load less than the capacity each time.
[0132] Perform mutation operation on the current population. Compared with the perturbation in local search, the mutation operation has a larger degree of change. Two mutation strategies are designed for batch processing stage encoding, and one mutation strategy is designed for discrete processing stage encoding.
[0133] Fig.12 The figure is a schematic diagram of the variation of the discrete processing stage encoding in the genetic algorithm framework constructed according to the preferred embodiment of the present invention. P is the original encoding, and O is the new encoding. For the original encoding, several pairs of different types of workpieces are randomly selected and their positions are exchanged to obtain the new encoding.
[0134] Fig.13 1 is a schematic diagram of a mutation method 1 of batch processing stage encoding in a genetic algorithm framework constructed according to a preferred embodiment of the present invention. P is the original encoding, and O is the new encoding. For the original encoding, several pairs of batches of the same type are randomly selected (each pair of batches has at least one workpiece number greater than 0 or less than 0), and the number of workpieces in the two batches is +1 and -1 respectively.
[0135] Fig.14 2 is a schematic diagram of a mutation method 2 of batch processing stage encoding in a genetic algorithm framework constructed according to a preferred embodiment of the present invention. P is the original encoding, and O is the new encoding. For the original encoding, several pairs of batches of the same type (each pair of batches has a different number of workpieces) are randomly selected and the positions of the two batches are exchanged.
[0136] S5: Problem restoration: Based on the optimal production plan in S4, disassemble and determine the start and completion time and processing resources of each original process.
[0137] To verify the practical application effect of the present invention, a simulation experiment is conducted below. Using 6 products (A, B, C, D, E, F) existing in an actual workshop as data input, 6 examples of 3 scales are generated. At the same time, according to the lower bound estimation method of the mixed flow workshop and considering the influence of the batch processing machine, the optimization lower bound of this problem is estimated. Since it is difficult to consider the raw material life constraint when estimating the lower bound, this lower bound is actually underestimated.
[0138] Table 1 Workpiece number settings and calculation results for six production examples
[0139]
[0140] From the above experiments, it can be seen that in the six examples generated using the actual product processing data, the maximum completion time of the optimal solution obtained by the present invention is close to the theoretical lower bound, with an error between 0.18% and 11.02%. Among them, the Gantt chart and batching scheme of the optimal solution of the 100 (1) example are as follows Fig.15 and Fig.16 It can be seen that for most resources, the processes are closely arranged and the scheduling effect is good.
[0141] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A production scheduling method for an aviation composite material workshop, characterized in that: The steps include: Step 1: Recombine the original process, simplify the process, and establish the corresponding mathematical model as follows: Objective function: min C max , C max The latest completion time of the last process of all workpieces; Constraints: C max ≥C im ,i=1,2,…,n Where i represents the workpiece index number, n represents the number of workpiece index numbers; v represents the workpiece class index number, f represents the number of workpiece class index numbers; l iv Indicates whether workpiece i belongs to category v. If l iv =1, if l iv =0, it does not belong; j represents the stage index number, m is the number of stage index numbers; k j represents the parallel machine index of stage j, k j ∈{1,2,…,p j }, p j is the number of parallel machine indexes in stage j; t ij represents the processing time of workpiece i in stage j; u jv C represents the capacity of the machine at stage j for type v workpieces; ij represents the completion time of job i in stage j; C b represents the completion time of batch b, b max is the batch quantity; C max Indicates the latest completion time of the last process of all workpieces; x jki Indicates whether workpiece i is processed on machine k in stage j. If it is 1, it is processed; if it is 0, it is not processed; y ji1i2 Indicates whether workpiece i1 is processed before workpiece i2 at stage j, if it is 1, it is, if it is 0, it is not; z jib Indicates whether workpiece i is processed in batch b at stage j, if it is 1, it is, if it is 0, it is not; γ jvb Indicates whether workpiece type v is processed in batch b at stage j, if it is 1, it is, if it is 0, it is not; Step 2: Set algorithm parameters and initialize the population for the problem in step 1. The initialized population is the initial coding population. Use a hybrid genetic algorithm to optimize the initial coding population to obtain the optimal production plan.
2. The method for scheduling production of aviation composite materials workshop according to claim 1, characterized in that: The method for recombining the original process and simplifying the process in step 1 is as follows: (1) For the parallel processes of honeycomb processing, raw material and tooling processing, and curing, the tooling preparation start time is calculated based on the curing time; (2) For the less frequently used honeycomb processing parts, merge them with the raw material and tooling processing parts and use the same team. When they are needed at the same time, they are processed serially to avoid the honeycomb processing team being idle; (3) The inspection process after each process is incorporated into the process, and the duration of the process is increased accordingly; (4) Combine consecutive processes processed by the same team at the same stage into one large process; (5) As for workpiece batching and tooling batching, since in actual production an autoclave can hold multiple toolings, and each tooling can hold multiple workpieces depending on the type of workpiece, the two batchings are simplified to one batching based on the actual number of toolings and workpieces.
3. A production scheduling method for aviation composite material workshop according to claim 1 or 2, characterized in that: The mathematical model of step 1 also includes encoding and decoding, and the basic encoding method is two-dimensional encoding, including discrete processing stage encoding and batch processing stage encoding.
4. The method for scheduling production of aviation composite materials workshop according to claim 3, characterized in that: For each individual decoding, the process is as follows: First, in the first discrete processing stage, the discrete processing stage encoding is used, and the process is decoded step by step. The machine is selected using the FCFS and FAM rules, and the order of the subsequent stages is determined by the end time of the previous stage. The first stage to end is the first stage to be processed. Then the batch processing stage is decoded, and the order is arranged according to the end time of the discrete processing stage according to the type of workpiece, and the batches are processed according to the batch processing stage code; The order of different batches is determined by the order of the last workpiece in each batch; in machine selection, the FCFS and FAM rules are also used; In the second discrete processing stage, the order in which the workpieces are put on line is determined according to the end time of the batch processing stage; for the same batch, the order in which the workpieces are put on line is determined according to the end time of the first discrete processing stage, and the machine selection is determined according to FCFS and FAM.
5. The method for scheduling production of composite materials for aviation workshops according to claim 1 or 2, characterized in that: The mathematical model of step 1 also includes raw material life constraints. Specifically, the optimization target is set to obj = M-makespan-k*n, where M is a preset sufficiently large positive integer, makespan is the maximum completion time, k is the penalty function coefficient, and n is the number of scrapped raw material workpieces.
6. The method for scheduling production of aviation composite materials workshop according to claim 1 or 2, characterized in that: In the hybrid genetic algorithm of step 2, multiple populations are used for global optimization and taboo search is used for local optimization. After the optimization reaches a certain stage, the second discrete processing stage sequence is optimized by taboo search. The neighborhood structure of taboo search should be selected from the already set neighborhood structures. The neighborhood structures of discrete processing stage encoding and batch processing stage encoding should be reflected, or the neighborhood structure should be redesigned according to the optimization goal.
7. The method for scheduling production of aviation composite materials workshop according to claim 6, characterized in that: The steps of the hybrid genetic algorithm are as follows: First, set the algorithm parameters and initialize the population. After initializing the coding group, a population of size p is obtained, in which each individual has two codes, namely, the discrete processing stage code and the batch processing stage code; then check whether the number of iterations has reached the preset value, if not, perform iterative search; Iterative search uses a genetic algorithm framework, including tabu search, local search, selection, crossover, and mutation; In the taboo search, the discrete processing stage code and the batch processing stage code are slightly perturbed in turn, and at the same time, it is continuously checked whether both have failed to improve the historical optimal solution of the current taboo search for n consecutive times; if these two conditions are not met, the taboo search continues, and if they are met, it ends.
8. An aviation composite material workshop production scheduling program product, characterized in that: When called for execution, the aviation composite material workshop production scheduling method according to any one of claims 1 to 7 is implemented.
9. A computer-readable storage medium, characterized in that: The product of the production scheduling program for an aviation composite material workshop as claimed in claim 8 is stored.
10. An aviation composite material workshop production scheduling system, characterized in that: A production scheduling program product for an aviation composite material workshop comprising the product described in claim 8.
Citation Information
Cited By
Method for improving curing efficiency of autoclave of aeronautical composite component
CN119599212A
Electronic product manufacturing process grouping method, device, equipment and program product
CN120087719A
Aviation composite material solidification scheduling method and system considering multilayer space and time window constraints
CN121787872A
