Multi-target flexible workshop scheduling method and device, equipment and storage medium
Through the improved differential evolution algorithm (IADE), using adaptive search strategies and new coding methods, the problem of limited productivity improvement in traditional algorithms in the multi-objective flexible workshop scheduling model is solved, and more efficient flexible workshop scheduling optimization is achieved.
Patent Information
- Application Number
- CN202410212242.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-02-27
- Publication Date
- 2024-11-08
AI Technical Summary
Traditional differential evolution algorithms are not ideal for the optimization of multi-objective flexible workshop scheduling models, resulting in limited improvement in production efficiency in manufacturing operations.
An improved differential evolution algorithm (IADE) is proposed to generate feasible solutions and optimize the scheduling scheme through adaptive search strategies, using successful and failed parameter pairs, adopt new encoding methods and local searches based on critical paths.
The IADE algorithm performs better than other algorithms in benchmark examples, can effectively optimize the flexible workshop scheduling model, improve production efficiency in manufacturing operations, and the generated solution sets have good diversity and convergence.
Smart Images

Figure CN118915629A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of manufacturing scheduling, and in particular to a multi-objective flexible workshop scheduling method, device, equipment and storage medium. Background Art
[0002] The job shop scheduling problem (JSP) is a highly complex scheduling problem often encountered in manufacturing systems and has been proven to be an NP-hard problem. In manufacturing systems, planning or scheduling schemes have a vital impact on the production of the shop floor. Its core function is to coordinate the overall supply, including capabilities, resources, manpower and materials, to match the current demand. The demand is determined by the specific resources and capabilities required to complete each job, which is closely linked to the actual work orders within the manufacturing facility. In essence, the scheduling scheme is the hub for coordinating the smooth coordination between available resources and production demand, which can ensure the efficiency and optimization of the manufacturing process. A properly managed scheduling scheme can not only improve production efficiency, but also help to shorten the manufacturing cycle, and ultimately improve the overall operational efficiency of the manufacturing facility. Therefore, developing effective scheduling strategies and systems remains the focus of manufacturers in today's competitive environment. The goal of job shop scheduling is to effectively and optimally integrate these internal supply and demand components. Many job shop manufacturers rely on outdated and inefficient scheduling methods that are incompatible or unreliable in the chaotic shop environment, which has a significant negative impact on the production process and puts them at a disadvantage in the competition. Overall, using the right job shop scheduling solution will effectively handle unstable shop floor conditions and ultimately improve the productivity of your manufacturing operations.
[0003] The flexible job shop scheduling problem (FJSP) is a more complex version of the classic job shop scheduling problem, which presents a combinatorial optimization challenge. In response to this problem, people have invested a lot of time and money to explore and design more effective methods to find more effective ways to solve the increasingly severe and advanced FJSP problem. Generally speaking, FJSP involves two or more machines that can handle workpiece processing, and each machine has a unique processing time for each workpiece processing. Each workpiece includes multiple processing processes with a fixed processing route, and each machine can perform multiple processing processes for multiple jobs. Therefore, in FJSP, in addition to arranging the order of processing processes on each machine for each workpiece, it is also necessary to appropriately allocate machines to perform the processing of each workpiece. The flexible job shop scheduling problem (FJSP) is usually defined as N workpieces processed on M machines. According to the process route requirements of the workpiece, each workpiece consists of one or more processes with sequential constraints, and each process may have multiple machines that can be selected for processing. During the processing process, the following constraints need to be met:
[0004] 1) All machines are available at time 0;
[0005] 2) All processes of all workpieces can be processed at 0 time;
[0006] 3) At the same time, each process can only be processed by one machine at most. Once the processing starts, it cannot be interrupted until the processing is completed;
[0007] 4) Each machine can only process one process at a time;
[0008] 5) There are sequence constraints between the processes of the same workpiece, but there are no sequence constraints between the processes of different workpieces;
[0009] 6) Different workpieces have the same processing priority;
[0010] 7) Equipment failure is not considered;
[0011] 8) Not considering the time required for process turnaround.
[0012]
[0013] The table above gives an example of FJSP, where the rows correspond to the workpiece processes and the columns correspond to the machines. In this example, there are three machines, namely, M 1 , M 2 and M 3 Each item in the table represents the processing time of the corresponding process on the corresponding machine. For example, 12 , listed as M 2 The item "3" means that the processing time of workpiece 1, process 2 on machine 2 is 3 units of time, and the mark "-" means that the machine cannot perform the corresponding operation.
[0014] When solving the flexible job shop scheduling problem (FJSP), the traditional differential evolution algorithm initializes the population by randomly generating individuals in a given search space. i , (i=1,2,…,N) all contain D genes as variables, where N is the number of individuals or the size of the population. In each generation or iteration, each individual becomes a parent in turn, and each parent x i The processing is as follows: perform a mutation operation. In addition to the parent x i In addition, several individuals are selected from the population, and the first individual is the basis vector. All other individuals are paired and the difference vector is obtained. The difference vector is scaled by the scaling factor F and added to the basis vector to obtain the mutation vector m i However, these traditional DE series algorithms are not ideal in optimizing the multi-objective flexible shop scheduling model, and their effects on improving production efficiency in manufacturing operations are limited.
[0015] Therefore, how to provide a multi-objective flexible workshop scheduling method to improve the production efficiency of manufacturing operations is a technical problem that needs to be solved urgently. Summary of the invention
[0016] The main purpose of the present invention is to provide a multi-objective flexible workshop scheduling method, device, equipment and storage medium, aiming to solve the above problems.
[0017] To achieve the above object, the present invention provides a multi-objective flexible workshop scheduling method, comprising the following steps:
[0018] S1: Encode the scheduling scheme of multi-objective flexible workshop scheduling;
[0019] S2: Initialize the population;
[0020] S3: Use the improved differential evolution algorithm to perform mutation processing, crossover processing and selection processing on the population, and adjust the parameters of the differential evolution algorithm according to the processing results;
[0021] S4: Perform local search on the selected population to obtain a new population;
[0022] S5: Execute the restart strategy on the population after local search and the original population;
[0023] S6: Repeat S3-S5 until the maximum number of evaluations is reached, obtain the optimized population search range, and determine the optimal solution based on the condition of minimizing the preset goal.
[0024] Optionally, step S1: encoding the scheduling scheme of the multi-objective flexible workshop scheduling, specifically including:
[0025] Process coding: Generate a process sequence vector based on the processing order of each workpiece in each process;
[0026] Machine encoding: Generate a machine allocation vector based on the machine serial number that can execute the processing task of the target workpiece in the target process.
[0027] Optionally, step S2: initializing the population, specifically including:
[0028] Randomly select a target selection strategy from a preset machine selection strategy group and a target process sorting strategy from a preset process sorting strategy group, and combine them to obtain an initialization strategy;
[0029] The population is initialized using the initialization strategy. After obtaining half of the chromosome of the machine selection part according to the machine selection strategy, the other half of the chromosome of the process sorting part is arranged according to the process selection strategy based on the information of this half of the chromosome.
[0030] Optionally, step S3: using an improved differential evolution algorithm to perform mutation processing, crossover processing and selection processing on the population, specifically including:
[0031] Mutation processing: Use a specific mutation strategy to perform mutation processing on each individual in the population. The expression of the specific strategy is specifically:
[0032]
[0033] v i,G is the mutation vector, x i,G is the current individual, x best,G is an individual randomly selected from a previously preset proportion of individuals, x r1,G is an individual randomly selected from the current population, is an individual randomly selected from the current population and the external archive, and the parameter F i ∈[0,1] is the scaling factor, which is used to control the size of the mutation operator;
[0034] Crossover treatment: after mutation treatment, individuals are crossed with their parents; exponential crossover is used in the process part, and binomial crossover is used in the machine part;
[0035] Selection processing: If the crossover succeeds, individual u i,G Dominate the original individual or with the original individual x i,G Do not dominate each other, then F i With CR i Store in the success collection and set x i,G Save to external archive; otherwise, F i With CR i Save to the failed collection.
[0036] Optionally, step S3: adjusting the parameters of the differential evolution algorithm according to the processing results, specifically includes: adjusting the parameters of the differential evolution algorithm using the JADEFI strategy based on the success set and the failure set.
[0037] Optionally, step S4: performing a local search on the selected population to obtain a new population, specifically comprising:
[0038] (1) Local search of machine parts:
[0039] Reassignment of the machine with the largest load: Determine the machine with the largest load, randomly select any process on it, and reassign it to other machines that can process. If there is only one machine that can process, reselect the process;
[0040] Reassign the kq machines with the largest load: Determine the top k machines with the largest load, and randomly select q of them. Randomly select a process assigned to these machines and reassign it;
[0041] Random redistribution: randomly generate an integer I from 1 to M (M is the number of machines), randomly select I processes, and redistribute them. If there is only one machine that can process, no redistribution is performed.
[0042] (2) Partial search of process:
[0043] Insertion operation: Randomly select two processes and insert the previous process after the latter, or insert the latter process before the former.
[0044] Exchange operation: Randomly select two processes. If the two processes belong to the same process, reselect and exchange the positions of the two processes.
[0045] Optionally, step S5: executing a restart strategy on the population after the local search and the original population, specifically including: after differential evolution and local search, reconstructing duplicate individuals in the population.
[0046] In addition, in order to achieve the above object, the present invention also provides a multi-objective flexible workshop scheduling device, comprising:
[0047] Coding module: used to encode the scheduling scheme of multi-objective flexible workshop scheduling;
[0048] Initialization module: used to initialize the population;
[0049] Differential evolution module: used to use the improved differential evolution algorithm to perform mutation, crossover and selection processing on the population, and adjust the parameters of the differential evolution algorithm according to the processing results;
[0050] Local search module: used to perform local search on the selected population to obtain a new population;
[0051] Restart module: used to execute restart strategy on the population after local search and the original population;
[0052] Loop module: used to repeat differential evolution and local search until the maximum number of evaluations is reached, to obtain the optimized population search range, and to determine the optimal solution under the condition of minimizing the preset goal.
[0053] In addition, in order to achieve the above-mentioned objectives, the present invention also provides a multi-objective flexible workshop scheduling device, which includes: a memory, a processor, and a multi-objective flexible workshop scheduling program stored in the memory and executable on the processor, and when the multi-objective flexible workshop scheduling program is executed by the processor, the steps of the multi-objective flexible workshop scheduling method as described above are implemented.
[0054] In addition, in order to achieve the above-mentioned purpose, the present invention also provides a storage medium, on which a multi-objective flexible workshop scheduling program is stored, and when the multi-objective flexible workshop scheduling program is executed by a processor, the steps of the above-mentioned multi-objective flexible workshop scheduling method are implemented.
[0055] The beneficial effects of the present invention are as follows: a multi-objective flexible workshop scheduling method, device, equipment and storage medium are proposed, and a new adaptive differential evolution algorithm (IADE) is proposed to solve the multi-objective flexible workshop scheduling problem. In the algorithm, not only the parameter pairs of successful mutation crossover are utilized, but also the parameter pairs of failure are utilized; in addition, the search strategy based on IADE is adaptive, and it has strong development capabilities while retaining strong exploration capabilities; and the algorithm adopts a new encoding method, which can enable individuals to directly generate feasible solutions in the search space; at the same time, the algorithm also performs a local search based on the critical path for each individual; in order to evaluate the performance of IADE, IADE is compared with some advanced algorithms on a benchmark instance; experimental results show that IADE is superior to other algorithms. BRIEF DESCRIPTION OF THE DRAWINGS
[0056] Figure 1 A schematic diagram of the device structure of the hardware operating environment involved in the embodiment of the present invention;
[0057] Figure 2 A schematic diagram of a flow chart of a multi-objective flexible workshop scheduling method according to an embodiment of the present invention;
[0058] Figure 3 It is a schematic diagram of the principle of the multi-objective flexible workshop scheduling method according to an embodiment of the present invention;
[0059] Figure 4 A Gantt chart of a feasible scheduling solution is shown for an embodiment of the present invention.
[0060] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION
[0061] In order to make the purpose, technical solution 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.
[0062] like Figure 1 As shown, Figure 1 It is a schematic diagram of the device structure of the hardware operating environment involved in the embodiment of the present invention.
[0063] like Figure 1 As shown, the device may include: a processor 1001, such as a CPU, a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory, or a stable memory (non-volatile memory), such as a disk memory. The memory 1005 may optionally be a storage device independent of the aforementioned processor 1001.
[0064] Those skilled in the art will understand that Figure 1 The structure of the device shown in the figure does not constitute a limitation of the device, and may include more or less components than shown in the figure, or combine certain components, or arrange the components differently.
[0065] like Figure 1 As shown, the memory 1005 as a computer storage medium may include an operating system, a network communication module, a user interface module, and a multi-objective flexible shop scheduling program.
[0066] exist Figure 1 In the terminal shown, the network interface 1004 is mainly used to connect to the backend server and perform data communication with the backend server; the user interface 1003 is mainly used to connect to the client (user end) and perform data communication with the client; and the processor 1001 can be used to call the multi-objective flexible workshop scheduling program stored in the memory 1005 and perform the following operations:
[0067] Encode the scheduling scheme for multi-objective flexible shop scheduling;
[0068] Initialize the population;
[0069] Use the improved differential evolution algorithm to perform mutation, crossover and selection processing on the population, and adjust the parameters of the differential evolution algorithm according to the processing results;
[0070] Perform local search on the selected population to obtain a new population;
[0071] Execute the restart strategy on the population after local search and the original population;
[0072] Repeat differential evolution and local search until the maximum number of evaluations is reached to obtain the optimized population search range, and determine the optimal solution based on the condition of minimizing the preset goal.
[0073] The specific embodiments of the present invention applied to the device are basically the same as the embodiments of the following multi-objective flexible shop scheduling method, and will not be described in detail here.
[0074] The embodiment of the present invention provides a multi-objective flexible workshop scheduling method, referring to Figure 2 , Figure 2 It is a flow chart of an embodiment of the multi-objective flexible workshop scheduling method of the present invention.
[0075] In this embodiment, the multi-objective flexible workshop scheduling method includes the following steps:
[0076] S1: Encode the scheduling scheme of multi-objective flexible workshop scheduling;
[0077] S2: Initialize the population;
[0078] S3: Use the improved differential evolution algorithm to perform mutation processing, crossover processing and selection processing on the population, and adjust the parameters of the differential evolution algorithm according to the processing results;
[0079] S4: Perform local search on the selected population to obtain a new population;
[0080] S5: Execute the restart strategy on the population after local search and the original population;
[0081] S6: Repeat S3-S5 until the maximum number of evaluations is reached, obtain the optimized population search range, and determine the optimal solution based on the condition of minimizing the preset goal.
[0082] It should be noted that in the traditional differential evolution algorithm, the population is initialized by randomly generating individuals in a given search space. i , (i=1,2,…,N) all contain D genes as variables, where N is the number of individuals or the size of the population. In each generation or iteration, each individual becomes a parent in turn. Each parent x i The processing is as follows: perform a mutation operation. In addition to the parent x i In addition, several individuals are selected from the population. The first individual is the basis vector. All other individuals are paired and a difference vector is obtained. The difference vector is scaled by the scaling factor F and added to the basis vector to obtain the mutation vector m i The following are several commonly used mutation strategies:
[0083] DE / rand / 1,
[0084] v i,G =x r1,G +F·(x r2,G -x r3,G ),
[0085] DE / rand / 2,
[0086] v i,G =x r1,G +F·(x r2,G -x r3,G )+F·(x r4,G -x r5,G ),
[0087] DE / best / 1,
[0088] v i,G =x best,G +F·(x r1,G -x r2,G ),
[0089] DE / best / 2,
[0090] v i,G =x best,G +F·(x r1,G -x r2,G )+F·(x r3,G -x r4,G ),
[0091] DE / rand-to-best / 1,
[0092] v i,G =x r1 +F·(x best,G -x r1 )+F·(x r2,G -x r3,G ),
[0093] DE / current-to-best / 1,
[0094] v i,G =x i,G +F·(x best,G -x i,G )+F·(x r1,G -x r2,G ),
[0095] Among them, individual x r1 、x r2,G 、x r3,G 、x r4,G and x r5,G is randomly selected from the current population and is not repeated. i,G is the current vector, x best,G is the best individual in the G generation population. i,G Then, make it equal to the parent x i,G Perform cross selection to generate the test vector u i,GThere are generally two types of crossover, binomial crossover and exponential crossover. The individuals produced by binomial crossover come discretely from the parent individuals or variant individuals in each dimension, while the individual components produced by exponential crossover come continuously from the parent individuals or variant individuals, so the impact of population individual differences on exponential crossover is greater than that on binomial crossover. The binomial crossover operation compares the size of the random number with the crossover probability. If the random number is less than the crossover rate or j=j rand When the test vector u i,G The value of the current dimension is determined by the mutation vector v i,G To provide, otherwise by the parent x i,G Provide, set j = j rand The purpose is to prevent all dimensions from coming from the parent x i,G , resulting in no change in this evolution. The specific formula is as follows:
[0096]
[0097] Here, rand[0,1] represents a random number uniformly randomly selected in [0,1). rand is the decision variable index uniformly randomly selected in [1,D], ensuring that at least one gene comes from the mutation vector. The parameter CR∈[0,1] is the crossover operator, which is used to control the crossover rate.
[0098] DE uses the greedy selection method to perform the selection operation, which reflects the idea of natural selection and survival of the fittest in nature. According to the fitness value, the better individual is greedily selected one-to-one from the parent individual and the test individual to become the parent individual of the next generation. The specific operation is as follows:
[0099]
[0100] In order to solve the problem that these traditional DE series algorithms are not ideal in optimizing the multi-objective flexible workshop scheduling model and have limited effect on improving the production efficiency in manufacturing operations, the embodiment of the present application proposes an improved differential evolution algorithm based on restart strategy and local search. The principle is as follows: Figure 3 shown.
[0101] Step S1: Encode and decode the scheduling scheme;
[0102] Step S2: population initialization;
[0103] Step S3: Apply the improved differential evolution algorithm to mutate, crossover and select the population, and adjust the parameters of the differential evolution algorithm according to the results;
[0104] Step S4: Perform local search on the selected population to obtain a new population;
[0105] Step S5: Execute the restart strategy on the population after local search and the original population;
[0106] Step S6: Repeat S3-S5 until the maximum number of evaluations is reached.
[0107] Furthermore, the coding scheme of step S1 consists of two parts: the process sequence vector [process order sequence (OS) of the process] and the machine allocation vector [process machine allocation (MA)]. This embodiment adopts a new encoding method suitable for the mutation and crossover operators of the differential evolution algorithm, which can directly generate feasible solutions during the search process. In the process coding part, the gene is encoded with an integer, and the value is the processing priority of the process corresponding to the array subscript. The order of the machine coding is arranged in sequence according to the workpiece process, and the value represents the machine number in the process processable machine set.
[0108] Furthermore, in step S2, in order to ensure that the initial population has high quality and diversity, the following initialization method is used. Common initialization strategies for machine selection:
[0109] (1) Global Minimal Workload: A workpiece is randomly selected from all workpieces. All processes are arranged on the processing machine with the smallest cumulative processing load according to the processing sequence of the workpiece. After all processes of the workpiece are arranged, other workpieces are randomly selected.
[0110] (2) Local Minimal Workload: A workpiece is randomly selected at the beginning. When all the processes of this workpiece are randomly arranged on the machines, a workpiece is randomly selected from the remaining workpieces.
[0111] (3) Random Assignment: Each process of each workpiece randomly selects a machine from the available machine set.
[0112] (4) Minimum Work Time: Each process of each workpiece is arranged on the machine with the shortest processing time in the optional machine set.
[0113] Common initialization strategies for process sequencing:
[0114] (1) Most Work Remaining: Sort all workpieces by their remaining processing time, select the workpiece with the largest remaining processing time to be processed, and update the remaining processing time array.
[0115] (2) Most Number of Operations Remaining: Sort all workpieces according to the number of remaining operations and select the operation with the most remaining operations to be processed.
[0116] (3) Random Dispatching: Randomly arrange the order in which each process appears in the chromosome of the process sorting part.
[0117] This embodiment adopts an initialization strategy that combines different machine selection strategies and process sorting strategies. When initializing the population, this paper randomly selects a combination for initialization. After obtaining half of the chromosome of the machine selection part according to the machine selection strategy, the other half of the chromosome of the process sorting part is arranged according to the process selection strategy based on the information of this half chromosome.
[0118] Furthermore, the mutation strategy in step S3 is:
[0119]
[0120] v i,G is the mutation vector, x i,G is the current individual, x best,G is an individual randomly selected from a previously preset proportion of individuals, x r1,G is an individual randomly selected from the current population, is a random selection of individuals from the current population and the external archive. In the absence of an external archive, no external archive is used and individuals are randomly selected from the current population. The external archive is set to empty when initially generated. Each parent replaced by a child is added to the external archive. Its size is fixed to np. If the number of individuals in the external archive exceeds np, randomly selected individuals are deleted to maintain the archive size. Parameter F i ∈[0,1] is a scaling factor used to control the size of the mutation operator. After the mutation and crossover are completed, it is necessary to pay attention to the constraint violation caused by the disturbance. After the end, the internal process processing priority of the workpiece is reordered to ensure the legitimacy of the processing plan.
[0121] Furthermore, the crossover strategy in step S3 is: after the mutation is completed, the individual is crossed with the parent. In order to make the solution more effective, a crossover operator similar to exponential crossover is used in the crossover of the process part. The continuous crossover part is all the processes of a workpiece, that is, the crossover is performed in units of workpieces, and the binomial crossover represented by the formula is still used in the machine part.
[0122] Furthermore, the selection strategy in step S3 is: if the individual u that crosses successfully i,G Dominate the original individual or with the original individual x i,G Do not dominate each other, then F i With CR i Store in the success collection and set x i,G Save to external archive; otherwise, F i With CR iStored in the failed set. The crossover successful individual set is generated in all the trial vectors u i,G Afterwards, the set of individuals that successfully crossover is merged with the original population to obtain the first np individuals in the non-dominated sort.
[0123] Furthermore, the parameter adaptation strategy in step S3 is JADEFI, which not only uses cross-valid information, but also uses failure information to control parameter value pairs. Among them, the JADEFI strategy proposes a new method that uses both success information and failure information, specifically: a failure probability is defined to represent the degree to which the generated parameter value pair will lead to failure. According to the failure probability, the parameter value pair is probabilistically rejected, and a new parameter value pair is regenerated until a good pair is generated.
[0124] Furthermore, the local search strategy in step S4 is:
[0125] (1) Partial search operation of the machine:
[0126] Reassignment of the machine with the largest load: Determine the machine with the largest load, randomly select any process on it, and reallocate it to other machines that can process. If there is only one machine that can process, reselect the process.
[0127] Reassign the kq machines with the largest load: Determine the top k machines with the largest load, randomly select q of them, and randomly select a process assigned to these machines and reassign it.
[0128] Random redistribution: randomly generate an integer I from 1 to M (M is the number of machines), randomly select I processes, and redistribute them. If there is only one machine that can process, no redistribution is performed.
[0129] (2) Partial search operation of the process:
[0130] Insertion operation: Randomly select two processes and insert the previous process after the latter, or insert the latter process before the former.
[0131] Exchange operation: Randomly select two processes. If the two processes belong to the same process, reselect and exchange the positions of the two processes.
[0132] Furthermore, the restart strategy in step S5 is that after differential evolution and local search, a certain number of repeated individuals appear in the population. Here, individuals with the same three target values are regarded as repetitive individuals. In order to avoid local convergence of the algorithm and improve the diversity of the solution set, the repetitive individuals are reconstructed to increase the diversity of the population. Local search is used here to reconstruct the repeated solutions. On the one hand, this reconstruction operation draws on useful information that has undergone iterative search solutions. On the other hand, the reconstructed solution is not necessarily better than the initial solution. These individuals increase the diversity of the population, expand the algorithm search range, and increase the possibility of obtaining the optimal solution in the solution space.
[0133] It should be noted that the maximum completion time, total machine load and critical machine load in FJSP are very close to real life. This article minimizes these three goals as the three goals in the algorithm. The maximum completion time is the longest time to complete all processing, in other words, the total length of the processing time. Minimizing the maximum completion time can speed up the completion of processing, which is conducive to promoting the rapid response to market needs. The total machine load represents the operating time of all machines, and the critical machine load is the maximum operating time of all machines. Minimizing the total machine load can reduce the use of machines; minimizing the critical machine load can balance the load between machines. The processing completion time for each workpiece is the sum of the processing time of all processes processed on the machine. These three goals can be defined as follows:
[0134] Maximum completion time: f 1 =max{C i |i=1,2,…,n};
[0135] Total Machine Load:
[0136] Key machine load: f 3 =max{W k |k=1,2,…,m}.
[0137] Furthermore, this embodiment is tested on well-known benchmark instances, including 10 BRdata instances. These test sets will be used to comprehensively evaluate the IADE algorithm in the experiment. The population size NP is set to 200. The maximum function evaluation FEs is used as the termination criterion, which is set to 500000, and each instance is optimized through 20 independent runs respectively. In order to evaluate the performance of the proposed DEDMP algorithm, the inverse generation distance, hypervolume, and coverage are used as indicators in the experiment.
[0138] The inverse generation distance IGD is the reverse mapping of the generation distance GD. It calculates the average distance from the individuals in the true Pareto frontier PFtrue to the non-dominated solution set PFknown obtained by the algorithm. The solution set with a smaller IGD value is better. In addition to reflecting the convergence of the solution set, it can also well reflect the uniformity and extensiveness of the distribution. The smaller the IGD value, the better the diversity and convergence. The formula is as follows
[0139]
[0140] where F is the Pareto frontier obtained, F * is a set of solutions uniformly distributed along the true Pareto front. dis(x,F) is the Euclidean distance between each solution on the obtained Pareto front and the nearest member of the true Pareto front. Since the true Pareto front is unknown here, F * is the non-dominated solution extracted from all experimental results.
[0141] Hypervolume measures the size of the area dominated by non-dominated solution sets, and has good mathematical properties in theory. It reflects the convergence and diversity of the corresponding Pareto front to a certain extent. The larger the HV value, the better the algorithm performance. Among all univariate measures, it is the only one that can detect that a non-dominated solution set A is not worse than another non-dominated solution set B. For discrete multi-objective optimization problems, hypervolume is maximized if and only if the obtained solution set contains only Pareto optimal solutions.
[0142] HV(A)=∪ a∈A vol(a),
[0143] Here vol(a) is the area or volume covered by a set of non-dominated solutions compared to the reference point. The objective value of the non-dominated solution is constrained to the range [0, 1] by normalization. The reference point for all benchmark instances is set by constructing a vector of the worst objective function values (1.1, 1.1, 1.1)T.
[0144] In this example, IADE is compared with the traditional DE series algorithm and some existing advanced algorithms. These algorithms include: DE, JADE, SaDE, SaJADE, TDE and NSGA-II. It should be noted that the Pareto optimal frontiers of these examples are not fixed and known. Therefore, the three metrics are calculated from the collection of Pareto optimal frontiers collected from all the running results of each algorithm.
[0145] First, the IGD calculation results are shown in Table 1. It reveals that MADE achieved 8 optimal IGD values in all 10 instances, and MK3 and MK4 were not much different from the optimal IGD values. DE and SaJADE achieved the optimal IGD values on MK3 and MK8 instances, but lagged behind MADE on the other 8 instances. MADE performed worse than TDE on MK4, but was at least as good as TDE on the remaining instances. On the MK8 instance, all algorithms except NSGA-II achieved the optimal IGD value. However, the reason why NSGA-II performed worse than other algorithms was that its fast non-dominated sorting algorithm lost some advantages after being used by other algorithms. The same situation also appeared in the calculation results of HV.
[0146] Table 1: Comparison of inverse generation distance (IGD) values between MADE and other algorithms
[0147]
[0148] The HV calculation results are shown in Table 2. It reveals that MADE achieved the best HV values in 9 of the 10 instances, and the HV value of the MK3 instance was only lower than SaJADE and ranked second. Although SaJADE had the best IGD value calculation results for the MK4 instance, its HV value was lower than MADE. The data proves that the MADE algorithm can effectively optimize the flexible workshop scheduling model on the BRdata instance and improve the production efficiency in manufacturing operations. Figure 4 A feasible scheduling plan is shown in the form of a Gantt chart, with the horizontal axis representing time and the vertical axis representing the machine number. Each rectangular block represents the step of the target machine processing the target workpiece in the target process at the target time.
[0149] Table 2: Comparison of hypervolume (HV) values between MADE and other algorithms
[0150]
[0151]
[0152] This embodiment solves the multi-objective flexible shop scheduling problem by proposing a new differential evolution algorithm. In this algorithm, not only the parameter pairs of successful mutation crossover are utilized, but also the parameter pairs of failure are utilized. In addition, the search strategy based on IADE is adaptive, retaining strong exploration capabilities while having strong development capabilities. Moreover, the algorithm adopts a new encoding method, which enables individuals to directly generate feasible solutions in the search space. At the same time, the algorithm also performs a local search based on the critical path for each individual. In order to evaluate the performance of IADE, IADE is compared with some advanced algorithms on a benchmark instance. Experimental results show that IADE is better than other algorithms or at least not much different from other algorithms.
[0153] The embodiment of the present invention further provides a multi-objective flexible workshop scheduling device, which includes:
[0154] Coding module: used to encode the scheduling scheme of multi-objective flexible workshop scheduling;
[0155] Initialization module: used to initialize the population;
[0156] Differential evolution module: used to use the improved differential evolution algorithm to perform mutation, crossover and selection processing on the population, and adjust the parameters of the differential evolution algorithm according to the processing results;
[0157] Local search module: used to perform local search on the selected population to obtain a new population;
[0158] Restart module: used to execute restart strategy on the population after local search and the original population;
[0159] Loop module: used to repeat S3-S5 until the maximum number of evaluations is reached, to obtain the optimized population search range, and to determine the optimal solution based on the condition of minimizing the preset target.
[0160] Other embodiments or specific implementations of the multi-objective flexible workshop scheduling device of the present invention can refer to the above-mentioned method embodiments and will not be described in detail here.
[0161] In addition, the present invention also proposes a multi-objective flexible workshop scheduling device, which includes: a memory, a processor, and a multi-objective flexible workshop scheduling program stored in the memory and runable on the processor. When the multi-objective flexible workshop scheduling program is executed by the processor, the steps of the multi-objective flexible workshop scheduling method described above are implemented.
[0162] The specific implementation of the multi-objective flexible workshop scheduling device of the present application is basically the same as the various embodiments of the multi-objective flexible workshop scheduling method described above, and will not be repeated here.
[0163] In addition, the present invention also proposes a readable storage medium, the readable storage medium includes a computer readable storage medium, on which a multi-objective flexible workshop scheduling program is stored. The readable storage medium may be Figure 1 The memory 1005 in the terminal may also be at least one of a ROM (Read-Only Memory) / RAM (Random Access Memory), a magnetic disk, and an optical disk. The readable storage medium includes a number of instructions for enabling a multi-objective flexible workshop scheduling device with a processor to execute the multi-objective flexible workshop scheduling method described in each embodiment of the present invention.
[0164] The specific implementation methods in the readable storage medium of the present application are basically the same as the various embodiments of the above-mentioned multi-objective flexible workshop scheduling method, and will not be repeated here.
[0165] It is understood that, in the description of this specification, the description with reference to the terms "one embodiment", "another embodiment", "other embodiments", or "first to Nth embodiments" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.
[0166] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including 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, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.
[0167] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.
[0168] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present invention.
[0169] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.
Claims
1. A multi-objective flexible workshop scheduling method, characterized in that: The following steps are involved: S1: Encode the scheduling scheme of multi-objective flexible workshop scheduling; S2: Initialize the population; S3: Use the improved differential evolution algorithm to perform mutation processing, crossover processing and selection processing on the population, and adjust the parameters of the differential evolution algorithm according to the processing results; S4: Perform local search on the selected population to obtain a new population; S5: Execute the restart strategy on the population after local search and the original population; S6: Repeat S3-S5 until the maximum number of evaluations is reached, obtain the optimized population search range, and determine the optimal solution based on the condition of minimizing the preset goal.
2. The multi-objective flexible workshop scheduling method according to claim 1, characterized in that: Step S1: Encode the scheduling scheme of multi-objective flexible workshop scheduling, specifically including: Process coding: Generate a process sequence vector based on the processing order of each workpiece in each process; Machine encoding: Generate a machine allocation vector based on the machine serial number that can execute the processing task of the target workpiece in the target process.
3. The multi-objective flexible workshop scheduling method according to claim 2, characterized in that: Step S2: Initialize the population, including: Randomly select a target selection strategy from a preset machine selection strategy group and a target process sorting strategy from a preset process sorting strategy group, and combine them to obtain an initialization strategy; The population is initialized using the initialization strategy. After obtaining half of the chromosome of the machine selection part according to the machine selection strategy, the other half of the chromosome of the process sorting part is arranged according to the process selection strategy based on the information of this half of the chromosome.
4. The multi-objective flexible workshop scheduling method according to claim 3, characterized in that: Step S3: Using the improved differential evolution algorithm to perform mutation processing, crossover processing and selection processing on the population, specifically including: Mutation processing: Use a specific mutation strategy to perform mutation processing on each individual in the population. The expression of the specific strategy is specifically: v i,G is the mutation vector, x i,G is the current individual, x best,G is an individual randomly selected from a previously preset proportion of individuals, x r1,G is an individual randomly selected from the current population, is an individual randomly selected from the current population and the external archive, and the parameter F i ∈[0,1] is the scaling factor, which is used to control the size of the mutation operator; Crossover treatment: After mutation treatment, individuals are crossed with their parents. The crossover in the process part adopts exponential crossover, and the crossover in the machine part adopts binomial crossover. The parameter CR i ∈[0,1] is the crossover factor, which is used to control the size of the crossover operator; Selection processing: If the crossover succeeds, individual u i,G Dominate the original individual or with the original individual x i,G Do not dominate each other, then F i With CR i Store in the success collection and set x i,G Save to external archive; otherwise, F i With CR i Save to the failed collection.
5. The multi-objective flexible workshop scheduling method according to claim 4, characterized in that: Step S3: adjusting the parameters of the differential evolution algorithm according to the processing result, specifically including: adjusting the parameters of the differential evolution algorithm using the JADEFI strategy based on the success set and the failure set.
6. The multi-objective flexible workshop scheduling method according to claim 5, characterized in that: Step S4: Perform a local search on the selected population to obtain a new population, specifically including: (1) Local search of machine parts: Reassignment of the machine with the largest load: Determine the machine with the largest load, randomly select any process, and reassign it to other machines that can process. If there is only one machine that can process, reselect the process; Reassign the kq machines with the largest load: determine the first k machines with the largest load, randomly select q of them, and randomly select a process assigned to these machines and reassign them; Random redistribution: randomly generate an integer I from 1 to M, randomly select I processes, and redistribute them. If there is only one processable machine, no redistribution will be performed; M is the number of machines; (2) Partial search of process: Insertion operation: Randomly select two processes and insert the previous process after the latter, or insert the latter process before the former. Exchange operation: Randomly select two processes. If the two processes belong to the same process, reselect and exchange the positions of the two processes.
7. The multi-objective flexible workshop scheduling method according to claim 6, characterized in that: Step S5: Execute a restart strategy on the population after the local search and the original population, specifically including: reconstructing duplicate individuals in the population after differential evolution and local search.
8. A multi-objective flexible workshop scheduling device, characterized in that: include: Coding module: used to encode the scheduling scheme of multi-objective flexible workshop scheduling; Initialization module: used to initialize the population; Differential evolution module: used to use the improved differential evolution algorithm to perform mutation, crossover and selection processing on the population, and adjust the parameters of the differential evolution algorithm according to the processing results; Local search module: used to perform local search on the selected population to obtain a new population; Restart module: used to execute restart strategy on the population after local search and the original population; Loop module: used to repeat differential evolution and local search until the maximum number of evaluations is reached, to obtain the optimized population search range, and to determine the optimal solution under the condition of minimizing the preset goal.
9. A multi-objective flexible workshop scheduling device, characterized in that: The multi-objective flexible workshop scheduling device includes: a memory, a processor, and a multi-objective flexible workshop scheduling program stored in the memory and executable on the processor. When the multi-objective flexible workshop scheduling program is executed by the processor, the steps of the multi-objective flexible workshop scheduling method as described in any one of claims 1 to 7 are implemented.
10. A storage medium, characterized in that: The storage medium stores a multi-objective flexible workshop scheduling program, which, when executed by a processor, implements the steps of the multi-objective flexible workshop scheduling method according to any one of claims 1 to 7.
