A method and device for scheduling optimization of a thin plate planar section flow water plant

Abstract

The present application belongs to the technical field of workshop production scheduling and intelligent manufacturing, and discloses a scheduling optimization method and equipment for a thin plate planar segmented flow shop, wherein a scheduling model is constructed by considering three stages of a pre-processing process, an assembly processing stage and a post-assembly processing stage, taking unified constraints obtained by converting parallel equipment constraints of the pre-processing stage, multi-predecessor synchronization constraints of the assembly processing stage and process connection constraints of the post-assembly processing stage as model constraints, without omitting or weakening assembly constraints, and iteratively optimizing the scheduling scheme by using a variable neighborhood search mechanism, so that the maximum completion time can be effectively shortened under the premise of ensuring the feasibility of the scheduling scheme, the scheduling optimization efficiency is improved, and the feasibility and optimization effect of the scheduling result are further improved.

Description

Technical Field

[0001] This invention belongs to the technical field of workshop production scheduling and intelligent manufacturing, and more specifically, relates to a scheduling optimization method and equipment for a thin-plate planar segmented assembly line workshop. Background Technology

[0002] Thin-plate planar sections are key intermediate products in the manufacturing process of large ships, and their production is usually concentrated in the thin-plate planar section workshop. Taking the thin-plate planar section manufacturing workshop in the shipbuilding industry as an example, this type of workshop is mainly responsible for the processing and assembly of thin-plate planar sections of the hull, and is characterized by large production scale, long process flow, and strong assembly constraints.

[0003] In a sheet metal panel segmentation workshop, panel segments typically consist of multiple types of workpieces, primarily sheet metal and profiles. Sheet metal workpieces require multiple pre-processing steps, including cutting, splicing, and grinding, while profiles require pre-processing steps such as blanking, forming, and straightening. After these two types of workpieces complete their pre-processing on different parallel production lines or in different processing stages, they need to undergo longitudinal rigging or welding in the assembly stage to form complete panel segments, which then proceed to subsequent welding, straightening, and handling processing stages.

[0004] From a scheduling model perspective, this type of production system shares some similarities with the hybrid flow shop scheduling problem in form, namely, the operation needs to go through multiple processing stages sequentially, and some stages are equipped with multiple parallel machines. However, unlike the traditional hybrid flow shop, the final product in the thin plate planar segmented shop is not processed by a single operation sequence, but rather by multiple types of workpieces that have completed pre-processing separately, and then combined for processing in the assembly stage. The assembly operation must wait for the completion of its corresponding multiple preceding workpiece operations, resulting in significant assembly synchronization constraints.

[0005] Existing hybrid flow shop scheduling methods typically focus on single operations, making it difficult to directly describe the synchronous dependencies of assembly operations on multiple preceding operations. Furthermore, sequence-based encoding struggles to characterize the actual process of local search based on critical paths. When applied to thin-plate planar segmented workshops, these methods often simplify the model by ignoring or weakening assembly constraints, thereby affecting the feasibility and optimization effectiveness of the scheduling results.

[0006] Therefore, for production systems of thin plate flat section assembly line workshops, such as those of large cruise ships, there is an urgent need for a scheduling method that can uniformly model multiple types of constraint relationships and perform efficient optimization based on key structures, so as to improve the overall scheduling efficiency and production organization level of the thin plate flat section production process. Summary of the Invention

[0007] In view of the above-mentioned defects or improvement needs of the existing technology, the present invention provides a scheduling optimization method and equipment for a thin plate planar segmented assembly line workshop. It aims to solve the problem that the scheduling method of the existing thin plate planar segmented assembly line workshop ignores or weakens the assembly constraints to simplify the model, thereby affecting the feasibility and optimization effect of the scheduling results.

[0008] To achieve the above objectives, according to one aspect of the present invention, a scheduling optimization method for a thin-plate planar segmented flow workshop is provided, comprising the following steps: A genetic algorithm is used to solve the scheduling model of a thin-plate planar segmented flow shop for the scheduling problem to obtain an initial feasible scheduling scheme. Then, a variable neighborhood search strategy is used to search for the obtained initial feasible scheduling scheme to obtain the optimal scheduling scheme. The thin-plate planar segmented flow shop scheduling model considers the pre-processing stage, the assembly stage, and the post-assembly processing stage. The model constraint is a unified constraint obtained by transforming the parallel equipment constraint of the pre-processing stage, the multi-predecessor synchronization constraint of the assembly stage, and the process connection constraint of the post-assembly processing stage.

[0009] Furthermore, the start time of the assembly operation is expressed by coupling the workpiece selection decision variable and the completion time variable, and the start time of the assembly operation satisfies the multi-predecessor synchronization constraint of the assembly processing stage.

[0010] Furthermore, under model constraints, a metaheuristic search algorithm is used to generate an initial feasible scheduling scheme, and an assembly disjunction graph is constructed based on the initial feasible scheduling scheme. In the assembly disjunction graph, critical paths containing synchronous assembly nodes and their synchronous predecessor chain structures are identified, and a variable neighborhood search containing four types of neighborhood structures is performed based on the critical paths and synchronous predecessor chains. The initial feasible scheduling scheme is optimized by sequentially calling each neighborhood structure, and then the optimal scheduling scheme is output.

[0011] Furthermore, the multi-predecessor synchronization constraint in the assembly and processing stage is expressed as:

[0012] in, J Represents a set of jobs. j Indicates the processing operation sequence number. Indicates pre-processing operations P gather, Indicates pre-processing operations B gather, Indicates assembly operation AJ gather, s Indicates the processing stage number, the range of which is: Where p is the number of stages for P-type machining workpieces, b is the number of stages for B-type machining workpieces, a is the assembly stage, and m is the total number of stages; STs,j Indicates assembly operation J During the processing stage s The start time of processing CT s,j Indicates assembly operation J During the processing stage s The completion time, y i,j Indicates if workpiece J Can be selected for assembly If it is 1, then it is 1; otherwise it is 0. H This represents a sufficiently large constant.

[0013] Furthermore, the assembly parsing diagram includes the following three types of arc structures: job concatenation arcs, machine parsing arcs, and assembly parsing arcs; wherein, the job concatenation arcs are used to represent the process sequence constraints between pre-processing operations, assembly operations, and post-assembly processing operations at each processing stage; the machine parsing arcs are used to represent the mutually exclusive processing relationships between different operations on the same processing equipment; and the assembly parsing arcs are used to represent the synchronous dependency relationship between the assembly operation and its predecessor workpiece.

[0014] Furthermore, in the constructed assembly disjunction graph, critical paths are calculated and identified based on the earliest and latest start times of nodes, and critical path substructures containing assembly disjunction arcs are selected. At the same time, the synchronous precursor chain structure formed by back propagation along the assembly disjunction arc is extracted.

[0015] Furthermore, the four types of neighborhood structures are as follows: Neighborhood Structure 1: For operation nodes on the critical path in the assembly disjunction graph, exchange their relative processing order in the same processing stage or on parallel processing equipment; Neighborhood Structure 2: For consecutive operation blocks located on the critical path in the assembly disjunction graph, adjust the overall position of the operation blocks in the scheduling sequence; Neighborhood Structure 3: For the critical path and its predecessor non-critical operations in the assembly disjunction graph, adjust the process execution order of the predecessor non-critical operations in the pre-assembly processing stage; Neighborhood Structure 4: For operations in the critical path or bottleneck stage, redistribute these operations among the parallel processing equipment in their respective processing stages.

[0016] Furthermore, a two-layer scheduling coding structure is adopted to represent the initial feasible scheduling scheme, wherein: the first layer is a job sequence vector, which describes the order of jobs in the overall scheduling sequence to represent the relative processing priority of each job in the multi-stage production process; the second layer is a stage-job equipment mapping, which represents the equipment selection relationship of each job in each processing stage; the stage-job equipment mapping uses the stage-job combination as the index key to realize independent equipment allocation decision in the parallel equipment environment.

[0017] The present invention also provides a scheduling optimization system for a thin-plate planar segmented production line workshop. The system includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to perform the scheduling optimization method for the thin-plate planar segmented production line workshop as described above.

[0018] The present invention also provides a computer-readable storage medium storing machine-executable instructions, which, when invoked and executed by a processor, cause the processor to implement the scheduling optimization method for a thin-plate planar segmented production line as described above.

[0019] In summary, compared with the prior art, the scheduling optimization method and equipment for thin-plate planar segmented flow workshop provided by the present invention have the following beneficial effects: 1. The scheduling model of this invention considers three stages: pre-processing, assembly, and post-assembly. It uses a unified constraint derived from the parallel equipment constraints of the pre-processing stage, the multi-predecessor synchronization constraints of the assembly stage, and the process connection constraints of the post-assembly stage as the model constraint. It does not omit or weaken the assembly constraints, and uses a variable neighborhood search mechanism to iteratively optimize the scheduling scheme. Under the premise of ensuring the feasibility of the scheduling scheme, it can effectively shorten the maximum completion time, improve the scheduling optimization efficiency, and thus improve the feasibility and optimization effect of the scheduling results.

[0020] 2. To address the issues of strong coupling between workpiece operations and assembly operations, complex assembly synchronization constraints, and diverse configurations of parallel processing equipment in thin-plate planar segmented assembly workshops, a unified multi-stage assembly scheduling model for thin-plate planar segmented workshops is constructed, which can accurately depict the operation and processing relationships in different processing stages and parallel equipment environments.

[0021] 3. Map the scheduling scheme to an assembly disjunction graph that conforms to multiple precursor assembly, and uniformly represent the process sequence constraints, equipment mutual exclusion constraints and assembly synchronization constraints. This provides structured support for the feasibility determination and optimization of the scheduling scheme, and avoids generating infeasible scheduling schemes during the search process.

[0022] 4. Based on the critical path information identified in the disjunctive graph, and combined with the production structure characteristics of the thin-plate planar segmented flow workshop, multiple neighborhood structures are designed, and a variable neighborhood search is performed on the disjunctive graph. This enables the scheduling optimization process to focus on key operations and bottleneck stages, improving the search efficiency and solution stability of scheduling optimization while ensuring the feasibility of the scheduling scheme. It has good engineering applicability. Attached Figure Description

[0023] Figure 1This is a flowchart of a scheduling optimization method for a thin-plate planar segmented assembly line provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the encoding and decoding scheme according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the Gantt chart corresponding to the disjunctive diagram in an embodiment of the present invention; Figure 4 This is a flowchart of the variable neighborhood operation involved in the embodiments of the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0025] This invention provides a scheduling optimization method for a thin-plate planar segmented assembly line workshop. The method unifies the modeling of work processing constraints, equipment constraints, and assembly synchronization constraints, introduces a homogeneous workpiece completion pool, constructs a synchronization precursor chain in the assembly disjunction graph, obtains the critical path and its precursor path based on this chain, and uses a variable neighborhood search mechanism to iteratively optimize the scheduling scheme, thereby effectively reducing the maximum completion time and improving the feasibility and optimization effect of the scheduling results.

[0026] The method considers scheduling modeling of the production process of the ship's thin plate planar segmented assembly line workshop, and uses assembly disjunction graph based on critical path and variable neighborhood search to optimize and solve the obtained scheduling model, which can improve scheduling optimization efficiency while ensuring the feasibility of the scheduling scheme.

[0027] Specifically, the method first requires constructing a thin-plate flat-panel segmented production line scheduling model for the manufacturing process of large cruise ships. Secondly, it proposes a method based on critical path disjunctive graph and variable neighborhood search to optimize and solve the obtained scheduling model. The constructed scheduling model needs to consider three stages: pre-processing, assembly processing, and post-assembly processing. The optimization objective is to minimize the maximum completion time, which is the maximum value of the completion time of the last process of all products.

[0028] Please see Figure 1 The method includes the following steps: Step 1: Construct a thin-plate planar segmented flow workshop scheduling model that considers the pre-processing stage, assembly stage, and post-assembly processing stage for the thin-plate planar segmented flow workshop scheduling problem. The thin-plate planar segmented flow workshop scheduling model uses a unified constraint obtained by transforming the parallel equipment constraint of the pre-processing stage, the multi-predecessor synchronization constraint of the assembly stage, and the process connection constraint of the post-assembly processing stage as the model constraint.

[0029] The production process is divided into three stages: pre-processing, assembly, and post-assembly. The pre-processing stage comprises multiple independent production lines for processing P-type and B-type workpieces (profiles and sheets). Assembly operations can only begin after all pre-processed workpiece operations are completed, creating multiple precursor synchronization constraints. The model's optimization objective is to minimize the maximum completion time. The scheduling model's inputs are workpiece information and equipment information (including the number of workpieces and the processing equipment's operating time); the output is the scheduling scheme.

[0030] In the thin plate planar segmented flow shop scheduling problem, the following assumptions need to be met: In the assembly processing stage, there are multiple precursor workpiece operations, and their start time is jointly determined by the completion time of the precursor workpiece operations that meet the assembly requirements; In the pre-processing stage, there is a set of homogeneous workpiece operations, and for the same assembly operation, its precursor workpiece is not limited to a fixed individual, but rather workpieces that meet the quantity constraints are selected from the corresponding set of homogeneous workpieces to participate in the assembly.

[0031] The start time of the assembly operation is expressed by coupling the workpiece selection decision variable and the completion time variable, and the start time of the assembly operation satisfies the assembly synchronization constraint. When a workpiece is selected for assembly, its completion time participates in the assembly synchronization constraint, which is expressed as follows:

[0032] in, J Represents a set of jobs. j Indicates the processing operation sequence number. Indicates pre-processing operations P gather, Indicates pre-processing operations B gather, Indicates assembly operation AJ gather, s Indicates the processing stage number, the range of which is: Where p is the number of stages for P-type machining workpieces, b is the number of stages for B-type machining workpieces, a is the assembly stage, and m is the total number of stages; ST s,j Indicates assembly operation J During the processing stage s The start time of processing CTs,j Indicates assembly operation J During the processing stage s The completion time, y i,j Indicates if workpiece J Can be selected for assembly If it is 1, then it is 1; otherwise it is 0. H This represents a sufficiently large constant.

[0033] Step 2: After solving the thin-plate planar segmented flow workshop scheduling model using a genetic algorithm to obtain an initial feasible scheduling scheme, a variable neighborhood search strategy is used to search for the obtained feasible scheduling scheme to obtain the optimal scheduling scheme.

[0034] Please see Figure 2 , Figure 3 and Figure 4 Under the constraints of the aforementioned scheduling model, a metaheuristic search algorithm is used to generate an initial feasible scheduling scheme. Based on this scheme, an assembly disjunction graph is constructed. The assembly disjunction graph identifies critical paths containing synchronous assembly nodes and their synchronous predecessor chains. Based on these critical paths and synchronous predecessor chains, a variable neighborhood search strategy incorporating four types of neighborhood structures is executed to optimize the scheduling scheme by sequentially calling each neighborhood structure until a termination condition is met, outputting the optimal scheduling scheme. Specifically, the assembly disjunction graph introduces synchronization constraint arcs on top of process constraint arcs and equipment mutual exclusion arcs to represent the multi-predecessor synchronization relationship of assembly operations.

[0035] In one implementation, step two includes the following sub-steps: Step 1: Construct a two-layer scheduling coding structure that includes job sequence vectors and stage-job equipment mapping. Use the two-layer scheduling coding structure to encode and represent the initial feasible scheduling scheme to form a feasible initial solution and form an initial population.

[0036] In one implementation, a feasible solution to the scheduling model is represented using sequence encoding to describe the processing order of a job at each processing stage. Specifically: Encoding: n The final product is broken down into 3 according to the assembly relationship. n There are n scheduled jobs. P Pre-processing operations: ; n indivual B Pre-processing operations: and n assembly operations The chromosome length is 3 n Based on this, the individual adopts a two-layer scheduling coding structure of job sequence-device mapping. The corresponding job sequence code (OS) is a 3-bit / 4 ...5-bit / 4-bit / 4-bit / 5-bit / 4-bit / 4-bit / 5-bit / 4-bit / n An integer permutation is used to represent the global scheduling priority of all jobs; the device mapping code (MS) is used to represent the device allocation result of each job at its processable stage. For example, when n When =2, a feasible OS encoding is: OS=[2, 3, 1, 6, 4, 5]. This OS encoding only indicates the order in which jobs are scheduled. Furthermore, if the device mapping encoding is given as: ,For example This indicates that task 2 was assigned to the equipment in stage 1. Task 3 is assigned to equipment in Phase 1. Assignment 2 is assigned to stage 6. Therefore, the OS and MS together constitute the coded representation of the complete scheduling solution.

[0037] Decoding: First, jobs are selected sequentially according to the OS (Operation Sequence) encoding. The earliest feasible start time of each job in the current stage is calculated based on the serial scheduling generation rules. The priority device for each job in the corresponding stage is determined according to the MS (Schedule Mapping) encoding. If a device is unavailable or cannot obtain the earliest completion time, a device with the earliest completion time is selected from the available device set for correction. Second, during the pre-processing stage, workpieces are classified into homogeneous categories to construct a homogeneous workpiece job set and establish a corresponding homogeneous workpiece completion pool. The homogeneous workpiece job set refers to a set of workpieces that are consistent in structural attributes, processing technology, and processing resource requirements, and can be substituted for each other in the assembly stage. When an assembly job is scheduled, a workpiece that meets the constraints is selected from the corresponding homogeneous workpiece completion pool, and the synchronous start time of the assembly job is determined based on the completion time of the selected workpiece. Finally, the available time of the equipment and the completion time of the job are updated, and scheduling continues according to the job sequence encoding until all jobs are completed, thus obtaining a complete scheduling scheme.

[0038] Step 2, set the parameters of the hybrid optimization algorithm. Specifically, set the maximum number of iterations for the metaheuristic hybrid algorithm to [value missing]. The population size is Popsize The number of evolutionary learning operators is L ; P c Let be the probability of a crossover operation occurring. P m The probability of mutation operation occurring; the number of candidate neighborhood structures is V The proportion of elite individuals to the total number of individuals in the population is EliteRate .

[0039] Step 3: Update the population based on the evolutionary search mechanism.

[0040] In this step, evolutionary learning operations need to be performed on each individual in the population towards the optimal individual. The following are the three operation operators proposed in this embodiment: Operation Operator 1: Individual initialization uses a random generation method. Each individual in the population consists of two parts: a job sequence code and a device mapping code. The job sequence code represents the scheduling order of jobs, and the device mapping code represents the priority device allocation of a job in the corresponding stage. Subsequently, the target value of each individual is calculated according to the decoding process in Step 1.

[0041] Operation operator 2: Select the best individual in the population S best and the individual to be operated S For the optimal individual S best and the individual to be operated S The sequence encoding is subjected to POX cross-interleaving. Taking the process encoding layer as an example, select... S One of the workpiece numbers is selected, ensuring its position remains unchanged; the remaining positions are filled in sequentially. S best The process codes for other products in the solution. Assume the current solution... S The sequence encoding is [1, 7, 3, 8, 4, 2, 9, 6, 5], the optimal solution is... S best The process code is [7, 2, 9, 1, 6, 8, 3, 5, 4]. If, for example, all AJ jobs are selected as the retain set: Ω={7, 8, 9}, then the process code layer after POX crossover is [2, 7, 1, 8, 6, 3, 9, 5, 4].

[0042] Operation operator 3: Perform a mutation operation on the code. Specifically, randomly select the individual to be operated on. S Two positions in the sequence are swapped, and the corresponding job numbers are exchanged. Assuming the current batch code of the solution is [6, 4, 5, 8, 3, 2, 1, 9, 7], if two positions 6 and 3 are randomly selected and swapped, the code layer after the operation will be [3, 4, 5, 8, 6, 2, 1, 9, 7].

[0043] Simultaneously, the aforementioned operators also need to perform device crossover and device mutation operations on the device mapping encoding. Device crossover is achieved by randomly inheriting the device allocation results of the parent individual in the corresponding job stage; device mutation involves randomly selecting processing equipment from the set of available devices, thereby achieving evolutionary updates at the device level. Finally, the relative merits of the new individuals and the original individuals will be evaluated, and a better scheduling scheme will be added to the population.

[0044] Step 4: Construct an assembly disjunction graph containing process constraint arcs, equipment mutual exclusion arcs, and assembly synchronization constraint arcs for the candidate scheduling schemes in the updated population.

[0045] Specifically, an assembly disjunction graph is constructed based on the feasible scheduling scheme, such as... Figure 3 As shown, the assembly disjunction diagram includes the following three types of arc structures: job conjunction arcs, machine disjunction arcs, and assembly mate disjunction arcs. The job conjunction arcs represent the process sequence constraints between pre-processing operations, assembly operations, and post-assembly processing operations at each processing stage. To clearly express the operation sequence of pre-processing precursor P-type and B-type workpieces, different arc forms can be used. The machine disjunction arcs represent the mutually exclusive processing relationships between different operations on the same processing equipment. The assembly mate disjunction arcs represent the synchronous dependency relationships between assembly operations and their precursor workpieces. The assembly mate disjunction arcs, unlike traditional disjunction diagram structures, are used to depict the synchronous completion constraints of multiple workpiece nodes on the same assembly node.

[0046] Step 5: Identify the critical path containing the assembly synchronization node and its corresponding synchronization precursor chain structure in the assembly parsing diagram.

[0047] Specifically, in the constructed assembly disjunction graph, critical paths are calculated and identified based on the earliest and latest start times of nodes, and critical path substructures containing assembly disjunction arcs are selected. At the same time, the synchronous precursor chain structure formed by back propagation along the assembly disjunction arc is extracted.

[0048] Step 6: Based on the critical path and the synchronous predecessor chain, perform cross-stage linkage neighborhood operations to make local improvements to the candidate scheduling scheme.

[0049] Based on the structural characteristics of the shipbuilding thin plate planar section production workshop, four types of cross-stage linkage neighborhood structures are constructed, including: Neighborhood Structure 1: For operation nodes on the critical path in the disjunction graph, their relative processing order in the same processing stage or parallel processing equipment is exchanged; Neighborhood Structure 2: For continuous operation blocks located on the critical path in the disjunction graph, the overall position of the operation blocks in the scheduling sequence is adjusted; Neighborhood Structure 3: For the critical path and its predecessor non-critical operations in the disjunction graph, the execution order of the preceding non-critical operations in the pre-assembly processing stage is adjusted; Neighborhood Structure 4: For operations in the critical path or bottleneck stage, these operations are redistributed among the parallel processing equipment in their respective processing stages. The above neighborhood structure operations are executed sequentially according to a preset order. When the candidate scheduling scheme improves the maximum completion time index, the current scheduling scheme is updated. When no neighborhood structure produces improvement, the variable neighborhood search is terminated. When the preset termination condition is met, the globally optimal scheduling result is output as the final scheduling scheme of the thin plate planar section assembly line workshop.

[0050] Step 7: Iterate and update repeatedly until the termination condition is met, then output the optimal scheduling scheme.

[0051] If the hybrid optimization algorithm satisfies the termination condition, it outputs the optimal individual of the population; otherwise, it jumps to step 4 and iterates repeatedly until the termination condition is met.

[0052] The encoding adopts a two-layer scheduling encoding structure, wherein: the first layer is a job sequence vector, which describes the order of jobs in the overall scheduling sequence to represent the relative processing priority of each job in the multi-stage production process; the second layer is a stage-job equipment mapping, which represents the equipment selection relationship of each job in each processing stage. The stage-job equipment mapping uses the "stage-job" combination as the index key to realize independent equipment allocation decision in the parallel equipment environment.

[0053] The decoding process includes the following steps: First, based on the job sequence vector, the serial scheduling generation rules are executed sequentially at each processing stage to determine the candidate processing start time of the job; second, in the pre-processing stage of the workpiece, a set of homogeneous workpiece jobs is constructed and a corresponding homogeneous workpiece completion pool is established. The set of homogeneous workpiece jobs refers to a set of workpieces that are consistent in structural attributes, processing technology flow, and processing resource requirements, and can be substituted for each other in the assembly processing stage; when the assembly job enters the assembly processing stage, a workpiece that meets the constraints is selected from the corresponding homogeneous workpiece completion pool, and the synchronous start time of the assembly job is determined based on the completion time of the selected workpiece; finally, in the equipment selection process, the pre-coded equipment is used first. When the earliest completion condition is not met, the equipment that can obtain the earliest completion time is selected from the set of available equipment for correction.

[0054] The present invention will be further described in detail below with reference to specific embodiments.

[0055] Example 1 Please see Figure 1 , Figure 2 , Figure 3 and Figure 4 In a thin plate flat segmented assembly line workshop, there are 6 processing machines and 3 products to be processed. The length of the sequence code is 2+2+2=6.

[0056] This embodiment 1 includes the following steps: Step 1: Construct a multi-stage assembly scheduling model for the thin plate planar segmented workshop.

[0057] Based on the processing workshop and other conditions in the embodiments, a multi-stage assembly scheduling model for the thin plate planar section workshop in the manufacturing process of thin plate planar sections of ships is constructed.

[0058] Step 2: Optimize the solution.

[0059] Step two mainly includes the following sub-steps: Step 1: Encoding and Decoding.

[0060] Figure 2 The left half of the diagram is a schematic diagram of the sequence coding scheme, where 1-2 represents... B Pre-machined workpieces, 3-4 indicates P Type 5 represents pre-processed workpieces, and type 6 represents assembly workpieces. Meanwhile, the equipment mapping code is used to represent the priority equipment allocation relationship of the operation in the corresponding stage, where the triple (s,j,m) indicates that operation j is preferentially allocated to equipment m in stage s. Figure 2 The right half is a Gantt chart after decoding an example of the encoding scheme. An example of the decoding process is: Let the sequence encoding be 1-3. n The full permutation is: S=[P2,B1,P1,AJ2,B2,AJ1], Assembly operations AJ1 consist of workpieces P1 and B1, and AJ2 consists of workpieces P2 and B2. Activity decoding is based on a combination of job sequence coding and device mapping coding: when a job is processed, its scheduling order is first determined according to the job sequence coding, then the priority device at the corresponding stage is determined according to the device mapping coding, and the earliest possible start time is calculated. For processing ordinary workpieces (such as P2, B1, P1, B2), if a feasible slot exists for the priority device, it is directly inserted; otherwise, the earliest completing device is selected from the available devices for scheduling, thus obtaining the completion time of each workpiece. CT (P1) , CT (B1) , CT (P2) , CT (B2) After pre-processing, the workpieces are assigned to the corresponding homogeneous workpiece completion pool based on their structural attributes and technological characteristics. When decoding to the assembly operation, workpieces that meet the constraints are selected from the corresponding completion pool, and their latest completion time is used as the assembly synchronization readiness time. For example, the readiness time of AJ1 is... R (AJ1) =max{ CT (P1) , CT (B1) This indicates that assembly is only allowed to begin after all required parts are completed, and then during the assembly phase, the equipment will start from... R (AJ1) We will continue to use the rule of "earliest available slot and insert" to arrange AJ1.

[0061] Similarly, AJ2 with R (AJ2) = max{ CT (P2) , CT (B2)Starting with the first phase, we insert scheduling; then, we insert scheduling on the equipment in the assembly phase according to the same rules, and finally obtain the activity scheduling solution that satisfies the phase sequence, equipment constraints and assembly synchronization constraints.

[0062] Step 2: Setting parameters for the hybrid algorithm.

[0063] The maximum number of iterations for the hybrid algorithm is set to 200, the population size is 60, and the number of evolutionary learning operators is 3; the probability of crossover operation is... P c =0.7, the probability of mutation operation occurring. P m =0.15; the number of candidate neighborhood structures is 4, and the proportion of elite individuals to the total number of individuals in the population is EliteRate =0.1.

[0064] Step 3: Population update operation operators.

[0065] Evolutionary learning operators: Evolutionary learning operations need to be performed on each individual in the population in the direction of the optimal individual. The following are three operation operators proposed in this embodiment.

[0066] Operation operator 1: Select the best individual in the population S best and the individual to be operated S For the optimal individual S best and the individual to be operated S The sequence encoding is subjected to POX cross-interleaving. Taking the process encoding layer as an example, select... S One of the workpiece numbers is selected, ensuring its position remains unchanged; the remaining positions are filled in sequentially. S best Process codes for other products in the series.

[0067] Operation operator 2: Perform a mutation operation on the code. Randomly select two positions in the individual S to be operated on and swap the positions of the corresponding job numbers. At the same time, perform a device mutation operation on the device mapping code and reselect the processing device from the set of available devices.

[0068] Variable Neighborhood Search Operator: After evolutionary learning, a local search is needed for some elite individuals to further explore better scheduling schemes. Before neighborhood exploration, the elite individuals need to be transformed into a disjunctive graph. Based on the disjunctive graph, the critical path and the predecessors corresponding to the workpieces assembled on the critical path (e.g., ...) are identified. Figure 3 (As shown).

[0069] Perform a variable neighborhood search operation on the critical path and its predecessor. The specific process is as follows: Figure 4As shown, the following are the four neighborhood structures proposed in this embodiment: The primary function of neighborhood structure 1 is to change the relative processing order of process nodes on the critical path within the same processing stage or on the same equipment. (This is illustrated in the Gantt chart.) P Line equipment PM 2,1 For example, its processing sequence is as follows: O 1,2 →O 2,2 Assume the critical path passes through node . u = O 1,2 Then, adjacent nodes on the same device can be selected. v = O 2,2 and u Perform an exchange. Then execute an arc-breaking exchange operation: disconnect ( O 1,2  → O 2,2 ), and reconnect as ( O 2,2  → O 1,2 ), thus PM 2,1 The above sequence becomes O 2,2 →O 1,2 If the reconnection does not violate any processing sequence constraints or assembly synchronization constraints, and no cycles are generated in the disjunctive graph, then the exchange yields a feasible neighborhood solution; otherwise, the exchange is rejected.

[0070] The primary function of neighborhood structure 2 is to adjust the overall position of consecutive work blocks on the critical path. (This is related to) assembly equipment. AM 4,1 For example, its device sequence is O 3,4 →O 2,4 →O 1,4 If the critical path forms a contiguous block on this device. B = [ u 1, u 2] = [ O 2,4 , O 1,4 If there are other work blocks on the critical path, the current contiguous block can be... B After the entire assembly is moved to another work block, the assembly sequence changes. If, after the move, the required workpiece is still present at the start of the assembly process... P line or BIf the line is completed and the disjunctive graph does not produce cycles, then a feasible neighborhood solution is obtained; otherwise, the move is canceled.

[0071] The main function of neighborhood structure 3 is to adjust the execution order of non-critical operations in the critical path and its predecessors during the pre-assembly stage, thereby freeing up assembly readiness time in advance. This is based on critical nodes in the assembly stage. u = O 2,4 For example, according to the BOM table, u The commencement of work requires waiting for the corresponding workpiece to be ready. P Line and B All lines are complete; if backtracking reveals... B Line equipment BM 3,1 On the workpiece process O 2,3 It is a "slow front-end", and the device sequence is O 3,3 →O 1,3 →O 2,3 Then select non-critical nodes that are not on the critical path. v = O 3,3 or O 1,3 To perform an insertion or swap operation, put v Adjust to a more rearward position to enable slow front-wheel drive. O 2,3 Completed earlier, thus enabling assembly nodes u This allows for earlier fulfillment of assembly conditions and shorter assembly wait times. This operation only changes... BM 3,1 The device sequence adjacency relationship is determined without changing the process precursor and assembly synchronization constraints. If the adjusted disjunctive graph does not generate loops and the synchronization constraints are still satisfied, a feasible neighborhood solution is obtained; otherwise, the adjustment is rejected.

[0072] The primary function of neighborhood structure 4 is to redistribute critical path or bottleneck stage processes among parallel devices to balance the load. P Linear parallel devices PM 2,1 and PM 2,2 For example, if the critical path process u = O 2,2 exist PM 2,1 If queuing causes the completion time to be delayed, then... u from PM 2,1 Remove from the device sequence and insert into PM 2,2A feasible location. Specifically, find the position ranked on the original device. u Previous processes O 1,2 And the processes that follow. u MS Find the insertion boundary on the target device. O 3,2 Then disconnect ( O 1,2  → O 2,2 ), and reconnect as ( O 3,2  → O 2,2 The process involves reassigning data across different equipment. If the reassignment does not violate the process precursor / assembly synchronization constraints and the disjunctive graph does not generate cycles, then the operation yields a feasible neighborhood solution; otherwise, the reassignment is revoked.

[0073] The above four neighborhood structures are executed sequentially, progressing layer by layer. When any preceding neighborhood structure can obtain a better individual than the original scheme, the variable neighborhood search process stops and the population is updated.

[0074] The present invention also provides a scheduling optimization system for a thin-plate planar segmented production line workshop. The system includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to perform the scheduling optimization method for the thin-plate planar segmented production line workshop as described above.

[0075] The present invention also provides a computer-readable storage medium storing machine-executable instructions, which, when invoked and executed by a processor, cause the processor to implement the scheduling optimization method for a thin-plate planar segmented production line as described above.

[0076] Those skilled in the art will readily understand that the above description is merely 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 within the scope of protection of the present invention.

Claims

1. A scheduling optimization method for a thin-plate planar segmented flow workshop, characterized in that, The steps are as follows: A genetic algorithm is used to solve the scheduling model of a thin-plate planar segmented flow shop for the scheduling problem to obtain an initial feasible scheduling scheme. Then, a variable neighborhood search strategy is used to search for the obtained initial feasible scheduling scheme to obtain the optimal scheduling scheme. The thin-plate planar segmented flow shop scheduling model considers the pre-processing stage, the assembly stage, and the post-assembly processing stage. The model constraint is a unified constraint obtained by transforming the parallel equipment constraint of the pre-processing stage, the multi-predecessor synchronization constraint of the assembly stage, and the process connection constraint of the post-assembly processing stage.

2. The scheduling optimization method for a thin-plate planar segmented flow workshop as described in claim 1, characterized in that: The start time of the assembly operation is expressed by coupling the workpiece selection decision variable and the completion time variable. The start time of the assembly operation satisfies the multi-predecessor synchronization constraint of the assembly processing stage.

3. The scheduling optimization method for a thin-plate planar segmented flow workshop as described in claim 1, characterized in that: The multi-predecessor synchronization constraint during the assembly and processing stage is expressed as: in, J Represents a set of jobs. j Indicates the processing operation sequence number. Indicates pre-processing operations P gather, Indicates pre-processing operations B gather, Indicates assembly operation AJ gather, s Indicates the processing stage number, the range of which is: Where p is the number of stages for P-type machining workpieces, b is the number of stages for B-type machining workpieces, a is the assembly stage, and m is the total number of stages; ST s,j Indicates assembly operation J During the processing stage s The start time of processing, CT s,j Indicates assembly operation J During the processing stage s The completion time, y i,j Indicates if workpiece J Can be selected for assembly If it is true, then it is 1; otherwise, it is 0. H This represents a sufficiently large constant.

4. The scheduling optimization method for a thin-plate planar segmented assembly line workshop as described in claim 1, characterized in that: Under model constraints, a metaheuristic search algorithm is used to generate an initial feasible scheduling scheme. Based on this initial feasible scheduling scheme, an assembly disjunction graph is constructed. In the assembly disjunction graph, critical paths containing synchronous assembly nodes and their synchronous predecessor chain structures are identified. Based on the critical paths and synchronous predecessor chains, a variable neighborhood search containing four types of neighborhood structures is performed to optimize the initial feasible scheduling scheme by sequentially calling each neighborhood structure, thereby outputting the optimal scheduling scheme.

5. The scheduling optimization method for a thin-plate planar segmented flow workshop as described in claim 4, characterized in that: The assembly parsing diagram includes the following three types of arc structures: job concatenation arc, machine parsing arc, and assembly merging arc; wherein, the job concatenation arc is used to represent the process sequence constraints between pre-processing operations, assembly operations, and post-assembly processing operations at each processing stage; the machine parsing arc is used to represent the mutually exclusive processing relationship between different operations on the same processing equipment; and the assembly merging arc is used to represent the synchronous dependency relationship between the assembly operation and its predecessor workpiece.

6. The scheduling optimization method for a thin-plate planar segmented flow workshop as described in claim 5, characterized in that: In the constructed assembly disjunction graph, the critical path is calculated and identified based on the earliest start time and latest start time of the node, and the critical path substructure containing the assembly disjunction arc is filtered out. At the same time, the synchronous precursor chain structure formed by back propagation along the assembly disjunction arc is extracted.

7. The scheduling optimization method for a thin-plate planar segmented flow workshop as described in claim 4, characterized in that: The four types of neighborhood structures are as follows: Neighborhood Structure 1: For operation nodes on the critical path in the assembly disjunction graph, exchange their relative processing order in the same processing stage or on parallel processing equipment; Neighborhood Structure 2: For consecutive operation blocks located on the critical path in the assembly disjunction graph, adjust the overall position of the operation blocks in the scheduling sequence; Neighborhood Structure 3: For the critical path and its predecessor non-critical operations in the assembly disjunction graph, adjust the execution order of the preceding non-critical operations in the pre-assembly processing stage; Neighborhood Structure 4: For operations in the critical path or bottleneck stage, redistribute these operations among the parallel processing equipment in their respective processing stages.

8. The scheduling optimization method for a thin-plate planar segmented flow workshop as described in any one of claims 1-7, characterized in that: A two-layer scheduling coding structure is adopted to represent the initial feasible scheduling scheme, wherein: the first layer is a job sequence vector, which describes the order of jobs in the overall scheduling sequence to represent the relative processing priority of each job in the multi-stage production process; the second layer is a stage-job equipment mapping, which represents the equipment selection relationship of each job in each processing stage; the stage-job equipment mapping uses the stage-job combination as the index key to realize independent equipment allocation decision in the parallel equipment environment.

9. A scheduling optimization system for a thin-plate planar segmented assembly line workshop, characterized in that: The system includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it performs the scheduling optimization method for the thin-plate planar segmented production line as described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores machine-executable instructions, which, when invoked and executed by a processor, cause the processor to implement the scheduling optimization method for the thin-plate planar segmented flow workshop as described in any one of claims 1-8.

Images