A Distributed Rescheduling Method for Aircraft Pulsating Assembly Lines Based on CNP-NSGA-II

Abstract

This invention relates to a distributed rescheduling method for aircraft pulsed assembly lines based on CNP-NSGA-II, comprising the following steps: acquiring real-time disturbance state information; using a trained scene recognition model based on gated recurrent units to obtain the rescheduling scene type; responding to the rescheduling scene type being a first type, performing right-shift rescheduling on the aircraft pulsed assembly line; responding to the rescheduling scene type being a second type, using a non-dominated sorting genetic algorithm II to optimize the assembly process ordering and / or assembly worker allocation at the aircraft pulsed assembly line stations, achieving collaborative scheduling within the stations of the aircraft pulsed assembly line; responding to the rescheduling scene type being a third type, generating bidding information including human resources requiring assistance from other stations through a contract network algorithm, optimizing the allocation of assembly workers at the stations, and achieving collaborative scheduling between stations of the aircraft pulsed assembly line.

Description

Technical Field

[0001] This invention relates to the field of assembly manufacturing, and in particular to a distributed rescheduling method for aircraft pulsed assembly lines based on CNP-NSGA-II (Contract Network-Non-Dominated Sorting Genetic Algorithm II). Background Technology

[0002] In the aircraft manufacturing process, the assembly stage is the final critical link, which includes several important operations such as the docking of large components, cable laying, installation of key components, and system testing. The execution of these operations directly determines the final production efficiency and supply capacity of the aircraft.

[0003] Pulsed assembly means that each assembly station, following the sequence of assembly instructions, begins its assembly task at the current station, completes the designated assembly process within a fixed cycle, and then moves to the next station at the same time to begin the next assembly cycle, ultimately completing all assembly processes. Compared to traditional fixed assembly lines, pulsed assembly lines have a detailed and clear division of labor, a high degree of specialization, and a stable production cycle. With changes in the domestic and international situation, the demand for various types of aircraft, including commercial, military, and general aviation aircraft, is increasing daily. At the same time, the number of aircraft components is increasing, and the processes are becoming more complex. These changes have brought significant challenges to the aircraft manufacturing industry. Therefore, optimizing the management and processes of the assembly stage, and improving operational efficiency and quality, is key to enhancing the level of aircraft manufacturing.

[0004] In the actual assembly process of aircraft pulsed assembly lines, unpredictable emergencies often arise. In uncertain environments, the manufacturing system generates a new scheduling strategy based on the original strategy, adapting to the current state, according to the real-time changes in the system state. This scheduling method is called rescheduling. Currently, actual assembly lines lack effective identification mechanisms and response measures for emergencies. When disturbances occur, they are usually addressed through overtime work and other excessive workloads, resulting in low assembly efficiency and high worker fatigue. Therefore, efficiently identifying the source of emergencies and adjusting the scheduling strategy promptly based on their characteristics and impact has become a pressing issue.

[0005] In Chinese patent application "An Online Simulation Method for Aircraft Assembly Line Based on Working Condition Data" (Publication No. CN118898115A), Zhang Jie et al. proposed an online simulation method for aircraft assembly lines based on working condition data. This method collects real-time on-site production information from the aircraft assembly line, constructs an online simulation model, and outputs simulation results to predict the planned operation status of ongoing and inactive operations on the aircraft assembly line. It also performs anomaly judgment and source tracing for delayed operations in the planned operation status, providing strong support for decision-making by on-site management personnel. In Chinese patent application "An Intelligent Monitoring System for Aircraft Assembly Process Based on Simulation" (Publication No. CN106647336A), Guo Yu et al. proposed an intelligent monitoring system for aircraft assembly process based on simulation. This system provides visualized monitoring of the real-time assembly process. Based on real-time data updates, combined with intelligent optimization algorithms and assembly knowledge, it achieves functions such as manufacturing behavior prediction, abnormal manufacturing behavior identification, fault analysis, and material status tracking, effectively improving assembly management efficiency and assembly quality. Its advantage is that it takes into account some abnormal situations and makes predictions and tracks them, but it does not take into account situations such as worker absenteeism, nor does it propose corresponding rescheduling plans, and still relies on the decision-making of on-site management personnel.

[0006] In summary, there is currently a lack of a rescheduling method for aircraft pulsed assembly lines to solve or partially solve the aforementioned problems. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of the existing technology by providing a distributed rescheduling method for aircraft pulse assembly lines based on CNP-NSGA-II, so as to solve or partially solve the problems of how to efficiently identify the source of emergencies and how to reschedule for emergencies such as worker absenteeism.

[0008] The objective of this invention can be achieved through the following technical solutions:

[0009] This invention provides a distributed rescheduling method for aircraft pulsed assembly lines based on CNP-NSGA-II, comprising the following steps:

[0010] Obtain real-time disturbance state information and use the trained scene recognition model based on gated recurrent units to obtain the rescheduling scene type;

[0011] In response to the rescheduling scenario type being the first type, the aircraft pulse assembly line is rescheduled to the right.

[0012] In response to the rescheduling scenario being of the second type, the assembly process ordering and / or assembly worker allocation of the aircraft pulse assembly line station are optimized by constructing a non-dominated sorting genetic algorithm II, thereby achieving collaborative scheduling within the station of the aircraft pulse assembly line.

[0013] In response to the rescheduling scenario being of type three, the contract network algorithm generates bidding information that includes human resources requiring assistance from other stations, optimizes the allocation of assembly workers at stations, and achieves collaborative scheduling between stations on the aircraft pulse assembly line.

[0014] As a preferred technical solution, the training process of the scene recognition model includes the following steps:

[0015] By simulating rescheduling scenarios, disturbance state information under different rescheduling scenarios is obtained. The corresponding labels are obtained through annotation, and the disturbance state information is normalized to construct training samples.

[0016] A densely connected network with sequential connections, a gated recurrent neural network, and a normalized exponential function layer are constructed as a scene recognition model.

[0017] The scene recognition model is trained using the training samples.

[0018] As a preferred technical solution, the method of optimizing the process sequencing and / or worker allocation of aircraft pulsed assembly line stations using a non-dominated sorting genetic algorithm II includes the following steps:

[0019] In response to the rescheduling scenario being of the second type, for the station on the aircraft pulse assembly line, the rule weights corresponding to the assembly process optimization and assembly worker allocation are used as chromosome codes to initialize the population.

[0020] By decoding the contract collaboration, the optimization target values ​​of the actual completion time of the station and the time for dispatching workers to the bidding station are calculated. Subgroups are generated through fast non-dominated sorting, selection, crossover and mutation. The parent group and subgroups are merged to minimize the actual completion time of the station and the time for dispatching workers to the bidding station. A new parent generation is generated through fast non-dominated sorting and crowding calculation. This step is repeated until the preset iteration round is reached.

[0021] Based on the Pareto front solution set of the iterative population, a rescheduling scheme is obtained. It is then determined whether the rescheduling scheme exceeds the cycle time. If so, the scheduling process corresponding to the third type of rescheduling scenario is executed. If not, the station-based collaborative scheduling of the aircraft pulse assembly line is realized based on the rescheduling scheme.

[0022] As a preferred technical solution, the rules corresponding to assembly process optimization include: the more subsequent processes there are, the higher the priority; the longer the sum of the times of the subsequent processes of the current process, the higher the priority; and the longer the time from the current process to the point where there are no subsequent processes, the higher the priority.

[0023] As a preferred technical solution, the rules for allocating assembly workers include: workers with shorter actual assembly time are given priority, workers with shorter completion time are given priority, and workers with higher assembly efficiency are given priority.

[0024] As a preferred technical solution, the process of calculating the optimization target value of the rule weights includes the following steps:

[0025] For all assembly processes in the work group of the disturbed worker, they are divided into completed process sets, process sets in progress, and process sets to be assembled, based on the disturbance.

[0026] For each assembly process in the set of processes to be assembled, based on the current population, the assembly process with the highest priority is calculated according to the rules for optimizing the assembly process, and the worker with the highest priority is calculated according to the rules for assigning assembly workers. The assembly process set and worker status are then updated.

[0027] After all assembly processes in the worker's work group that caused the disturbance have been updated, the optimization target value is calculated.

[0028] As a preferred technical solution, the process of generating bidding information that includes human resources requiring assistance from other stations through the contract network algorithm and optimizing the allocation of assembly workers at stations includes the following steps:

[0029] Determine if the negotiated resource lock in the current booth is accessible. If not, suspend and wait, repeating this step until the negotiated resource lock is accessible.

[0030] In response to the negotiated resource lock becoming accessible, lock the negotiated resource lock;

[0031] Multiple candidate scheduling solutions for process sequencing and assembly worker allocation are obtained using the NSGA-II algorithm.

[0032] The multiple candidate scheduling solutions are sorted, and bidding information corresponding to the candidate scheduling solutions is constructed according to the sorting. The bidding information corresponding to the candidate scheduling solutions with higher sorting is sent to other stations of the machine pulse assembly line. If other stations reject the bid, the bidding information corresponding to the next candidate scheduling solution is sent to other stations of the machine pulse assembly line. The bidding is repeated until the information of acceptance of the bid is received or the number of negotiation rounds reaches the preset limit, and the negotiation resource lock is released.

[0033] As a preferred technical solution, in response to the negotiation rounds reaching a preset upper limit, the assembly process optimization and assembly worker allocation rules for all stations on the machine pulse assembly line are retrained.

[0034] As a preferred technical solution, the bidding information includes bidding station, bidding station, bidding virtual worker number, and virtual worker number of the bidding station that meets the requirements of the bidding document.

[0035] As a preferred technical solution, the aircraft pulse assembly line includes multiple stations, and a distributed decision-making station agent is constructed for each station. Each station agent includes a scheduling agent, and a task agent and a worker agent that perform assembly process optimization and assembly worker allocation under the guidance of the scheduling agent. The distributed rescheduling method is applied to each station agent.

[0036] Compared with the prior art, the present invention has at least one of the following beneficial effects:

[0037] (1) Achieve efficient identification of the source of emergencies: This invention obtains real-time disturbance state information and uses a trained scene recognition model based on gated loop units to obtain the rescheduling scene type, thereby achieving efficient and accurate scene identification and preparing for subsequent rescheduling.

[0038] (2) Realizing rescheduling for emergencies such as worker absenteeism: This invention adopts a distributed scheduling approach, allowing each station to make independent decisions and execute tasks, reducing communication burden and improving scheduling efficiency. Furthermore, unlike existing static scheduling, this invention is based on a GRU neural network model, which can accurately identify and classify the impact of disturbance events on the pulse beat, preparing for subsequent rescheduling. By employing improved CNP and NSGA-II algorithms, dynamic adjustment of resources between stations and optimization of process sequencing and worker allocation within stations are achieved, improving collaborative efficiency and realizing global optimization. Attached Figure Description

[0039] Figure 1 This is a flowchart illustrating the distributed rescheduling method for aircraft pulsed assembly lines based on CNP-NSGA-II in the embodiment.

[0040] Figure 2 This is a flowchart illustrating the rescheduling scenario identification process in the embodiment.

[0041] Figure 3 This is a flowchart illustrating the intra-station collaborative rescheduling process based on NSGA-II in the embodiment.

[0042] Figure 4 This is a flowchart illustrating the improved contract website inter-position negotiation and rescheduling process in the embodiment.

[0043] Figure 5 This is a schematic diagram of the electronic device in the embodiment. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0045] Example 1

[0046] To address the aforementioned problems in the existing technology, this embodiment provides a distributed rescheduling method for aircraft pulsed assembly lines based on CNP-NSGA-II (Contract Network-Non-Dominated Sorting Genetic Algorithm II). This method aims to overcome the shortcomings of existing scheduling methods in handling dynamic disturbance events, achieve accurate identification and effective response to sudden worker absenteeism events, and ensure the stable operation and production efficiency of the assembly line in a dynamic environment.

[0047] Step S0 involves setting up distributed decision-making station agents at each station on the assembly line. These agents include three sub-agents: a task agent, a worker agent, and a scheduling agent. The task agent and worker agent each handle two sub-tasks: optimizing the assembly sequence and allocating assembly workers. Guided by the scheduling agent, these two types of agents collaborate in real-time to optimize the overall performance of the station. When a task agent or worker agent detects a sudden disturbance event, it inputs the disturbance information into the information perception module of the scheduling agent for rescheduling scenario identification.

[0048] Step S1, see Figure 1 The scheduling agent uses pre-trained Gated Recurrent Units (GRUs) to analyze real-time information, identify the impact of disturbance events on the pulse beat, and categorize them into three scenarios: Scenario 1 (the first type of rescheduling scenario): minimal impact on station positions, requiring only right-shift rescheduling; Scenario 2 (the second type of rescheduling scenario): still exceeding the original beat after right-shift rescheduling, requiring intra-station coordinated rescheduling; Scenario 3 (the third type of rescheduling scenario): still exceeding the beat after right-shift rescheduling and intra-station coordinated rescheduling, requiring inter-station negotiated rescheduling.

[0049] For details, see Figure 2 Rescheduling scenario identification based on GRU. Specifically, it includes steps S101-S103:

[0050] Step S101, Sample Generation and Processing. First, simulation methods are used to simulate various rescheduling scenarios, and state data under different rescheduling scenarios are collected. Each sample is then labeled. Specifically, for each scenario: if direct right-shift rescheduling can resolve the issue, it is marked as Scenario 1, which does not require rescheduling; if the cycle time is exceeded, the reassignment to other workers is attempted according to the original rules; if the cycle time is not exceeded, it is marked as Scenario 2, which requires in-station rescheduling; if the cycle time is exceeded, it is marked as Scenario 3, which requires assistance from other stations.

[0051] Each sample ultimately formed contains the state X: x1, x2, ..., x n The labels y = 1, 2, 3. X describes the disturbance information from three aspects: process, worker, and station. The state includes information on process, worker, and station. Specifically, the process aspect includes two state variables: the total time of the subsequent processes of the disturbed worker and the number of subsequent processes of the disturbed worker; the worker aspect includes four state variables: the disturbed shift number, the expected duration of the disturbance, the remaining task time of the normal workers in the same shift as the disturbed worker, and the remaining time of the disturbed worker's process; the station aspect includes two state variables: the remaining time of the current distance from the beat and the ID of the disturbed station. Next, the sample labels are processed using one-hot encoding. In this embodiment, the rescheduling scenario is divided into three scenarios. To eliminate the influence of the category values, one-hot encoding is used to change the labels 1, 2, 3 to [1, 0, 0], [0, 1, 0], and [0, 0, 1], respectively. Finally, to eliminate the influence of dimensional imbalance on training, the state variables are normalized, letting

[0052] Step S102, Network Training. First, design the network. The input layer uses a dense network, the middle layers include a GRU neural network, and the last layer uses a normalized exponential function (softmax) for classification. Next, randomly initialize the parameters and set the loss function as shown in the equation. The loss function is iteratively calculated and the parameters are updated for training. The trained network is then saved to the scene recognition module of the scheduling agent.

[0053] Step S103: When performing rescheduling scenario identification, the real-time disturbance state information is input into the trained model, and the rescheduling scenario category is output.

[0054] In step S2, after identifying the scenario as rescheduling scenario 1, the scheduling agent directly issues a right-shift rescheduling instruction and guides the task agent and worker agent to make local adjustments. This does not involve changes to the process sequence or personnel allocation. The scheduling agent then issues the finalized scheduling plan to the task agent and worker agent, guiding them to execute the actual scheduling.

[0055] Specifically, for rescheduling scenario 1, the impact of the disturbance is relatively small. After the right-shift rescheduling, the predetermined cycle time is not exceeded, and it does not affect the subsequent normal operation. Therefore, no further adjustments are required.

[0056] In step S3, after identifying the rescheduling scenario as Scenario 2, the scheduling agent guides the task agent to optimize the work process sequencing and worker allocation within the workstation using the Non-dominated Sorting Genetic Algorithms-II (NSGA-II) algorithm to adapt to the impact of disturbance events. The decision execution module of the scheduling agent directly issues the adjusted scheduling strategy to the scheduling execution modules of the task agent and the worker agent. The agents within the two workstations then collaboratively execute this strategy to convert it into the actual scheduling plan.

[0057] See Figure 3 This step addresses rescheduling scenario 2 by providing a station-based collaborative rescheduling method based on NSGA-II. The station-based collaborative rescheduling problem is as follows: station z, shift b has a set of assembly operations to be completed. Some workers in shift b are absent. Station z adjusts the operation sequencing rules and personnel allocation rules for shift b to enable shift b to complete the assembly within a specified cycle time under disturbance conditions. This problem requires considering both production performance and stability optimization objectives. That is, it must ensure optimized performance after rescheduling while minimizing the difference between the changed scheduling scheme and the original scheme to avoid excessive costs in resource reallocation. Optimization objectives include minimizing the actual completion time of the shift and the number of worker operation changes. This section combines NSGA-II with the assembly rules in the pulse line to adjust the station-based scheduling scheme and improves the encoding and decoding methods based on the specific characteristics of the research object.

[0058] Improved Coding Method: A coding method combining assembly rules is adopted. In the research object, the number of rescheduled processes is relatively small, but the range of process numbers varies greatly, and the workers to be assigned have multiple attributes such as station positions and work groups. If an end-to-end coding method is used, it is necessary to re-code according to the process set, worker work group type, and worker number for each rescheduled process, which would cause coding complexity. Therefore, a combined rule method is adopted to improve chromosome coding. Each chromosome code w represents the weight of the rule used. [w1, w2, w3] belongs to the rule weight of the task agent, and [w4, w5, w6] belongs to the rule weight of the worker agent. i∈(0,1). The rules for task agents include higher priority for tasks with more subsequent tasks, higher priority for tasks with a longer sum of their times, and higher priority for tasks with a longer time to reach a point where no subsequent tasks remain. The rules for worker agents include higher priority for workers with shorter actual assembly times, shorter completion times, and higher assembly efficiency. In the combined scheduling rules, the priority CP of the scheduling object o is... o The mathematical expression for P can be summarized as follows: ow The priority of the scheduling object o obtained by single rule calculation.

[0059]

[0060] Actions based on combined scheduling rules can be expressed as: Each value is the weight of the corresponding single scheduling rule, W. i For the number of scheduling rules, This represents the weight of rule w in the task agent. This represents the weight of the worker agent rule w.

[0061] Improved Decoding Method: A cooperative decoding approach is adopted. After collaborative execution by the task agent and worker agent, the chromosome can be decoded to obtain the optimized target value. First, the task agent divides the process set, checks all assembly process tasks in the shift of the disturbed worker, and divides them into three sets based on the disturbance status: completed process sets, process sets in progress, and process sets to be assembled. Completed processes do not require adjustment, process sets in progress maintain their original scheduling order, and process sets to be assembled are rescheduled based on the rules output by NSGA-II. Then, the population output by the scheduling agent is decoded, i.e., the highest priority process is calculated according to the task ranking rules and input into the worker agent, subsequently updating the process sets to be assembled; the worker agent then calculates the highest priority worker according to the allocation rules in the chromosome and updates the worker status. After all processes are allocated, two optimized target values ​​are calculated.

[0062] Specifically, the NSGA-II-based intra-station collaborative rescheduling method may include steps S301-S304:

[0063] Step S301, Population Initialization. First, the scene recognition part of the scheduling agent outputs the adjustment results within the station, and the station-based collaborative rescheduling module performs population initialization for the NSGA-II algorithm.

[0064] Step S302: Decode the population input into the task agent and worker agent, and calculate the optimization target value.

[0065] Step S303, Iterative Optimization. After fast non-dominated sorting and selection, crossover, and mutation to generate subgroups, the parent and offspring generations are merged, and a new parent generation is generated through fast non-dominated sorting and crowding calculation. Specifically, the selection operator is based on a binary tournament strategy, selecting individuals with smaller non-dominated order and larger crowding distance; the crossover operator randomly generates two crossover points on the two parent chromosomes each round; the mutation operator sets a mutation distribution parameter μ, generates a random number ε each round, and performs mutation operations on the parent generation according to polynomial mutation, with the mutation of each chromosome shown in the following formula.

[0066]

[0067] Step S304: Considering both stability and efficiency, select the final rescheduling scheme from the Pareto front solution set of the last generation population. Given the limitations of rescheduling scenario prediction accuracy, if the obtained scheduling scheme still exceeds the tick time, proceed to step S4, entering rescheduling scenario 3: station-to-station negotiation rescheduling based on improved CNP.

[0068] Step S4: After identifying the rescheduling scenario 3, the scheduling agent uses the Contract Net Protocol (CNP) algorithm to generate a tender document containing human resource information that requires assistance from other stations. This document is then input into the information perception module of the scheduling agents at other stations, and the agents select from the tender documents returned by other stations to adjust the human resources between stations.

[0069] See Figure 4 For rescheduling scenario 3, resource adjustments between stations are required. Due to information asymmetry among stations in the distributed scheduling system, each station lacks access to resource information from other stations. Therefore, station agents need to negotiate and interact to adjust resources. Among existing negotiation algorithms, the Contract Network algorithm is a classic algorithm for solving dynamic interactions between different parts of a distributed system. CNP's logical structure is relatively simple, effectively decomposing large-scale production tasks and dynamically finding the optimal allocation scheme. This section improves the Contract Network algorithm to achieve negotiated rescheduling between stations.

[0070] First, we make the following assumptions. Due to the randomness of sudden events and the efficiency of algorithm processing, there is only one negotiation thread at any given time; each negotiation thread considers one work group, that is, the workers who are absent each time belong to the same work group, but there can be multiple workers from the same work group.

[0071] The CNP method comprises two key elements: agent structure design and bidding / tendering strategy. Based on a distributed agent network, an improved CNP method is proposed by designing a dynamic role-playing mechanism for bidding / tendering agents and a win-win-oriented bidding / tendering negotiation strategy.

[0072] Dynamic Role-Playing Mechanism for Bidding Agents: Regarding the structure of bidding agents, traditional contract networks typically require a unified bidding party to manage tasks. However, in the research object of this paper, considering the need to adjust the personnel structure at each station, such a model would require collecting real-time data on all relevant work groups and workers at all stations for global rescheduling and simulation calculations. This significantly increases the workload, fails to fully utilize the advantages of distributed scheduling, and is not conducive to timely adjustments to the scheduling scheme. To address this issue and improve decision-making efficiency, a dynamic role-playing mechanism for bidding agents is designed. Each station agent can act as a bidding party to create bidding documents and also as a bidder to accept bids from other stations. Human resource adjustments can be achieved through negotiation between station agents. To avoid conflicts between negotiation threads, a negotiation resource lock is first set up. When the rescheduling scenario identification module of a station's scheduling agent outputs rescheduling scenario 3, it checks whether the negotiation resource lock is accessible. If it is inaccessible, the resource is suspended and waits; if it is accessible, the bidding process begins, and the resource remains locked throughout the bidding process until the bidding ends and the resource is released.

[0073] Win-win oriented bidding negotiation strategy: Traditional contract network algorithms often only consider the performance optimization of the bidding party. Since each station agent dynamically plays the role of bidding agent in this paper, the relationship between each agent is equal. Therefore, when a station requests resource assistance from other stations, it should also consider the performance of other stations and try not to affect the normal assembly of other stations. This part designs three sub-strategies for the three most important steps in the negotiation process: tender document design, bid document design, and tender document selection. The tender document design method based on NSGA-II, the bid document production method based on idle time matching, and the bid document selection method based on worker combination efficiency scoring are designed respectively.

[0074] First, the tender document design is based on NSGA-II. To reduce invalid tenders and ensure that the performance of other stations is not affected as much as possible, the specific information of the assistance required by the disturbed station is first calculated, including the number of people assisting and the assistance time. This calculation is still based on the NSGA-II algorithm. Through the collaboration between the scheduling agent, task agent, and worker agent, the final rescheduling scheme is obtained, thereby determining the tender document. Considering minimizing the completion time of the bidding station while minimizing the dispatch time of other stations, the objective function of the NSGA-II algorithm is designed as minimizing the actual completion time f1 of the current station and minimizing the dispatch time f2 of the bidding station. A preference-based selection operator is adopted, where preference refers to prioritizing the comparison of the optimization objective f2 of the two chromosomes when selecting two parent chromosomes based on a binary tournament strategy. In the chromosome decoding calculation by the worker agent, considering the role of dispatched workers, virtual workers need to be set: if NoW workers are absent, NoW virtual workers are set. The Pareto front solutions are sorted in ascending order according to the sum of the assembly times of dispatched workers from external stations (it may be necessary that more than one external worker is dispatched), and tender documents are generated for each. Each solution corresponds to a tender document of NoW dimension, with each dimension containing one tender document, represented by a quintuple. <Z 0 ,B,W 0 ,St,Ft> means that Z 0 B represents the bidding station number, and B represents the work group number that needs to be dispatched, i.e., the work group number of the absent worker; W 0 This represents the ID of the virtual worker that needs to be dispatched, where St represents worker W. 0 The expected start time of dispatch, Ft represents the worker W 0 Expected end time of dispatch.

[0075] Next, bids are generated based on available time slots. After receiving the tender documents, other workstations query the current status of the workers in the work group and generate bids from the worker information that meets the tender requirements. Each bid is represented by a six-tuple. <Z 0 ,Z,Agr,W 0 ,W,Effi>:Z 0 Z represents the bidding station, and W represents the tendering station. 0 W is the virtual worker ID in the tender document, and W is the worker ID in the bidding station that meets the requirements of the tender document; Agr represents whether the bidding station agrees to the bid, and the bidding station approves each tender information <Z>. 0 ,B,W 0 The `[St,Ft]` method iterates through the status of all workers in group B. If there are no idle workers in the range [St,Ft], then Agr is 0, and W and Effi are empty. If there are workers that meet the requirements of the tender information, then Agr is 1. W and Effi are the worker numbers and efficiencies that meet the requirements.

[0076] Finally, the selection of bids for worker combination efficiency scoring. (Station agent) i After receiving bids from other stations, the optimal combination of dispatched workers is found by iterating through them.

[0077] Specifically, the inter-station negotiation rescheduling method based on the improved VNP may include steps S401-S407:

[0078] Step S401: Set a negotiated resource lock when the agent is in position. i When the rescheduling scenario identification module outputs rescheduling scenario 3, the negotiation module first checks whether the negotiation resource lock is accessible. If it is inaccessible, the resource is suspended and waited for; if it is accessible, it enters 32) and locks the negotiation resource.

[0079] Step S402, station agent i The negotiation module and task agent, along with the worker agent, obtain several sets of rescheduling solutions based on the NSGA-II algorithm.

[0080] Step S403: Sort these scheduling solutions according to certain preferences, and design tender documents for these solutions in turn according to the sorting order. First, broadcast the tender document corresponding to the first sorted solution to other stations.

[0081] Step S404, Positioning agent j The scheduling agent information processing module receives information from the agent. i The agent receives the tender document, matches and generates tender documents based on available time, and then sends them back to the agent. i .

[0082] Step S405, station agent i The dispatch agent information processing module receives bids from other stations and inputs them into the negotiation module. The negotiation module selects a suitable station based on worker combination efficiency scores and notifies it, signs a contract with it, dispatches workers to the corresponding bidding station, terminates the current negotiation, and releases the negotiation resource lock. If all other stations reject the bid, the next round of bidding begins, and the process proceeds to step 46.

[0083] Step S406, Positioning agent i The negotiation module selects the next solution according to the sorting of solutions in Step 3 to generate the tender document, repeating steps S403-S406 until the upper limit of negotiation rounds is reached. If a bidding position is still not found, proceed to step S407.

[0084] Step S407: Start global network training, retrain to generate assembly rules for all stations, and release the negotiated resource lock.

[0085] The effectiveness of this method will be illustrated below using an actual aircraft pulsed final assembly line capable of assembling multiple aircraft models. The assembly line layout and total human resources have been determined. The assembly line is linear, divided into 5 stations, each responsible for assembling different parts. The aircraft completes its assembly tasks sequentially at each station and is transported to the next station at a predetermined pace.

[0086] The final assembly line currently receives final assembly orders for various aircraft models. This embodiment focuses on a specific aircraft model for research. The process route for this aircraft model has been planned in the early stage, with a total of more than 3,000 processes. For ease of management, the processes are roughly divided into 12 process packages AL according to the assembly parts. The process packages strictly follow a linear pre- and post-process relationship in ascending alphabetical order. The sub-processes within the process package exhibit a non-linear and sparse pre- and post-process relationship.

[0087] Table 1 presents basic information for some processes. The standard assembly time, required assembly team, and number of workers for each process are determined by the process characteristics. Each process requires several workers from one team for assembly. The assembly time distribution is relatively uneven, with 94% of the processes taking less than 10 hours to assemble. The number of assembly workers ranges from 1 to 5, and most assembly processes require two or more workers to complete. Furthermore, this set of processes exhibits a sparse predecessor-successor relationship, with 80% of processes having no predecessor process and 84% having no successor process. Based on production experience, the initial cycle time is set at 290 hours.

[0088] The assembly line has been planned in advance: it is divided into 17 teams according to their specialties. Each team has 5 workers, and the workers' skill proficiency is evenly distributed between [0.8, 1.2]. For the same process, the workers with higher proficiency values ​​have shorter assembly times. The actual assembly time for this process is the standard assembly time multiplied by the proficiency value.

[0089] Table 1. Partial Process Schematic Table

[0090]

[0091] Using this actual assembly line, and incorporating random worker absence disturbances, the rescheduling optimization method of this invention is applied. The specific steps are as follows:

[0092] Step S1, as follows Figure 2 The process involves identifying GRU-based rescheduling scenarios. The specific steps are as follows:

[0093] Step S101: Randomly set up worker absence disturbances in different shifts at each station, and collect multiple sets of disturbance state data during the simulation process. Label each sample. Specifically, for each scenario: if direct right-shift rescheduling can resolve the issue, it is marked as Scenario 1 (no rescheduling required); if the time limit is exceeded, attempt to reassign workers according to the original rules; if the time limit is not exceeded, it is marked as Scenario 2 (requiring in-station rescheduling); if the time limit is exceeded, it is marked as Scenario 3 (requiring assistance from other stations). Each final sample contains the state X: x1, x2, ..., x n The labels y = 1, 2, 3. X describes the disturbance information from three aspects: process, worker, and station. The process aspect includes two state variables: the total time of subsequent processes for the disturbed worker and the number of subsequent processes for the disturbed worker. The worker aspect includes four state variables: the disturbed shift number, the expected duration of the disturbance, the remaining task time of normal workers in the same shift as the disturbed worker, and the remaining time of the disturbed worker's process. The station aspect includes two state variables: the remaining time at the current distance from the beat and the ID of the disturbed station. Next, the sample labels are processed using one-hot encoding. Labels 1, 2, 3 are transformed into [1, 0, 0], [0, 1, 0], and [0, 0, 1], respectively. Then, the state variables are normalized, letting...

[0094] Step S102, Network Training. The network is designed with a Dense network for the input layer, a GRU neural network for the intermediate layers, and a softmax layer for classification in the final layer. Next, the parameters are randomly initialized, and the loss function is set as shown in the equation. The loss function is iteratively calculated and the parameters are updated for training. The trained network is then saved to the scene recognition module of the scheduling agent.

[0095] Step S103: When performing rescheduling scenario identification, real-time disturbance state information is input into the trained model, and the rescheduling scenario category is output. Table 2 shows the comparison between the GRU rescheduling identification results of this embodiment and the general classification method KNN. The deep learning network constructed in this embodiment has higher accuracy than general classification algorithms and is more suitable for classifying disturbance events in aircraft pulse assembly.

[0096] Table 2. Accuracy comparison between GRU method and KNN method

[0097] GRU KNN Test set accuracy 0.90 0.77 Validation set accuracy 0.80 0.54

[0098] Step S2: Implement the disturbance handling method under rescheduling scenario 1, namely, the implementation of right-shift rescheduling. Suppose that a disturbance occurs at time 187h, and workers 3 and 5 of shift 8 at station 1 are absent until time 237h. At this time, right-shift rescheduling is adopted, and the actual completion time of the shift is moved to time 246.4h, but it does not exceed the cycle time. Therefore, right-shift rescheduling can complete the rescheduling.

[0099] Step S3, as follows Figure 3 The disturbance handling method in rescheduling scenario 2 is implemented, namely, the implementation of station-based collaborative rescheduling based on NSGA-II. At 350 hours, worker 2 of shift 1 at station 2 leaves his post, and the remaining rescheduling operations include: [712 822 786 750 765 763 688 800 828 760 715 824 795 731 713 808 816 719 820 722 852 779 825 739 771 735]. The specific process of station-based collaborative rescheduling based on NSGA-II is as follows:

[0100] Step S301, Population Initialization. First, the scene recognition part of the scheduling agent outputs the station adjustment results, and the station-based collaborative rescheduling module performs population initialization for the NSGA-II algorithm. A combined rule approach is used to improve chromosome coding. Each chromosome code w represents the weight of the adopted rule; [w1, w2, w3] belongs to the rule weight of the task agent, and [w4, w5, w6] belongs to the rule weight of the worker agent. i ∈(0,1). In the combined scheduling rule, the priority CP of the scheduling object o is... o The mathematical expression for P can be summarized as follows: ow The priority of the scheduling object o obtained by single rule calculation.

[0101]

[0102] Actions based on combined scheduling rules can be expressed as: Each value is the weight of the corresponding single scheduling rule, W. i For the number of scheduling rules, This represents the weight of rule w in the task agent. This represents the weight of the worker agent rule w.

[0103] Step S302: The population is decoded in the task agent and worker agent to calculate the optimization objective value. The task agent divides the process set, checks all assembly process tasks in the shift of the disturbed worker, and divides them into three sets of processes based on the different disturbance conditions: completed process set, process set in progress, and process set to be assembled. Completed processes do not need to be adjusted, the process set in progress maintains its original scheduling order, and the process set to be assembled is rescheduled based on the rules output by NSGA-II. Then, the population output by the scheduling agent is decoded, that is, the highest priority process is calculated according to the task sorting rules, and the process is input into the worker agent, and then the process set to be assembled is updated; the worker agent then calculates the highest priority worker according to the allocation rules in the chromosome and updates the worker status. After all processes are allocated, two optimization objective values ​​are calculated. Considering both production performance and stability optimization objectives, that is, to ensure that the performance after rescheduling is optimized, and to minimize the difference between the changed scheduling scheme and the original scheduling scheme, so as to avoid excessive costs in resource reallocation, two types of optimization objectives are designed as follows:

[0104] Optimization objective 1 is the actual completion time f1 of the work team, defined as follows, where O L For processes that require rescheduling:

[0105] f1=min(t O ),o∈O L

[0106] Optimization objective 2 is the number of worker process changes f2, defined as follows, O zbw O represents the process set of worker w in shift group z. ′ zbw This represents the worker's assembly process set after the adjustment, and f2 represents the changes in the worker's assembly process set before and after the adjustment.

[0107]

[0108] Step S303, Iterative Optimization. After fast non-dominated sorting and selection, crossover, and mutation to generate subgroups, the parent and offspring generations are merged, and a new parent generation is generated through fast non-dominated sorting and crowding calculation. Specifically, the selection operator is based on a binary tournament strategy, selecting individuals with smaller non-dominated order and larger crowding distance; the crossover operator randomly generates two crossover points on the two parent chromosomes each round; the mutation operator sets a mutation distribution parameter μ, generates a random number ε each round, and performs mutation operations on the parent generation according to polynomial mutation, with the mutation of each chromosome shown in the following formula.

[0109]

[0110] Step S304: Considering both stability and efficiency, the final rescheduling scheme is selected from the Pareto front solution set of the last generation population. The optimization objective values ​​are shown in Table 3. Considering both stability and assembly efficiency, the rule corresponding to f1 = 449.1 and f2 = 155 is selected as the final scheme. At this time, the adjusted weights of the two types of rules are [0.41, 0.31, 0.10, 0.16, 0, 0.16]. The final scheduling result still does not exceed the cycle time (560h), proving the effectiveness of the scheduling.

[0111] Table 3 Rescheduling Scenario 2 - Optimization Target Values

[0112] f1 f2 453.7 146 452.7 148 449.4 149 449.1 155 449.1 159

[0113] Step S4, as follows Figure 4 The implementation of the rescheduling scenario 3, specifically the inter-station negotiation rescheduling method based on the improved CNP, involves designing a disturbance event. Assume that 250 hours after the completion cycle, two workers, worker 2 and worker 3, leave their posts in shift 7 of station 1. Worker 2 has remaining work processes [554, 639, 649], and worker 3 has remaining work processes [557, 639, 649]. The rescheduling scenario is identified as scenario 3. In this case, the improved CNP inter-station negotiation rescheduling method is needed to negotiate resource adjustments between stations through bidding. The specific process is as follows:

[0114] Step S401: Set a negotiated resource lock when the agent is in position. i When the rescheduling scenario identification module outputs rescheduling scenario 3, the negotiation module first checks whether the negotiation resource lock is accessible. If it is inaccessible, the resource is suspended and waits; if it is accessible, it proceeds to step 2) and locks the negotiation resource.

[0115] Step S402, station agent i The negotiation module, task agent, and worker agent calculate the specific information needed for assistance at the disturbed station, including the number of people requiring assistance and the assistance time. Several rescheduling solutions are obtained based on the NSGA-II algorithm. Considering minimizing the completion time of the bidding station while also minimizing the dispatch time of other stations, the objective function of the NSGA-II algorithm is designed as minimizing the actual completion time f1 of the current station and minimizing the dispatch time f2 of the bidding station. In the chromosome decoding calculation by the worker agent, considering the role of dispatched workers, virtual workers need to be set: if NoW workers are absent, NoW virtual workers are set.

[0116] Step S403: Sort these scheduling solutions according to certain preferences, and design tender documents for each solution in order of sorting. Broadcast the tender document corresponding to the top-ranked solution to other stations. A preference-based selection operator is used; when selecting two parent chromosomes based on a binary tournament strategy, the optimization objective f2 of the two chromosomes is compared first. The Pareto front solutions are sorted in ascending order according to the sum of the assembly times of workers dispatched from external stations (which may require more than one worker), and tender documents are created for each. The tender document corresponding to each set of solutions is NoW-dimensional, with each dimension containing one tender message, using quintuples. <Z 0 ,B,W 0 ,St,Ft> means that Z 0 B represents the bidding station number, and B represents the work group number that needs to be dispatched, i.e., the work group number of the absent worker; W 0 This represents the ID of the virtual worker that needs to be dispatched, where St represents worker W. 0 The expected start time of dispatch, Ft represents the worker W 0 The expected end time of the dispatch. As shown in Table 4, in order to minimize the dispatch cost of other stations, the assembly rules corresponding to the solutions f1 = 280 and f2 = 18.6 are taken, and the tender document [[1,7,1,268,277],[1,7,2,268,273]] is calculated.

[0117] Table 4. Tender Document Preparation - Pareto Front Solution

[0118] f1 f2 268.8 36.8 280.0 18.6

[0119] Step S404, Positioning agent j The scheduling agent information processing module receives information from the agent. i The agent receives the tender document, matches and generates tender documents based on available time, and then sends them back to the agent. i After receiving the tender documents, other workstations will check the current status of the workers in the work group and create tender documents from the worker information that meets the requirements of the tender documents. Each tender document is represented by a six-tuple. <Z 0 ,Z,Agr,W 0 ,W,Effi>:Z 0 Z represents the bidding station, and W represents the tendering station. 0 W is the virtual worker ID in the tender document, and W is the worker ID in the bidding station that meets the requirements of the tender document; Agr represents whether the bidding station agrees to the bid, and the bidding station approves each tender information <Z>. 0 ,B,W 0The `[St,Ft]` method iterates through the status of all workers in team B. If there are no idle workers in the range [St,Ft], then Agr is 0, and W and Effi are empty. If there are workers that meet the requirements of the tender, Agr is 1. W and Effi are the worker numbers and efficiencies that meet the requirements. The returned tender documents are shown in Table 5.

[0120] Table 5. Preparation of Bidding Documents

[0121] <![CDATA[Z 0 ]]> Z Agr <![CDATA[W 0 ]]> W Effi 1 2 0 0 0 0 1 3 1 1 3 0.8 1 3 1 2 4 0.9 1 4 1 1 3 0.8 1 4 1 2 4 0.9 1 5 1 1 1 0.8 1 5 1 2 3 1

[0122] Step S405, station agent i The scheduling agent information processing module receives bids from other stations and inputs them into the negotiation module. The negotiation module selects a suitable station based on worker combination efficiency scores, notifies it, signs a contract, dispatches workers to the corresponding station, terminates the negotiation, and releases the negotiation resource lock. Following a traversal principle, the two most efficient workers in the bids are selected for calculation. The final completion time for both is 280.0 hours. The two most efficient bids are selected: worker 4 from station 3 and worker 4 from station 4.

[0123] Step S406: Start global network training, retrain to generate assembly rules for all stations, and release the negotiated resource lock.

[0124] In summary, this method employs a distributed scheduling approach, allowing each station to make independent decisions and execute tasks, reducing communication burden and improving scheduling efficiency. Furthermore, unlike existing static scheduling methods, this invention, based on a GRU neural network model, can accurately identify and classify the impact of disturbance events on the pulsating rhythm, preparing for subsequent rescheduling. By using improved CNP and NSGA-II algorithms, dynamic adjustment of resources between stations and optimization of process sequencing and worker allocation within stations are achieved, improving collaborative efficiency and realizing global optimization.

[0125] Example 2

[0126] Based on Embodiment 1, this embodiment provides an electronic device, including: one or more processors and a memory, wherein the memory stores one or more programs, the one or more programs including instructions for executing the CNP-NSGA-II-based distributed rescheduling method for aircraft pulse assembly lines as described in Embodiment 1.

[0127] like Figure 5 At the hardware level, the electronic device includes a processor, internal bus, network interface, memory, and non-volatile memory, and may also include other hardware required for the business operations. The processor reads the corresponding computer program from the non-volatile memory into memory and then runs it to achieve the above-mentioned functions. Figure 1The method described herein. Of course, in addition to software implementation, this invention does not exclude other implementation methods, such as logic devices or a combination of hardware and software, etc. That is to say, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0128] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0129] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0130] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A distributed rescheduling method for aircraft pulsed assembly lines based on CNP-NSGA-II, characterized in that, Includes the following steps: Obtain real-time disturbance state information and use the trained scene recognition model based on gated recurrent units to obtain the rescheduling scene type; In response to the rescheduling scenario type being the first type, the aircraft pulse assembly line is rescheduled to the right. In response to the rescheduling scenario being of the second type, the assembly process ordering and assembly worker allocation of the aircraft pulse assembly line station are optimized by constructing a non-dominated sorting genetic algorithm II, thereby achieving collaborative rescheduling within the station of the aircraft pulse assembly line. In response to the rescheduling scenario being of type three, the contract network algorithm generates bidding information that includes human resources requiring assistance from other stations, optimizes the allocation of assembly workers at stations, and realizes negotiated rescheduling between stations on the aircraft pulse assembly line. The process of generating bidding information that includes human resources requiring assistance from other stations through the contract network algorithm and optimizing the allocation of assembly workers at stations includes the following steps: Determine if the negotiated resource lock within the current station is accessible. If not, suspend and wait, repeating this step until the negotiated resource lock becomes accessible. In response to the negotiated resource lock becoming accessible, lock the negotiated resource lock; The negotiation module and task agent and worker agent of the current station calculate the specific information of the disturbance station that needs assistance, including the number of people to assist and the assistance time, and obtain several sets of rescheduling solutions based on the NSGA-II algorithm; The objective function of the NSGA-II algorithm is to minimize the actual completion time of the current station and the time for dispatching workers to the bidding station. In the chromosome decoding calculation of the worker agent, considering the role of dispatched workers, virtual workers need to be set: if NoW workers are absent, NoW virtual workers are set. These scheduling solutions are sorted according to certain preferences, and tender documents are designed for these solutions in turn according to the sorting order. The tender document corresponding to the first sorted solution is broadcast to other stations. The tender document corresponding to each set of solutions is NoW dimension, and each dimension contains one tender information. When the dispatch agent information processing module of other stations receives the tender document from the current station, it queries the current status of the workers in the shift and creates a tender document based on the idle time and the worker information that meets the requirements of the tender document. The current station's dispatch agent information processing module receives bids from other stations and inputs them into the negotiation module. The negotiation module selects a suitable station for bidding based on the worker combination efficiency score and notifies it, signs a contract with it, dispatches workers to the corresponding bidding station, terminates the current negotiation, and releases the negotiation resource lock. If all other stations reject the bid, the bidding information corresponding to the next candidate scheduling solution is sent to other stations on the aircraft pulse assembly line. The bidding is repeated until the information of acceptance of the bid is received or the number of negotiation rounds reaches the preset limit, and the negotiation resource lock is released. The aforementioned aircraft pulse assembly line includes multiple stations. For each station, a distributed decision-making station agent is constructed. Each station agent includes a scheduling agent, as well as task agents and worker agents that optimize assembly processes and allocate assembly workers under the guidance of the scheduling agent. The distributed rescheduling method is applied to each station agent.

2. The distributed rescheduling method for aircraft pulsed assembly lines based on CNP-NSGA-II according to claim 1, characterized in that, The training process of the scene recognition model includes the following steps: By simulating rescheduling scenarios, disturbance state information under different rescheduling scenarios is obtained. The corresponding labels are obtained through annotation, and the disturbance state information is normalized to construct training samples. A densely connected network with sequential connections, a gated recurrent neural network, and a normalized exponential function layer are constructed as a scene recognition model. The scene recognition model is trained using the training samples.

3. The distributed rescheduling method for aircraft pulsed assembly lines based on CNP-NSGA-II according to claim 1, characterized in that, The method of optimizing the process sequencing and / or worker allocation of aircraft pulse assembly line stations using a non-dominated sorting genetic algorithm II includes the following steps: In response to the rescheduling scenario being of the second type, for the station on the aircraft pulse assembly line, the rule weights corresponding to the assembly process optimization and assembly worker allocation are used as chromosome codes to initialize the population. By decoding the contract collaboration, the optimization target values ​​of the actual completion time of the work team and the number of changes in the worker's work process are calculated. Subgroups are generated by fast non-dominated sorting, selection, crossover and mutation. The parent group and subgroups are merged to minimize the actual completion time of the work team and the number of changes in the worker's work process. A new parent generation is generated by fast non-dominated sorting and crowding calculation. This step is repeated until the preset iteration round is reached. Based on the Pareto front solution set of the iterative population, a rescheduling scheme is obtained. It is then determined whether the rescheduling scheme exceeds the cycle time. If so, the scheduling process corresponding to the third type of rescheduling scenario is executed. If not, the station-based collaborative scheduling of the aircraft pulse assembly line is realized based on the rescheduling scheme.

4. The distributed rescheduling method for aircraft pulsed assembly lines based on CNP-NSGA-II according to claim 3, characterized in that, The rules for optimizing assembly processes include: the more subsequent processes there are, the higher the priority; the longer the sum of the times of the subsequent processes of the current process, the higher the priority; and the longer the time from the current process to the point where there are no subsequent processes, the higher the priority.

5. The distributed rescheduling method for aircraft pulsed assembly lines based on CNP-NSGA-II according to claim 3, characterized in that, The rules for assigning assembly workers include: workers with shorter actual assembly time are given priority, workers with shorter completion time are given priority, and workers with higher assembly efficiency are given priority.

6. The distributed rescheduling method for aircraft pulsed assembly lines based on CNP-NSGA-II according to claim 3, characterized in that, The process of calculating the optimization objective value of the rule weights includes the following steps: For all assembly processes in the work group of the disturbed worker, they are divided into completed process sets, process sets in progress, and process sets to be assembled, based on the disturbance. For each assembly process in the set of processes to be assembled, based on the current population, the assembly process with the highest priority is calculated according to the rules for optimizing the assembly process, and the worker with the highest priority is calculated according to the rules for assigning assembly workers. The assembly process set and worker status are then updated. After all assembly processes in the worker's work group that caused the disturbance have been updated, the optimization target value is calculated.

7. The distributed rescheduling method for aircraft pulsed assembly lines based on CNP-NSGA-II according to claim 1, characterized in that, In response to the negotiation rounds reaching the preset limit, the assembly process optimization and assembly worker allocation rules for all stations on the machine pulse assembly line are retrained.

8. The distributed rescheduling method for aircraft pulsed assembly lines based on CNP-NSGA-II according to claim 1, characterized in that, The bidding information includes bidding positions, bidding positions, bidding virtual worker numbers, and virtual worker numbers in the bidding positions that meet the requirements of the bidding documents.

Images