Research On Scheduling Methods For Panel Block Assembly Line Under Uncertainties

Modern shipbuilding model is oriented to intermediate products.Shipyards realize flow production and batch production by establishing intermediate-product lines.Panel block assembly line is representative of this kind of production lines.With the development of large-scale ships,the demand for panel blocks is increasing and the task of scheduling is becoming more difficult.However,conventional scheduling is more extensive because of insufficient consideration of reducing working hours,JIT requirements and uncertainties.Studying the scheduling problems of panel block assembly is of great significance to improve the management level and production efficiency.The scheduling methods for panel block assembly line can provide reference for the scheduling of other production lines of shipbuilding.The scheduling of panel block assembly makes reasonable arrangements for the panel blocks to be processed,which mainly based on the processing time and due date of each panel block.The processing time is usually measured by standard working hours.In the actual production enviroment,the uncertainties arising from both machine and human factors result in a deviation between actual and standard hours.Therefore,the processing time can not be accurately predicted.That is,the processing time is uncertain.The due date is usually not a fixed point in time,but a time interval associated with the satisfaction of demand-side.Therefore,the due date is also uncertain.There will be cases where new tasks are inserted during production,affecting the original scheduling scheme.This dissertation aims to reduce working hours of panel block assembly and to meet JIT requirements.It studies the scheduling methods for single panel block assembly line and parallel panel block assembly lines under the aforementioned uncertainties.The main contents are listed as follows:(1)Fuzzy scheduling method for panel block assembly and its verification.In order to describe the uncertainties of the processing time and the due date,the processing time of each panel block is considered a triangular fuzzy number and the due date is considered a trapezoidal fuzzy number.Fuzzy scheduling is carried out based on the two fuzzy temporal parameters.It focuses on the indexes such as the maximum fuzzy completion time(fuzzy makespan),and the agreement index of fuzzy completion time and fuzzy due date(AICD).Minimizing the fuzzy makespan can reduce working hours,and maximizing the average AICD helps to meet JIT requirements.Numerical calculations and simulation experiments are conducted to compare the scheduling schemes based on exact processing time or fuzzy processing time.The results show that fuzzy scheduling is effective with uncertain processing time.(2)Multi-objective fuzzy scheduling method for single panel block assembly line.There is no buffer between two consecutive stations.Each panel block is processed sequentially on station 1,station 2,and so on until the final station.The scheduling problem of single panel block assembly line is formulated based on these characteristics,with bi-objective of minimizing the fuzzy makespan and maximizing the average AICD.To solve this problem,a hybrid multi-obective grey wolf optimization(HMOGWO)algorithm is proposed.The HMOGWO introduces two novel heuristics to generate initial solutions,and incorporates two local search methods based on the insert neighborhood structure to improve the exploitation capability.The performance of the HMOGWO is tested on the production instances of panel block assembly in shipbuilding.Computational comparisons of the HMOGWO with other well-known multi-objective evolutionary algorithms demonstrate the feasibility and effectiveness of the HMOGWO.The reliabilities of the obtained scheduling schemes are verified through simulation experiments.(3)Multi-objective fuzzy scheduling method for parallel panel block assembly lines.There are two identical assembly lines.Each line can produce panel sheets and panel blocks.A panel sheet is transferred out of the assembly line through the transfer station after finishing the processes.A panel block is processed sequentially on station 1,station 2,and so on until the final station.The scheduling problem of parallel assembly lines can be decomposed into two subproblems:(i)assigning each panel block to one of the assembly lines;(ii)scheduling the panel blocks in each line.This problem is formulated based on these characteristics.The objectives are minimizing the fuzzy makespan and maximizing the average AICD.A multi-obective memetic algorithm(MOMA)is proposed to solve this scheduling problem.In the MOMA,a vector constructed by the index of blocks and a separator * is applied to represent a solution.The heuristic procedures for initial solutions,the crossover operator,the mutation operator and the local search method are designed based on the form of solutions.The performance of the MOMA is tested on the production instances of panel block assembly.Computational comparisons of the MOMA with other well-known multi-objective evolutionary algorithms demonstrate that the MOMA can provide better scheduling schemes.The reliabilities of the obtained scheduling schemes are verified through simulation experiments.(4)Reactive fuzzy scheduling method for new tasks insertion.Taking into account the stability requirement of rescheduling,AISS,the agreement index of the fuzzy starting time of each panel block in the re-scheduling scheme and the original scheduling scheme is defined.The reactive fuzzy scheduling problems for single assembly line and parallel assembly lines are formulated with the objectives of minimizing the fuzzy makespan,maximizing the average AICD and maximizing the average AISS.A hybrid multi-objective discrete differential evolution(HMODDE)algorithm is proposed for solving these problems.The HMODDE introduces novel heuristics to generate initial solutions,incorporates a local search method based on the insert neighborhood structure to improve the exploitation capability,and employs a dynamic population updating strategy to improve the evolution efficiency.The HMODDE is effective for the scheduling problems of single assembly line and parallel assembly lines.The reliabilities of the obtained scheduling schemes is verified through simulation experiments.(5)A decision support system for the scheduling problem of panel block assembly line.A decision support system is developed based on the research results of this dissertation.The system can be used for solving the scheduling problem,selecting the scheduling scheme and simulating the scheme.