Research On Dynamic Parallel Multi-machine Open-shop Scheduling Based On Simulation

With the deeply implementation and application of China 2025 strategy,workshop scheduling is gradually developing in the direction of digitization and intelligence.The digital twin workshop is a new mode of workshop operation in the future,and building a high-fidelity model of the production scheduling problem is one of the basic conditions for realizing the digital twin workshop.The Open Shop Scheduling Problem(OSSP)widely exists in all aspects of production and life,and it has been proven to be an NP-Hard problem.Methods based solely on mathematical programming and swarm intelligence algorithms are more suitable for solving deterministic steady-state problems.For dynamic problems considering the existence of uncertain events,simulation-based methods can not only achieve rapid selection of multiple solutions,but also provide a basis for the establishment of fidelity models for digital twin workshops.Therefore,it has important theoretical and practical significance to carry out the research on the Open Shop Dynamic Scheduling Problem(OSDSP)based on simulation.This article takes the parallel multi-computer OSDSP as the research object,and the main research contents are as follows:Firstly,this thesis reviews the current research status of parallel multimachine OSDSP both domestic and overseas,and analyzes suitable scenarios of mathematical optimization methods,heuristic algorithms,modeling simulation methods and intelligent algorithms,which provide basis for the method design in this thesis.Secondly,study the single-target parallel multi-machine OSDSP with random arrival of workpieces and uncertain processing time.With the goal of Total Weighted Completion Time(TWC),a complete response based on FlexSim simulation model and heuristic algorithm is designed.Type rescheduling method.A simulation model based on the production data of the grain picking workshop is constructed.The experiment compares the target value of the heuristic algorithm of different workpiece selection machine rules and machine selection workpiece rules under different workpiece arrival time distributions.The experimental results show that the load rule is better than the random available and first available rules in the workpiece selection machine rules;the WSPT rule is better than the FCFS and the maximum weight priority rule in the machine selection rules;the arrival time distribution of different workpieces The change trend of the target value is similar,indicating that the proposed heuristic algorithm can adapt to the solution of OSDSP problems under different dynamic levels.Finally,this thesis studies Bi-objective parallel multi-machine OS Finally,study the dual-objective parallel multi-machine OSDSP with random arrival of workpieces and uncertain processing time.Aiming at TWC and Mean Flow Time(MFT),an adaptive periodic rescheduling method integrating FlexSim simulation model and improved NSGA-Ⅱalgorithm is designed..Aiming at the premature convergence of the basic NSGA-Ⅱ algorithm to solve the parallel multi-machine OSSP problem,an improved NSGA-Ⅱ algorithm is proposed to control the independent solution density;for the insufficient flexibility of fixed-period scheduling,a flexible rescheduling period is designed to generate a flexible rescheduling period according to the dynamic load of the workshop The method;a scheduling system integrating FlexSim simulation model and improved NSGA-Ⅱalgorithm is constructed on the dual-objective parallel multi-machine OSDSP.Taking the actual fitting data of the die picking workshop as an example,the simulation experiment compares the target value and the number of reschedules in scenarios such as adaptive periodic rescheduling,non-rescheduling,and fixed periodic rescheduling with intervals of 10 and 30,respectively.And the degree of process deviation.Experimental results show that adaptive periodic rescheduling can obtain a satisfactory solution while minimizing the deviation of the scheduling scheme.This thesis only considers the dynamic factor of random job release time and random processing time,future research needs to take other factors such as equipment failure and urgent orders into consideration.Furthermore,only NSGA-Ⅱ algorithm is designed,the efficiency and effective are uncertain,other swarm intelligence algorithms should be researched in the future.Last but not least,this thesis just realizes the logic simultation model of parallel multi-machine OSDSP,three-dimensional visualization model and the interaction with physical workshop still need to be researched to realize digital twin workshop in the future.