Study On Intelligent Scheduling Approaches For Flexible Handling Systems

Material handling is the highest frequency operations in the process of enterprise internal logistics,it needs to take up a lot of time and consumes a lot of labor.Because of the advantages of high degree of automation and superior performance,more and more Flexible Handling Systems(FHS) typically represented by Automated Guide Vehicle systems,Rail Shuttle Vehicle systems,etc,have been widely used in the enterprises of a variety of industries such as tobacco,electronics,manufacture,warehousing and services.Since scheduling problem is one of the key factors which greatly influence the performances of the FHS,it is of great theoretical and practical importance to study its schedule issue.Based on the current research status of related issues around the world,the dissertation systematically analyzes the related scheduling methods of FHS and aims to perform in-depth investigation of the scheduling problem for FHS,the shortages of current research and several problems needing further investigation are pointed out.The research contents include the FHS scheduling approaches based on dispatching rules,rolling horizon strategy and distributed scheduling method based on multi-agent system.For all proposed FHS scheduling approaches,simulation study have been carried out by the simulation software.On the basis of the simulation study,the experimental tests have been carried out to verify some of the FHS scheduling approaches proposed in this paper.The detailed contents studied in the paper are as follows:(1)The research of the FHS scheduling approach based on dispatching rules.Dispatching rules are common scheduling approaches for the FHS with highly dynamics and randomness.In view of the fact that the current dispatching rules do not consider consistency of performing tasks of the same vehicle,Modified Nearest Vehicle First with Time Priority(MNVFTP) and Modified Dispatching rule based on Distance and Waiting time(MDDW) are proposed,they can reduce the distance of vehicle running unloaded,thus improve the FHS performance;In order to consider both system efficiency and quality of service,a Hybrid Dispatching rule(HD) is present.In this rule,when job intensity is high,Nearest Vehicle First(NVF) is used,otherwise an Adaptive Multi-attribute Dispatching rule(AMD) is adopted,the schedule approach can be adapted to the system dynamics by neural network predicting the weights of multi attributes periodically.For AMD,The distance between job and vehicle,job’s waiting time,and job related buffer status are simultaneously considered in the cost function,taking job-vehicle ratio,the jobs’waiting times balance factor and station buffer status balance factor as the inputs of the neural network,the weights of the distance between job and vehicle,job’s waiting time,and job related buffer status as the outputs of the neural network,the multi attributes’weights are adjusted periodically by the neural network,cost function value which is the basis of dispatching decision was calculated.Finally,dispatching decision was made by Hungarian algorithm based on cost function value matrix.Simulation results show that HD can improve the performances of FHS effectively and has good robustness under different conditions.(2) The research of the FHS scheduling approach based on rolling optimization strategies.The dispatching rules are easy to implement and can be earily adapted to the dynamics and randomness of FHS,but they belong to instant scheduling and lack of long-term planning,the dispatching rules can not use the pre-arrival information to serve the FHS scheduling effectively,so rolling optimization scheduling approach based on heuristic algorithm is present.A mathematical model which take minimizing the average job waiting time as the objective for FHS scheduling problem is established.we propose an algorithm which called Hybrid Genetic Annealing Algorithm(HGAA) to solve the scheduling problem,a rolling optimization approach based on HGAA is present. Simulation results show that the proposed scheduling approach can utilize the pre-arrival information to serve the FHS scheduling more effectively than the dispatching rules.In comparsion with the basic genetic algorithm in the process of scheduling,the HGAA shows better performances.(3) The research of FHS distributed scheduling approach based on Multi-agent Systems(MAS).Since the centralized scheduling approaches such as the dispatching rules and rolling optimization basd on mathematical model have the advantages of simple structure, easy to get overall coordination,etc,they have some problems of poor reliability,low efficiency in information transmission and difficult to applied in large-scale systems.To solve the problems,we propose a FHS distributed scheduling approach based on MAS,the The related elements of MAS such as the architecture of MAS model for FHS scheduling problem,the agents’ behaviors,consultation mechanism,bid evaluation,mechanism of special situations handling are designed carefully. We improve the performance of the proposed distributed scheduling approach by allowing vehicle agents exchange jobs and allowing job agent reauction conditionally.Through the comparision of the approach based on MAS and traditional scheduling approaches in an extensive simulation experiment,it is demonstrated that the proposed approach has better environmental adaptability,more stable performance under the different environments.(4)Experimental tests.The intelligent Rail Shuttle Vehicle(RSV) system served for the filter production line of a tobacco enterprise and the RSV control system of a company are introduced. Some scheduling approaches are implemented and tested.In comparsion with some common dispatching rules,the results show that the proposed HD scheduling approach can effectively improve the performances of the FHS.The research of this paper has promoted the interaction and inosculation among Simulation Technology,Operational Research,Artificial Intellegence and Control Technology.It is beneficial to the exploration for solving schedule problem of different types of FHS.The research results can provide decision supports for internal logistics system decision-makers of enterprises and have important theoretical value and applied engineering meaning for reducing enterprises’costs of material handling.