Research On AGV Scheduling In Hybrid Flow Shop

Compared with the general serial flow shop,the hybrid flow shop has a higher production capacity and is extensively used in food,electronics,textile,and other industries.With the comprehensive promotion of Made in China 2025,the shortening of product life cycles,and the increase of labor costs,enterprises have a surging demand for introducing automated transportation equipment and strengthening the efficient coordination operation of production and logistics.As key transportation equipment,AGV can replace the manual handling of line-side material,which can ensure the flexibility and timeliness of the production system.Especially in the hybrid flow shop,the area of the material transportation network accounts for a large proportion.Therefore,it is of great theoretical reference and engineering application value to study the AGV scheduling problem in the hybrid flow shop,so that AGV can cooperate with the production system efficiently and improve the production efficiency of the workshop.This thesis studies the AGV path planning problem,the power management problem,and the task scheduling problem in the hybrid flow shop,and the main study contents are as follows:(1)For the path planning problem of AGVs in a hybrid flow shop,this thesis designs an improved A*algorithm,proposes the dynamic detection and resolution of conflicts and deadlocks,and perfonns the global and local path planning of AGVs respectively.The grid method is used for the creation of the workshop environment model,and the improved A*algorithm is designed to solve the problem of many path inflection points and long computation time when solving large environment maps for global static path planning of AGVs.To solve the conflict and deadlock problems that occur dynamically in local road networks,the deadlock dynamic detection algorithm based on segment reservation strategy and the deadlock dynamic resolution algorithm based on the Breadth First Search are proposed.Simulation experiments show that the improved A-algorithm outperforms the A*algorithm,the Ant Colony algorithm,and the Dijkstra algorithm in terms of solution time and results.The proposed conflict and deadlock dynamic detection and resolution method,which can avoid conflict,and dynamically detect and release deadlocks,is more widely applicable and faster than the coupled A*algorithm,WHCA*algorithm,and delayed collision avoidance strategy.(2)For the problem of AGV power management in a hybrid flow shop,this thesis proposes an automatic charging model for AGVs and uses an improved particle swarm algorithm for solving the model to schedule the charging tasks of AGVs and increase the available production time of AGVs.The widely used industrial threshold method only considers the power threshold problem,and cannot decide a more reasonable charging time,location,and duration according to the tasks that AGVs need to perform.Therefore,this thesis establishes an automatic charging model to minimize the total time for AGVs to perform their tasks.Considering the multivariable and nonlinear constraints in the model,this thesis designs an improved particle swarm optimization algorithm to optimize the model.The experimental results show that,in comparison with the industrial threshold method,the proposed charging model and algorithm are not affected by the initial power of AGVs and can adjust the power limit range according to the actual needs of enterprises,significantly reducing the total task execution time and having a strong power management effect.(3)For the AGV task scheduling problem in a hybrid flow shop,this thesis constructs an integrated task scheduling model of machines and AGVs and proposes a discrete water wave optimization algorithm based on a genetic algorithm to solve the model.The integrated task scheduling model takes the workpiece processing sequence,machine selection,and AGV task assignment as decision variables and aims to minimize the maximum completion time.Based on the genetic algorithm to discretize the water wave optimization algorithm,the AGV tasks are allocated based on the market bidding method with full consideration of the road network conflict deadlock and AGV power management,thus solving the combined optimization problem of integrated scheduling of hybrid flow shop.Three experiments are designed:Experiment 1 verified that the discrete water wave optimization algorithm based on a genetic algorithm is superior to the other four intelligent optimization algorithms commonly used in job shop scheduling,which can obtain better integrated task scheduling results and has strong robustness;Experiment 2 verified that the AGV task allocation method based on market bidding has a strong ability to adapt to complex problems in largescale tasks,and the solution results and calculation time are better than the brute force method and genetic algorithm;Experiment 3 verified that the proposed method of considering AGV’s conflict deadlock and power management in the task allocation process has a smaller maximum completion time than the threestage scheduling method.In summary,this thesis designs a hybrid flow shop AGV scheduling scheme from three aspects to ensure the fastest arrival of AGVs at their destinations without conflicts and deadlocks,to schedule the charging tasks of AG Vs to be executed at the optimal time and place,to perform integrated scheduling of production processing and material transportation,to achieve the goal of improving workshop productivity,and to reduce the risks that may occur when enterprises apply automated transportation systems.