Automated-Guided Vehicle(AGV)is a flexible material handling device that is automatically steered to accomplish its assigned task.AGV has been widely used in manufacturing systems and attracted a lot of attention from researchers.The efficient operation of manufacturing system involving AGV requires optimizing conflict-free AGV path and job-shop scheduling simultaneously.AGV path planning and job-shop scheduling were proved to be NP-Hard problems,traditional algorithms have some drawbacks such as high computational cost,local minima problem and poor efficiency when applied to address high-dimensional problems.To overcome such limitations,this thesis focuses on new intelligent algorithms for AGV path planning and the integrated scheduling of job-shop with AGV,including:(1)The AGV path planning problem is described briefly.The path length,the path smoothness and path collision times with obstacles are considered as three sub-objectives.Simple addictive weight method is used to map the sub-objectives into a single objective.Due to the merits of Grid map method,such as simplicity,insensitivity for the shape of obstacles and decision variable deduction,it is exploited for the environment modeling in AGV path planning.(2)The Genetic Algorithm(GA)with traditional Elitism Tactic strategy is enhanced with Grey Wolf Optimization(GWO)to prevent population diversity decreasing.To maintain population diversity in the later stage of the evolution,the self-adaptive mutation and crossover probability adjustment strategy based on the entropy of chromosome is proposed,when population diversity gets worse,mutation and crossover probability will automatically increase.To prevent GA being trapped into stagnation and increase the ability of GA in finding the global optimum with better convergence speed as well as the solution accuracy for solving AGV path planning problem,neighborhood mutation operator and path fine-tuning operator are developed.As the basic GWO easily suffers from the premature,to improve exploration and exploitation capabilities of GWO in solving AGV path planning problems with complex environment,the neighborhood mutation operator and path fine-tuning operator are introduced to its architecture.An AGV path planning platform which is improved by the proposed algorithms is developed by MATLAB GUI tools.(3)For verifying the proposed methods,its performance is compared to traditional GA with single or multi-population strategy under three different complex static environment.The experimental results confirm that improved GA and GWO perform better than its rivals.To verify the performance of the improved GA in solving AGV path planning with complex dynamic environment,we conduct experiments with three complex unknown environments which contain three dynamic obstacles,experimental results show the superiority of the proposed method.(4)To enhance the performance of Teaching-Learning-based Optimization(TLBO)in solving job-shop scheduling,student learning reflection operator motivated by Particle Swarm Optimization(PSO)is proposed as a feedback phase in TLBO.To improve the quality of the global best solution gradually,an effective self-adaptive initialization strategy is proposed based on Chromosome Similarity Matrix(CSM)which evaluates the diversity of population.Experimental results reveal the effectiveness of the proposed method.Flower Pollination Algorithm(FPA)is a new population-based intelligent algorithm inspired from flower pollination behavior.To avoid FPA being trapped into the local optima,mutation operator and crossover operator are introduced into the basic FPA.To balance the proportion of local pollination and global pollination better,the Dynamic Switching Probability Strategy(DSPS)is adopted.Improved FPA exhibits better performance in solving job-shop scheduling problem.(5)The prototype platform of integrated scheduling of job-shop and AGV were developed to confirm the effectiveness of AGV path planning algorithm,job-shop scheduling algorithm and integrated scheduling of job-shop with AGV. |