Seagull Optimization Algorithm Combining Reinforcement Learning And Non Dominant Sorting And Its Application In FJSP

The manufacturing industry is the lifeblood of the national economy.For the manufacturing industry,deepening the integration of information technology and manufacturing industry,and improving resource utilization efficiency are urgent matters.Therefore,in-depth research on workshop scheduling problems is becoming increasingly important.Developing scheduling plans suitable for workshop processing is beneficial for reducing costs and increasing efficiency,and improving resource utilization.This article focuses on the common flexible job shop scheduling problem in the manufacturing industry and proposes a seagull optimization algorithm bases on fusion reinforcement learning to optimize and solve the problem;Further consideration is given to the multi-objective flexible job shop scheduling problem with machine failures,and a multi-objective improved seagull optimization algorithm and dynamic rescheduling framework are proposed.The main research contents of this paper are as follows: Firstly,according to the shortcomings of seagull optimization algorithm,chaos mapping and Reverse learning are used to establish the initial population,and an adaptive local search model based on iteration is designed,and an improved seagull optimization algorithm(ISOA)is designed.Furthermore,on the basis of ISOA,by integrating the Q-learning algorithm and the non-dominated sorting algorithm,the Q-learning fusion seagull optimization algorithm(QSOA)and the multi-objective improved seagull optimization algorithm(MO-ISOA)based on the integrated non-dominated sorting were proposed respectively.Then,the performance tests of QSOA and MO-ISOA were conducted respectively,which verified the superiority and effectiveness of two types of improved seagull optimization algorithms.Secondly,QSOA is used to solve the flexible job shop scheduling problem with the goal of minimizing the maximum completion time.Two sets of simulation experiments are conducted to simulate the flexible job shop scheduling problem,and the experimental results are analyzed in detail.The results show that when solving the flexible job shop scheduling problem,the completion time of QSOA solution is small,and the optimal solution can be achieved through a few iterations.Meanwhile,two sets of progressive simulation experiments have demonstrated that QSOA has the ability to improve the ability to solve a type of scheduling problem by training and learning optimization experience through partial scheduling cases.Finally,facing the multi-objective dynamic scheduling problem with machine failures in the workshop production process,MO-ISOA was applied to solve it,and an event driven dynamic rescheduling framework was designed.The effectiveness of the MO-ISOA algorithm in solving multi-objective problems was demonstrated through experiments,and tests were conducted in dynamic scheduling scenarios to verify the algorithm’s rescheduling ability in the event of machine failures.