Research On Dynamic Job-shop Scheduling Rule Decision In Complex Environment

As an important basis of the real economy,manufacturing industry is the foundation of our country and the foundation of our country.With the intensification of global competition,domestic manufacturing industry faces new challenges.With the development of new manufacturing concept,it is necessary for us to solve the problems of dynamic uncertainty,complexity and real-time response in the real manufacturing process.However,as the basis of manufacturing system,production scheduling optimization is the core of advanced manufacturing technology and modern management technology.Job shop scheduling problem is a kind of basic and complex production scheduling problem.Aiming at the above problems,this paper completes the following tasks based on dynamic Job shop scheduling problem.1.The feature selection of dynamic job shop scheduling problem is optimized.This paper improves the firefly algorithm for feature selection in dynamic job shop scheduling.First,the position of the firefly is updated by binary string coding and discretization.Then,in order to solve the problem that the firefly algorithm is easily trapped in local optimum,the mutation operation and Cauchy mutation step-size operation are used to achieve the balance.In addition,dynamic mutation factor is introduced to prevent premature convergence.Finally,the effectiveness of the proposed algorithm is verified by experiments.2.Aiming at the problem of mining dynamic job shop scheduling rules.In this paper,the genetic programming algorithm is improved.Firstly,the tree structure is used to represent the scheduling rules,which makes the genetic programming algorithm suitable for mining the rules of dynamic job shop scheduling.Then,a virtual objective function is set to evaluate the population diversity,which avoids the disadvantage of traditional genetic programming algorithm that is prone to local optimization.Finally,the effectiveness and robustness of combining scheduling rules to solve the dynamic job-shop scheduling problem are verified by simulation experiments.3.A complete reactive scheduling method based on reinforcement learning and combinatorial rules is proposed for dynamic job-shop scheduling with real-time production data in complex environment.Firstly,the job shop scheduling problem is transformed into a Markov decision process to initialize the dynamic job shop scheduling environment under complex environment.Then,a state space and a reward function are designed based on the combination rules obtained by the above rule mining method,and a dynamic scheduling mechanism is used to motivate agents to learn each other.Finally,the feasibility and effectiveness of applying reinforcement learning to dynamic job-shop scheduling are verified by simulation experiments.