Application of Machine Learning and Rule Scheduling in a Job-Shop Production Control System
Yanchun Zhao, H. Zhang
Abstract
As intelligent and precision manufacturing becomes the trend of industrial production, it is of practical significance to study the job-shop production control. However, the existing studies have not provided an evaluation mechanism to reasonably measure the control efficiencies of different plans. The desired control objectives are not easily achieved for job-shop production control problems with dynamic changes. Therefore, this paper probes into the dynamic job-shop production control problem based on deep reinforcement learning and rule scheduling. Firstly, a multi-objective optimization model was established for the production control system of dynamic job-shop. Then, deep reinforcement learning was introduced to job-shop production control system to transform the dynamic job-shop production control problem. After that, the authors proposed a dynamic job-shop production control method based on deep reinforcement learning, and explained the collaboration strategy for multiple subsystems. The proposed method was proved effective through experiments.