An Improved Q-learning Algorithm for Solving Integrated Process Planning and Scheduling Problems

An integrated process planning and scheduling (IPPS) is a key issue in improving manufacturing system efficiency and resource utilization, involving collaborative optimization of process planning and scheduling. However, traditional methods are difficult to deal with its complexity and are prone to fall into local optimization. This paper proposes an improved Q-learning algorithm (iQ-learning) to enhance adaptability to constraints by redefining state and action spaces and combining immediate reward and long-term penalty mechanisms. The improved algorithm considers short-term gains and global optimization, guides agents to make more reasonable decisions, and improves scheduling efficiency and optimization.