A learning and searching method for integrated process planning and scheduling

The integrated process planning and scheduling problem combines process route planning and shop scheduling to achieve high efficiency in manufacturing and maximize resource utilization, which is crucial for modern manufacturing systems. Traditional approaches, such as mixed-integer linear programming and heuristic algorithms, often struggle to balance solution quality and computational efficiency, particularly in large-scale scenarios. To address this, we propose learning-then-searching, an efficient framework combining deep reinforcement learning with heuristic search to solve integrated process planning and scheduling problems. We formulate the problem as a Markov decision process and develop a deep reinforcement learning method tailored to capture the complex relationships among operations, machines, and jobs. With the learned policy distributions, Monte Carlo tree search employs learned priors and problem-specific rollouts to further refine scheduling strategies, effectively balancing exploration and exploitation to achieve better solutions. Experimental results show that the proposed framework improves solution efficiency and quality over traditional methods across benchmark and large-scale instances, with pronounced advantages when route ambiguity is high. It also offers practitioners a flexible tradeoff between rapid policy-based decisions and further refinement through tree search when additional computation time is available.