Optimization Method for the Propagation Path of Complex Product Design Change Based on Reinforcement Learning

Design changes are a common situation in the update and replacement of complex products, but the parts of complex products have characteristics such as large quantity, high coupling degree, and complex assembly relationships, which leads to significant decision-making difficulties for designers when a part has a design change. To address the above problems, this paper proposes a reinforcement learning-based method for optimizing the propagation path of complex product design changes. Taking parts as state nodes and design change propagation impact as actions, the reward function is determined by considering three objectives: product change time, change cost, and assembly difficulty. A Markov decision process (MDP) model for designing change propagation path optimization is constructed, and Monte Carlo reinforcement learning is used to solve the model. This method provides optimal design change propagation paths for designers. Finally, a robot suspension system design change is used as an example to verify the feasibility and effectiveness of the method.