A Deep Reinforcement Learning Method for Orthogonal Turn-Milling Parameters Optimization

Turn-milling is an advanced manufacturing technology which has been widely applied to various industries due to its high-speed and interrupted cutting. However, there are few literatures studying on the multi-objective optimization of turn-milling, which can effectively improve overall performance of turn-milling to realize profit maximization. In this paper, we presented a multi-objective optimization method based on deep reinforcement learning (DRL) to solve the multi-objective optimization problem of orthogonal turn-milling parameters. This method adopts a decomposition strategy to convert the optimization problem into multiple subproblems of single objective. Then all subproblems are modelled by neural networks, and collaboratively optimized by means of a DRL algorithm and a neighborhood-based parameter transfer strategy. Finally, the Pareto optimal solutions are obtained from these optimized neural networks by means of the greedy strategy. Extensive experimental results show the effectiveness and competitiveness of the proposed method for solving the multi-objective optimization problem.