Aero-engine life limit parts replacement policy optimization: Reinforcement learning method

摘要

An optimization method for aero-engine life limit parts (LLPs) replacement policy is proposed based on reinforcement learning method, aiming at optimizing the aero-engine LLP-s replacement policy. In the proposed LLPs replacement policy optimization method, the real-life LLPs replacement rules are adopted as the constraints and the minimum long-term LLPs replacement discount cost is regarded as the optimization objective. In reinforcement learning framework, the Q-learning algorithm is adopted to optimize the LLPs replacement policy. Compared with the traditional methods, the proposed optimization method is simple in structure, and it can achieve better optimization results. To validate the proposed aero-engine LLPs replacement policy optimization method, the LLPs list of a civil turbofan aero-engine is adopted as the sample data. And the existing particle swam optimization algorithm is adopted as the comparative experimental method. The comparison experiment results show that the proposed LLPs replacement policy optimization method achieves obvious advantages. The proposed optimization method is able to provide decision-making supports for aero-engine LLPs replacement.