Transition state performance optimization of propfan engine based on DDPG algorithm

Abstract

As the “heart” of an aircraft, aero-engines work in harsh environments of high temperature and high pressure for a long time. In order to ensure that the engine can operate safely and reliably within the entire flight envelope, a large safety margin needs to be reserved in the design of the control system. This design idea limits the full play of the engine performance, so it is necessary to carry out the research on the performance search control (PSC) of aero-engine. This paper studies the performance optimization control of propfan engine based on deep reinforcement learning algorithm. The Deep Deterministic Policy Gradient (DDPG) algorithm, which is more suitable for continuous action space, is used to optimize the acceleration process of the propfan engine. The simulation results show that, compared with the unoptimized adjustment process, the transition state adjustment time of the engine is reduced by 32.5% under the action of the optimal adjustment law obtained by the DDPG algorithm. Therefore, the DDPG algorithm can be applied to the performance optimization of the engine acceleration process, and has a good transition state performance optimization effect.