Design and Realization of Intelligent Aero-engine DDPG Controller

Abstract As the artificial intelligence technology advances, intellectualization has become an important development trend for future aero-engine industry. The intelligent will bring a new round of revolution in aviation Since is complex system with strong nonlinearity and coupling, it difficult to achieve optimal control performance traditional PI control. Therefore, DDPG (Deep Deterministic Policy Gradient, DDPG) algorithm belonging family Deep Reinforcement Learning hereby proposed designing controller. Based on non-linear polynomial state-space mathematical model JT9D turbofan engine, controller designed then compared can effect through training neural network. simulation results demonstrates that control, method this paper achieves superior response speed, overshoot anti-interference ability.