Control law design of variable cycle engine based on DQN

In order to solve the on-line optimization problem of variable cycle engine, a control law of variable cycle engine based on depth Q network (DQN) is proposed, the minimum fuel consumption of variable cycle engine in cruise phase is achieved by reinforcement learning algorithm In this paper, the con...

Full description

Saved in:
Bibliographic Details
Published in2020 3rd International Conference on Unmanned Systems (ICUS) pp. 937 - 941
Main Authors Nie, Lingcong, Mu, Chunhui, Yin, Zhongjie, Jiang, Weiyu
Format Conference Proceeding
LanguageEnglish
Published IEEE 27.11.2020
Subjects
control law
depth certainty strategy
Engines
Fuels
Optimization
Regulation
Reinforcement learning
variable cycle engine
Online AccessGet full text

Cover

More Information
Summary:In order to solve the on-line optimization problem of variable cycle engine, a control law of variable cycle engine based on depth Q network (DQN) is proposed, the minimum fuel consumption of variable cycle engine in cruise phase is achieved by reinforcement learning algorithm In this paper, the control law of the variable cycle engine in the cruise phase is constructed based on the reinforcement learning algorithm, the pressure ratio can be adjusted online to change the engine working point, so as to reduce the fuel consumption. The unit fuel consumption rate of subsonic cruise can be reduced by 7.45%, and the self-learning time from pulling deviation to fuel consumption can be reduced to the lowest 80s. The performance optimization algorithm can realize the selfregulation of engine control law and achieve the lowest fuel consumption with the existence of the engine performance deviation.
DOI:10.1109/ICUS50048.2020.9274888