eduzhai > Applied Sciences > Engineering >

A Reinforcement Learning Approach for Transient Control of Liquid Rocket Engines

  • Save

... pages left unread,continue reading

Document pages: 14 pages

Abstract: Nowadays, liquid rocket engines use closed-loop control at most near steadyoperating conditions. The control of the transient phases is traditionallyperformed in open-loop due to highly nonlinear system dynamics. This situationis unsatisfactory, in particular for reusable engines. The open-loop controlsystem cannot provide optimal engine performance due to external disturbancesor the degeneration of engine components over time. In this paper, we study adeep reinforcement learning approach for optimal control of a genericgas-generator engine s continuous start-up phase. It is shown that the learnedpolicy can reach different steady-state operating points and convincingly adaptto changing system parameters. A quantitative comparison with carefully tunedopen-loop sequences and PID controllers is included. The deep reinforcementlearning controller achieves the highest performance and requires only minimalcomputational effort to calculate the control action, which is a big advantageover approaches that require online optimization, such as model predictivecontrol. control.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...