05 - Propulsion and Energy

INTELLIGENT FULL-ENVELOPE OPTIMIZATION OF TRANSIENT CONTROL LAWS FOR ADAPTIVE CYCLE ENGINES VIA TRANSFER LEARNING-ENHANCED DEEP REINFORCEMENT LEARNING

M. Qiao¹, Y.Z. Huang, Shenzhen Research Institute of NPU , China; L.F. Gou¹; ¹Northwestern Polytechnical University, China

This work proposes a closed-loop full-envelope optimization method for transient control laws of adaptive cycle engine using deep reinforcement learning and transfer learning. By reusing control knowledge across operating conditions, the method reduces training cost while maintaining safe transient performance. It is also applicable to other complex industrial systems with wide operating envelope.

View Paper

DOI: 10.71945/icas2026_0584