Abstract: A hybrid predictive control scheme for a turbo-shaft engine is proposed by combining the cascade PID control with the nonlinear model predictive control (NMPC). Through adding the predictive controller, i.e., the engine compensator, to the cascade PID controller in the main loop, the engine control system has a better control quality. The predictive controller is composed of an online predictive model and a rolling optimizer. Satisfactory model accuracy within 3‰ is achieved by online sliding least squares-support vector regression (OSLS-SVR), and the rolling optimizer is implemented using the sequential quadratic programming (SQP) algorithm. Finally, based on an integrated UH-60A helicopter/T700 turbo-shaft engine simulation platform, a large number of bob-up and bob-down flight simulation is carried out to verify the proposed control mode. The results show that the hybrid predictive control has better dynamic disturbance rejection ability and robustness, which enables the helicopter to exhibit much greater maneuverability than the conventional control method does.

Key words: turbo-shaft engine, control system, hybrid control, predictive control, support vector machine