The power turbine rotor speed controller based on internal model principle was designed for turbo-shaft engine.To suppress the disturbance of the torque variation of the main rotor to the power turbine rotor speed

a torque prediction method based on extreme learning machine was proposed.The training of extrem learning machine(ELM)was based on the data of dynamic simulation

and the input variables were obtained by correlation analysis approach.The internal model based on power turbine rotor speed controller was designed by pole placement.The internal model of input signals was embedded into the controller to obtain robust tracking.In order to achieve effective compensation of the engine's load variation

the proportional-differential feed-forward compensator was adopted.The simulation results indicate that the torque prediction based on ELM has excellent accuracy

and the relative prediction error is less than 1.5‰.Compared with the control without torque feed-forward

the proposed control method reduces both the overshoot and droop of the power turbine rotor speed more than 30% during maneuvering flight.