Nonlinear Model Predictive Control For Turbo-Shaft Engine

With the development of new generation of turbo-shaft engine,more requirements of its control system are required.The coupling dynamic effects between engine system and helicopter system also increase,the traditional casecard PI controller has been unable to meet the requirements of engine control system because of the effect of delay characteristic of the rotor.Based on the predictive model,nonlinear model predictive control has ability to realize the optimal control,which can deal with the system delay and limit constraints in comparision with the casecard controller.Thus,the integrated helicopter/turbo-shaft engine simulation model is eastablished,and nonlinear model predictive control of turbo-shaft engine is researched based on the integrated platform.In order to describe the coupling characteristic of helicopter system and the time delay of the helicopter rotor system,combining the nonlinear aerodynamic and thermodynamic model of turbo-shaft engine with the nonlinear aerodynamic model of helicopter,the helicopter/turbo-shaft engine integrated simulation model is built based on the helicopter and turbo-shaft engine characteristic data,which provides a simulation platform for the design of tubo-shaft engine control system.A nonlinear model predictive control method based on multi-output recursive reduced kernel extreme learning machine algorithm is proposed.Aiming at the shortage of generalization ability and sparseness of large training samples for kernel extreme learning machine,the multi-output recursive reduced kernel extreme learning mahine algorithm is presented,which can effectively simplify the model structure and ensure the accuracy of the model.The proposed algorithm is used to identify the multi-output predictive model of turbo-shaft engine,and the nonlinear model predictive controller is designed based on sequential quadratic programming.The simulation results show that the model predictive controller can effectively decrease the overshoot or droop of power turbine rotor speed,and the control system has better dynamic response performance.A nonlinear model predictive control method based on multi-output online sliding sequence kernel extreme learning machine algorithm is proposed to deal with the lack of robustness problem of offline kernel extreme learning machine algorithm.Combining the approximate linear dependency anlaysis,the online training samples are selected to identify predictive model of turbo-shaft engine,the structure of model is simplied and the precision is guaranteed.Nonliner model predictive control of turbo-shaft engine is researched based on the online kernel extreme machine algorithm.The simulation results show that the proposed controller has faster response performance and better steady response performance with the speed of calculation was guaranteed.A robust H_?control for helicopter flight control based on the guardian maps theory was put forward is researched.Combined with the nonlinear model predictive control for turbo-shaft engine,the intergrated control for helicopter/turbo-shaft engine system is designed.Based on the integrated simulation platform of helicopter with turbo-shaft engine,the maneuvering flight simulation results show that the proposed controller has better control quality,and the overshooting of power turbine speed is smaller compared with the traditional cascade PI controller.