Research On Switching Models Control Of Turbofan Engines

Because turbofan engine is an aerodynamic thermodynamic system with strong nonlinearity,and the research of turbofan engine is closely related to the system model,so it is very important for the design of control algorithm and actual engineering practice to find an applicable engine model.In different working environment and working conditions,the system characteristics of engine have significant differences.Therefore,this thesis establishes switched system model for the.engine,and designs the controller by using this model.Based on the nonlinear equilibrium manifold expansion(EME)model of engine,this thesis offers an on-line parameter identification algorithm,and using the results to design a feedback linearization controller.The main contribution of this thesis is as follows:(1)The thesis designs the controllers which meet the requirements for the linear switched model.A controller is designed to stabilize the engine under arbitrary switching law according to stability conditions of switched systems.The mixed sensitivity method is also used to design the dynamic controller based on frequency-domain index considering the disturbance problem.The H? disturbance attenuation problem is studied using the linear matrix inequality(LMI)method.A dynamic output feedback controller is given.(2)The parameters of the EME model are identified,and using the result,a controller is designed for the model by the feedback linearization method.This thesis presents a finite-time parameter estimation method and on-line parameter estimation method for the EME model.Theoretically,the former estimator can provide real estimation,while in reality,estimation error always exists due to the presence of disturbance,and this estimation error is also related with the estimation time.On the other hand,the estimation accuracy of the later estimation method is independent with the estimation time.Hence,in order to improve the estimation speed and estimation accuracy,the finite-time parameter estimation result can be used as the initial value of the on-line parameter estimation.