Modeling And Nonlinear Control Of Variable Cycle Engine

In order to grasp the development of aviation power plant, variable cycle engine istookas the research project in this article, and the research isfocused on the physical structure, modeling and the nonlinear multivariable control techniques.Firstly, the physical structure is designed in this dissertation. The final scheme is determined through a comparative analysis, by adjusting the mode selector valve and the rear variable area bypass injector; the forward variable area bypass injector is adjusted passively. Engine fan is modeled as splition of fan tip and fan hub. Modeling of the mode selector valve and obtained the performance curve of the variable inlet guide vane is also studied. Based on Matlab Simulink, a component-level model isbuilt using the volume dynamics. By comparing with the data from NASA, the correctness and the effectiveness of the model areensured.Secondly, nonlinear model identification based on component-level model isstudied. First, identification based on NARX model isstudied, where sigmoid networks and waveletnetworks havebeen chosen to identify the model. The simulation result shows the identification accuracy heavily depends on the choice of parameters, andit is hard to improve the accuracy. Second, time series prediction based on dynamic neural network isstudied, the simulation shows that the maximum relative error is within0.7%, and this method has high accuracy but the analyticity is poor. At last identification based on Hammerstein-Wiener model isstudied, and the result is similar with the NARX model, they all need percipient parameters to improve the accuracy of the model. A recursive model based on Hammerstein-Wiener isproposed in this paper by summarizing all these methods above. Genetic algorithm isused to realize the parametric identification, and this method can not only improve the accuracy, but also makes the model havea certain analyticity, and the maximum relative error is below 0.5%. Simulation research in the whole envelope isstudied based on this method, and by analyzing the non-identification points data, envelope segmentation method isproposed.Finally, nonlinear control of variable cycle engine isstudied in this article. Generalized predictive control with minimized performance index isobtained to solve the optimal control problem. The controller isdesigned by using the Kalman filter and the saturation function module, and the influence of the parameters selection isalso studied. Finally a set of parameters ischosen to control the variable cycle engine. The result shows that comparing with the PID controller, this method is more effective.