航空发动机模型参考自适应控制

Aircraft Engine MRAC Based on Elman Neural Network

针对航空发动机数学模型的复杂、多变,建立了以Elman反馈神经网络为基础的模型参考自适应控制系统。为提高控制系统的响应性能,神经网络控制器和辨识器同时采用动态Elman网络,利用动态反向传播(DBP)算法实现参数的在线调整和辨识,并利用李雅普诺夫函数对算法的收敛性进行了证明。采用某涡喷发动机的地面模型和高空模型作为控制对象进行了大量的仿真研究;结果表明:此控制系统具有自适应能力强、响应速度快、稳态误差小等优点;理论分析与仿真结果一致,证明算法和结果是正确有效的,对复杂动态系统控制具有参考价值。

In order to deal with the complexity and variability of the math model of aircraft engine, a self adaptive system of control based on Elman neural network is established. For improving the response results, the DBP of Elman network is adopted simultaneously in NNC and NNI to realize online adjusting and identification of parameters and the convergence of its calculation is testified by Lyapunov Function. The thesis further simulates the control of aircraft turbojet engine of both on - the - ground and in - the - air model with step signal as input, which proves several advantages of this control system such as better adaptive capability, sensitive response and minimal stable error. And theoretic analysis and simulation results are tallied, which prove that the arithmetic and result are correct and effectual, can provide reference for controlling complex dynamic system.