航空发动机鲁棒自适应反演控制

摘要

针对航空发动机非线性和不确定性的特点,提出了一种基于神经网络的多输入多输出反演控制方法.采用径向基神经网络逼近系统中的不确定性,在控制中引入自适应鲁棒项,以克服系统中不确定性的影响.在递推过程中,虚拟控制量和实际控制量的求取始终基于Lyapunov稳定性原理,从而保证了闭环系统的一致渐近有界.最后针对某型涡扇发动机非线性模型设计了转速控制器.仿真结果验证了该方法的有效性.%A multi-input/multi -output ( MIMO) backstepping control strategy based on neural network was presented in view of nonlinearity and uncertainty of aero-engine. Radial basis function (RBF) neural networks was used as approximation models for the uncertain of the system, and overcome the uncertainty by introduce a adaptive robustness. Using Lyapunov stability analysis, uniform boundedness of the MIMO nonlinear control system is proved, and simulation results for a certain type of turbofan engine further validate the effectiveness and performance of the proposed control method.