研究故障观测器优化设计,针对一类非线性动态系统,在考虑系统的输入输出包含外部扰动及建模误差等不确定性项[1]的情况下,为了提高所设计观测器对系统数学模型的在线跟踪能力从而进一步提高故障诊断的鲁棒性减少系统的误报警率,提出了基于模糊神经网络的诊断方法.利用神经网络以及模糊系统对非线性函数的无限逼近能力,设计了基于T-S模糊模型[2]的神经网络自适应观测器来拟合系统的非线性模型和系统的非线性故障特性.由Lyapunov稳定性方法获得调整观测器权重的规律.对所用改进方法的收敛性进行了证明,并通过仿真实例说明了诊断方法的有效性和使用性.%Study fault diagnosis method based on neural network fuzzy model for a class of nonlinear dynamic system. The inputs and outputs of the system include external disturbance and modeling errors and other uncertain parts. In order to improve the observer online tracking ability and further improve the robustness of the fault diagnosis, a diagnosis method based on fuzzy neural network was proposed. Neural network and fuzzy system can infinite approximate nonlinear function, the neural network adaptive observer based on T—S fuzzy model was utilized to estimate the unknown model in the nonlinear system, simultaneously, to estimate the fault value. The method of turning the weights of the observer was realized based on Lyapunov stability theory. The stability of the system was analysised in detail. A simulation example was given to illustrate the effectiveness of the approach.
展开▼
