具有干扰解耦能力的鲁棒在线故障诊断

Robust On-Line Fault Detection with Disturbance Decoupling Ability

针对一类带有未知输入干扰和模型不确定性系统 ,提出了一种新型鲁棒在线故障诊断方法 .该方法综合应用带有未知输入状态观测器技术和 RBF神经网络的在线建模能力 ,利用状态观测误差实时调整 RBF神经网络的权值 ,使其不但能在线实时检测、分离、估计故障信号 ,而且对未知输入干扰具有解耦能力 ,对有界模型不确定性具有鲁棒性 .给出了该方法的鲁棒性和灵敏度的分析结果 .仿真结果表明了该方法的有效性 .

A new robust on-line fault detection method for systems with unknown input disturbance and model uncertainty was proposed. It uses the technology for unknown input observer design and on-line modeling ability of RBF neural network comprehensively, and by adjusting the weight of RBF neural network, it can not only real-time detect, isolate and estimate the fault signal, but also has both the ability of decoupling unknown input disturbance and the robustness to the bounded modeling uncertainty. Analysis results for robustness and sensitivity were also given. The simulation result indicates that this method is effectiveness.