基于自适应观测器的非线性系统鲁棒故障诊断.pdfVIP

of tile24tfl Chinese Proceedings ConbolOonference Guanf鲫ou,P．R．Chitin 15童2005 July 基于自适应观测器的非线性系统鲁棒故障诊断 罗小元，关新平，佘君 (燕山大学电气工程学院，秦皇岛，066004) E-mail：xyluo@ysu．edu．cn 摘要：针对一类特殊的不确定非线性动态系统，本文提出了一种基于神经网络在线估计的鲁棒故障检测和诊断 方法。该方法通过在逼近系统反馈回路中采用神经网络在线逼近器来提取非线性故障信息而实现系统故障检测 和诊断。通过设计和分析验证该方法对系统中存在的建模误差和外部扰动等不确定性因素具有良好的鲁棒性。 最后，针对柔性机器人用三相直流电动机常见故障的仿真研究验证了该故障诊断方法的有效性。 关键词：非线性，神经网络，自适应观测器，故障诊断，鲁棒性 Robustfault basedonon—line diagnosis adaptive for nonlinear approximatorsystems Luo Jun Xiaoyuan，GuanXinping，She (Institute ofElectricalEngineering,YanShan University,Qinhuangdao，066004) Abstract：Inthis robustfaultdetectionand which on networkon．1ine iSbasedneural paper．a diagnosis(FDD)scheme be to foraclassof can able adaptive nonlinearuncertain proposed．Thescheme approximator systems，is proposed realize iflthe the FDD theneuralnetworkfeedbaekto thenonlinearfaultinformationof byusing pathcaptureonly estimated is thatthe schemehas robustnessuncertainties system．Itproved proposed good against includingmodeling errorand a was flexiblelink unknownexternal simulationof CUrrentmotor．whichutilizedin