基于神经网络的旋转机械故障诊断研究_陈长征.pdf

2 2 2 　　　　　　 Vol.22 No.2 2000 6 JOURNAL　OF　MECHANICAL　STRENGTH June 2000 FAULT DIAGNOSIS FOR ROTATING MACHINERY BASED ON NEURAL NETWORKS 　　 　　 (沈阳工业大学 建筑工程系, 沈阳 110023)　(东北大学 设备诊断工程中心, 沈阳 110006) CHEN Changzheng (Department of Architecture Engineering , Shenyang University of Technology , Shenyang 110023, China) Z HANG Sheng 　YU Heji (Center of Equipment Diagnosis, Northeastern University , Shenyang 110006, China ) 　, 。 , 。 ;, 、、 。 , , 。 , , , , , 。 　　　　　 　TP206.3　TP18 Abstract　Generally, it is difficult to determine in advance a suitable network structure w en a multi-layer perceptron neural net- works is used for a special fault diagnosis problem.For t is reason, t e fault diagnosis and pattern recognizing ability of multi-layer per- ceptron was analyzed, t e met od of increasing or decreasing t e numbers of idden layer and idden unit was adopted to determining t e structure (including t e numbers of idden layer and idden unit)of t e neural networks.T e fault diagnosis met od based on t is ap- proac for rotating mac inery was studied, and t e intelligent fault diagnolsis system was developed.T e system consist of networks learning, fault diagnosis, data management and result inquiring.T e knowledge sub-c unkwas adopted and used in it;system asfriend interface, powerful interc ange and reliable diagnosis results.T is system was used in fault diagnosis of t e blower, it s ows t at t is system is convenient to manipulate wit precise result and powerful intelligence, and in some case t e diagnosis can automatically be car- ried out.It is p