Wear Fault Dynamic Analysis And Fault Diagnosis Research Of Reciprocating Machinery

Reciprocating machinery is widely used in all kinds of industrial fields, on which the fault diagnosis is one of study focuses in domestic and overseas. Many people do some work on Reciprocating machinery fault diagnosis , but the level on Reciprocating machinery fault diagnosis is low and mismatch with the present situation on the application for its complex structure and multiple activation. So it's necessary to reinforce reciprocating mechanical fault diagnosis research.The wear fault dynamic model of reciprocating membrane pump is built, which based on the contact-separation state contact force model. The wear fault kinetic equation is solved by the Runge-Kutta method. The wear fault dynamic response for slider-crank of membrane pump is researched on the different parameter. The Poincare mapping is rendered , which proves that the movement of membrane pump on the wear fault is chaotic motion.To solve the problem of reciprocating mechanical fault diagnosis, the fault diagnosis method of correlation dimension and Lyapunov exponent is researched based on the fractal and chaos theory. Three different membrane pump simulation signal which is wear fault is analyzed. For three different signals, the correlation dimension is calculated by the G-P algorithm and the Lyapunov exponent is calculated by the wolf algorithm. For three different signals, the correlation dimension and Lyapunov exponent is different, so it is easy to judge the fault state by the correlation dimension and Lyapunov exponent. This research shows that the method based on the correlation dimension and Lyapunov exponent is effect in the field of reciprocating mechanical fault diagnosis.The other development direction of Equipment fault diagnosis technology is integration with the latest signal processing methods and modern methods of intelligent. This paper introduces EMD technology and the AR model and RBF neural network and proposes the RBF neural network fault diagnosis method based on the EMD-AR model. At last, taking rolling bearing of reciprocating compressor as an example, the RBF neural network fault diagnosis method based on the EMD-AR model is applied on reciprocating mechanical fault diagnosis, the result is shows that the test samples state is identified and this method is effect to reciprocating mechanical fault diagnosis. The RBF neural network fault diagnosis method based on the EMD-AR model supplies a new method to reciprocating mechanical fault diagnosis.