Development And Research Of Fault Rotor Diagnosis System Based On Wavelet And Neural Network

Rotating machinery is widely used in engineering.Timely and effective fault diagnosis is very important for preventing accidents and reducing property losses.The early mechanical fault diagnosis method adopts artificial diagnosis,which is relatively primitive but very basic diagnostic form.At present,it is still used in practical engineering.With the development of artificial intelligence,electronic technology and computer technology,the mechanical vibration fault diagnosis provides the prerequisite for the development of automation and intelligence.This paper built around research extraction method and BP neural network,wavelet packet feature vector of mechanical vibration signal,using the graphical programming.language Lab VIEW programming features a unique and powerful test capabilities,develop a set of fault diagnosis system.The main contents of this paper include:(1)By analyzing the characteristics of wavelet packet function,the wavelet function used to decompose and reconstruct the signal by wavelet packet function is determined.The wavelet packet decomposition and reconstruction of signals are carried out by Lab VIEW,and the correctness of wavelet packet decomposition and reconstruction scheme and the arrangement order of each frequency band are verified by simulation signals.From the energy point of view,the wavelet packet is used to extract the feature vector of the original signal and normalize it.(2)This paper studies the use of Lab VIEW to establish the BP neural network,then establishes the establishment method of input layer,hidden layer and output layer and the number of neurons needed.This paper optimizes the performance of neural networks through the assignment algorithm of weight matrix and the introduction of momentum terms,and trains the data that has been extracted.(3)The overall framework of the development of the fault diagnosis system and the development process are studied,and the software is developed by using Lab VIEW.The system mainly includes three modules:data acquisition and storage,wavelet packet analysis and neural network.,The wavelet packet analysis module includes the wavelet packet extraction feature to the quantum module and the use of the extracted feature vector to establish a fault database module which can be updated and extended at any time.The neural network module contains neural network training sub module and fault identification sub module.(4)A multi span rotor test bench is built to simulate faults,and the adjustable speed control is realized.Through the multi span rotor test-bed,the normal working conditions and the three kinds of faults,such as looseness of bearing,misalignment of rotor and rub impact,are simulated.Followed by the sensor,preamplifier,NI data acquisition platform will be uploaded to the host computer science after conditioning the signal,the fault diagnosis system for analysis,processing and fault diagnosis of the signal in the host computer,to verify the accuracy and effectiveness of the system.