Rolling Bearing Fault Diagnosis Based On Wavelet Transform And PNN Neural Network

Rolling bearing is the most widely used mechanical and electrical equipment mechanical, is one of the most easily damaged. With the progress of modern industry and science and technology, high-speed centralized and large-scale rolling bearing is moving. Rolling bearing fault will cause the device to decrease or loss of function, or even cause the entire device is not working properly, causing huge economic losses. And causing serious accidents. Rolling lifetime prevalence for discrete, high fault-occupancy ratio, a wide range of issues such as failure of the working status of the rolling bearing vibration signal acquisition, monitoring, analysis and processing in order to identify the characteristics of vibration signals, selecting a appropriate signal processing method and neural network fault diagnosis, with good economic and scientific significance.This study based on the vibration signal of the rolling bearing vibration characteristics, fault mechanism, fault signal characteristics. First, used wavelet denoising processing the signal, improved the data SNR, then wavelet packet decomposition to obtain each wavelet packet frequency band for the processed data, and then normalized the data, realized the feature vector extraction of rolling bearing inner ring, outer ring and a rolling body fault vibration signal. With the extracted feature vectors for training and validation samples, input it into the probability neural network, to achieve the purpose of rolling bearing fault condition recognition.Research results show that wavelet analysis combined with probabilistic neural network diagnosis method can effectively rolling bearing fault detection, diagnosis.