A Simple SOM Neural Network Based Fault Detection Model for Fault Diagnosis of Rolling Bearings

Abstract

Rolling bearings are common parts in the transmission systems and have been widely used in various kinds of applications. The normal operation of the rolling bearings hence plays an important role on the efficiency of the system performance. However, due to hostile working environment the rolling bearings are prone to failures. The transmission systems may break down when there occurs faults in the rolling bearings. As a result, it is essential to detect the faults of rolling bearings. However, when use artificial intelligence method to diagnose the rolling bearings faults the signal processing is extensively complex while very few works have been done on the simplification of the artificial neural network (ANN) models for the rolling bearings fault detection. To deal with this problem, a simple self-organized map (SOM) neural network method together with a principal component analysis (PCA) based feature reduction procedure is proposed to diagnosis rolling bearings faults in this work. The vibration data of the normal and faulty rolling bearings was acquired from an experimental test bed. The PCA was firstly used to extract distinct fault features. Then the SOM was employed to train and learn the fault features to identify the fault patterns. The fault detection results show that the proposed method is feasible and effective for the fault diagnosis of rolling bearings. The fault detection rate is beyond 89.0%.

Learning features from High Speed Train vibration signals with Deep Belief Networks

“…Zhao et al introduced a method based on Empirical Mode Decomposition (EMD) and Fuzzy Entropy to extract the features of HST vibration signals, and Back Propagation (BP) neural network was used as the model for the fault diagnosis of HST [7]. Li proposed a Self-Organized Map (SOM) neural network together with Principal Component Analysis (PCA) for the fault diagnosis of rolling bearing [8]. In [9], feature selection was done with wavelet entropy, and then Support Vector Machine (SVM) was used as the model of fault recognition.…”

Fault Detection through Vibration Signal Analysis based on HSM with TRIPPY Classifier

“…This notion of neural networks brings together the grouping with the principal component analysis (PCA) of the Self-Organized Map (SOM) for error diagnosis [15]. The stuck in inefficient feature extraction by PCA brought wavelet entropy along with SVM [16].…”