Intelligent Method for Bearings Diagnosis Using Statistical Filters and Deep Learning

Abstract

Since bearings are important components in rotating machinery equipment, it is important to regularly monitor and diagnose the condition of bearings in critical equipment in order to prevent sudden and serious accidents. In recent years, research on intelligent and automatic diagnosis technology using AI technology has been conducted for the diagnosis of abnormalities in bearings. In particular, deep learning has been attracting attention in the field of equipment diagnosis because it is an AI technology with high feature extraction capability. In this study, we proposed a method to automatically perform feature extraction and state classification using convolutional neural network (CNN), a type of deep learning, after removing noise from vibration signals measured for bearing diagnosis using statistical filters. As a result of various verification experiments, it was found that the proposed method can achieve highly accurate diagnosis of bearing abnormalities even in noisy environments.