Extracting symptoms of bearing faults from noise using a non-linear neural filter

Abstract

Improving the signal-to-noise ratio is an important feature for the early detection of faults in bearings subject to large amounts of environmental noises. A method is proposed for improving the signal-to-noise ratio by adaptive neural filtering (ANF). A comparison of failure detection capabilities of a linear adaptive filter using the least mean square (LMS) algorithm and a non-linear adaptive filter using the ANF algorithm in conditions of large amounts of environmental noise is made. Experimental results show that an adaptive filter using a neural filtering algorithm is an effective means for extracting the symptoms of a bearing fault under such conditions.