An Automatic Filtering Method Based on an Improved Genetic Algorithm—With Application to Rolling Bearing Fault Signal Extraction

• DOI: 10.1109/jsen.2017.2738152
• Publication Date: 2017-08-10T18:42:42Z
• Authors: Zhiqiang Liao; Liuyang Song; Peng Chen; Shilun Zuo
• Keywords: Rolling-element bearing

Fault feature extracted from fault signals plays a vital role in detection and diagnosis. However, the early stage of defect, (amplitudes energy) are very weak overwhelmed by large noise, it is difficult to extract effectively. To handle this problem, paper, an automatic filtering method based on improved genetic algorithm (IGA) for extracting presented. This improves binary coding generation GA, so that each value matches frequency signal spectrum. Based designed "statistical information evaluation function" as IGA fitness function, similarity between normal state (noise signal) noise evaluated, employs operations search optimal string. Negating string can obtain It will then be multiplied abnormal vibration spectrum, yield new spectrum only has components. proposed does not need seek center bandwidth, automatically remove noise. The effective advanced performance evaluated three different rolling bearing defects (outer race inner roller defect) rotating machine experimental bench gearbox. Compared with characteristic results denoising methods bandpass wavelet decomposition, filter effectively signal. proves feasible, effective, advanced.