Abstract:When rolling bearings rotate at high speed, they generate vibration and friction, which can easily cause minor wear and damage to the bearing surface. In harsh working environments such as high temperature, high pressure, and corrosion, it can exacerbate the wear and corrosion of the bearing, making surface defects more complex and difficult to distinguish. In order to accurately monitor and evaluate the condition of bearings and detect potential signs of faults early, a rolling bearing full life cycle fault detection method based on mathematical morphology and LMD algorithm is proposed. Based on the fault mechanism and characteristics of rolling bearings, set fault detection standards for rolling bearings and simulate the entire life cycle working process of rolling bearings. Collect surface image data and internal vibration data of rolling bearings separately, and complete the preprocessing of initial working parameters through filtering, enhancement, and other operations. Using mathematical morphology to extract small features of rolling bearing surface images based on shape features, including surface shape and small detail structures, and using LMD algorithm to decompose complex signals into multiple single frequency modulation and narrowband frequency modulation components, key features such as kurtosis and frequency are extracted. The combination of mathematical morphology and LMD algorithm can comprehensively extract the fault characteristics of rolling bearings at different life cycle stages, providing more comprehensive information for fault diagnosis, and using feature matching to obtain detection results of rolling bearing fault types, positions, and amounts. The conclusion drawn from performance testing experiments is that compared with current fault detection methods, the optimized design method significantly reduces the false detection rate of fault types and has good fault detection capabilities.
