Clustering-based regularized orthogonal matching pursuit algorithm for rolling element bearing fault diagnosis

The sparse representation, which is based on the orthogonal matching pursuit (OMP) algorithm, is a useful technique for identifying defect characteristics in rolling element bearings. However, OMP is easily influenced by noise interference and is prone to choosing irrelevant atoms during the sparse decomposition process, resulting in a reduction in reconstruction accuracy. A clustering-based regularized orthogonal matching pursuit (CROMP) algorithm is proposed for bearing fault diagnosis. The clustering technique can successfully eliminate redundant atoms from the dictionary, improving the system’s stability and performance, while regularization can enhance the program’s capacity to recover sparse signals. The suggested technique may successfully recover transient signals from loud noise, according to simulation simulations. The approach performs well in extracting notable fault impacts, according to actual testing. The suggested approach takes less time to run and extracts early defect information more effectively than the OMP algorithm.