Abstract:
Early detection of changes, that are from normal to abnormal in dynamical degradation process, is essential to enhance the function of bearings during their long-term ser...Show MoreMetadata
Abstract:
Early detection of changes, that are from normal to abnormal in dynamical degradation process, is essential to enhance the function of bearings during their long-term service time. This article presents a new and practical method for this problem based on hierarchical graph model (HGM) coupled with adaptive inputs weighting (AIW) fusion. Hierarchical decomposition (HD) is first adopted on the raw vibration signal, such that the inspection of bearing health condition can be allowed over different multiple scales. HGMs are then constructed from each decomposed components that are resulted from HD. AIW is employed to aggregate all constructed HGMs by taking into account their difference in characterization of dynamical bearing health condition. A common hypothesis testing is finally adopted to make decision. Experiments are conducted on two publicly available datasets. Results demonstrate the effectiveness of the proposed method; meanwhile, comparisons with benchmarking methods suggest its good potential in real applications.
Published in: IEEE Transactions on Industrial Informatics ( Volume: 17, Issue: 5, May 2021)