LAUSR

In condition monitoring for rolling bearings, it has achieved good diagnostic performance and clear mechanistic interpretation based on vibration data. The high sampling frequency of data collection preserves fault characteristics but brings the problem of big data. An effective way to reduce this problem is to apply data compression. However, in order not to affect the diagnostic performance of data, it is difficult to improve the compression ratio further. Inspired by the binarization method, the compression dimension of the bit cost of a single sample point is first introduced into the fault-mechanism-based method in this article. On this basis, a three-dimensional data compression method is proposed, and it is subsequently validated with two real-bearing datasets. Two performance metrics, including a newly defined one, are utilized to compare the proposed method with the five existing methods. The comparison results show that the proposed method significantly improves the compression ratio of data but maintains good diagnostic performance.