A Fast and Accurate Lempel-Ziv Complexity Indicator Based on Data Compression and Multiscale Coding for Recognition of Bearing Fault Severity.

Lempel-Ziv complexity indicator (LZCI), as one of the complexity indicators, is effectively used for identifying the bearing fault severity due to its own advantages. However, it has suffered from issues of low computational efficiency and inaccurate coding during the calculation process. To address these problems, this paper proposed a fast and accurate LZCI for the recognition of bearing fault severity. The proposed method focused on enhancing the computational efficiency of LZCI by incorporating a data compression algorithm. This compression process not only improved the computational efficiency of LZCI, but also preserved the original shape of the signal. Then, the compressed signal was subjected to multiscale encoding. Finally, the Lempel-Ziv complexity value (LZCV) was calculated and standardized to reduce the influence of data length on LZCV. The proposed method was verified by two single-point fault datasets and two life-cycle datasets. The results showed that the data compression method can effectively compress data while maintaining the original shape, and the proposed method can improve the calculation efficiency and accuracy of LZCI.