Health condition monitoring and diagnosis of rotating machinery based on Manhattan entropy

• Published in: Measurement : journal of the International Measurement Confederation Vol. 227; p. 114243
• Main Authors: Tan, Hongchuang, Xie, Suchao, Yang, Dalian, Cheng, Jiaqi, Zhou, Ziqiang
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 15.03.2024
• Subjects: Health condition monitoring and diagnosis; Manhattan entropy; Rotating machinery; Serial Manhattan distance; Time series; Serial Manhattan distance; Health condition monitoring and diagnosis; Rotating machinery; Manhattan entropy; Time series
• ISSN: 0263-2241; 1873-412X
• DOI: 10.1016/j.measurement.2024.114243

•The serial Manhattan distance is designed to measure the distance between subseries.•The proposed ME is able to accurately evaluate the complexity of time series.•A new health condition monitoring and diagnosis method based on ME is advanced.•The effectiveness of the developed method is verified by two experimental cases. Information entropy has been used for machinery fault diagnosis. However, most of the entropy methods suffer from some shortcomings, which are not in accordance with the original definition of information entropy. To this end, a Manhattan entropy (ME) is proposed to estimate the complexity of time series and to realize the health condition monitoring and diagnosis of machinery. Compared with other methods, the estimation of the proposed ME can maintain better uniformity with the change in complexity of the simulated signals and the complexity of the logistic map, and the results are more accurate and reasonable. Then, real signals from the aviation bearings and reducer gears show that ME can accurately estimate the complexity of mechanical signals, thus realizing the health monitoring of machinery. Furthermore, for bearing and gear fault diagnosis, ME can extract excellent features with an accuracy of more than 99.10%, which is an encouraging performance.