Title: Gear faults identification based on big data analysis and CatBoost model

Authors: Yongsheng Qi; Xiaoda Zhang; Jianxin Zhang

Addresses: Inner Mongolia Key Laboratory of Mechanical and Electrical Control, College of Electric Power, Inner Mongolia University of Technology, Hohhot 010080, China ' Micron Intelligent Manufacturing Systems Science and Technology (Beijing) Co., Ltd., Beijing 100086, China ' Inner Mongolia Key Laboratory of Mechanical and Electrical Control, College of Electric Power, Inner Mongolia University of Technology, Hohhot 010080, China

Abstract: The gear faults identification based on big data analysis and CatBoost is investigated. The big datasets with nine and ten features for five gear faults are constructed, respectively. The CatBoost models based on the above two datasets are constructed and trained, respectively. The testing results show that CatBoost, XGBoost, and LGBM models based on the dataset with ten features are better than one with nine features, and the fault identification accuracy and time obtained by CatBoost are better than the other two models. By calculating the influence of features to the identification results, it can be found that four features play the crucial roles. The CatBoost based on the dataset with the above four characteristics and five faults is verified to achieve identification accuracies and time of 100% and 680 s, respectively, which are better than ones obtained by using XGBoost and LGBM.

Keywords: gear faults identification; big data analysis; CatBoost; classification prediction; feature importance.

DOI: 10.1504/IJMIC.2022.128312

International Journal of Modelling, Identification and Control, 2022 Vol.41 No.4, pp.334 - 342

Received: 21 Nov 2021
Accepted: 11 Jan 2022
Published online: 17 Jan 2023 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article