文献类型：期刊文献

英文题名：Research on Sample Imbalanced Fault Diagnosis of Ship Water Pump Bearing Dataset Based on HCAB-SMOTE

作者：Yan, Zhijia[1]; Cai, Renchao[1]; Huang, Lin[1]; Jia, Baozhu[1]; Song, Xuewei[1]; Liao, Zhiqiang[1]; Chen, Feipeng[1]

机构：[1] Guangdong Ocean University, Naval Architecture and Shipping College, Zhanjiang, China

年份：2025

起止页码：81

外文期刊名：2025 2nd International Conference on Intelligent Ships and Electromechanical System, ICISES 2025

收录：EI(收录号：20262420909195)

语种：英文

外文关键词：Bandpass filters - Fault detection - Filtration - Navigation - Pattern recognition - Pumps - Ships - Signal processing - Systems analysis

外文摘要：Aiming at the case of imbalanced dataset in ship water pump bearing fault diagnosis, which lead to low diagnostic efficiency, this paper proposes a hybrid clustering-adaptive borderline synthetic minority oversampling technique (HCAB-SMOTE) for fault diagnosis. First, the proposed method applies band-pass filtering to filter the original vibration signals noise and extract fault-related time-frequency features. Then, HCAB-SMOTE is utilized to intelligently oversample minority fault samples, mitigating the impact of imbalance dataset on classifier performance. The effectiveness and feasibility of this method have been demonstrated through validation using laboratory bearing datasets and analysis of comparative experiments. The results indicate that the proposed method outperforms other methods in all evaluation metrics and achieves a maximum diagnostic accuracy of 98 %. These findings indicate that the proposed method is effective for diagnosing faults in ship water pump bearings. ?2025 IEEE.