华中科技大学学报（自然科学版）2025，Vol.53Issue(4)：1-7,7.DOI:10.13245/j.hust.250151
基于相关熵和GMSVM的推进轴系故障诊断方法
Correntropy and GMSVM-based fault diagnosis method for propulsion shaft system
摘要
Abstract
Aiming at the problem that deep learning-based fault diagnosis methods focus on single modality and require a large amount of training data,a novel correntropy and Gram matrix support vector machine(GMSVM)-based fault diagnosis method for ship propulsion shaft system was proposed.First,the correntropy matrix between the multi-modal monitoring signal fragments was calculated to reveal the spatial correlation of different monitoring signals and to weaken the interference caused by outliers in the monitoring signals.Then,using matrix logarithmic operation,the correntropy matrix was mapped from Riemannian manifold space to Euclidean metric space,which enhanced the information representation ability and extracted the key features of faults.Finally,a Gram matrix support vector machine was constructed to achieve the fault identification under small samples.Experimental results show that for 15 different working conditions,the average diagnostic accuracy of the proposed method reaches 94.31%and 99.68%with 1 and 5 training samples per class,respectively,which is significantly better than other deep learning-based methods.关键词
推进轴系/故障诊断/相关熵/格拉姆矩阵支持向量机/多传感器/小样本Key words
propulsion shaft system/fault diagnosis/correntropy/Gram matrix support vector machine/multi-sensor/small sample分类
交通运输引用本文复制引用
邓琪,汪承杰,万海波,吴军..基于相关熵和GMSVM的推进轴系故障诊断方法[J].华中科技大学学报（自然科学版）,2025,53(4):1-7,7.基金项目
国家自然科学基金资助项目(523B2100) （523B2100）
华中科技大学交叉研究支持计划资助项目(2024JCYJ028). （2024JCYJ028）
