船舶柴油机增压器转子故障的监测参数较多，不同故障参数之间存在重合的情况，导致计算耗时增加和分类准确率下降。为提高增压器转子故障识别率及诊断实时性，本文提出一种基于支持向量机递归消除法（SVM-RFE）与改进Hu不变矩结合的增压器转子故障轴心轨迹特征提取选择方法，即Hu-SVM-RFE算法。首先，针对不同故障类型转子轴心轨迹提取改进Hu不变矩，然后通过SVM-RFE方法进行特征排序与选择，筛选出分类识别率较高的特征组合，最后分别采用PNN、BP神经网络和SVM三种诊断算法对Hu-SVM-RFE算法筛选组合的特征矩阵进行分类识别诊断。结果表明：采用Hu-SVM-RFE方法获得的最优特征子集能够在简约特征数量的同时保证增压器转子故障信息的丰富性，结合PNN算法可以获得较高的分类识别率并且耗时较少。
Due to the large number of parameters monitored for turbocharger rotor faults and the overlap between different fault parameters, this results in increased calculation time and reduced classification accuracy. To improve the recognition rate of turbocharger rotor faults, a recursive elimination method based on support vector machine (SVM-RFE) combined with improved Hu invariant moments is designed for the selection of turbocharger rotor fault axial trajectory feature extraction. Firstly, improved Hu-invariant moments are extracted for different rotor fault axis trajectories, then the feature ranking and selection is carried out by the SVM-RFE method to filter out the feature combinations with higher classification recognition rates. Finally, the feature matrix of the Hu-SVM-RFE algorithm screening combination was identified for classification using each of the three diagnostic algorithms. The results show that the optimal feature subset obtained by the Hu-SVM-RFE method ensures a richness of information on the fault information of the booster rotor while reducing the number of features, and in combination with the PNN algorithm achieves a high classification rate with low time consumption.
2025,47(9): 108-113 收稿日期：2024-5-13
DOI：10.3404/j.issn.1672-7649.2025.09.019
分类号：TH133
作者简介：刘琦（1988-），男，博士，工程师，研究方向为动力机械及热力系统设计、仿真与优化
参考文献：
[1] SINGH A, GUPTA T C. Effect of rotating unbalance and engine excitations on the nonlinear dynamic response of turbocharger flexible rotor system supported on floating ring bearings[J]. Archive of Applied Mechanics, 2020, 90: 1117-1134.
[2] KHORRAMI H, RAKHEJA S, SEDAGHATI R. Vibration behavior of a two-crack shaft in a rotor disc-bearing system[J]. Mechanism and Machine Theory, 2017, 113: 67-84.
[3] XU H, WANG N, JIANG D, et al. Dynamic characteristics and experimental research of dual-rotor system with rub-impact fault[J]. Shock and Vibration, 2016, 2016: 1-11.
[4] 李鑫, 赵坤鹏, 朱凌寒. 改进的不变矩和PNN相结合的多品种产品识别算法[J]. 传感器与微系统, 2019, 38(8): 132-135.
[5] 符书楠, 许枫, 刘佳, 等. 结合区域提取和改进卷积神经网络的水下小目标检测[J]. 应用声学, 2023, 42(6): 1280-1288.
[6] 遆佳, 李霁. 船体局部姿态图像的配准算法[J]. 舰船科学技术, 2021, 43(24): 13-15.
TI J, LI J. Design of registration algorithm for hull local attitude image[J]. Ship Science and Technology, 2021, 43(12A): 13-15.
[7] KOZODOI N, LESSMANN S, PAPAKONSTANTINOU K, et al. A multi-objective approach for profit-driven feature selection in credit scoring[J]. Decision Support Systems, 2019, 120: 106-117.
[8] GOTTWALT F, CHANG E, DILLON T. CorrCorr: a feature selection method for multivariate correlation network anomaly detection techniques[J]. Computers & Security, 2019, 83: 234-245.
[9] KORNYO O, ASANTE M, OPOKU R, et al. Botnet attacks classification in AMI networks with recursive feature elimination(RFE) and machine learning algorithms[J]. Computers & Security, 2023, 135: 103456.
[10] KUSHWAHA N L, RAJPUT J, SUNA T, et al. Metaheuristic approaches for prediction of water quality indices with relief algorithm-based feature selection[J]. Ecological Informatics, 2023, 75: 102122.
[11] ZHANG T, MARINO A, XIONG H, et al. A ship detector applying principal component analysis to the polarimetric notch filter[J]. Remote Sensing, 2018, 10(6): 948.
[12] 汪飞翔, 杨亚东, 田书冰, 等. 基于SVM的水上交通事故严重程度的影响因素研究[J]. 交通信息与安全, 2018, 36(2): 18-23+32.
[13] 石荣丽, 林亦舒. 基于特征优选和SVM的船舶航行事故致因分析[J]. 运筹与管理, 2023, 32(12): 99-105.
[14] 杨永平. 基于遗传算法优化支持向量机的船舰目标识别分类[J]. 舰船科学技术, 2024, 46(4): 174-178.
YANG Y P. Ship target recognition and classification based on genetic algorithm optimization of support vector machine[J]. Ship Science and Technology, 2024, 46(4): 174-178.
