轴承是舰船故障发生的常见位置，针对现有机器学习方法在舰船轴承故障诊断领域中存在多分类精度差、运算效率低等问题，提出一种基于Cat Boost(category boosting)算法的轴承诊断技术。首先，对振动信号进行时域分析、频域分析以及EMD(empirical mode decomposition)分解，得到截选振动信号段的特征指标；其次，利用Cat Boost算法在所提取特征中进行筛选，通过基尼指数快速建立树结构并进行排序。最后，选取不同维数特征输入进行模型算法评价，并与传统方法分类的准确率进行对比。试验结果表明，该方法在处理滚动轴承故障多分类问题上故障特征提取更为有效，识别效果明显高于其他传统算法。
Bearing is a common location for ship failure, and a bearing diagnostic technology based on Cat Boost (category boosting) algorithm is proposed for the existing machine learning methods in the field of ship bearing fault diagnosis, such as poor multi-classification accuracy and low computational efficiency. Firstly, the vibration signals are analyzed in time domain, frequency domain and EMD decomposition, and the characteristics of the selected vibration signal segment are obtained. Secondly, Cat Boost algorithm was used to filter the extracted features, and Gini index was used to quickly establish the tree structure and sort it. Finally, the input of different dimension features is selected to evaluate the model algorithm, and the accuracy of classification is compared with that of traditional methods. Experimental results show that the proposed method is more effective in fault feature extraction for multi-classification of rolling bearing faults, and the recognition effect is obviously better than other traditional algorithms.
2022,44(23): 117-122 收稿日期：2021-09-23
DOI：10.3404/j.issn.1672-7649.2022.23.023
分类号：U672.74
基金项目：国家科技重大专项资助项目（J2019-IV-0021）；湖北省自然科学基金资助项目（2020CFB536）
作者简介：邢芷恺(1997-)，男，硕士研究生，研究方向为舰船动力及热力系统的监测、控制与故障诊断
参考文献：
[1] 吴国文, 田杨阳, 毛文涛. 基于特征融合的滚动轴承在线故障诊断[J]. 控制与决策, 2018:2-7.
[2] 李俊, 刘永葆, 余又红. 卷积神经网络和峭度在轴承故障诊断中的应用[J]. 航空动力学报, 2019, 34(11):2423-2431
[3] 李郅琴, 杜建强, 聂斌, 等. 特征选择方法综述[J]. 计算机工程与应用, 2019, 55(24):10-19
[4] 崔宇佳, 张一迪, 王培志, 等. 基于多评价标准融合的医疗数据特征选择算法[J]. 复旦学报(自然科学版), 2019, 58(2):250-255+268
[5] 苏庆, 林华智, 黄剑锋, 等. 结合CNN和Catboost算法的恶意安卓应用检测模型[J]. 计算机工程与应用, 2021, 57(15):140-146
[6] MCCLELLAND J L, RUMELHART D E, Hinton G E. Une nouvelle approche de la cognition:le connexionnisme[J]. Le Débat, 1987, 47(5).
[7] CORINNA C, VLADIMIR V. Support-vector networks[J]. Machine Learning, 1995, 20(3).
[8] 崔立明. 中介轴承寿命预测方法与寿命试验研究[D]. 大连:大连理工大学, 2016.
[9] 洪振麒. 基于多特征融合的滚动轴承剩余寿命预测方法研究[D]. 沈阳:沈阳大学, 2021.
[10] 周小龙, 杨恭勇, 梁秀霞. 基于EMD重构和SVM的滚动轴承故障诊断方法研究[J]. 东北电力大学学报, 2016, 36(6):71-6
[11] VEER L J, DAI H, VIJVER M J, et al. Gene expression profiling predicts clinical outcome of breast cancer[J]. Nature, 2002, 415(6871):530-536
[12] PROKHORENKOVA L, GUSEV G, VOROBEV A. CatBoost:unbiased boosting with categorical features[C]//Advances in Neural Information Processing Systems, 2018:6638-6648.
[13] 姜琦刚, 杨秀艳, 杨长保, 等. 基于CatBoost算法的面向对象土地利用分类[J]. 吉林大学学报(信息科学版), 2020, 38(2):185-191
[14] GOLDSTEIN B A, POLLEY E C, BRIGGS F B S. Random forests for genetic association studies.[J]. Statistical Applications in Genetics and Molecular Biology, 2011, 10(1).
