ReliefF与QPSO结合的故障特征选择算法

PDF(925 KB)

振动与冲击 ›› 2020, Vol. 39 ›› Issue (11) : 171-176.

论文

薛瑞，赵荣珍

作者信息 +

The fault feature selection algorithm of combination of ReliefF and QPSO

XUE Rui, ZHAO Rongzhen

Author information +

文章历史 +

摘要

为提高故障数据集的分类精度，将ReliefF算法与量子粒子群算法(QuantumParticle Swarm Optimization，QPSO)进行结合，提出一种能够降低故障数据集维度的敏感故障特征选择方法。首先，在对经滤波消噪后的故障信号进行多域量化特征提取基础上，设定时域与频域特征、经小波包分解得到的各频带能量特征作为描述转子系统故障状态的初始故障特征集，并用转子系统的典型故障模拟信号集合得到了一种原始的故障数据集。随后，用Relief F算法通过迭代计算得到的权值对故障数据集各特征向量进行加权、并设定阈值剔除不相关特征，据此实现了对原始故障数据集各特征的第一次筛选。最后，引入量子粒子群算法(QPSO)对特征集合进行二次筛选，剔除不利于实施分类的冗余特征并同时实现优化支持向量机的参数，通过处理得到了一种精简的最优特征子集和最合适的一组支持向量机参数。用得到的原始故障数据集对所建立的方法性能进行了计算验证。结果表明，本方法可有效地筛选出规模较小且故障模式辨识度高的低维故障数据集，它可显著提高故障分类器的辨识准确率。

Abstract

In order to improve the classification accuracy of fault data sets, the combination of ReliefF algorithm and Quantum Particle Swarm Optimization (QPSO) is adopted to propose a sensitive fault feature selection method that can reduce the dimension of fault data sets. First of all, the fault signal of after filtering de-noising multi-domain quantitative feature extraction, based on the setting time domain and frequency domain characteristics, the characteristics of each frequency band energy of wavelet packet decomposed as to describe the state of the rotor system fault initial fault feature set, and the typical fault of rotor system analog signal collection got a primitive failure data set. Then, the weight obtained through iterative calculation by the Relief F algorithm is used to weight each feature vector of the fault data set, and the threshold is set to eliminate the irrelevant features, so as to realize the first screening of each feature of the original fault data set. Finally, a quantum particle swarm optimization (QPSO) algorithm is introduced to filter feature sets twice, eliminate redundant features that are not conducive to classification, and optimize the parameters of support vector machines at the same time. By processing, a simplified optimal feature subset and the most appropriate set of support vector machines parameters are obtained. The method performance is verified by the original fault data set. The results show that this method can effectively screen out low-dimensional fault data sets with small size and high fault pattern recognition, which can significantly improve the identification accuracy of fault classifier.

Key words

feature selection, ReliefF algorithm, irrelevant features / Quantumparticle swarm optimization, support vector machine

引用本文

EndNote

Ris (Procite)

Bibtex

导出引用

薛瑞，赵荣珍. ReliefF与QPSO结合的故障特征选择算法[J]. 振动与冲击, 2020, 39(11): 171-176

XUE Rui, ZHAO Rongzhen. The fault feature selection algorithm of combination of ReliefF and QPSO[J]. Journal of Vibration and Shock, 2020, 39(11): 171-176

参考文献

[1] Zhang L, Zhang F, Hu Y. A Two-phase Flight Data Feature Selection Method Using both Filter and Wrapper[C]// IEEE International Conference on Systems, Man, and Cybernetics, 1999. IEEE Smc '99 Conference Proceedings. IEEE, 1999:132-136 vol.2.
[2] 姚旭, 王晓丹, 张玉玺,等. 特征选择方法综述[J]. 控制与决策, 2012, 27(2):161-166.
By, Yao Xu,Wang Xiaodan, Zhang Yuxi, et. A review of feature selection methods [J]. Control and decision, 2012, 27 (2): 161-166.
[3] Zhang Y, Gong D, Hu Y, et al. Feature selection algorithm based on bare bones particle swarm optimization[J]. Neurocomputing, 2015, 148(1):150-157.
[4] Bolón-Canedo V, Sánchez-Maroño N, Alonso-Betanzos A. Feature selection for high-dimensional data[J]. Progress in Artificial Intelligence, 2016, 5(2):65-75.
[5] Apolloni J, Alba E. Two hybrid wrapper-filter feature selection algorithms applied to high-dimensional microarray experiments[M]. Elsevier Science Publishers B. V. 2016.
[6] Solorio-Fernández S, Carrasco-Ochoa J A, Martínez-Trinidad J F. A new hybrid filter–wrapper feature selection method for clustering based on ranking[J]. Neurocomputing, 2016, 214:866-880.
[7] 赵荣珍, 李坤杰, 邓林峰. 基于ReliefF和遗传算法的故障特征筛选方法[C]// 全国设备监测诊断与维护学术会议、全国设备故障诊断学术会议暨2014年全国设备诊断工程会议. 2014.
Zhao Rongzhen, Li Kunjie, Deng Linfeng. Fault feature screening method based on Relieff and genetic algorithm [c]//National equipment monitoring and diagnosis and maintenance academic conference, national conference on Equipment Fault diagnosis and 2014 National Equipment Diagnostic Engineering Conference. 2014.
[8] 何涛, 胡洁, 夏鹏,等. 基于ReliefF算法与遗传算法的肌电信号特征选择[J].上海交通大学学报,2016, 50(2):204-208..
He Tao, Hujie, Xia Peng, et. EMG signal Feature selection based on Relieff algorithm and genetic algorithm [J]. Journal of Shanghai Jiaotong University, 2016, 50 (2): 204-208.
[9] 肖艳, 姜琦刚, 王斌,等. 基于Relief F和PSO混合特征选择的面向对象土地利用分类[J]. 农业工程学报, 2016, 32(4):211-216.
Xiao Yan, Jiang Qigang, Wang Bin, et al. Object oriented land use classification based on Relief F and PSO hybrid feature selection [J]. Acta agronomy Sinica, 2016, 32 (4): 211-216..
[10] 孙俊. 量子行为粒子群优化算法研究[D]. 江南大学, 2009.
CMa. Research on particle swarm optimization algorithm for quantum behavior [D]. Jiangnan University, 2009.
[11] Yuan Q, Zhou Z, Yang K. Ultrasonic Non-destructiveClassification for Wood Materials Based on ReliefF and PSO-SVM[J]. Journal of Green Science & Technology, 2016.
[12] 霍天龙, 赵荣珍, 胡宝权. 基于熵带法与PSO优化的SVM转子故障诊断[J]. 振动.测试与诊断, 2011, 31(3):279-284.
Huo Tianlong, Zhao Rongzhen, Hu Baoquan. SVM Rotor Fault Diagnosis Based on Entropy Band Method and PSO Optimization [J]. Vibration. Testing and Diagnosis, 2011, 31(3): 279-284.
[13] 张钦礼, 王兢, 王新民. 基于核主成分分析与PSO-SVM的充填管道失效风险性分级评价模型[J]. 黄金科学技术, 2017, 25(3):70-76.
Zhang Qinli, Wang Jing, Wang Xinmin. Based on nuclear principal component analysis and PSO-SVM classification evaluation model of filling pipeline failure risk [J]. gold science technology, 2017, 25 (3): 70-76.
[14] 柳一鸣. 自适应量子行为粒子群算法及其在图像分类中的应用研究[D]. 浙江大学, 2011.
Liu. Adaptive quantum behavior particle swarm optimization and its application in image classification [D]. Zhejiang University, 2011.
[15] 张燕君, 张芳草, 付兴虎,等. 基于QPSO-MLSSVM算法的拉曼光谱检测四组分调和油含量[J]. 光谱学与光谱分析, 2018, 38(5).
Zhang Yanjun, Zhang Cao, Shinghung, et. Determination of four-component blending oil content by Raman spectroscopy based on QPSO-MLSSVM algorithm [J]. Spectroscopy and spectral analysis, 2018, 38 (5).