摘要
针对以往模式识别方法的不足及特征值贡献度的问题,提出了基于特征加权的代理判别模型(agent discriminate model based feature weighted,ADMFW)模式识别方法。该方法的核心在于利用加权因子获取加权特征,并采用代理模型建立加权特征之间的关系函数,即首先计算特征值的权值因子,评估特征值的显著度,进而对每个特征值予以权值;然后利用加权特征和代理模型建立预测模型;最后采用预测模型对未知样本进行识别诊断。对滚动轴承实测数据的分析结果表明,ADMFW可以有效地对滚动轴承的工作状态和故障类型进行识别。
In view of the shortcomings of the previous pattern recognition methods and the saliency problem of features,this paper proposed a pattern recognition method—agent discriminate model based feature weighted( ADMFW). The core of this method is to use the weighting factor to obtain the weighted feature and utilize the agent discriminate model to establish the weighting function of weighted feature. Firstly,it calculated the weighting factors of the features to evaluate the saliency of the feature and assigning the weights to features. Secondly,it used the weighting features and the agent discriminate model to establish the prediction model. Finally,it applied the prediction model to identify the unknown samples. The analysis of the experimental data of rolling bearing shows that this method can identify the working states and fault types of rolling bearings effectively.
引文
[1] Sugumarm V,Ramachandran K I. Fault diagnosis of roller bearring using fussy classifier and histogram features with focus on automatic rule learning[J]. Expert Systems with Applications,2011,38(5):4901-4907.
[2] Wang Huaqing,Chen Peng. Intelligent diagnosis method for rolling element bearing faults using possibility theory and neural network[J].Computer&Industrial Engineering,2011,60(4):511-518.
[3] Fei Shengwei,Zhang Xiaobin. Fault diagnosis of power transformer based on support vector machine with genetic algorithm[J]. Expert Systems with Applications,2009,36(8):11352-11357.
[4] Zhang Xiaoyuan,Liang Yitao,Zhou Jianzhong,et al. A novel bearing fault diagnosis model integrated permutation entropy,ensemble empirical mode decomposition and optimized SVM[J]. Measurement,2015,69:164-179.
[5] Liu Xiaofeng,Bo Lin,Luo Honglin. Bearing faults diagnostics based on hybrid LS-SVM and EMD method[J]. Measurement,2015,59(1):145-166.
[6] Raghuraj R,Lakshminarayanan S. Variable predictive models:a new multivariate classification approach for pattern recognition applications[J]. Pattern Recognition,2009,42(1):7-16.
[7] Yang Yu,Pan Haiyang,Ma Li,et al. A roller bearing fault diagnosis method based on the improved ITD and RRVPMCD[J]. Measurement,2014,55(9):255-264.
[8] Zheng Jinde. Rolling bearing fault diagnosis based on partially ensemble empirical mode decomposition and variable predictive model-based class discrimination[J]. Archives of Civil and Mechanical Engineering,2016,16(4):784-794.
[9] Gou Peng,Liu Wei,Cui Weicheng. A comparison of approximation methods for multidisciplinary design optimization of ship structures[J]. Journal of Ship Mechanics,2007,11(6):913-923.
[10]郭金维,蒲绪强,高祥,等.一种改进的多目标决策指标权重计算方法[J].西安电子科技大学学报,2015,41(6):118-125.(Guo Jinwei,Pu Xuqiang,Gao Xiang,et al. Improved method on weights determination of indexes in multi-objective decision[J]. Journal of Xidian University,2015,41(6):118-125.)
[11]Zhou Wenhua,Wang Rusong. An entropy weight on the fuzzy synthetic assessment of Beijing urban ecosystem health[J]. Acta Ecologica Sinica,2005,25(12):3245-3251.
[12]程军圣,郑近德,杨宇.基于局部特征尺度分解的经验包络解调方法及其在机械故障诊断中的应用[J].机械工程学报,2012,48(19):87-94.(Cheng Junsheng,Zheng Jinde,Yang Yu. Empirical envelope demodulation approach based on local characteristic-scale decomposition and its applications to mechanical fault diagnosis[J].Chinese Journal of Mechanical Engineering,2012,48(19):87-94.)