摘要
为消除实际工况中振动信号时变异常点对故障特征的干扰影响,提出一种利用CRITIC权重法量化估计信号被干扰程度,改进冗余属性投影(nuisance attribute projection,NAP)的故障特征处理方法。首先,采集多时段样本数据,利用CRITIC误差评价矩阵,量化估计离散度,自适应构造权重矩阵;其次,通过权重矩阵更新NAP投影矩阵,对故障特征进行冗余属性投影;最后,随机抽取已知运行状态投影后的样本作为SVM训练集,剩余样本作为测试集进行故障诊断。通过故障特征曲线对比实验,得出处理后的特征曲线基本完全重合,消除了同状态不同样本间故障特征差异性,验证了所提方法的有效性。通过相关性实验,与传统NAP做对比,4类状态特征相关性均有提高,平均相关系数高达近1,验证了所提方法的优越性。
In order to eliminate the interference of time varying anomaly points of vibration signals on fault characteristics in actual working conditions,a fault feature processing method using CRITIC weight method to quantitatively estimate signal interference degree and improve nuisance attribute orojection(NAP)is proposed.First,sample data of multi-period were collected,error evaluation matrix of CRITIC was used to quantify the discrete degree,and weight matrix was constructed adaptively.Secondly,the NAP projection matrix was updated by weight matrix,and the fault characteristics were projected with nuisance attribute projection.Finally,samples with known running state projection were randomly selected as SVM training set,and the remaining samples were used as test set for fault diagnosis.Through the comparison experiment of fault characteristic curves,it is concluded that the treated characteristic curves basically coincide completely,eliminating the fault characteristic difference between different samples in the same state,and verifying the effectiveness of the proposed method.The correlation experiment showed that compared with traditional NAP,the correlation of four state features was improved,and the average correlation coefficient was as high as 1,which verified the superiority of the proposed method.
作者
张玉劼
姜宏
章翔峰
ZHANG Yujie;JIANG Hong;ZHANG Xiangfeng(School of Mechanical Engineering,Xinjiang University,Urumqi 830017,China)
出处
《组合机床与自动化加工技术》
北大核心
2023年第11期128-132,137,共6页
Modular Machine Tool & Automatic Manufacturing Technique
基金
国家自然科学基金项目(51865054)。
