振动与冲击
振動與遲擊
진동여충격
JOURNAL OF VIBRATION AND SHOCK
2015年
1期
63-67
,共5页
螺栓松动%概率密度%主元分析%故障诊断
螺栓鬆動%概率密度%主元分析%故障診斷
라전송동%개솔밀도%주원분석%고장진단
bolt looseness%probability density%principal component analysis%faults diagnosis
螺栓松动是一种常见且具有潜在危害的机械故障。考虑到螺栓松动会导致被联接件结合部动力参数发生变化，提出了一种基于两被联接件振动信号的松动识别方法。所提方法首先计算两信号的概率密度，并对概率密度曲线进行网格化处理生成概率矩阵，继而对概率矩阵进行主元分析（PCA），在合并两路信号经主元分析后所得投影矩阵之后，再次进行主元分析和投影。设计了两种识别方式，方式1首先按上述过程进行已知样本训练以得到各松动状态投影点，识别时根据所得投影点与各状态投影点间的欧式距离进行判断；方式2使用螺栓紧固状态时所得样本数据和现场实测数据直接按上述过程进行计算，并根据 PCA 特性设计了松动判别条件。试验验证表明所提方法能够准确区分不同松动状态，且识别方式2操作简便，无需故障样本，易于实际应用。
螺栓鬆動是一種常見且具有潛在危害的機械故障。攷慮到螺栓鬆動會導緻被聯接件結閤部動力參數髮生變化，提齣瞭一種基于兩被聯接件振動信號的鬆動識彆方法。所提方法首先計算兩信號的概率密度，併對概率密度麯線進行網格化處理生成概率矩陣，繼而對概率矩陣進行主元分析（PCA），在閤併兩路信號經主元分析後所得投影矩陣之後，再次進行主元分析和投影。設計瞭兩種識彆方式，方式1首先按上述過程進行已知樣本訓練以得到各鬆動狀態投影點，識彆時根據所得投影點與各狀態投影點間的歐式距離進行判斷；方式2使用螺栓緊固狀態時所得樣本數據和現場實測數據直接按上述過程進行計算，併根據 PCA 特性設計瞭鬆動判彆條件。試驗驗證錶明所提方法能夠準確區分不同鬆動狀態，且識彆方式2操作簡便，無需故障樣本，易于實際應用。
라전송동시일충상견차구유잠재위해적궤계고장。고필도라전송동회도치피련접건결합부동력삼수발생변화，제출료일충기우량피련접건진동신호적송동식별방법。소제방법수선계산량신호적개솔밀도，병대개솔밀도곡선진행망격화처리생성개솔구진，계이대개솔구진진행주원분석（PCA），재합병량로신호경주원분석후소득투영구진지후，재차진행주원분석화투영。설계료량충식별방식，방식1수선안상술과정진행이지양본훈련이득도각송동상태투영점，식별시근거소득투영점여각상태투영점간적구식거리진행판단；방식2사용라전긴고상태시소득양본수거화현장실측수거직접안상술과정진행계산，병근거 PCA 특성설계료송동판별조건。시험험증표명소제방법능구준학구분불동송동상태，차식별방식2조작간편，무수고장양본，역우실제응용。
Considering that the bolt looseness induces the variation of dynamic parameters of joint of bolt connection,a detection method was proposed based on vibration signals of two connected parts.The probability densities of the two signals were calcuated,the probability density curves were processed by a mesh to obtain probability matrixes, and were the probability matrixes were transformed by using principal component analysis (PCA)method.After PCA,two projection matrixes of the probability matrixes were merged to one matrix.To this matrix,PCA and projection were implemented again.According to the peculiarity of bolt looseness,tow detection modes are designed based on the proposed method.The mode 1 carries out training work using foregone samples to generate projection points of each looseness condition employing the proposed method,then in detection,Euclidean distance between projection point and each foregone projection point was used as estimation criterion.The mode 2 directly detects looseness condition by using proposed method based on the signals of tight bolt connection condition and field measurement,and a criterion for judging bolt looseness was designed.The experimental verification shows that the proposed method can distinguish different looseness conditions,and the detection mode 2 is easy and simple to be handled without foregone fault samples.
