Working Performance Evaluation of Rolling Bearings Using Modern Statistics

用近代統計學評估滾動軸承工作性能

徐永智(Yongzhi Xu) ； 夏新濤(Xintao Xia) ； 南翔(Xiang Nan)

《中國機械工程學刊》 40卷1期 (2019/02) Pp. 63-72

摘要

The performance assessment of time series with unknown distributions, which belongs to the category of problems with poor information, is a key challenge for modern statistics. On the basis of modern statistics, the fusion method of histograms and a normality test to judge the robustness and the direction of unsteady data of time series, and the fusion method combines the median estimate and Huber (M) estimate obtains robust data, unsteady data and the significance level of the time series. These methods are used in the vibration analysis of rolling bearings to verify their effectiveness, and the results show that unsteady data exist in time series at both ends of the order statistics. The reliability reflects the significance level of the rolling bearing vibration data and avoids error due to artificial factors. The intrinsic interval and the variation ratio accurately represent the working performance of rolling bearings, even in cases of complex and diverse running states. Additionally, the above fusion method provides a valuable solution to robustness problem for unknown distribution, the significance level test data and the boundary value of the Huber (M) estimate in modern statistical methods. of the histogram and normality test provides rules to judge the performance variation of the rolling bearing; the fusion method combining the median estimate and Huber (M) estimate of the

關鍵字

robust theory ； significance level ； variation ratio ； intrinsic interval ； working state

並列摘要

未知分佈時間序列的性能評估屬於乏資訊範疇，是近代統計學中的重要問題，一種融合方法被提出評估未知分佈時間序列的工作性能。基於近代統計學，採用長條圖與正態性檢驗相融合方法判斷時間序列的穩健性及不穩健資料方向；採用中位數估計與Huber（M）估計相融合方法獲取時間序列的穩健資料、不穩健資料以及顯著性水準.上述方法對滾動軸承振動資料分析，結果顯示滾動軸承振動資料中存在不穩健資料並且在次序統計量兩端；在工作狀況複雜性多樣性條件下，可靠度可以避免人為誤差反映時間序列的置信水準；本征區間和變異率可以很好的評估滾動軸承的工作性能和變異性能。此外，長條圖與正態性檢驗融合方法給出滾動軸承性能變異的判斷準則，中位數估計與Huber（M）估計相融合方法為分佈未知、置信水準未知試驗資料的穩健處理以及Huber（M）估計的邊界值確定提供了一種方法。