The discussion of the rolling bearing performance data

Discussion: the performance data of the rolling element bearing condition of discrete data and some error brings great difficulties to data analysis, and even serious damage data and error analysis, reality under the condition of the phenomenon do exist. The median is - in data analysis A very robust parameters, when the data is not discrete data, the numerical data set close to the average; When data are discrete data, the average will be away from the real value, and the median hardly deviates from the value of the real, notes digits is steady. So, you can use the difference between the median and mean to judge the existence of discrete data in the data. Using the improved Huber M method processing data, if more and more close to the median, average shows that there is no discrete data; If the average deviation from the median more and more, data are discrete data. Significance level according to the data requirements, according to recent statistics, the recommended values range of 0 ~ 0. 1. Improved Huber processed data M method, the discrete data affects overall data reduction, increased significance level, provide a reliable guarantee for the progress of the data analysis. Need to improve the progress of some theory, such as the median and the mean similarity, such as how to determine the significance level of the data, it is lack of theoretical basis. According to the improved Huber M method of rolling bearing vibration data and C of A and B rolling bearing friction torque data processing results, it can be seen in the significance level 0 ~ 0. Within 1 C of rolling bearing vibration data variance change complex, A and B of rolling bearing friction torque data variance monotone decreasing.