基于乏信息的滚动轴承振动与噪声的模糊预报

Fuzzy prediction for vibration and noise of rolling bearing based on poor information

根据非统计原理,提出了一种新的预报技术——模糊预报技术,以解决样本少且概率分布未知条件下滚动轴承的动态性能预报问题。从小样本数据序列入手,基于乏信息隶属函数,建立轴承振动与噪声的经验概率密度函数与预报函数,并将小样本数据序列分成尺度信息与随机信息两个子序列,挖掘更多的滚动轴承动态系统的信息,然后用模糊区间数运算法则进行综合,得到滚动轴承振动与噪声的预报区间。试验研究表明,所提出的模糊预报方法可以有效地预报滚动轴承振动与噪声的波动区间,置信度与可靠度可以达到95%-100%。

By the aid of the non-statistical theory, a novel fuzzy prediction technique was proposed to predict the dynamic characteristics of rolling bearings under the condition of small sample and unknown probability distribution. Starting with a data sequence with small sample, an empirical probability density function and the prediction function of vibration and noise of rolling bearings were established based on a membership function with poor information. The data sequence with small sample was divided into two sub-sequences, the measure information and the random information, in order to mine further information about the dynamic system of the rolling bearing, then the predicted interval of vibration and noise of the rolling bearing was obtained by synthesis with the algorithm of fuzzy interval number. Experiments showed that the fluctuation interval of vibration and noise of rolling bearings can be predicted effectively, and the confidence level and reliability from 95% to 100% achieved by the proposed fuzzy prediction technique.