基于Alpha稳定分布特征参数的滚动轴承故障诊断研究

发布时间：2017-12-31 10:35

本文关键词：基于Alpha稳定分布特征参数的滚动轴承故障诊断研究　出处：《哈尔滨工业大学》2011年硕士论文　论文类型：学位论文

【摘要】：滚动轴承作为机械设备中最常用的部件之一,其运行状态是机械设备能否正常工作的重要影响因素。因此,对于滚动轴承的性能监测以及故障诊断,其重要性不言而喻。当滚动轴承在出现点蚀等故障时,很明显的特点就是会在振动采样信号中出现故障脉冲成份。在轴承故障初期,故障脉冲幅值较小,一般都会淹没在采样信号的噪声之中,但到故障中后期时,故障脉冲幅值会越来越大,采样信号本身也会呈现出明显的脉冲特性。本文首先从统计信号处理的角度出发,通过分析滚动轴承故障仿真信号及其实际故障信号的非高斯性和脉冲特性,引入Alpha稳定分布这一可用来描述具有明显脉冲特性信号的统计模型。作为广义化的高斯分布,Alpha稳定分布拟合滚动轴承故障信号概率密度分布的精度更高,也更加的合理。通过滚动轴承故障仿真信号的数学表达式,本文分析故障信号中轴承故障程度等因素对Alpha稳定分布特征指数α及峭度值的影响,并明确了α值及峭度值只是反映信号本身的脉冲特性,与信号中的实际故障脉冲大小,即轴承故障程度并无直接关系这一结论。然后,通过进一步分析对比α值及峭度值在滚动轴承性能评估的特点,提出一个新的基于Alpha稳定分布概率密度分布的性能评估参数Lambda,通过仿真和实验分析,指出其相对于α值及峭度值,具有对早期故障脉冲敏感度更高,晚期故障不会出现性能退化,且自身的敏感度可以根据实际需求调节的优势。本文基于Alpha稳定分布的特征参数,提出了结合自适应小波的滚动轴承早期故障信号检测方法,以及结合神经网络的滚动轴承故障分类方法,通过实验数据分析,表明将现代信号处理方法同Alpha稳定分布结合的故障信号诊断方法,可以有效地分析滚动轴承故障信号,并且表现出了一定的优势。
[Abstract]:Rolling bearing is one of the most common parts of the mechanical equipment, its running status is an important factor influencing the mechanical equipment can work normally. Therefore, the rolling bearing performance monitoring and fault diagnosis, and its importance is self-evident. When there is pitting in rolling bearings fault, it's very clear that in the sampling pulse vibration fault ingredients appear in rolling bearing fault signal. The early fault pulse amplitude is small, usually submerged in the noise signal, but in the late fault, the fault pulse amplitude will become increasingly large, the sampling signal itself will clearly show the pulse characteristics.
Firstly, from the angle of statistical signal processing, non Gauss and pulse characteristics simulation fault signal of rolling bearings and the actual fault signal by analyzing the Alpha stable distribution which can be used to describe a statistical model was introduced. The characteristics of pulse signal distribution as a generalized Gauss Alpha stable distribution of the probability density distribution of rolling bearing fault the signal of higher accuracy and more reasonable.
The mathematical expression of fault simulation signal of the rolling bearing, the factors of bearing fault degree of fault signal analysis of the impact on the Alpha stable distribution characteristic exponent and kurtosis, and the alpha value and kurtosis value only reflect the pulse characteristics of the signal itself, and the actual size of fault pulse signal, namely bearing fault degree there is no direct relationship between this conclusion. Then, through further comparative analysis of alpha value and kurtosis in the characteristics of the performance evaluation of rolling bearings, put forward a Lambda performance evaluation of Alpha stable distribution of probability density distribution based on the new, through simulation and experimental analysis, pointed out that compared with the alpha value and kurtosis value is on early fault pulse sensitive late fault no performance degradation, and the sensitivity can be adjusted according to the actual needs of the advantage.
The characteristic parameters based on Alpha stable distribution, is presented based on adaptive wavelet of rolling bearing early fault signal detection method and fault classification method combined with neural network, by analyzing the experimental data show that the modern signal processing method with fault signal diagnosis method combining with Alpha stable distribution, can effectively analyze the fault signals of rolling bearing, and showed certain advantages.

【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：TH165.3

【参考文献】