基于循环统计量的盲源分离方法研究及其在旋转机械故障特征提取中的应用

项目名称： 基于循环统计量的盲源分离方法研究及其在旋转机械故障特征提取中的应用

项目编号： No.50875162

项目类型： 面上项目

立项/批准年度： 2009

项目学科： 轻工业、手工业

项目作者： 陈进

作者单位： 上海交通大学

项目金额： 36万元

中文摘要： 机械设备状态监测和故障诊断学科中的关键问题之一是故障特征提取技术，其直接关系到故障诊断的准确率和故障早期预报的可靠性。然而相对较强的背景噪声或其它振源干扰信号会严重影响故障特征的提取，因此利用盲源分离(BSS, Blind Source Separation)技术分离传感器测得的振源混合信号，以获得单一振源信号，势必会明显提高故障诊断的准确度。因此，盲源分离技术近几年在机械设备状态监测与故障诊断学科中的应用也逐渐表现出活跃的研究态势。但是根据盲源分离技术的假设条件，盲源分离算法用于振动信号源分离还存在源信号数目未知、传感器数目限制及源信号卷积混合等许多问题需要解决。基于循环平稳信号模型对于旋转机械振动信号的普遍适用性，利用一阶、二阶及高阶循环统计量理论可以有效设计新的盲源分离算法，同时改进后的算法对于设备状态监测和故障诊断学科又不失一般性，无疑是一个很有前景的研究方向。

中文关键词： 循环平稳；盲源分离；旋转机械；故障诊断；特征提取

英文摘要： Feature extraction technique is one of the key works in machinery condition monitoring and fault diagnosis(CMFD), it has direct effect on the accuracy of fault diagnosis and the reliability of early fault pognosis. However, the relatively stronger background noise and other vibration disturbance sources will affect the real fault feature extraction greatly, thus the idea of using blind source separation(BSS) techniques to decease the influences of unwanted disturbances is being paid more attention in current study of CMFD. Although it is a good idea, based on the assumptions of BSS techniques, the successful application of the BSS techniques in the mechnical vibration source seperation has met many problems such as the unkonwn source numbers, limited sensors, convolutive mixing and so on. Because of generality of the cyclostationary signal model for viration signals of rotating machinery, we can use the theory of first order, second order, and higher order cyclic statistics to design new BSS algorithms for vibration source seperation. The improved BSS algorithms based on cyclic statistics do not loose generality for the subject of CMFD, and will be a promising research direction.

英文关键词： cyclostationarity; blind source seperation; rotating machinery; fault diagnosis; feature extraction