振动信号分频带分形维数及其在汽轮机故障诊断中的应用  

Fractal Dimensioning of Vibration Signals and Its Application to Steam Turbine Fault Diagnosis

作　　者：胥永刚[1] 何正嘉[1] 

机构地区：[1]西安交通大学机械工程学院,陕西西安710049

出　　处：《发电设备》2003年第5期22-25,共4页Power Equipment

基　　金：国家自然科学基金资助项目(编号:50175087)

摘　　要：二进小波包作为一种有效的非平稳信号处理工具,在对信号进行频带划分时,具有频带不能任意设置、数据点减少等不足,论文提供了一种简单而实用的卷积滤波划分频带的方法。应用于离散振动信号分频带分形盒维数的计算,可以定量表征信号不同频带的复杂性和不规则性。通过汽轮发电机组转子碰摩故障信号分析表明,分频带分形维数为机械设备故障诊断提供了一种有效的新方法。Wavelet packet transformation, as an effective implement for analyzing nonstationary vibration signals, has the shortcoming that, in the course of dividing frequency bands, these can not be arbitrarily fixed and that the number of datapoints are reduced etc.. The paper presents a simple and practical way of frequency band division with the help of convolution filters, which, when used to frequency band division and fractal dimensioning of discrete vibration signals, is able to reflect quantitatively the complexity and irregularity of diverse frequency bands. An analysis of rubbing fault signals of a steam turbinegenerator rotor shows, that frequency band division and fractal dimensioning provides an effective new way for fault diagnoses of mechanical equipment.

关 键 词：汽轮发电机组 故障诊断 离散振动信号 分形维数 

分 类 号：TK269[动力工程及工程热物理—动力机械及工程]