匹配追踪信号分解与往复机械故障特征提取技术研究

Extraction of Fault Features in Reciprocating Machinery Based on Matching Pursuits

提出了一种自适应提取往复机械振动信号冲击特征的信号处理方法 .方法的核心是基于特别选定的基元函数将信号展开 .在采用匹配追踪算法分解信号时 ,应用指数衰减正弦波函数作为信号分解的基元函数 ,能够十分理想地匹配信号中的冲击响应波形 ,与往复机械振动信号有最好的相似性 .往复机械的故障诊断实例证明 ,利用该方法提取信号中的瞬态冲击响应特征是可行的 .研究结果为机械冲击故障的特征提取与诊断提供了一条新途径 .

A method for extracting impulse features from vibration signals of reciprocating machinery is presented. The signals are composed of event-related impulse response waveforms with varying shapes and arrival times. The extraction method is based on series expansion of the signal by a set of elementary functions. The exponentially decayed sinusoidal functions are suitable for simulating transient impulse responses of mechanical system. Hence, they are used as the elementary functions. A matching pursuit algorithm is applied to decompose the signal into linear composition of event-related impulse responses that could be correlated to the elementary functions. Fault diagnosis for a reciprocating compressor using this new method is illustrated. The result shows that the extracted features can effectively represent faults in the reciprocating machinery.