ICA及其在往复泵多源信号分离中的应用

ICA and its application to separation of multi-sourcesignals in reciprocating pumps

研究了往复泵泵阀故障诊断中多信源振动信号分离的独立分量分析方法。针对往复泵泵阀振动信号成分复杂和相互干扰等特点,将盲源分离技术中独立分量分析方法引入到泵阀振动信号的预处理中,试验结果表明,这种方法能够有效地消除目标泵阀振动信号中混杂的其他泵阀振动分量,从而使后续分析中得到的监测诊断信息的准确性和可靠性得到保障。

This paper presents an independent component analysis based approach for separation of multisource vibration signals in reciprocating pump valves fault diagnosis.For the complexity and interference of vibration signals collected from reciprocating pump valves,independent component analysis(ICA)in blind sources separation is introduced into vibration signals preprocessing of pump valves.The experimental results show that vibration signal components from other valves can be efficiently eliminated from the target valve signal by ICA.Therefore,the correctness and reliability of monitoring and diagnosis information acquired from following analysis can be guaranteed.