往复泵轴承故障诊断的神经网络方法

Neural Net Work Method of Reciprocating Pump Bearing Fault Diagnosis

研究了基于BP网络的往复泵轴承故障诊断方法.利用频域和倒频域进行特征提取,采用集成BP网络进行故障诊断和识别,解决了往复泵轴承故障特征提取困难、多故障识别困难的问题.试验结果表明,利用BP网络可以有效地诊断与识别往复泵轴承多故障模式,并且具有很高的成功率.

This article studies the method of bearing trouble-shooting of reciprocating pump based on BP net work. The feature is extracted by using frequency and mel-frequency. Trouble-shooting and distinguishing are done by using integration BP net work. The difficulties of bearing fault feature extraction and of distinguishing multi-faults have been soled out in reciprocating pump. The experimental results have proved that BP network can efficiently diagnose and distinguish the multi-faults of bearing in reciprocating pump, and the successful rate is high.