基于KPCA的核电厂隔膜压缩机缸盖螺栓松动故障诊断研究

Research on bolts fault diagnosis of diaphragm compressor cylinder head in nuclear power plant based on KPCA

针对核电厂隔膜压缩机容易发生缸盖螺栓松动的问题,通过分析螺栓松动的影响因素,搭建了隔膜压缩机缸盖螺栓松动故障诊断试验系统,掌握了螺栓松动前、后缸盖所有螺栓轴向力的差异变化、分布特征和影响规律,基于KPCA算法和故障重构贡献率构建了一种螺栓松动故障识别与定位方法。不同松动状态螺栓的训练与测试结果表明,轴向力是识别螺栓松动状态的关键特征,该方法依据缸盖螺栓轴向力载荷实时数据和变化规律,可以快速识别螺栓松动的故障行为,准确判断松动螺栓的故障位置,当轴向力衰减系数为5%和10%时,最低识别准确率分别为96%和98%。研究结果为隔膜压缩机螺栓早期松动预警和在线实时检测提供了新的方法,对确保核电厂隔膜压缩机安全稳定运行具有重要意义。

Aiming at the problem of loose bolts in the cylinder head of diaphragm compressor in nuclear power plant,by analyzing the influencing factors of loose bolts,an experimental system for diagnosing loose bolts in the cylinder head of diaphragm compressor was established.The variation,distribution characteristics and influence laws of axial forces of all bolts in the cylinder head before and after bolt loosening were mastered.A method for identifying and locating loose bolts was constructed based on KPCA algorithm and fault reconstruction contribution rate.The training and testing experiments were carried out on bolts in different loose states.The results showed that axial force was the key feature to identify the loose state of bolts.According to the real-time data and change law of cylinder head bolt axial force load,this method could quickly identify the bolt loose fault behavior and accurately determine the fault location of loose bolts.When the axial force attenuation coefficient was 5%and 10%,the minimum recognition accuracy was 96%and 98%respectively.It provided a new method for early warning and online realtime detection of diaphragm compressor bolts,which was of great significance to ensure the safe and stable operation of diaphragm compressor in nuclear power plant.