摘要
采用频谱及轴心轨迹图的方法提取仿真台得到的故障振动信号特征,分别建立子BP神经网络,并采用D-S证据理论对子BP神经网络的输出进行融合(多层信息融合)方法,从不同侧面对故障进行诊断。结果表明:采用多层信息融合方法的故障诊断置信度比频谱方法提高约0.03,比轴心轨迹图方法提高0.4,效果显著;对故障类型的识别准确率具有显著提高。
The fault vibration signal feature obtained from the simulation platform was extracted by using the method of spectrum and shaft center orbit.Moreover,the BP sub neural network was established,and the D-S evidence theory was employed to fuse the output of each neural network and diagnose the faults from different aspects.The results indicate that,this multi-level information fusion method can identify the various typical faults effectively,of which the confidence was increased by 0.03 and 0.4,compared with the spectrum method and shaft center orbit method,respectively.Furthermore,the identification accuracy of this method was increased significantly.
出处
《热力发电》
CAS
北大核心
2014年第8期140-142,146,共4页
Thermal Power Generation
基金
河北省自然科学基金(F2012402021)
关键词
汽轮机
故障诊断
频谱
轴心轨迹
子BP神经网络
D-S证据理
多层信息融合
置信度
steam turbine
fault diagnosis
spectrum
shaft center orbit
BP-neural network
D-S evidence theory
multi-level information fusion
confidence