摘要
针对汽车发动机气门间隙异常故障,探讨了应用小波分解后求各尺度小波系数信息熵,和RBF神经网络对发动机进行不解体故障诊断的方法。由此,对某汽油发动机进行了故障模拟试验,分别在正常工况和三种故障工况下测取了缸盖表面振动信号。对所采集信号进行Stein无偏估计消噪处理,利用小波系数信息熵提取特征向量,进行归一化处理,然后用RBF神经网络对处理后的振动信号进行分类识别。发动机气门间隙故障的诊断实例表明,在不同工况下利用小波系数信息熵提取故障特征向量、进行基于RBF神经网络的故障诊断方法现实可行,对实现发动机不解体故障诊断具有一定的应用价值。
Aiming at valve clearance fault in abnormal automotive engine, a method of engine fault diagnosis withoutdisassembly was studied with the application of information entropy of wavelet coefficients of different scales after waveletdecomposition and RBF neural network. The failure simulation test was performed on a gasoline engine. The vibration signalsin cylinder head surface were measured in a normal working condition and three abnormal working conditions respectively.The collected signals were processed to eliminate the noise by Stein unbiased estimation. The characteristic vectorswere extracted and normalized by using the wavelet coefficient information entropy. Then, the processed vibration signalswere classified and identified by means of RBF neural network. The application example of diagnosis of the engine valveclearance faults shows that the proposed method is practical and feasible. It can be used for fault diagnosis of engines withoutdisassembly.
出处
《噪声与振动控制》
CSCD
2015年第1期214-218,239,共6页
Noise and Vibration Control
基金
内蒙古自然基金资助项目(2012MS0704)
内蒙古高校科研基金重点项目(NJZZ11070)
关键词
振动与波
故障诊断
发动机
RBF神经网络
信息熵
vibration and wave
fault diagnosis
engine
RBF neural network
information entropy
