摘要
利用神经网络诊断模型来识别故障发展的不同程度,并以柴油机连杆铜套磨损故障为例进行分析.首先在295柴油机上进行了设定及待定工况实验,获取各工况下的缸盖振动信号;然后利用基于神经网络和小波分析的故障诊断方法进行故障程度识别;最后利用训练后的模型对待定工况进行故障程度的判定.实验和仿真结果表明:对于各设定工况,诊断模型可以定量地识别出来,准确率达到100%;对于待定工况,诊断模型也可以给出定量的故障程度描述.从而使操作者能及时了解故障的发展情况,并根据网络模型的定量输出结果对故障部件进行相应的维修或更换处理.
Different faults grade were identified by the diagnosis model of neural networks based on faults of the bushing of connecting rods on the diesel engine. The experiments of the setting and pending status were set to measure the vibration signals on the cylinder head. Then the fault diagnosis method based on the neural networks and wavelet analysis was used to identify different grades. At last, the pending status is estimated by the training model. According to the experiment and simulation result, for the setting status, diagnosis model can identify different grades quantificationally and accurately. For the pending status, the model also can describe the quantificational fault grades. So the operator can know the development of faults in time and maintain or change the parts by the output of network model.
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2007年第6期105-107,共3页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
关键词
柴油机
故障程度
振动分析
神经网络
diesel engine
fault grade
vibration analysis
neural network
