摘要
现代信息技术和人工智能技术的飞速发展,为日趋复杂的机械设备故障诊断提供了技术支持;人工神经网络具有良好的分类特性,适合故障诊断;为了克服单神经网络故障诊断缺陷,研究了一种基于信号类型和信息融合思想的集成神经网络的智能故障诊断方法,降低了网络的复杂度,同时由于综合考虑了各种信息对故障特征的反映,提高了故障确诊率;为了提高学习效率,采用了一种快速的网络训练算法—免疫遗传算法;以某型兵器传动装置中的齿轮故障诊断为例,采用振动和声发射检测等技术获取训练样本训练网络,专家给出各网络对故障类型的置信权矩阵,将现场采集数据输入到诊断系统中,系统准确快速地得到了设备故障类型,结果符合实际,表明该方法行之有效。
The quick development of modern information and artificial intelligence technology provides a technical surport for fault diagnosis of complex mechanical equipment.Artificial neural network,with good classifying ability,is suiltable for fault diagnosis.In order to overcome the limitation of fault diagnosis using single neural network,an intelligent fault diagnosis method of integrated neural network based on signal type and information integration is studied.The method reduces the complex degree of neural network,at the meantime,because of the synthetical consideration of various information reflecting fault character,it increases preciseness of the fault diagnosis.In order to increase learning efficiency,a kind of quick learning method—Immune Genetic Algorithm is used.Take fault diagnosis of the gear in transmission equipment of a certain weapon for example,use modern technology such as vibration and acoustic emission testing to get sample to train the network,and the expert provides the confidence weight matrix of each network against fault type,after inputing the data aquired on real time to the diagnosis system,it presents the fault type of the equipment quickly and exactly,the result obeys the fact,which indicates that the method is feasible.
出处
《计算机测量与控制》
CSCD
北大核心
2010年第10期2219-2221,2231,共4页
Computer Measurement &Control
关键词
机械设备
智能故障诊断
信息融合
学习算法
mechanical equipment
intelligent fault diagnosis
information integration
learning method