基于SOM神经网络的发动机故障诊断  

Engine fault diagnosis based on SOM neural network

作　　者：王书提 巴寅亮[1,2] 郭增波 

机构地区：[1]新疆农业大学机械交通学院,新疆乌鲁木齐830052 [2]新疆农业工程装备创新设计重点实验室,新疆乌鲁木齐830052

出　　处：《自动化与仪器仪表》2015年第8期52-54 57,共4页Automation & Instrumentation

摘　　要：自组织特征映射神经网络(SOM)是一种无教师学习神经网络,主要用于对输入向量进行模式识别和区域分类。提出了基于SOM神经网络的发动机电控系统故障诊断的方法,介绍了SOM神经网络及其学习算法,以北京现代05款途胜G4GC型发动机电控系统为实验对象,让发动机在怠速工况下,并对其进行故障设置,利用金德KT600故障诊断仪采集发动机故障数据流,运用SOM神经网络建立诊断模型,诊断结果表明,SOM神经网络能对故障进行识别和分类,具有较好的聚类功能,具有一定的工程应用价值。Self-organizing feature map(SOM)neural network is a kind of non teacher learning neural network, mainly using for pattern recognition and classification of the input vector area. Proposing a fault diagnosis method of the electronic controlled system of engine based on SOM neural network, introducing SOM neural network and its learning algorithm, taking the electronic controlled system of Beijing Hyundai Tucson Paragraph 05 G4 GC engine as the experimental object, keeping the engine at idle speed condition, setting up some fault assumption for the engine, collecting the failure data flow of the engine by kinder KT600 fault diagnosis instrument, using SOM neural network to establish diagnosis model, the diagnosis results show that SOM neural network can identify fault and sort fault, with clustering function better, and hasing a certain value of engineering application.

关 键 词：SOM神经网络 发动机 电控系统 故障诊断 

分 类 号：TK407[动力工程及工程热物理—动力机械及工程] TP183[自动化与计算机技术—控制理论与控制工程]