基于自组织竞争神经网路的发动机故障诊断  

Fault Diagnosis of Engine Based on Self-organizing Competitive Neural Network

作　　者：巴寅亮[1,2] 王书提 谢鑫[1,2] BA Yinliang WANG Shuti XIE Xin(College of Mechanic and Traffic, Xinjiang Agricultural University, Urumqi 830052, China Xinjiang Key Laboratory of Agricultural Engineering Equipment Innovative Design, Urumqi 830052, China)

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：《机械制造与自动化》2017年第1期210-212,219,共4页Machine Building & Automation

摘　　要：自组织竞争神经网络自适应的学习能力拓宽了人工神经网络在模式识别和分类方面的应用,提出了基于自组织竞争神经网络的发动机电控系统故障诊断的方法。介绍了自组织竞争神经网络及学习规则,以长城哈佛GW2.8TC发动机为实验对象,让发动机在怠速状态下,对发动机进行故障设置,利用金德KT600故障诊断仪采集发动机的故障数据流,运用自组织竞争神经网络建立诊断模型,诊断结果表明自组织竞争神经网络能对故障进行识别和分类,准确率比较高。This paper uses the adaptive learning ability of self-organizing competitive neural network to broaden the application of artificial neural network in pattern recognition and classification, proposes a fault diagnosis method of the electronic controll system of the engine based on the self-organizing competitive neural network, introduces this network and its learning rule, takes the Great Wall Harvard GW2.8TC engine as the experimental object, keeps the engine at idle speed condition, sets up some faults for the en- gine, collects its failure data flow by kinder KT600 fault diagnosis instrument and uses the self-organizing competitive neural network to establish the diagnosis model. The diagnosis results show that this neural network can be used to identify and sort faults, the ac- curacy rate is relatively high.

关 键 词：自组织竞争神经网络 电控系统 故障诊断 

分 类 号：TK428[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]