基于神经网络的电喷发动机传感器故障诊断的研究  

Research into the Electrical Injection Engine Sensor Fault Diagnosis Based on the Neural Network Technology

作　　者：杨金玉[1] 仝秋红[1] 蹇小平[1] 肖锋[1] 

机构地区：[1]长安大学汽车学院,陕西西安710061

出　　处：《电子科技》2008年第4期20-23,共4页Electronic Science and Technology

摘　　要：基于传统专家系统存在的缺点,文中以人工神经网络技术在电喷发动机故障诊断方面的应用为基础,以MATLAB7.0为开发平台,在怠速工况条件下,对无故障、氧传感器故障和空气流量传感器故障3种状态进行数据采集,并对数据进行预处理;基于BP神经网络的优点,构建了3层BP网络模型,经过测试样本的检验,证明此模型误差小、精度高,具有可应用性。In view of the shortcomings of traditional expert systems, this paper presents the application of artificial neural network technology in the fault diagnosis of electrical ejection engine. Developed on the platform of MATLABT. 0, the system collects data in the states of no fault, oxygen sensor fault and air flow sensor fault under idle speed of work instance condition, and the data are pretreated. Based on the advantages of the BP neural network, a three-layer model BP network is constructed. By verifying the testing stylebooks, the model is proved to be practical with little error and high precision.

关 键 词：神经网络 电喷发动机 传感器 故障诊断 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]