基于神经网络的电控汽油发动机的智能故障诊断研究

详细信息    本馆镜像全文|  推荐本文 |  |   获取CNKI官网全文

英文题名：Study of Intelligent Diagnosis about Electronic Control Engine Fault Based on Neural Network 作者：李明钊 论文级别：硕士 学科专业名称：控制理论与控制工程 中文关键词：电控汽油机 ; 智能故障诊断 ; BP神经网络 ; 模糊BP神经网络 英文关键词：electric control gasoline engine ; intelligent fault diagnosis ; BP neural network ; fuzzy BP neural network 学位年度：2008 导师：杨承志 学科代码：081101 学位授予单位：昆明理工大学 论文提交日期：2008-03-01

摘要

随着汽车工业的迅速发展,汽车生产量和保有量不断增长,而且汽车制造技术的不断进步,使汽车发动机的越来越复杂。2008年将在全国实施机动车污染物排放国Ⅲ标准(相当于欧Ⅲ)标准,并强制要求安装车载诊断系统(简称OBD)。
     在现代化社会中,车用汽油机的故障诊断技术越来越受到重视,如果车用汽油机的某些部位出现故障而未能及时地发现和排除,其结果不仅会导致汽油机本身的损坏,甚至可能会造成车毁人亡的严重后果。近年来,我国汽车工业得到了迅速的发展,给人们生活带来了极大的方便,但是由于设备更复杂与数量更多给汽车维修人员带来了难题。因此研究车用汽油机的智能故障诊断技术具有实际意义。而且汽车的安全运行问题受到越来越多的关注,加强汽车的安全技术检测,成为有待研究解决的重要课题。
     在这样的背景下,本文针对传统故障诊断专家系统获取知识的瓶颈,不具备自学习的功能,采用人工神经网络和模糊理论来研究电控汽油机的智能故障诊断。针对电控汽油机的的怠速或怠速控制阀故障、点火线圈故障、点火正时不对、火花塞故障、节气门故障、进气管漏气、空气滤清器故障、喷油器故障、燃油供给系统故障、冷却系统故障及润滑系统故障,设计了BP诊断网络和模糊BP诊断网络。根据神经网络的特点,指出采用神经网络进行故障诊断的可行性。仿真结果表明对于电控汽油机的故障诊断而言,BP网络确实为一种较为实用的神经网络,它具有很强的模式识别和分类能力。由于电控汽油机故障的复杂性和模糊性,采用传统的以布尔代数为基础的二值逻辑显得过于粗糙不精确,因此引入模糊逻辑的概念,构造了模糊神经网络,并用它进行电控汽油机的故障诊断。仿真结果表明,将模糊逻辑引入神经网络后,对知识的表示更加准确,不仅对输入故障现象的描述更加细致,而且对输出故障的原因也有明确的解释,更符合人们的思维习惯。
With the rapid development of the auto industry, auto production and keep growing volume, and automobile manufacturing technology continues to progress, so that the automobile engine become increasingly complex. the standards III of emissions of motor vehicle will be implemented nationwide(equivalent to Euro III) , and mandating the installation of on-board diagnostic systems (OBD) in 2008.
     In modern society,it is payed more and more attention to the automotive gasoline engine fault diagnosis technology, if the automotive gasoline engine occur failure in certain areas but not in time to discover and remove, and the results will not only lead to damage to gasoline engine itself, and it might even cause a fatal car crash of the serious consequences. In recent years, China's auto industry has developed rapidly , so improve our lives, but because of more complex equipment and the number of vehicle become more , this is a problem to maintenance staff.It is practical significance to study the automotive gasoline engine intelligent fault diagnosis technology. And the safe operation of motor vehicles is more and more concerned,so enhance vehicle safety testing technology that would be important to study and solve the issue. Against such a backdrop, in this paper , traditional fault diagnosis expert system has knowledge acquisition bottleneck, and does not have the self-learning function, using artificial neural networks and fuzzy theory to study intelligent fault diagnosis electronic control gasoline engine. In allusion to idling or idling control valve failure, ignition coil failure , ignition timing is wrong, sparkplug fault ,throttle failure, leak into the trachea, the air filter failure, fault injector ,fuel supply system failure,cooling system failure, lubrication system failure of the electronic Control gasoline engine, I have designed BP diagnostic network and fuzzy BP diagnostic network. According to the characteristics of neural networks, indicated that neural network is feasible and inevitable for fault diagnosis. BP network is used for fault diagnosis for gasoline engine, simulation results show that BP network is a more practical neural network for fault diagnosis of electric control gasoline engine, its pattern recognition and classification is better. Because of complexity and ambiguity of diagnosis of electric control gasoline engine ,it is too rough and imprecise to adapt traditional two-valued logic based on Boolean algebra,therefore the concept of fuzzy logic is introduced, and fuzzy neural network is constructed, and it used to carry out fault diagnosis for electronic control gasoline engine. The simulation results show that fuzzy logic is used to neural networks, the expression of knowledge is more precise, it is not only more detailed description on the importation fault phenomenon, but also is a clear explanation for the reasons for the output fault, it is more in line with our thinking habits.

引文

[1]潘书澜,基于BP神经网络的电控汽油机传感器故障诊断与调节研究[硕士论文],浙江大学,2006.1
    [2]V.Krishnaswami,G.C.Luh and G..Rizzoni.Nonlinear Parity Equation based residual Generation For Diagnosis of Automotive Engine Faults.Control Engine Practice,Vol.3.No.10.pp,1385-1392,1995
    [3]肖云魁,汽车故障诊断学.北京:北京理工大学出版社,2001.7
    [4]楼顺天,施阳,基于MATLAB的系统分析与设计--神经网络.西安:西安电子科技大学出版社,1998.9
    [5]王文成,神经网络及其在汽车工程中的应用.北京:北京理工大学出版社,1998.7
    [6]胡德文,神经网络自适应控制.长沙:国防科技大学出版社,2006.1
    [7]董长虹,神经网络与应用.北京:国防工业出版社,2005.1
    [8]周志华,曹存根.神经网络及其应用.北京:清华大学出版社,2004.9
    [9]飞思科技产品研发中心.神经网络理论与MATLAB7实现。北京:电子工业出版社,2005.3
    [10]鄂加强.智能故障诊断及其应用.长沙:湖南大学出版社,2006,12
    [11]吴今培,肖健华.智能故障诊断与专家系.北京:科学出版社,1997.9
    [12]田宏伟.加热炉钢坯温度场智能预报系统研究[硕士论文],昆明理工大学,2002.3
    [13]高隽.人工神经网络原理及仿真实例.北京:机械工业出版社,2003.7
    [14]刘金琨,先进PID控制MATLAB仿真.北京:电子工业出版社,2004.9
    [15]曹龙汉。柴油机智能化故障诊断技术.北京:国防工业出版社,2005.1
    [16]李国勇.智能控制与MATLAB在电控发动机中的应用.北京:电子工业出版社,2007.6
    [17]张静.MATLAB在控制系统中的应用.北京:电子工业出版社,2007.5
    [18]肖阳,智能故障诊断系统设计与应用研究[硕士学位论文],北京:北京航空航天大学,2004.1
    [19]孙颖楷.内燃机智能故障诊断系统的研究及应用[博士学位论文],重庆:重庆大学,2001.9
    [20]王晓根.基于神经网络与专家系统的智能故障诊断系统研究[硕士学位论文],南京: 东南大学,2005.4
    [21]张捷.基于神经网络的齿轮箱智能故障诊断的技术研究[硕士学位论文],南京:江苏大学,2003.6
    [22]王政.基于人工神经网络的摊铺机智能故障诊断系统研究[硕士学位论文],长沙:中南大学,2004.12
    [23]花伟.锅炉混合智能故障诊断技术的研究[硕士学位论文],南京:南京工业大学,2006.6
    [24]李先民.现代汽车电子控制系统故障诊断与检修.北京:人民邮电出版社,2004
    [25]王道平,张义忠.故障智能诊断系统的理论与方法.北京:冶金工业出版社,2001
    [26]邹长庚.现代汽车电子控制系统构造原理与故障诊断.北京:北京理工大学出版社,1999
    [27]王遂双.汽车电子控制系统的原理与检修.北京:北京理工大学出版社,2002
    [28]庄继德.汽车电子控制系统工程.北京:北京理工大学出版社,1998
    [29]周东华,叶银忠.现代故障诊断与容错控制.北京:清华大学出版社,2000
    [30]蔡自兴.智能控制.北京:电子工业出版社,2004
    [31](美)Patrick Henry Winston著;崔良沂、赵永昌译.人工智能:第3版.北京:清华大学出版社,2005.9
    [32]史忠植,王文杰.人工智能.北京:国防工业出版社,2007.2