基于缸压高频ARMA模型的爆震识别方法  

Engine knock recognition based on ARMA model high frequency

作　　者：康健[1] 张力[1] 郭刚[1] 

机构地区：[1]重庆大学机械工程学院,重庆400030

出　　处：《传感器与微系统》2010年第11期64-66,共3页Transducer and Microsystem Technologies

基　　金：国防基础科研基金资助项目(C1020060350);重庆市自然科学基金资助项目(2007BA6026);重庆市科技攻关计划资助项目(2007AA6006-6)

摘　　要：研究以汽油机燃烧测试平台为基础,通过调整点火提前角以获得较大的爆震强度,并获得实验分析数据。选择低通燃烧压力信号峰值前10°后20°曲轴转角范围最做为爆震窗口。对随机选择的一系列爆震窗口燃烧压力高频振动信号构建时间序列模型,并以模型参数向量均值的马氏距离为爆震识别标识。用蒙特—卡罗法对非爆震工况序列马氏距离进行选择,获得爆震阈值。并通过工程上广泛使用的VDO算法验证这种爆震识别方法的可行性。This research is based on spark ignition （SI）engine combustion test bench, in which, the ignition advance angle of the engine be changed to get high knock intensity load cases. 10° and 20° crank angle before and after low-pass combustion pressure peak value was selected as knock sensitive window（KSW）. ARMA time serial model was chosen to simulate high-frequency vibration signal of a bench of KSW. Mahalanobis distance of vector, created by mean value of ARMA model＇ s parametric, was selected to indentify knock phenomenon. Threshold value,designed by Monte-Carlo approach on mahalanobis distance of non-engine knock cycles, is suitable to indentify knock case, and was tested by experiment data from SI engine combustion test. The feasibility of this method was verified by VDO method ,which was widely used in engineering.

关 键 词：爆震识别 燃烧测试 时间序列 ARMA模型 

分 类 号：TK464[动力工程及工程热物理—动力机械及工程]