基于爆震因子统计特性的爆震燃烧识别方法

Knock Diagnosis Based on Statistical Characteristics of Knock Factor

汽油机燃烧诊断广泛采用Siemens VDO算法识别非正常爆震燃烧,爆震识别及强度评价严重依赖主观性经验所确定的爆震阈值.所进行的试验测试分析认为,爆震因子具有随机性特征,其值在很大程度上仅具有相对意义而非绝对性地指征爆震事件.基于爆震因子统计特性的分析,采用D检验、Lilliefors检验等正态检验手段进行爆震事件的识别.研究表明:非爆震工况爆震因子服从正态分布,爆震工况爆震因子不服从正态分布;通过依次剔除爆震工况爆震因子峰值个体可使样本正态化,并据此识别爆震事件.基于爆震因子统计特性的爆震识别方法能够避免爆震阈值的主观性经验取值,同时该爆震识别方法对爆震窗口宽度不具敏感性.

Siemens VDO method is widely used in knock detection in gasoline engine.However,identification of knock event and evaluation of knock intensity strongly depend on the threshold of Knock Factor which is subjectively determined by experience.Research demonstrates that Knock Factor signal is random,and the signal gives a knock event in the relative way rather than the absolute way.Statistical knock diagnosis method based on distribution characteristic of Knock Factor sample was investigated by normality test such as the Lilliefors test and the D test.It is confirmed that the Knock Factor sample with-out knock event obeys the law of normal distribution,and the Knock Factor sample with knock event does not but can be normalized by step-by-step kicking out the peak individual of the Knock Factor sample until the sample satisfies the normality test.The kicked out individual can be identified as knock event.Statisti-cal knock diagnosis method can avoid deciding subjectively the empirical threshold of Knock Factor,and is insensitive to the Knock Window width of Siemens VDO algorithm.