运用Vold-Kalman阶比跟踪的发动机失火故障在线诊断  

Method for Diagnosing Misfiring Fault of Engine Online Based on Vold-Kalman Order Track

作　　者：孙宜权[1,2] 张英堂[1] 李志宁[1] 程利军[1] 李志伟[2] 

机构地区：[1]军械工程学院车辆工程系 [2]66267部队

出　　处：《振动．测试与诊断》2013年第6期1014-1018,1095,共5页Journal of Vibration,Measurement & Diagnosis

基　　金：国家自然科学基金资助项目(50775219)

摘　　要：发动机一个工作循环内,其燃爆次数等于气缸数,燃爆产生的振动信号频率与转速相关。由于排气噪声的频率与转速也相关,并且排气噪声强度随燃爆振动强度变化,因此,提出了一种通过测量排气噪声间接检测发动机失火故障的方法。利用等角度采样阶比跟踪方法分析了各阶比分量幅值的变化规律,通过Vold-Kalman阶比跟踪方法提取了特定的阶比分量,以其能量与原信号总能量比值作为特征参量,利用极限学习机,实现了发动机失火故障的在线诊断。应用实例分析表明,发动机失火故障诊断准确率接近100%,诊断耗时不超过0.1s。The number of engine combustion is equal to its cylinder number within one working cycle.The frequency of vibration signal which is generated by engine combustion is correlation to engine speed.The frequency of exhaust noise is also correlation to engine speed,and its strength is varying as combustion strength changes.So a method for measuring the engine no burning fault utilizing exhaust noise is presented in this paper.The amplitude varying of of each order is analyzed through order track of constant angle resample.Some specific orders are extracted using Vold-Kalman method,and the order signal power divide the total power of original signal as feature value.The engine misfiring fault diagnosing online is implemented through extreme learning machine.The analysis of application example shows that the accuracy of engine misfiring fault diagnosis is approximate to 100%and the diagnosis time is no more than 0.1second.

关 键 词：发动机 排气噪声 等角度采样 Vold—Kalman阶比跟踪 极限学习机 

分 类 号：TK421[动力工程及工程热物理—动力机械及工程] TK413.4