CAJ | 학술논문

针对柴油机多发故障，提出了自适应奇异值标准谱和经验模态分解（Empirical Mode Decomposition ， EM D ）相结合的故障诊断模型。通过计算平均最近邻域发散度和奇异值标准谱的方法自适应地选择奇异值分解的嵌入维数和重构阶数，提高了奇异值分解降噪的精度。对降噪后的信号进行EM D分解，并利用调整余弦相似度标准提取反映信号真实特征的主固有模态函数（Intrinsic Mode Function ，IMF），进而提取故障特征参数。将此模型应用于F3L912柴油机进气门漏气、单缸失火和多缸失火等故障的诊断，通过提取峭度和过零率作为故障特征，获得了较高的故障分类准确率。
침대시유궤다발고장，제출료자괄응기이치표준보화경험모태분해（Empirical Mode Decomposition ， EM D ）상결합적고장진단모형。통과계산평균최근린역발산도화기이치표준보적방법자괄응지선택기이치분해적감입유수화중구계수，제고료기이치분해강조적정도。대강조후적신호진행EM D분해，병이용조정여현상사도표준제취반영신호진실특정적주고유모태함수（Intrinsic Mode Function ，IMF），진이제취고장특정삼수。장차모형응용우F3L912시유궤진기문루기、단항실화화다항실화등고장적진단，통과제취초도화과령솔작위고장특정，획득료교고적고장분류준학솔。
For the multiple faults of diesel engine ,fault diagnosis model consisted of adaptive singular value standard spectrum and empirical mode decomposition (EMD) was proposed .The average divergence of neighboring area and singular value stand‐ard spectrum were calculated to determine the embedding dimension and reconstruction order adaptively and hence the precision of noise reduction with the singular value decomposition improved .EMD of signal was conducted after the noise reduction ,the main intrinsic mode function (IMF) with the real characteristic was extracted according to the adjusted cosine similarity stand‐ard and the fault characteristic parameters were extracted .With the model ,diesel engine faults including inlet valve leakage , single cylinder misfire and multiple cylinder misfire of F3L912 were diagnosed .Kurtosis and zero crossing rate were extracted as the fault characteristic parameters and the classification accuracy improved .