Abstract: Crankshaft speed signals of engine are non-stationary, and it is difficult to extract misfire fault information from them effectively. For this purpose, a misfire detection method of engine based on EEMD (Ensemble Empirical Mode Decomposition) is proposed. The EEMD method can adaptively decompose a crankshaft signal into several IMFs ( Intrinsic Mode Function). The IMF component which contains fault information can be determined. Through observing the abnormal amplitude fluctuations of the IMF, the time range of engine misfire can be apparently estimated. Besides, a simulation model of engine is built by AMESim, and the crankshaft speed signals of three conditions are collected. Then these signals are decomposed by EEMD respectively to detect misfire fault. The results show that this method can effectively extract fault information to accomplish the off-line detection of misfire fault, and it can also be used as the foundation of on-line detection.

Key words: engine, misfire fault, crankshaft speed, ensemble empirical mode decomposition(EEMD)