提出一种新的柴油机瞬时转速信号的故障特征的提取方法,柴油机瞬时转速信号经过小波降噪处理,有效地剔除柴油机瞬时转速信号的噪声干扰,提高信号的信噪比。提取单缸作功冲程峰谷值差值作为特征值,通过时域曲线和特征值变化曲线可以明显看出单缸断油的故障,建立起能量到柴油机故障的映射关系,最后利用BP神经网络训练可以得到很好的效果。 A new method to extract the fault characteristics of instantaneous speed signal of diesel engine is proposed. The instantaneous speed signal of diesel engine is processed by wavelet to reduce noise effectively and increase signal-to-noise ratio of diesel engine instantaneous speed signal. The single-cylinder power stroke peak-trough difference is extracted as the characteristic value. The time-domain curve and eigenvalue variation curve can clearly show the single-cylinder fuel cut failure and establish the mapping relationship of energy to diesel engine failure. Finally, BP neural network Training can be very good results.