基于自适应滤波器和径向基函数网络的内燃机气缸压力识别

Identification of Diesel Engine Cylinder Pressure Based on RLS and RBF

利用鲁棒的递推最小二乘自适应滤波,对非平稳时变性质的柴油机振动信号进行了降噪处理,然后设置软阈值,即将门限以下的信号设置为零,避免了长数据序列的累计误差;进一步去除信号的背景噪声,提高信噪比;最后,利用径向基函数网络强大的函数逼近功能,识别内燃机气缸压力。通过实验表明了该方法的有效性和鲁棒性。

Based on our experiments, the author firstly used the robust recursive least square (RLS) adaptive filter to denoise the sampled signal of diesel engine head vibration, which was nonstationary. Then, softthresholding was set to further removing the background noise, i.e. to set the value below the softthreshold to zero. Thus the cumulated error of the long data sequences is avoided and the signal noise ratio is improved. Finally, the cylinder pressure is identified by using radial base function (RBF) network, which is featured with the powerful functional approach. It is practically proved that the method is valid and robust.