基于时域统计平均预处理的RBF网络发动机气缸压力识别方法

Identification of Diesel Engline Cylinder Pressure Based on RBF Network with Time Statistic Average

在柴油发动机气缸压力的识别过程中 ,针对缸盖振动信号的信噪比低且呈非平稳特性 ,影响神经网络对气缸压力识别的问题 ,提出了在径向基函数 ( RBF)网络训练前 ,对训练样本进行时域统计平均降噪预处理 ,再对实验样本进行识别 .结果表明 ,时域统计平均有效提高了训练样本的信噪比 ,RBF网络经训练后 ,对柴油发动机气缸压力的识别具有计算速度快。

As the identification of diesel engine cylinder pressure is influenced by the low signal noise ratio and nonstationary of the cylinder cover vibration signal, it was proposed to preprocess the training samples with the denoising method of time statistic average before the training of radial basis function network, and then, to proceed to identify the testing sample. It is proved by practice that the signal noise ratios of the training samples are enhanced effectively by the method, and the identification of diesel engine cylinder pressure is characterized with quick computer rate and high precision.