在理解线性预测分析(LPC)方法及其自相关法和自协方差法两种高效解法基础上,根据舱音短时平稳特点,提出了基于LPC原理的Durbin迭代的舱音信号特征提取与重构方法,构建了舱音信号特征提取与重构方法的步骤,得到了详细图形和结果。获得了最佳预测阶数P、满足线性预测解稳定条件等有益结论。研究表明:基于线性预测分析方法得到的参数和结论为舱音信号特征识别提供了一条比较好的新途径。 Based on the LPC method and two efficient methods of autocorrelation and autocovariance, according to the short-term stability of the vocals, the Durbin iterative feature extraction and weighting based on LPC principle Structure method, the step of extracting and reconstructing the signal of cabin signal is constructed, and the detailed graph and result are obtained. Obtain the best prediction order P, and meet the beneficial conclusion such as the stability of linear prediction solution. The results show that the parameters and conclusions obtained based on the linear predictive analysis method provide a new and better way for feature recognition of the cabin signals.