基于双边线性预测的自适应时频分析

针对非平稳信号的时频分析,采用所谓的双边线性预测方法代替传统的单边线性预测。在双边线性预测中,采用若干过去值和将来值的线性组合形式来对信号建模;采用基于最小二乘的噪声梯度法来估计预测系数向量。数值仿真结果表明,双边线性预测方法比单边方法具有更高的分辨率,同时所需要的阶次比较小,因而是一种有效的计算方法。

Adaptive Time-Frequency Analysis Based on Two-Sided Linear Prediction

One-sided linear prediction (OLP) is often used for time-frequency analysis of nonstationary signals,an alternative approach based on two-sided linear prediction (TLP) is proposed.In the TLP,the signal is modeled as a linear combination of the past and future points,and the prediction coefficients are computed adaptively by the noisy gradient technique based on the least mean square.Numerical simulation results show that the TLP is able to achieve a higher resolution than OLP even with more small order,so TLP is an efficient algorithm.