Alpha稳定分布噪声条件下的广义S变换时频分析  

Time-Frequency Distribution of Generalized S-Transform under Alpha Stable Distributed Noise Conditions

作　　者：舒彤[1] 余香梅 查代奉[1] 邱天爽[3] 

机构地区：[1]九江学院电子工程学院,江西九江332005 [2]江西省数控技术与应用重点实验室,江西九江332005 [3]大连理工大学电子与信息工程学院,辽宁大连116024

出　　处：《信号处理》2014年第6期634-641,共8页Journal of Signal Processing

基　　金：国家自然科学基金(61261046;60372081);江西省教育厅科技项目(GJJ11244;GJJ11245;GJJ11621)

摘　　要：在分析Alpha稳定分布噪声干扰下的非高斯信号时,基于传统二阶广义S变换时频算法效果显著退化。针对该问题,提出了基于分数低阶的广义S变换时频分析(FLO-GST)算法。该算法将分数低阶谱理论和常规广义S变换时频分布算法有机结合,同时优化窗口调节因子改善时频聚集性,得到较好的时频图谱。通过计算机仿真,并与其他时频算法进行对比,结果表明,该算法能有效抑制脉冲噪声干扰,更好的描述稳定分布噪声下非平稳信号的时频图谱,具有良好的韧性。In order to solve the problem that generalized s-transform time-frequency distribution method is invalid under α-stable distribution noise environments. This paper proposes an improved fractional lower order generalized s-transform （FLO-GST） algorithm. It combines the theory of fractional lower order spectrum with conventional generalized s-transform time-frequency distribution algorithm, and at the same time introduces regulatory factor of window function to improve time-frequency concentration. The simulation results and analysis show that, compared to conventional methods of time-frequency analysis, the proposed method is more robust under s-stable distribution interference environments than conventional methods.

关 键 词：ALPHA稳定分布 广义S变换 时频分析 分数低阶谱 

分 类 号：TN911.7[电子电信—通信与信息系统]