基于组合优化的选择双谱在核爆地震特征提取中的应用

onventional methods of feature extraction of nuclear explosions and earthquakes only take advantage of up to second order statistics of the seismic signals, but there is a great deal of useful information contained in the higher order statistics. In this paper, higher statistic—bispectra is used in the procedure of feature extraction of nuclear explosions and earthquakes, which provides a novel approach for seismic feature extraction. After the validation of the effectiveness of radially integrated bispectrum and selected bispectrum, a feature extraction method called selected bispectrum of combinatorial optimization is proposed by utilizing the optimization method of clonal selection. Compared with the selected bispectrum based on best single features, the proposed method provides higher recognition accuracy and is more insensitive to path parameter dimension of feature space.