摘要
应用小波包技术,提出了融合能量代价函数的概念及基于此函数的水声信号识别算法。算法以融合能量代价函数为标准,在整个小波库中确定特征空间及特征值。在MATLAB仿真环境下进行的BP神经网络实验显示:与固定尺度小波包能量法及最大距离小波包能量法相比,算法对特征模糊的信号有较好的识别效果。
Propose concept of fusing energy cost function and a method that based on the function to underwater acoustic signal with wavelet pocket technique. The algorithm confirms characteristic space and eigenvalue in wavelet library with criterion of fusing energy cost function. The simulant experiment of BP neural network in MATLAB shows that this algorithm has better recognition effect to fuzzy-feature signal compared with the fixed-scale wavelet-pocket energy method and the maximal distance wavelet-pocket energy arithmetic.
出处
《计算机应用与软件》
CSCD
北大核心
2005年第8期26-27,124,共3页
Computer Applications and Software
基金
广东省科技攻关项目(A1020103)。
关键词
水声信号识别算法
仿真
模式识别
小波包技术
能量代价函数
Fusing energy cost function Wavelet pocket Underwater acoustic signal BP neural network Characteristic space
