基于ANN的声目标识别系统及其DSP实时实现
详细信息 本馆镜像全文    |  推荐本文 | | 获取馆网全文
摘要
介绍了一种基于人工神经网络 (ANN)的声目标识别系统 ,概述了用MATLAB专用工具箱对神经网络权值进行训练及仿真的过程 ,叙述了ANN目标识别系统的数字信号处理器 (DSP)实时实现过程 ,并着重分析定点实现过程中程序变量的定标、非线性运算的实现、溢出的处理等关键步骤。对不同字长的识别结果进行比较表明 ,基于定点实时实现的系统数据保持很高的精度 ,可以得到与浮点处理相同的识别率
An acoustic target recognition system based on an artificial neural network is introduced, and the training and simulation of neural network based on matlab are also discussed. Then, the DSP real time implementation of the system is presented in detail. Some key processes, such as variables scaling, nonlinear operation and overflow protection are especially analyzed. Finally, the comparison of all recognition results shows that DSP real time implementation is accurate.
引文
[1]聂伟荣,朱继南,郭亚军,等.地震动信号的分析与目标识别.电子科技大学学报,2003,32(1):26~30
    [2]边肇祺,张学工,等.模式识别.北京:清华大学出版社,1988
    [3]张雄伟,陈 亮,徐光辉.DSP芯片的原理与开发.北京:电子工业出版社,1997

版权所有:© 2023 中国地质图书馆 中国地质调查局地学文献中心