摘要
本文提出并实现了一种基于人工神经元网络理论的通信网信号识别系统的研制方案。该系统可以在复杂的线路及干扰情况下,实时完成传真、计算机Modem通信、BP机自动寻呼、拨号音、双音频、电子合成音、热线音乐、噪声和话音的自动识别工作。系统最大监测容量可达16384线,普通巡检单元的检出率超过90%,重点捕捉单元的检出率超过99.99%,系统虚警率低于10^(-5)。
This paper proposes and investigates the approach to distinguish signals in communication networks based on artificial neural networks. This system can be used to discriminate signals of FAX、 MODEM、Calling from BP machine, Dialling, DTMF, electrinical synthesized speech,busy-line music, noise and speech with complex line interference. The maximum capacity of 16384 lines, the patrolling unit recognition rate of abore 90%, the major unit recognition rate of above 99. 99%, and systems false alarm rate of 10-5 can be attained by the system.
出处
《通信学报》
EI
CSCD
北大核心
1995年第3期1-7,共7页
Journal on Communications
基金
武汉青年晨光计划项目
邮电部科研基金
国家博士后科研基金
关键词
神经网络
非话业务
通信网
监测
管理系统
signal processing, neural networks, network adaministering, pattern recognition,non-voice business