| 摘要: |
| 针对复杂电磁环境下传统信号检测阈值设定方法存在的设定困难、适应性差、误检/漏检率高以及弱信号检测难度大等问题，提出了一种电磁频谱信号检测的智能化门限阈值设计方法。利用深度强化学习构建决策智能体，实现对一维电磁频谱的自主认知，智能地以分段方式提供自适应的信号检测门限阈值，实现信号检测率91.9%，弱信号（信噪比小于等于5 dB）检测率提升1.8%的效果。该技术可自动化实现前端侦测装备对电磁信号的自主检测，整个过程无需人工参与，已在无人自主信号检测的实际项目中进行了测试，各项指标均满足实际工程应用的要求。 |
| 关键词: 信号检测 宽带频谱数据 自适应门限值 决策智能体 |
| DOI：10.20079/j.issn.1001-893x.240620009 |
|
| 基金项目: |
|
| An Intelligent Threshold Design Method for Electromagnetic Spectrum Signal Detection |
| HU Chao |
| (1.Southwest China Institute of Electronic Technology,Chengdu 610036,China;2.School of Electrical Engineering,Southwest Jiaotong University,Chengdu 611756,China) |
| Abstract: |
| In view of the problems existing in the traditional signal detection threshold-setting methods in complex electromagnetic environments,such as difficult threshold setting,poor adaptability,high false-alarm/miss-detection rates,and great difficulty in weak signal detection,an intelligent threshold design method for electromagnetic spectrum signal detection is proposed.A decision-making intelligent agent is constructed using deep reinforcement learning to achieve autonomous cognition of one-dimensional electromagnetic spectra.It can intelligently provide adaptive signal detection threshold values in a segmented manner,achieving a signal detection rate of 91.9% and a 1.8% increase in the detection rate of weak signals(signal-to-noise ratio≤5 dB).The technology can automatically enable front-end detection equipment to autonomously detect electromagnetic signals without human intervention throughout the process.It has been tested in practical projects of unmanned autonomous signal detection,and all indicators meet the requirements of practical engineering applications. |
| Key words: signal detection broadband spectrum data adaptive threshold value decision agent |
《电讯技术》编辑部 