低空复杂频谱环境中占用与干扰频段检测

打开文本图片集

中图分类号：TN92；TN972；TN98 文献标志码：B 文章编码：672-7274（2025）-003-04

Abstract： An integrated system for occupiedand interfering frequency band detection is proposed， basedon the fusion of signal processing and deep learning systems.In terms of signal processing，a spectrum-analysis-based correction method for occupied frequencybands is proposed to tackle frequency offset errors and detected bandwidth distortion.Additionall，an identification method based on fixed bandwidth segmentation and an interference classification model is proposed to detect interfering frequency spectrum intervals.In the feld of deep learning，a spectrum occupancydetection methodcombining Welchtransformand UNetnetwork is proposed forthe first time，andan improved RT-DETR model is introduced forspectrum interference detection.Finally，thedual systemsare fused toreduce the missed detectionrateand falsealarmrate.Experiments demonstrate that the dual-system fusionscheme significantly improves performance，achieving an Fl score of 94.19% in occupied frequency band detection tasks.

Keywords： signal detection; signal processing; deep learning; target detection

研究背景

随着低空经济的高速发展，商用无人机、Wi-Fi、蓝牙等各种窄带、宽带、突发式甚至非结构化无线信号广泛存在于电磁频谱中，导致频谱环境复杂且动态变化，干扰多样化，频谱的动态管理与智能监测成为通信系统的挑战之一。（剩余5237字）