摘要
在无线电多径传播环境下,信号强度周期性变化易与多径效应波动叠加,导致测向和定位系统难以区分正常多径与信号源位置变化,易产生虚假信息,数据挖掘效果不佳。为此,本文提出一种基于模糊均值聚类的无线电监测数据挖掘方法。先通过小波变换对无线电信号进行分解,运用脊线算法提取瞬时频率特征。然后利用模糊均值聚类算法对瞬时频率特征进行初步聚类,并结合空间邻域信息优化聚类结果。最后计算每个聚类的异常因子,识别异常数据点,实现有用信息挖掘。实验结果表明,应用本文方法在多个实验场景下可显著提升数据挖掘速度、资源利用率和覆盖率,应用效果较好。
In radio multipath propagation environments,periodic variations in signal intensity can overlap with multipath effect fluctuations,making it difficult for direction-finding and positioning systems to distinguish between normal multipath phenomena and signal source location changes.This often leads to the generation of false information and suboptimal data mining outcomes.To address this issue,this paper proposes a radio monitoring data mining method based on fuzzy C-means clustering.First,wavelet transform is employed to decompose radio signals,and a ridge detection algorithm is used to extract instantaneous frequency features.Subsequently,the fuzzy C-means clustering algorithm is applied to perform preliminary clustering of the instantaneous frequency features,and the clustering results are optimized by incorporating spatial neighborhood information.Finally,the anomaly factor of each cluster is calculated to identify abnormal data points,thereby achieving effective information mining.Experimental results demonstrate that the proposed method significantly enhances data mining speed,resource utilization rate,and coverage rate across multiple experimental scenarios,indicating its practical effectiveness.
作者
路峥
陈景宇
LU Zheng;CHEN Jingyu(National Radio Monitoring Center Testing Center,Beijing 100041,China)
出处
《微处理机》
2025年第3期10-16,共7页
Microprocessors
关键词
模糊均值聚类
无线电监测
数据挖掘
特征提取
空间邻域信息
fuzzy C-means clustering
radio monitoring
data mining
feature extraction
spatial neighborhood information
