基于t-SNE降维和DBSCAN算法的多参数雷达信号分选方法

Multi-Parameter Radar Signal Sorting Based on t-SNE Dimensionality Reduction and DBSCAN Algorithm

摘要:

随着雷达信号的复杂性增加，传统的信号处理方法逐渐暴露出诸多不足之处。本文提出了一种基于t-分布随机邻居嵌入（t-SNE）降维技术与基于密度的聚类算法（DBSCAN）的雷达信号分选算法，旨在解决多参数雷达信号分选中的挑战。t-SNE通过降低数据的维度，能够有效提取出数据的主要特征并减少噪声和冗余信息，从而为后续的DBSCAN聚类提供了更清晰的数据分界。实验生成了五种不同类型的雷达信号数据，并使用t-SNE和DBSCAN进行降维和聚类，实验结果显示：t-SNE降维结合DBSCAN聚类算法在纯度和轮廓系数等指标上均表现出色，验证了该方法在复杂雷达信号分选中的有效性。

Abstract:

This study explores the combination of t-Distributed Stochastic Neighbor Embedding (t-SNE) dimensionality reduction technique and the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm to address the challenges in multi-parameter radar signal sorting. As the complexity of radar signals has been increasing, traditional signal processing methods have revealed limitations. t-SNE effectively extracts essential features from the data by reducing dimensionality, eliminating noise and redundant information, and providing a clearer boundary for subsequent DBSCAN clustering. In the experiment, we generated five different types of radar signal data and conducted analyses using t-SNE and DBSCAN. The results show that the t-SNE dimensionality reduction combined with the DBSCAN clustering algorithm performs well in terms of purity and silhouette score, confirming the effectiveness of this method in complex radar signal sorting.