基于频域相关和SVM的宽带多信号测向方法研究

提出一种基于SVM的宽带测向新方法，利用频域相关技术提取宽带范围内感兴趣的信号的相位差，利用此相位差作为支持向量机的输入特征训练DOA（Direction Of Arrival）估计模型，对宽带范围内的多个窄带或宽带信号进行测向。训练后的模型针对的是一个较宽的频段范围，有效避免了其他算法要针对不同频率频繁调整模型的问题，大大降低了计算量。实验结果也验证了该方法具有较高的测向精度和较快的测向速度。

Multi-singles direction of arrival estimation in broadband based on frequency related and SVM

A novel direction of arrival estimation method based on support vector machine is proposed. This method, which uses the phase differences extracted by frequency related technique in a broadband as the inputs of the SVM（Support Vector Machine） and trains the DOA（Direction Of Arrival） estimation model, can estimate multi narrowband or wideband singles in a wide frequency band. Compared with other algorithm, as the model trained is for a wider range of frequency, the method can greatly reduce the computation efficiency by avoiding changing the algorithm model to match the single frequency. Experimental results prove this method has higher precision and quick direction finding speed.