基于稀疏编码特征的多传感器辐射源识别

Multi-sensor Radiation Source Identification Based on Sparse Coding Feature

针对单传感器在低信噪比下对辐射源信号的识别性能较差的问题,提出了一种基于稀疏自编码器的多传感器辐射源融合识别的方法。该方法首先采用稀疏自编码器进行特征提取,得到信号样本的特征模板;然后利用特征匹配方法,得到匹配差值;最后将匹配差值转化为D-S证据理论的基本概率赋值函数,通过改进的D-S证据对多个传感器进行融合得到最终结果。仿真实验结果表明,融合算法能够有效地提取信号特征,可进一步提高单传感器的识别性能,在小样本、低信噪比条件下,具有更强的识别优势。

Aimed at the low recognition rate of emitter signal for the single sensor when the signal-to-noise ratio is relatively low,an emitter fusion recognition of multi-sensor method based on Sparse AutoEncoder(SAE)was proposed.Firstly,the feature template for signal was obtained by SAE.Then,the method of feature matching was used to calculate the matching difference.Finally,the matching difference was translated into basic probability assignment function,and the recognition result was acquired through D-S evidence theory for multi-sensor.Experimental results showed that the fusion method could extract signal feature effectively and further improved the recognition performance of single sensor.The proposed method had a stronger recognition with small-sample-size and low signal-to-noise ratio.