基于时域融合的辐射源识别方法提高了单个传感器的识别准确度。文中提出了一种采用属性测度法获取基本概率赋值函数(BPAF),同时应用D-S(Dempster-Shafer)证据理论将多个量测周期识别结果进行有效融合的方法。该方法通过对已知辐射源库中的样本训练,获取样本特征参数的统计分布和权重来计算BPAF。当辐射源库中样本模式较多时,由于引入了统计的思想,在低信噪比情况下的正确识别率较其他时域融合方法得到提高。此外,由于大量的计算在样本识别之前已经完成,融合识别的速度很快。仿真和实验表明该算法是一种实时、有效的辐射源识别方法。

The method of emitter recognition based on the fusion in time domain improves the accuracy of single sensor. A new method to calculate the basic probability assignment function (BPAF) with attribute measurement is proposed. Further, D-S (DempsterShafer) evidence theory is applied to refuse the recognition results with multiple measuring circles in this method. By samples training to known emitter base, the BPAF can calculate through the statistical distributions of characteristic parameters of samples and weights. Because of introducing the statistical thought, the accuracy of recognition compared to other methods is improved in the lower signal-to-noise ratio while samples are multiple. Furthermore, a lot of calculation is finished before recognition, so the speed of recognition is fast. The results of experiments and simulation indicate that the recognition method is real-time and effective.