基于缺失数据填补的辐射源识别方法

A Missing Data Imputation-Based Emission Source Identification Method

针对模板雷达特征参数残缺而造成的测量辐射源不能正确识别问题,提出了一种基于缺失数据填补的辐射源识别算法。该算法利用矢量神经网络对缺失数据进行填补,并对填补后的训练样本进行重新训练,从而得到网络结构参数。仿真实验表明本文方法不仅能处理缺失数据,而且在噪声环境下也能识别区间类型和标量类型的输入矢量。

To deal with the problem of emitter identification caused by the fragmentary feature parameters of the template radars,this paper proposes a new missing data imputation（MDI） based emission source identification method,a vector neural network（VNN） is used to substitute the missing feature parameters and make use of substituted training samples for training VNN to obtain the structure parameters of the network.A number of simulations are presented to demonstrate the performance of the MDI algorithm.Simulation results indicate that the MDI algorithm can not only deal with the missing data,but also can identify the scalar input data and interval-value input data correctly in noisy environment.