交叉验证的GRNN神经网络雷达目标识别方法研究

Radar Target Recognition Based on Cross Validation GRNN

雷达目标识别是现代雷达技术的重要发展方向之一,在未来武器系统中具有重要的意义。针对高分辨一维距离像在预处理中平移敏感性的问题,使用了全局最小熵距离对齐算法,能够较准确快速对准距离像单元,提高了距离对齐的精度。为了提高雷达目标识别的准确率,提出了一种与广义回归神经网络模型(generalized regression neural network,GRNN)相结合的目标识别方法。利用K重交叉验证法对神经网络训练,并且根据最小均方误差寻找出GRNN神经网络光滑因子spread的最优值,同时获得目标识别训练样本的最优输入输出值。通过对比,取得最优光滑因子的GRNN神经网络将大幅度提高其收敛速度与泛化能力。仿真实验证明,基于改进GRNN神经网络的雷达目标识别可以获得较高较稳定的识别正确率。

As one of the important development directions of modern radar technology,radar target recognition is of great significance to the future weapon system. Aiming at the problem of translational sensitivity in high resolution range profile（ HRRP） in preprocessing,the global minimum entropy distance alignment algorithm is used to accurately align the range profile and improve the accuracy of distance alignment. To improve the accuracy of radar target recognition,a new method of target recognition based on generalized regression neural network（ GRNN） is proposed. The K-fold cross-validation method is used to train the neural network,and the optimal value of the smoothing factor spread of GRNN neural network is found according to the mean square error（ MSE）. The optimal input and output values of the target recognition training samples are obtained. By contrast,GRNN neural network with optimal smoothing factor spread will greatly improve its convergence speed and generalization ability. The simulation results show that the radar target recognition based on the improved GRNN neural network can obtain higher and more accurate recognition accuracy.