一种基于径向基神经网络的阵列误差非参数化校正新方法  

A New Nonparametric Method of Array Errors Calibration Based onRBFNeuralNetwork

作　　者：张兴良 王粲 曲建军 ZHANG Xing-liang;WANG Can;QU Jian-jun(A Unit of PLA,Quzhou,Zhejiang 324000,China)

机构地区：[1]中国人民解放军某部,浙江衢州324000

出　　处：《通信对抗》2024年第2期31-34,共4页Communication Countermeasures

摘　　要：当离散测量点较少时,已有的非参数化阵列误差校正方法插值精度不够。为此,提出一种新的基于径向基神经网络的宽带阵列误差拟合方法。首先在不同频率和方向上离散测量阵列误差的幅度和相位,然后通过径向基神经网络算法对阵列误差函数进行拟合,最后用拟合结果对阵列输出信号进行校正。该方法充分利用了径向基神经网络在非线性插值方面的优势,具有较高的插值精度。仿真试验结果表明该方法有效。When the discrete measurement points are few,the interpolation of the existing nonparametric calibration for array error is not precise enough.To settle this problem,a new nonparametric method based-on RBF neural network is proposed in this paper.First,the amplitudes and phases of array errors are measured discretely at different frequencies and directions.Then,the error functions are fitted using RBF neural net-work algorithm.Finally,the array signals are calibrated with the fitting results.Due to the advantages of RBF neural networks algorithm in nonlinear interpolation,the interpolation accuracy of the new method is very high.Some simulation experiments demonstrate the effectiveness of the new method.

关 键 词：阵列天线 径向基神经网络 非参数化方法 误差校正 非线性插值 

分 类 号：TN911.7[电子电信—通信与信息系统]