【摘要】 通过将自适应小波神经网络 (AWNN)中的小波基函数直接替换为 Gauss径向基函数 ,提出了一种适于对目标一维距离像信号直接进行分类的径向基函数神经网络(RBFNN)。对用于信号分类的 RBFNN网络结构的确定、RBFNN的训练以及最终判决规则的确定等问题 ,进行了深入的讨论。对 6个目标不同信噪比下的分类结果表明 ,提出的 RBFNN对距离像信号具有很强的分类能力 ,对于开发更加实用化的目标识别算法显示了很大的潜力更多还原

【Abstract】 A new radial basis function neural network (RBFNN) is introduced for classification of aircraft targets. The method directly replaces the wavelet basis function in an adaptive wavelets neural network (AWNN) with the Gauss radial basis function. The RBFNN was then applied to the classification of 6 aircraft targets. Some related problems are thoroughly discussed, such as the selection of the RBFNN network structure, RBFNN training and construction of the criteria for making a final decision. The recognition results for different SNR suggest that this RBFNN is more suitable for classification directly on one dimensional range profile signals, and show great potential for developing more practical recognition algorithms.更多还原