在介绍自组织特征映射神经网络及其学习算法的基础上，提出了一种结构自适应自组织特征映射神经网络。该网络能自适应地选择合适的网络规模，并对分类决策边界进行更为精细的调整，因此该网络区分微小差别的不同模式样本的能力进一步增强。最后给出了两类网络对实测高分辩雷达目标的识别结果。 On the basis of introducing the self-organizing feature map neural network and its learning algorithm, a self-organizing feature map neural network with structure adaptive is proposed. The network can adaptively select the appropriate network size and adjust the classification decision boundary more finely. Therefore, the ability of the network to differentiate the different mode samples with small differences is further enhanced. Finally, the recognition results of two kinds of networks on the target of high resolution radar are given.