本文利用神经元交互作用函数描述拓扑特征映射神经网络，探讨了这种网络的学习收敛性．本文首先给出一个网络收敛的一般性结论，并利用该结论证明网络输入满足平均分布时的收敛性．由此可进一步得到Ｋｏｈｏｎｅｎ网络自组织学习的收敛性．本文的结果修正并拓广了关于自组织学习收敛性已有的一些结果，并为完全证明特征映射的收敛性提供了一种新途径 In this paper, neuronal interaction function is used to describe the topological feature mapping neural network, and the learning convergence of this network is discussed. In this paper, we first give a general conclusion of network convergence, and use this conclusion to prove that the network input satisfies the convergence of average distribution. This can be further obtained Kohonen network self-learning convergence. The result of this paper amends and extends some existing results about the convergence of self-organizing learning, and provides a new way to fully prove the convergence of feature mapping