自组织多项式网络是采用神经网络的思路结合生物控制论和自组织特征映射理论而导出的一种新型网络算法，该算法在寻求模型参数的最优组合上的自组织特征及通过层层搜索误差最小点的功能，使其在用于非线性映射的拟合中体现了较强的优越性．开发的软件应用表明，该算法较ＧＭＤＨ算法及一般网络算法具有更高的精度拟合． A self-organizing polynomial network is a new type of network algorithm based on the idea of ​​neural networks combined with biological control theory and self-organizing feature mapping theory. The self-organizing feature of the algorithm seeking the optimal combination of model parameters and the search error The function of the minimum point renders it superior for fitting in nonlinear mapping. The developed software application shows that this algorithm has higher accuracy than GMDH algorithm and general network algorithm.