基于自组织理论的自组织多项式网络算法

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．

The Self-Organizing Polynomial Network Algorithm Based on the Self-Organizing Theory

The self organizing polynomial network is a new network algorithm which combines biocybernetics with self-organizing feature map theory by using the thought of neural network. With its self organizing feature of seeking the best model parameter combination and the function of seeking minimum error at each level, the algorithm reflected stronger superiority on nonliner fitting. The application of developed software software dclared that the algorithm had better fitting precision compared with GMDH algorithm and other common network algoithms.