首先提出一种双曲函数型神经网络 HFL ANN,设计出一类基于 HFL ANN网络的层次双曲型函数网络HHFL ANN,给出了 HHFL AN N的网络学习算法 ,使其在用于非线性的拟合中体现了较强的优越性 ,对于任意的Volterra级数使用 HHFL ANN网络来逼近是完全可行的 ,该算法较 GMDH算法和 SOP算法 ,具有快速简单的特性 ,它优于 GMDH算法 ,有规律地选取部分多项式 ;优于 SOP算法 ,在构造 SOP网络不需要太多的中间隐层 ,从而加快了学习过程 ,提高了网络的逼近性能 ,更适合于具有层次结构的应用领域 Firstly, a Hyperbolic Functional Neural Network (HFL ANN) is proposed to design a class of hyperbolic function network HHFL ANN based on HFL ANN. The network learning algorithm of HHFL AN N is given, The fitting shows a strong superiority. It is feasible to approximate the Volterra series using HHFL ANN. Compared with GMDH algorithm and SOP algorithm, this algorithm is faster and easier than GMDH algorithm. The partial polynomials are selected regularly. Compared with the SOP algorithm, it is not necessary to construct too many intermediate hidden layers in the SOP network so as to speed up the learning process and improve the approximation performance of the network, which is more suitable for applications with hierarchical structure