BP神经网络作为一种前馈性的神经网络，RBF神经网络由于其独特的联想记忆功能，常常用来用于识别和优化计算方问题上。分别对这两种算法用于对逼近非线性函数进行编程，观察其拟合情况后，用其他未训练的样本数据进行泛化能力分析。(BP neural network is a feed-forward neural network. RBF neural network is often used to identify and optimize the computation problem due to its unique associative memory function. These two algorithms are programmed to approximate nonlinear functions respectively. After observing their fitting conditions, the generalization ability of other algorithms is analyzed with other untrained samples.)