提出一种模糊神经网络 ( FNN)结构学习算法 ,根据输入样本动态构建 FNN的输入节点及其对应的输入隶属函数 ,从而实现动态确定 FNN的结构 ,大大减少了对初始学习样本数目的要求 ,提高了 FNN学习算法在实时控制中的适应能力 .仿真结果表明 ,这一算法很好地实现了对超出初始学习样本范围的其他样本的学习 . A structure learning algorithm based on fuzzy neural network (FNN) is proposed. The input nodes of FNN and its corresponding input membership functions are dynamically constructed according to the input samples, so as to dynamically determine the structure of FNN, greatly reducing the number of initial learning samples and increasing The adaptive ability of FNN learning algorithm in real-time control.The simulation results show that this algorithm can well realize the learning of other samples beyond the initial learning sample.