利用高斯型隶属函数和采样数据得到了三层模糊前向神经网络。该网络模型利用权值直接确定法得到了最优权值,并依据采样数据中的插值样本较好确定了单隐层神经元个数。该网络是近似插值神经网络。仿真实验表明,高斯型模糊前向神经网络具有逼近精度高、网络结构简单、良好的去噪性和实时性高等优点。 By using Gaussian membership function and sampling data, a three-layer fuzzy forward neural network is obtained. The network model obtains the optimal weight by the method of direct weight determination, and determines the number of single hidden layer neurons based on the interpolated samples in the sampled data. The network is an approximate interpolation neural network. Simulation results show that the Gaussian fuzzy forward neural network has the advantages of high approximation accuracy, simple network structure, good denoising and real-time performance.