It is trained by using the recursively updated sample date sets and thus GRNN became dynamic neural network.

• 为此改造泛回归神经网络(GRNN)，运用递推更新的样本数据集训练GRNN，构成动态泛回归神经网络。