二维经验模态分解，采用形态学求极值点，径向基神经网络拟合，速度较快。-Two-dimensional empirical mode decomposition, using morphology for extreme points, radial basis function neural network fitting, faster.