一种新型双权值人工神经元网络的数据拟合研究

Data Fitting Based on a New Double Weights Neural Network

在本文中提出了一种针对新型双权值神经元网络的数据拟合算法 .采用这种新型网络结构和算法 ,可以克服传统的通用前馈网络中BP算法易陷入局部极小的问题 .通过实验比较证明在相同的网络规模下 ,采用这种新型网络结构和算法可以取得比径向基 (RBF)网络更高的拟合精度和更少的迭代次数 .

We construct a new method in data fitting fields. Usually, in traditional BP neural network model, data fitting may become trapped at a local minimum. By using the new Double Weights Model, this algorithm can give the Direction Weight, also the Central Weight at the same time. Experimental results show that this algorithm can get more accurate fitting effects and use less generations to calculate, compared with the RBF (Radial Basis Functions) while using the same environment and equal network scale. Data fitting on it should be a new method to modern industry applications in data control and analyses and so on.