基于FARBF神经网络算法的资产评估统计模型

Asset Evaluation Statistical Model Based on FARBF Neural Network Algorithm

针对RBF神经网络算法因原始变量间强相关性带来的处理难度,与因子分析得分评估模型不能充分结合先验知识等缺陷,文章综合因子分析与RBF神经网络算法的各自优点,构建一种基于FARBF神经网络算法的企业资产质量评估模型,并给出了原始数据的同向化处理方法。实证案例分析与仿真试验结果表明该模型精度高于单纯的RBF网络算法,且该算法简化了神经网络结构,提高了网络训练速度与算法精度。

For RBF neural network algorithm to deal with the difficulty due to strong correlation between the original variables,the defects that Factor Analysis model in scoring evaluation cannot fully combined with prior knowledge.This paper synthesizes the respectively advantages of Factor Analysis and RBF neural network algorithm,builds a enterprise asset quality evaluation model based on FARBF neural network algorithm,and a same direction processing method of the original data is given.Empirical case analysis and the simulation test results show that the model precision is higher than the pure RBF network algorithm,and this algorithm simplifies the structure of neural network,improves the network training speed and precision of the algorithm,to provide fast,accurate and reliable reference for decision making.