自组织FPNN在宏观经济预测中的应用  

Application of Self-Organization FPNN in Macro-Economy Forecasts

作　　者：李小红[1] 

机构地区：[1]天津天狮学院电子计算机系

出　　处：《商业文化（学术版）》2008年第10期233-236,共4页Business Culture

摘　　要：论文针对经济预测通常表现为复杂的非线性这种特性,提出了一种基于自组织过程神经元网络(FPNN)和改进的BP神经网络建立的经济预测模型方法。自组织过程神经元网络(FPNN)由输入层、竞争层和输出层组成。FPNN筛选出对因变量(网络输出)最有影响作用的变量(自变量)之后作为改进的BP算法网络的输入节点,再用进行学习。该模型不仅克服了时间序列预测模型只能进行线性预测的不足,而且还避免了传统神经网络的固有缺陷。以2001年到2004年国内生产总值作为预测分析样本,并对预测结果和实际值进行了比较分析,结果验证了该方法的有效性。Aiming at the complex Non-linear Property and The Time-Varying Property in the actual economic system,put forward the method of self-organization process neural networks(FPNN)merged a Improved BP neural network to constitute economic forecasting model.The self-organization process neural networks(FPNN)consists of input layer and competition layer and output layer.The variables(independent variables)which were filtrated by FPNN are the best influence to the dependent variable(the output of network),then qua input node of the improved BP neural network to study.The model not only overcomes the shortage that timing series forecasting model just can do linear forecast,but also averts the disfigurements of tradition neural networks.Using the gross domestic product between 2001 and 2004 as the forecasting analytic stylebooks,compare the forecasting result to the actual result,finally the availability of the model is proved.

关 键 词：自组织过程神经网络 学习算法 正交基函数 经济预测 预测模型 

分 类 号：F224[经济管理—国民经济] TP183[自动化与计算机技术—控制理论与控制工程]