BP网络预测能力仿真与分析

Analysis on the Simulation of BP Network Predictive Ability

人工神经网络用于预测时有多高的可靠性和精度 ?是一个还没有得到很好探讨的问题 .本文以BP神经网络为对象 ,从数值仿真和欧氏空间理论的角度 ,分析了预测样本与训练样本之间的关系对网络预测能力的影响 。

How much of the reliability and precision are there when artificial neural network is used to predict a complex system? That is one hot point still getting heated discussion. Taking BP network as an object, from the value simulation and Euclidean space theory, the relations between forecast sample and training sample for the influence of network predictive ability are analyzed in order to obtain the way to estimate the precision of prediction.