基于主成分和BP神经网络的汽车保有量预测

针对汽车保有量预测建模中输入因子过多而导致神经网络规模过大、泛化能力差的问题,通过主成分分析法和贝叶斯正则化方法对BP神经网络进行改进,可简化网络结构,增强泛化能力。对此,以我国汽车保有量预测为例进行的仿真计算表明,结果令人满意,这也同时证明了该方法用于汽车保有量预测的可行性与有效性。

Vehicle Quantity Prediction Based on Principal Component and BP Neural Network

In allusion to the poor generalization and the problem that too many input factors in prediction model of vehicle quantity leads to the big scale of neural network,the principal component analysis method and Bayesian regularization are utilized to modify BP neural network,which can simplify network structure and strengthen generalization.Given this,the emulation calculations as for vehicle quantity prediction in China show that the results is satisfying and prove its feasibility and effectiveness of vehicle quantity prediction.