基于PCA-GRA-AdaBoost的交通流预测模型研究

Research on Prediction of Traffic Flow Based on PCA-GRA-AdaBoost

针对交通流数据非线性和时序性特点,综合考虑天气因素与交通流量的潜在关系,提出一种基于主成分分析法(PCA)、灰色关联分析法(GRA)和自适应提升(adaptive boosting,AdaBoost)算法相结合的预测模型.首先利用PCA对样本进行降噪处理,再采用GRA计算各天气因素和交通流的非线性关联度,将灰色关联系数大于0.6的相关性强的特征变量输入到AdaBoost集成模型中,进行了模型简化.实验结果表明:与长短期记忆神经网络(LSTM)、分类回归树(CART)、自回归积分滑动平均模型(ARIMA)以及未被优化的AdaBoost集成模型对比,提出的PCAGRA-AdaBoost模型在预测误差和确定系数等指标方面均优于其他传统算法,体现了较高的预测精度.

Aiming at the characteristics of nonlinearity and time series of traffic flow data,and considering the potential relationship between weather factors and traffic flow comprehensively,a prediction model based on the combination of principal component analysis(PCA),grey relational analysis(GRA)and adaptive boosting(Ada Boost)algorithm was proposed. First,PCA was used to denoise the samples,and then GRA was used to calculate the nonlinear correlation degree of each weather factor and traffic flow. The characteristic variables with strong correlation and grey correlation coefficient greater than 0.6 were input into the AdaBoost integrated model to simplify the model. The experimental results show that compared with LSTM,classification regression tree(CART),ARIMA and unoptimized AdaBoost integration model,the PCA-GRA-AdaBoost model proposed in this paper is superior to other traditional algorithms in terms of prediction error and determination coefficient,reflecting higher prediction accuracy.