基于高斯过程回归的短时交通量不确定性预测

Short-Term Traffic Volume Uncertainty Prediction Based on Gaussian Process Regression

智能交通预测对解决交通难题具有重要意义。针对现有模型预测不精准的问题,本文提出一种全新的短期交通量预测模型——高斯回归模型(GPR),这是基于贝叶斯理论和统计学习理论发展起来的机器学习算法。首先对数据进行归一化处理后,将数据划分为训练集和测试集,建立标准高斯回归方程,利用贝叶斯后验概率公式,建立高斯回归预测模型,并采用无效覆盖率(KP)和区间宽度(Width)两个评价指标,将GPR与GARCH、BOOTSTRAP模型做对比,结果显示本模型表现出的性能最优。

Intelligent traffic prediction plays an important role in solving traffic problems. To solve the problem of inaccurate prediction of existing models, this paper proposes a new short-term traffic volume prediction model—Gaussian regression model (GPR), which is a machine learning algorithm developed based on Bayesian theory and statistical learning theory. Firstly, the data was normalized and divided into training set and test set, and the standard Gaussian regression equation was established. The Gaussian regression prediction model was established by using Bayesian posterior probability formula. Two evaluation indexes of invalid coverage (KP) and interval Width were used to compare GPR with GARCH and BOOTSTRAP models. The results show that the performance of this model is optimal.