The invention provides a road traffic flow prediction method under a distributed system. The method comprises the following steps: building a Hadoop and Spark cluster environment; acquiring all historical vehicle data of one road and predicting real-time vehicle data of the current day; counting the traffic flow of each hour in one day; calculating the vector distance between all historical traffic flows and the predicted real-time traffic flow of the current day; calculating the distance deviation and variance of the K vectors to obtain an optimal K value; and selecting data to train a machine learning regression model according to the K value. K pieces of historical traffic flow data are selected and input into the regression model, and multiple sets of predicted values and corresponding root-mean-square error values are obtained. And predicting the road traffic flow in the next hour of the day according to a formula. According to the method, huge data are processed in parallel on a distributed system, and the calculation speed is increased; multiple machine learning regression models are fused in K-nearest neighbor matching, all historical data are effectively utilized, and the model accuracy is high.
本发明提供了一种分布式系统下道路车流量预测方法,包括:搭建Hadoop、Spark集群环境;获取一条道路所有历史车辆数据以及预测当天的实时车辆数据;统计出一天中每个小时的车流量;计算所有的历史车流量与预测当天的实时车流量的向量距离;计算K个向量距离偏差和方差,得到最优的K值;根据K值选择数据训练机器学习回归模型。选择K条历史车流量数据,输入回归模型,得到多组预测值和相应的均方根误差值。根据公式预测当天下一小时的道路车流量。本方法在分布式系统上并行处理庞大的数据,加快计算速度;在K近邻匹配中融合了多种机器学习回归模型,有效地利用了所有历史数据,模型准确率较高。
Road traffic flow prediction method under distributed system
一种分布式系统下道路车流量预测方法
2022-03-01
Patent
Electronic Resource
Chinese
Distributed traffic flow prediction method and system based on road segmentation
European Patent Office | 2023
|European Patent Office | 2016
|