In the traffic volume combination prediction,there are problems of inaccurate traffic volume pre-diction and large prediction error,therefore,a traffic volume combination prediction model based on visual feature fusion is proposed based on big data.The time-based bilateral traffic flow series are calculated,and the time-series traffic volume data is analyzed.The identified color components of the visual image of traffic volume are grayed,the spatiotemporal association information is extracted,and the space-time sequence symbols are obtained.The temporal characteristic traffic volume sequence and the spatial characteristic traf-fic volume series are integrated,the visual features are fused by input,and the time-space series prediction weighting factor is adaptively adjusted to construct a traffic volume combination prediction model.It can be seen from the experiment results that the maximum error of weekdays'traffic volume prediction is 1％,and the prediction error of rest day is 0,which has an accurate prediction effect.
关键词
大数据/视觉特征融合/交通量/时间序列/空间序列
Key words
big data/visual feature fusion/traffic volume/time series/space sequence
维普
万方数据