Urban traffic road section speed prediction method and system based on multi-road-section space-time correlation

Abstract

The invention discloses an urban traffic road section speed prediction method and a system based on multi-road-section space-time correlation. The method comprises the steps of acquiring the speeds ofthe latest p historical time points of all road sections in an optimal feature subset corresponding to a to-be-predicted road section; wherein p is a positive integer; and inputting the acquired speeds of the latest p historical time points into a pre-trained GRU neural network, and outputting the prediction speed of the (p + 1) th time point of the to-be-predicted road section. The prediction model quantitatively and dynamically considers space-time correlation among all road sections in a road network from the aspects of traffic parameters, road section connectivity, road grades and the like, and can select a road section subset which is beneficial to speed prediction of a road section to be predicted from the road network. The prediction model can realize accurate prediction of the speed of the urban traffic road section.

Description

technical field [0001] The present disclosure relates to the technical field of urban traffic section speed prediction, in particular to a method and system for urban traffic section speed prediction based on multi-section time-space correlation. Background technique [0002] The statements in this section merely mention background art related to the present disclosure and do not necessarily constitute prior art. [0003] Traffic congestion is a common problem in modern cities. Pre-accurate traffic information has a positive effect on improving travel efficiency and alleviating traffic congestion. Speed ​​is the core indicator that reflects the state of the road. The exponential growth of urban traffic data provides strong data support for mining the internal mechanism of traffic phenomena and realizing traffic parameter prediction. [0004] In the field of traffic flow parameter and state prediction, there are two mainstream prediction methods, one is the prediction method...

AI Technical Summary

Problems solved by technology

Many prediction studies do not consider or only consider the correlation between the upstream and downstream road sections and the road sections to be predicted, and cannot fully obtain the interaction between road sections, and the prediction accuracy is limited; and considering the spatial correlation of multiple road sections within the road network is currently a popular forecasting method. direction, most studies only consider the numerical correlation of traffic parameters when considering the correlation, but the spatiotemporal correlation between road segments is related to factors such as road distance and road direction

Some existing studies use various methods to realize the spatial correlation measurement of the state of multiple road sections in the road network. In the prior art, the distance function between road sections is directly used to measure the spatial correlation between road sections. For a road network containing 88 fixed detectors For analysis, but only considering the distance of road sections can only statically determine the spatial correlation between road sections; in the prior art, it is pointed out that more than 100 detector data are related to the detector data to be predicted, and the traffic state is divided into two states: congestion and smooth flow , proposed that the p detection score uses the state of the road section for feature selection to measure the state classification ability of each detector, determines the optimal number of features according to the prediction accuracy, and uses the Gaussian model to predict the congestion probability at a certain detector, but ignores the topological relationship of the road network ; In the prior art, the data of 35 detectors in the road network are used as the research object, the K nearest neighbor algorithm is used, and the LSTM model is based on the principle of optimal prediction results, and the optimal number of input features is determined by the relationship between the parameters to be predicted, but Only the reduction of the number of relevant road sections is considered, and the prediction effect of different road section combinations is ignored; in the prior art, a road network with 33 detection points is selected as the research object, and a one-dimensional convolutional neural network is used to obtain the spatial characteristics of traffic flow , using two long-short-term memory neural networks to mine the short-term variability and periodicity of traffic flow, but only considering the quantitative relationship of traffic flow, without considering the influence of other factors in the actual road network

Images