Predicted arrival time calculation method and system based on graph network

Abstract

The embodiment of the invention provides a method and a system for calculating predicted arrival time based on a graph network, and the method and the system are used for mining vehicle types and weather information of historical trajectory data at the same time. In the aspect of time and space information deep fusion, a most natural graph neural network is adopted for modeling, the mutual influence of traffic flow speeds of connected road sections in a local area is fully considered, and modeling is performed on road topology through a most appropriate graph structure. Track data are projected on a graph structure, and time information and space structure information are fully fused in a feature modeling mode; road attribute information influencing vehicle circulation in a high-precision map is introduced in the feature modeling process, such as road section road types, road section highest speed limit, road section lane number, intersection or not and the like, and the information is introduced into static graph network modeling in an information embedding manner; firstly, a cache machine is utilized to ensure the real-time performance of query, and secondly, the world is divided into local areas to ensure the stability and convergence speed of graph network training.

Description

technical field [0001] Embodiments of the present invention relate to the technical field of application of high-precision maps in trajectory data mining, and in particular to a method and system for calculating estimated time of arrival based on a graph network. Background technique [0002] The estimated time of arrival (ETA) prediction problem is a key business problem that map companies and travel companies need to solve. In the context of fast-paced society, the accuracy of ETA calculation is particularly important. At present, common map companies and travel companies predict ETA by mining historical trajectory data. Among them, Recurrent Neural Network (RNN) and Convolutional Neural Network (CNN) will be used for modeling, but common methods cannot deeply integrate temporal information and spatial information. ETA is in four-dimensional space, where time and space influence each other. In addition, road attribute information, weather information, vehicle type infor...

AI Technical Summary

Problems solved by technology

Among them, Recurrent Neural Network (RNN) and Convolutional Neural Network (CNN) will be used for modeling, but common methods cannot deeply integrate temporal information and spatial information.

Images