An Attention-Based Parallel Hybrid Prediction Model for Intersection Level Turning Movement Forecasting
31
by Yancheng Gong, Chunhao Liu, Ce Zhang, Liping Fu, Hao Wei, and Guangyuan Pan
Published onMay 27, 2024
An Attention-Based Parallel Hybrid Prediction Model for Intersection Level Turning Movement Forecasting
·
Abstract
The key to real-time control of intersection signals lies in the accurate prediction of intersection level turning movements, and current research on traffic flow prediction focuses on road segment flow prediction rather than vehicle turning movement forecasting (TMF) at intersections. Based on this, this paper proposes an attention-driven parallel hybrid prediction model (ATT-PHM) for predicting intersection turning movement count. This model consists of a bidirectional long short-term memory network, a convolutional neural network, a multilayer perceptron and an attention mechanism. Experiments are conducted using real world traffic data collected from city of Milton, Ontario, Canada during 2021-2022, in which several current mainstream models are compared, and the proposed model shows promising results in terms of spatio-temporal correlation feature learning in TMF.
Creative Commons Attribution 4.0 International License (CC-BY 4.0)