Motorway Traffic Flow Prediction using Advanced Deep Learning

Abstract

Congestion prediction represents a major priority for traffic management centres around the world to ensure timely incident response handling. The increasing amounts of generated traffic data have been used to train machine learning predictors for traffic, however this is a challenging task due to inter-dependencies of traffic flow both in time and space. Recently, deep learning techniques have shown significant prediction improvements over traditional models, however open questions remain around their applicability, accuracy and parameter tuning. This paper proposes an advanced deep learning framework for simultaneously predicting the traffic flow on a large number of monitoring stations along a highly circulated motorway in Sydney, Australia, including exit and entry loop count stations, and over varying training and prediction time horizons. The spatial and temporal features extracted from the 36.34 million data points are used in various deep learning architectures that exploit their spatial structure (convolutional neuronal networks), their temporal dynamics (recurrent neuronal networks), or both through a hybrid spatiotemporal modelling (CNN-LSTM). We show that our deep learning models consistently outperform traditional methods, and we conduct a comparative analysis of the optimal time horizon of historical data required to predict traffic flow at different time points in the future.

Social physics

“…For physicists, traffic-flow models became a topic in the 1990s. In the recent decade, this topic has cooled down somewhat, but nonetheless remains an active field of research in physics and elsewhere, for instance, machine learning [111]. It is probably fair to say that the main motivation for physicists has never been to provide practical advice for urban planners.…”

Short term traffic flow prediction based on improved LSTM model

“…For the missing data, the method of averaging adjacent data is used to supplement. 10) (11). RMSE is used to reflect the applicability of data and model, and the result represents the deviation between the observed value and the real value; MAE is used to reflect the actual situation of the predicted value error, and the result represents the real error between the observed value and the real value.…”

“…Ma et al [9] applied LSTM to short-term traffic flow prediction according to the time series characteristics of traffic flow data, and verified the effectiveness of the method. Shi et al [10] Mihaita et al [11] applied the spatiotemporal prediction model of CNN and LSTM to the traffic congestion prediction of road network. Nicholas et al [12] proposed a new deep learning structure for traffic flow prediction, which includes L1 regularized linear model and a series of tanh network layers for traffic flow prediction.…”

Machine‐learning prediction models for pedestrian traffic flow levels: Towards optimizing walking routes for blind pedestrians

“…Omer, Rofè, and Lerman (2015) assert additional features that affect pedestrian traffic flow, namely tourist sites and public transportation stations. Although various techniques are used to predict traffic flow, mainly implementing machine‐learning algorithms (Miglani & Kumar, 2019; Mihaita, Li, He, & Rizoiu, 2019), these mostly refer to vehicular traffic flow. To the best of our knowledge, none have tackled this problem using OSM data.…”