Short-Term Traffic Flow Uncertainty Prediction Based on Novel GM(1,1)

Anticipating uncertainty in short-term traffic flow is crucial for effective traffic management within intelligent transportation systems. Various methods for predicting uncertainty have been proposed and implemented. However, conventional techniques struggle to provide accurate forecasts when confronted with sparse data. Hence, this study focuses on developing an uncertainty prediction model for short-term traffic flow under limited data conditions. A novel grey model that considers the volatility of the traffic data is proposed, which extends the grey model (GM) by integrating two techniques: smooth pre-processing and background value construction. The performance of the proposed novel grey model is mainly illustrated by comparing the novel grey model with the traditional GM model. Our results, in terms of uncertainty quantification, demonstrate that the proposed model outperforms the GM model regarding mean kick-off percentage (KP), width interval (WI) and width amplitude.