A Mixed Arima and Grey Relational Analysis Model for Predicting Urban Traffic Congestion

Transportation is pivotal to urban development, playing an important role in economic expansion and cultural exchange. However, the design of current smart city transportation platforms is ineffective in predicting traffic congestion states. Traditional forecasting models, while able to achieve current predictive goals and tasks, lack relevance and completeness in practical applications, leading to errors in traffic congestion state predictions due to external environmental influences and specific factors. Additionally, the inefficiency of single-structured traffic congestion state predictions makes it difficult to accurately determine the prediction range, which introduces uncontrollable obstacles to the implementation of subsequent work. To address these issues, this study proposes a hybrid model that combines ARIMA (Autoregressive Integrated Moving Average) and gray correlation analysis for urban traffic congestion prediction. This model integrates time series analysis with gray system theory to improve the accuracy and reliability of forecasting. Experimental results demonstrate that the hybrid model has high accuracy in capturing the changing trends of traffic flow, with an actual traffic flow of 520 vehicles/hour at 8 a.m. and a predicted flow of 510 vehicles/hour, resulting in a minimal error of 10 vehicles/hour. These findings provide reliable decision-making support for urban traffic management and help alleviate urban traffic congestion problems.