Application Research of Big Data Technology in Intelligent Prediction of Urban Traffic Flow

Abstract

With the acceleration of urbanization and the proliferation of information sensing technologies, the data generated by urban traffic systems exhibits typical big data characteristics, such as massive volume, high dimensionality, heterogeneity, and strong spatiotemporal dependencies. This presents both new opportunities and challenges for the field of traffic flow prediction. Traditional prediction models have inherent limitations in processing such data and capturing complex nonlinear spatiotemporal dynamics. This study aims to systematically investigate the application framework of big data technology in the intelligent prediction of urban traffic flow. The paper first analyzes the core characteristics of urban traffic big data and the evolutionary principles of intelligent prediction algorithms. Subsequently, it constructs a multi-source data fusion processing method tailored for prediction tasks, a spatiotemporal prediction model architecture based on deep learning (particularly graph neural networks and attention mechanisms), and a dynamic prediction integration system supporting real-time stream processing. Finally, it analyzes the application value of this technology in short-term high-precision prediction, simulation and deduction of congestion propagation, and system performance and uncertainty assessment. The research indicates that big-data-driven intelligent prediction methods can more effectively mine the underlying patterns of traffic systems, providing more accurate and robust technical approaches for dynamic traffic state perception and management decision support.