Urban Traffic Congestion Prediction System Based on Big Data

Abstract

With the rapid advancement of urbanization, the problem of urban traffic congestion has become increasingly prominent, which has become an important constraint affecting the sustainable development of cities. In order to effectively alleviate traffic congestion, it is urgent to use advanced technological means to deeply tap the value and potential of traffic big data. In this paper, an urban traffic congestion prediction system based on big data is proposed. By extensively collecting heterogeneous traffic data from multiple sources, a set of adaptive reinforcement learning prediction model is constructed on the basis of which the traffic signal control problem is modeled as a Markov decision process, and the optimization goal is to minimize the average travel time of regional vehicles. The model uses DQN algorithm to train the signal timing strategy, and introduces dual network architecture, experience playback, priority sampling and other technical improvements to improve the stability and training efficiency of the algorithm.