【摘要】 提出一种新的智能城市交通信号控制方案推荐系统，用于实时推荐交叉口控制方案，缓解交通拥堵问题。该系统基于聚类算法(K-means++)构建历史动态子区数据库，在短时预测状态下实时动态划分交通子区，并使用协同过滤算法推荐动态子区内交叉口的控制方案。为了提高推荐方案的下发效率，系统记忆训练集中多维数据度量参数并具备可视化效果，便于信号工程师根据交通实际复杂需求进行修改。实验分析表明，采用系统推荐的区域控制方案能够有效提高道路运行效率。更多还原

【Abstract】 This paper proposes a new smart city traffic signal control scheme recommendation system for real-time recommendation of intersection control schemes to alleviate the problem of traffic congestion. The system built a historical dynamic sub-area database based on the clustering algorithm(K-means++). It dynamically divided the traffic sub-region in real-time under short-term prediction state, and used the collaborative filtering algorithm to recommend the control scheme of the intersection in the dynamic sub-region. To improve the delivery efficiency of the recommendation scheme, the system memorized the multi-dimensional data metrics in training set and had a visualization effect, which was convenient for the signal engineer to modify according to the actual complex demands of the traffic. The experimental evaluation shows that the system control scheme recommended by this system can effectively improve the road operation efficiency.更多还原