Research On Multiple Spatial-temporal Big Data Enabled Shortest Path Optimization

Transportation plays a much important part of human daily activities.Efficient transportation is vital to social production and operation.The existing traffic big data is mainly collected through a variety of transportation facilities(including closed-circuit televisions,ground coils,GPS sensors,microwave sensors,video monitors,etc.).In the era of big data,people continuously tweet and forward traffic related data through kinds of social networks(such as Twitter,Facebook,WeChat,Sina,micro-blog and so on).However,this part of newly generated traffic information has rarely employed in transportation modeling and optimization.Fully mining and conducting insightful analyses on these semi-structured and unstructured data for better path planning and decision-making within the traffic network systems has become the most effective way to achieve well-performed smart transportation systems with lower costs.In this paper,we study the problem of modeling,solving and analyzing the shortest path problem driven by multi temporal and spatial traffic data.Firstly,we proposed the "Big Data-Granular Computing-Decision Making" general research framework aiming at mining the indispensable decision making information from the available big data.Though the optimization of city navigation decisions with social network traffic data,we give some illustrative examples described in detail on the effectiveness of the proposed framework.In addition,in this study,by considering the generality and particularity of traffic modeling and decision making problems in the era of big data,especially for the time and space characteristics of traffic network,the temporal-spatial shortest path problem is proposed.After modeling and analyzing the problem of large data traffic network based on the loop detectors,we put forward the spatial-temporal Dijkstra algorithm which is used to solve this problem.The complexity of optimal quantization method of shortest path and space-time Dijkstra algorithm is also analyzed by providing both the theoretical theorems and corresponding examples in detail.The research contents of this study main include two parts:(1)The framework of "Big data-Granular Computing-Decision Making" and its application.In order to provide a general solution to the problem of modeling and solving the problem of large ssale data driven decision making,the general research framework of "Big data-Granular Computing-Decision Making" is put forward.Through the method of social network analysis and mining,the high precision fine granularity complex micro-blog text data is transformed into a low precision coarse-grained information structure which can be used for decision making.Based on the empirical study of the real map data in Beijing,it is proved that the research framework and robust path optimization model proposed in this paper can effectively improve the performance of traffic navigation system.(2)The study of traffic data driven adaptive rolling shortest path problem and its mathematical modeling.According to the spatial and temporal characteristics of the traffic network in the era of big data,the road network is defined as a dynamic network.Based on the dynamic traffic speeds by loop detectors,the adaptive rolling shortest path optimization method is proposed,and the optimal path is obtained by searching the dynamically updating speed curves of road network.The case study shows that the shortest path based on the rolling optimization method has helped the travelers shorten both their total travel time and distance significantly.The contents and ideas of this study can be extended and applied to the scheduling and optimization of various types of traffic management problems(such as vehicle routing problem,etc.).