Research On Passenger Flow Forecasting Model And Method Based On Trajectory Data

With the improvement of our living standard, the various lifestyle including tourism,shopping and entertainment has enriched our lives, and the public areas like tourist spot andshopping area have become the gathering place of people. The situation of crowds swarmingin these places may happen due to unreasonable arrangement. And it not only has an impacton the tourists mood, but also may bring great security risk. At present, the behavior oflimiting the number of visitors in closed areas can alleviate the condition of congestion to acertain extent. But in open areas, such method is no longer applicable. In short, therecognition of the public areas, as well as the prediction and analysis of the crowds canprovide strong evidence for management and decision-making and has important practicalsignificance.As the development and popularization of various positioning technology, users canobtain the personal locations quite conveniently and use all kinds of services based onlocations, like the global positioning system(GPS) and the wireless cellular networks. Thepaper presents a method of predicting the crowds on the basis of trajectory data. First of all,aimed at the problem that the area is not quite accurate in the algorithm of current stay areaidentification, it adopts the segmentation method based on the change point to propose aalgorithm of current stay area identification with BP neural network algorithm and DBScanclustering algorithm. Then using grid processing to deal with the stay areas and usingclustering algorithm to cluster the grid to develop a method of discovering public areas.Finally, under the circumstance of considering completely the data of current population andthe flowing, it proposes a method of estimating the flow of people in public areas with thecombination of Gauss fitting algorithm.The results of the experiment show that the method proposed in the paper can exactlyidentify the stay areas and discover the public areas from the spatio-temporal trajectory data.And when the sampling data satisfies some certain conditions, the method of estimating the flow of people can estimate the range of the flow in public areas to a certain extent.