Factors such as holidays, important sporting events and major celebrations may cause sudden large passenger flows in certain sections and stations of the urban rail transit system. Sudden inbound passenger flow easily leads to continuous congestion of the subway network, which has a great impact on the reliability and stability of the subway system operation. Due to the large magnitude of the swipe data and the high dimension of the time series, it is difficult to identify the burst large passenger flow, and the recognition accuracy of the existing methods cannot meet the operational monitoring requirements. In order to solve these problems, Symbolic Aggregate Approximation (SAX) is first introduced to reduce the dimensionality of subway passenger time-series. Secondly, subway passenger time-series are clustered as "words" by typical pattern matching algorithm to identify sudden large passenger flows, while pre-set cluster types and dynamic time warping (DTW) are proposed to enhance the matching rate. Taking the passenger flow in 2014 of the Beijing Subway Dongsishitiao Station as an example, the optimization algorithm proposed in the paper successfully identifies the sudden inbound passenger flow caused by activities, with an accuracy rate of 86%. Compared with the traditional K-means method, the mining accuracy is increased by an average of 30%, and the excavation time is shortened to one-sixteenth of the original.

CICTP