Short-Time Bus Route Passenger Flow Prediction Based on a Secondary Decomposition Integration Method

Abstract

Bus passenger flow is one of the decisive factors for the development of public transportation. Therefore, accurate prediction of real-time passenger flow on bus routes not only helps bus companies to make reasonable scheduling plans to meet the travel needs of passengers but also promotes the sound development of urban public transportation and reduces pollution. In this paper, we propose a secondary decomposition integration method that combines empirical modal decomposition (EMD), sample entropy (SE), and kernel extreme learning machine (KELM) to achieve a short-time prediction of bus route passenger flow. The EMD decomposes the original passenger flow data into several intrinsic mode functions, measures the complexity of the decomposed intrinsic mode functions using SE, and performs a secondary decomposition of the intrinsic mode functions with the highest complexity using EMD, followed by the prediction of the two decomposition results using KELM. The final predicted result is the sum of the two results. The model is verified by the real card-swiping data of two bus lines per minute. Each group of data has 300 data, with 80% of the data as the training set and the remaining 20% as the test set, which can predict the passenger flow per minute. The experimental results show that the short-term bus passenger flow forecasting method proposed in this paper has high accuracy and good robustness.