Short-Term Bus Passenger Flow Prediction Based on BiLSTM Neural Network

Abstract

In order to analyze the passenger flow characteristics of single line bus and improve the operation of public transportation vehicles through combination optimization, this paper establishes a short-term bus passenger flow prediction model based on existing research, data characteristics, and solving objectives, and selects indicators for comparison and analysis of results. The research is based on a long short-term memory (LSTM) network, bidirectional long short-term memory (BiLSTM) network, and gated recurrent unit (GRU) network for modeling, and public health event management is included as an important influencing factor in the model establishment process. Through comparative analysis of the model prediction results, a short-term bus passenger flow prediction method based on BiLSTM network is finally proposed. Compared with existing methods, this method not only ensures prediction accuracy, but also ensures solution speed and universality performance. The research results further improve the existing theoretical and methodological system for optimizing the operation of conventional public transportation and have certain practical value for formulating more efficient public transportation scheduling plans, achieving refined management of public transportation, and improving the decision-making level of urban public transportation management.