Research On Short-term Bus Passenger Flow Prediction Model Based On Big Data

Bus passenger flow is an important basis for the bus companies to conduct bus dispatch operations and planning.Prediction of short-term bus passenger flow can help bus managers and planners timely and accurately get the changes of the passenger flow so that they can make scientific and reasonable bus dispatch operation and planning,thereby improving the efficiency of bus companies and meeting passengers' bus travel needs.At present,there has been great progress in the prediction research of short-term bus passenger flow,but there are still many deficiencies,such as the complexity of the prediction model,the accuracy problem and the inefficiency of the prediction model in a stand-alone environment.Based on a city's intelligent transportation big data platform,this paper conducts in-depth research on the short-term bus passenger flow prediction model based on big data,which mainly includes the following aspects:(1)After analyzing the research status of short-term bus passenger flow prediction and various prediction models,a Spearman-LMBP(SLMBP)short-term bus passenger flow prediction model is established in this paper to predict short-term bus passenger flow.The model uses the Spearman rank correlation coefficient method to analyze the influencing factors of bus passenger flow,and then uses the traffic influencing factor data together with the historical passenger flow data as the input data.The Levenberg-Marquardt algorithm is used to train the BP neural network to avoid BP neural network getting stuck in local optimal solutions easily and prompt the convergence speed.Moreover,the dropout technology of deep learning is used to optimize the structure and training methods of the model to solve the problem of overfitting and improve the generalization ability.The experimental results show that optimized SLMBP short-term bus passenger flow prediction model has a high prediction accuracy and can effectively predict short-term bus passenger flow.(2)The parallel computation of SLMBP short-term bus passenger flow prediction model based on Hadoop platform in distributed cluster environment is realized.Parallel computing can meet the needs of mass data storage,training and real-time processing in short-term bus passenger flow prediction.The experimental results show that when SLMBP short-term bus passenger flow prediction model based on Hadoop platform facing massive data,it can greatly reduce the time-consuming of model learning and training,and improve the operational efficiency of the prediction model under the premise of ensuring the accuracy of short-term bus passenger flow prediction.