Research And Implementation Of Bus Dispatching Optimization Based On Machine Learning Passenger Flow Prediction

With the rapid development of the city and the continuous improvement of residents’ living standards,citizens’ travel demand is increasing daily.The development of public transport has brought great impetus to the development of the city.At present,bus dispatching often relies on historical data and manual experience for planning,which has the disadvantages of lag and low efficiency.It is difficult to arrange the departure schedule according to the travel situation of citizens.Through the mining and analysis of bus big data,this paper proposes a passenger flow prediction model of bus stops based on genetic algorithm(GA)to optimize the parameters of long-term and short-term memory network(LSTM);Based on the predicted passenger flow data,a bus departure schedule optimization model for optimizing transfer efficiency is proposed.The main results of this paper are as follows:(1)Passenger flow prediction model of bus stops.Based on the analysis and mining of bus big data,this paper proposes a bus stop passenger flow prediction model(ga-lstm).The model uses LSTM to predict bus stop passenger flow,and applies genetic algorithm to optimize the time step,neuron number and learning rate on the basis of LSTM,which improves the accuracy of bus stop passenger flow prediction and avoids falling into local optimization.(2)Bus departure schedule optimization model.In order to solve the problem of long waiting time of passengers and the impact of adjusting the departure schedule on passengers’ travel plan,this paper proposes a bus departure schedule optimization model aiming at maximizing the number of coordinated transfers and minimizing the change of schedule.The model belongs to NP hard problem.Therefore,in order to improve the solution speed,the model realizes multi-objective genetic algorithm to search the optimal departure schedule.This paper forecasts the passenger flow of bus stops by combining the bus big data analysis and machine learning model.Based on the prediction of passenger flow,the existing departure schedule is optimized by using the bus departure schedule optimization model.Through experimental verification and analysis,ga-lstm reduces the prediction error compared with cyclic neural network(RNN)and BP neural network,and provides data support for bus operation.After searching the Pareto optimal solution by multi-objective genetic algorithm,the departure schedule with the highest congestion is selected for analysis.After the schedule optimization,the number of cooperative transfer and the number of cooperative transfer trains of public transport are improved.To sum up,the bus timetable optimization model based on passenger flow prediction provides an effective means for public transport enterprises to timely adjust departure strategies in response to passenger flow changes,which can improve the decision-making efficiency of public transport enterprises.