Research On Cooperative Flow Control Strategy For Urban Rail Transit Normal Peak Lines Based On Short-Term Passenger Flow Forecasting

Urban rail transit has become the preferred mode for urban residents because of its green and high punctuality.However,with the increase of rail transit passenger flow,a large number of passenger influx during the peak period,making the transport capacity provided by rail transit far from meeting the passenger demand,resulting in a large number of passenger congestion,which will cause great security risks to the relatively closed station environment.It is of great significance to improve the operation level of rail transit and reduce the potential safety hazards that how to forecast the passenger flow efficiently and take into account the synergy between stations and establish a reasonable synergistic current limiting scheme.Firstly,this paper summarizes the factors and principles of rail transit passenger flow.Then,taking Beijing Metro Line 8 as an example,the spatial and temporal distribution characteristics of passenger flow are analyzed.Based on this,a short-term passenger flow forecasting model and a dynamic collaborative current limiting model based on short-term passenger flow forecasting are established,taking peak passenger flow as the research object.The work of this paper is as follows:(1)The influencing factors of passenger flow in rail transit are summarized and analyzed.The spatial and temporal distribution characteristics of passenger flow in Beijing Metro Line 8 are summarized,and the reasons for the occurrence are reasonably explained.At the same time,the theory of line cooperative current limiting is summarized,and the threshold of line start current limiting is defined.(2)Wavelet Neural Network(WNN)is used as the prediction model to predict the inbound passenger flow of a station on Line 8.The analysis results show that the wavelet neural network can effectively predict the incoming passenger flow,but its stability is poor and its accuracy needs to be improved.Because the initial weights and wavelet factors will greatly affect the forecast results of passenger flow in rail transit stations,the Grey Wolf Optimizer(GWO)algorithm is introduced to find a set of optimal initial weights before the forecast begins.Based on this,the neural network is established and then the prediction analysis is carried out.The prediction results show that GWO-WNN model can significantly improve the prediction accuracy and effect.(3)The mathematical model is used to describe the process of passenger queuing,platform waiting and transmission on the line.A dynamic collaborative current limiting model based on short-term passenger flow forecasting is constructed,in which the number of current-limiting people in each period of time is taken as the decision variable and the total travel delay time is taken as the objective function.The constraints of the model are platform carrying capacity,and train carrying capacity.Subsequently,according to the characteristics of passenger flow dynamics in rail transit,a model solving strategy based on rolling time domain optimization is formulated.Through on-line calculation,local optimal solution is solved continuously rolling,and finally the overall flow limiting strategy for the whole calculation period is obtained.(4)Choose the early peak(7:30-8:30)of Zhu Xinzhuang-Huoying five stations in downward direction of Beijing Metro Line 8 as the research object.Firstly,GWO-WNN model is used to forecast and analyze the passenger flow data of these stations,and the prediction results are imported into the dynamic collaborative current limiting model.The rolling time domain optimization method and genetic algorithm are used to obtain the final current limiting strategy through continuous rolling calculation.By comparing the final strategy with the non-cooperative strategy,it is verified that the cooperative strategy can reduce the total delay time of passenger flow,thus verifying the feasibility and optimization effect of the model.