| Road passenger operation is the main means of cross-regional short-distance travel.Its industry market is huge.The total number of trips of nearly 15 billion people per year is about5 times that of train travel and 11 times that of air travel.The road passenger transportation system is an important part of comprehensive transportation system.It has the advantages of large carrying capacity,high transportation efficiency,low energy consumption,and low pollution.Recently years,the road passenger transportation network ticketing has developed rapidly.The provincial intranet and the national network have advanced layer by layer,enabling passengers to conveniently purchase tickets through the Internet and mobile terminals without leaving their homes,and improving the operating efficiency and services of the island road passenger transportation system.At the same time,it provides accurate data transaction records and accumulates a large amount of passenger transportation data.In order to meet the increasing travel demand of the public,it is urgent to promote the digital transformation of road passenger transport enterprises and release the potential energy of the passenger transport industry.Therefore,this article takes the road passenger transportation system of Yunnan Province as the research object,and discusses the key technology in mining the value of existing information based on big data technology.The main work includes the following four aspects.(1)A passenger flow forecasting model based on machine learning is proposed.The thesis uses statistical methods to study the data of road passenger transport routes in Yunnan Province,uses a low-rank matrix complement model to repair missing data on passenger routes,and studies the passenger volume of different passenger stations,the characteristics of the temporal and spatial distribution of passenger routes,and the influence of passenger routes on passenger transportation.In order to further improve the prediction accuracy of the Long Short-Term Memory(LSTM)model,taking the average daily passenger volume as a constraint,the LSTME three-layer passenger flow prediction model that integrates external factors such as weather,holidays and weekends is proposed.The experimental results show that the proposed LSTM-E prediction model reduces the prediction error by an average of 10.56%,which further improves the accuracy of passenger traffic prediction.(2)A mixed integer nonlinear programming method for road passenger transportation system dynamic pricing under competitive disadvantage conditions is proposed.By considering the competitive relationship between road passenger transport and high-speed railways,a mixed-integer nonlinear dynamic pricing model with the objective of maximizing the revenue of passenger transportation companies is proposed,which uses the logit stochastic utility model to determine passenger travel choices.The demand impact factor is set according to the transportation characteristics of road passenger transport and high-speed railways,which realizes the modeling of the degree of impact of high-speed railways on intercity passenger transport.Demand impact factors,remaining seat factors,unit mileage price and operating period jointly affect road passenger fares and adversely affect passengers’ choice of travel mode.The model introduces integer variables to enable the sharing of vehicle seats by passengers in non-overlapping OD segments of the same trip.The experimental results show that the dynamic pricing model can calculate the optimal dynamic pricing mechanism and the time interval of shifts in different time periods,and improve the operating income of the road passenger transport company.The operating income of the passenger transport company in the example is increased by 13.71% compared with the current pricing mechanism.(3)A global optimization model for the departure interval of intercity passenger transport lines based on mixed integer programming is proposed.By considering the profitability of road passenger transport companies,passenger waiting time and related constraints,a single-type and dual-type vehicle and passenger vehicle scheduling model is proposed.The model combines maximizing the profit of the operating company and minimizing the passenger waiting cost to find a bus scheduling result with the maximum overall benefit.Thee experimental results show that the model is better and more stable in terms of the stability of data fluctuations and the range of fluctuations.After optimization,the data obtained by the model is better and more stable.Passenger waiting time is reduced by about 12%,and the revenue of road passenger transportation companies increases by about 15%.(4)A model of road passenger vehicle detection system based on Petri net and a heuristic algorithm for the optimal sequence of vehicles to be inspected are proposed.The current vehicle inspection methods actually applied in the road passenger transportation system are timeconsuming and inefficient.Taking the shortest vehicle detection turnaround time as the optimization goal,the resource-oriented Petri net is used to model the automobile detection system,and a ROPN-based automobile detection system model is proposed.The operation process is analyzed to propose a calculation method for the detection turnaround time,and a further method is proposed.Based on the heuristic algorithm of ROPN to generate the optimal sequence of vehicles to be inspected.The experimental results show that compared with the traditional FCFS-based scheduling algorithms and scheduling algorithms based on MQ and SJF,the ROPN-based heuristic algorithm proposed in this thesis shortens the detection turnaround time and greatly improves the operating efficiency of the vehicle detection system. |
PDF Full Text Request