Research On Recommendation And Efficiency Evaluation Of Online Car Hailing's Operational Area Based On User Data

Smart cities and intelligent transportation,as a global emerging technology,aim to promote sustainable economic development through intelligent management of resources.Compared with other major urban transport vehicles,online ride-hailing has a high degree of flexibility in operation route and schedule,which provides great convenience for urban people to travel.However,due to the size of cities and the increased number and demand for online taxis.there is also the problem of difficulty in managing them.Using large-scale trajectory data to improve online ride-hailing services has become a crucial research topic in the field of big data analytics within the fourth management revolution.The main research is as follows.(1)Based on the relationship between the number of grids and computational complexity.the data are meshed in spatial dimension.using a side length of 1.5 km.with a total of 753 square grids.The departure and arrival frequency of the users and the transfer probability between the grids of the idle network are calculated,and the computer simulation is carried out according to the actual data.(2)In order to evaluate the single dispatching strategy more accurately.based on Pareto principle,the study area is divided into core area,middle area and edge area.(3)Periodic processing of the ride-hailing data in the time dimension,taking into account the normalization of spatial relevance,and finally checking the stationarity of the order sequence.(4)With the consideration of the temporal and spatial characteristics of input data.A hybrid network model C-LSTM based on a convolutional layer and a long and short-term memory neural network system is constructed,where the convolutional layer mines the spatial correlation of grid area demands and the LSTM layer extracts temporal information.After training the model several times.single time-step prediction is used as the basis for the improvement of the dispatching rules.(5)In this paper,the grid region order number prediction is introduced into the dispatching improvement scheme,a forward-looking pre-dispatching scheme is proposed and the rule design is carried out.the number of idle drivers in the grid region is calculated,the subsidy under the split contract is clarified,and the system efficiency of the serial dispatching system and the proposed pre-dispatching scheme are compared by computer simulation.The simulation results show that the pre-assignment scheme based on the prediction of the number of orders in the grid area is feasible.Passengers' average waiting time has reduced by 66%,55%and 46%in the core,middle and edge zones respectively,and the variance of the number of orders received is reduced from 14.9 to 6.2The thesis shows that the C-LSTM online ride-hailing demand forecasting model,which takes into account spatial correlation and temporal characteristics.has high accuracy in predicting the number of orders in grid areas.On the basis of demand forecasting and chain dispatching,the improved and optimized dispatching scheme can effectively reduce the average waiting time of passengers and improve the uneven order taking of drivers,which can save passengers' travel costs and stabilize the number of drivers,and bring better user experience to both drivers and passengers while improving the efficiency of the whole service system.