The Research On Shared Bicycle Demand Forecasting And Static Scheduling Method

As a solution to the "last mile" problem,the shared bicycle system is in a period of rapid development.However,in the actual operation management,the uneven distribution of bicycles is common.The uneven distribution of bicycles will lead to imbalance of supply and demand at each site.In order to alleviate the impact of imbalance between supply and demand on the system,this paper based on the analysis of the use of bicycles on the site,through the establishment of a shared bicycle static scheduling model to alleviate the imbalance between supply and demand.Firstly,based on the actual operation data of Divvy Bikes public bicycle system in Chicago,the paper counts the usage frequency,duration and user distribution of the public bicycle system,establishes the classification model of site demand pattern and the method of judging the initial vehicle of the site,analyses the relationship between site borrowing and scheduling demand,and puts forward the potential demand of the site.Secondly,considering the factors such as weather,temperature,wind,site capacity and initial number of vehicles,the training network is constructed with historical daily normal data,and the daily data with potential demand is used as the test network,and three neural network methods are used to predict the amount of borrowing of potential demand sites.The predicted potential demand is added to the original data to obtain the combined data,and the combined data is used to import the time series ARIMA model to obtain the result of the scheduling demand.Then,considering the total cost of dispatching and the factors of the damaged car at the site,a static scheduling model with the lowest total system scheduling cost is established.The station scheduling demand in this model is determined according to the site scheduling demand model established by the paper,and the problem of damaged bicycle is added.In the scheduling model,the number of damaged bicycles at the site is determined according to the conditions.A combination algorithm of discrete particle swarm optimization and variable neighborhood search algorithm is designed to solve the scheduling model.The discrete particle swarm optimization is used to determine the optimal solution of the model.Then the variable neighborhood search algorithm is used to perform local search to determine whether the optimal solution is the local maximum.Excellent solution,after adding the variable neighborhood search algorithm,it can effectively avoid the problem that the discrete particle swarm is used to find the local optimal solution.Finally,taking OLD TOWN area in Chicago as an example,the scheduling problem of public bicycle system is studied.The paper uses the method of site demand forecasting and the method of judging damaged vehicles to get the demand of site dispatch and the number of damaged vehicles.The paper uses MATLAB software to compile the combination algorithm of discrete particle swarm optimization and variable neighborhood search method to solve the case dispatch model.At the same time,the actual dispatch scheme and the predicted dispatch scheme are compared and analyzed.The results show that the site demand forecasting problem is considered in this paper.Question and static scheduling model have some enlightenment to practical scheduling schemes.