Research On Demand Forecast And Dispatching Route Optimization Method Of Shared Bicycles Based On Data Analysis

With the end of the "13th Five-Year Plan",my country’s economic aggregate and industrial level have reached a new level,so the total number of private cars has also increased,and a series of traffic problems have become increasingly severe.To this end,the state has issued a series of policies to implement a public transport priority strategy,aimed at alleviating traffic congestion.In this context,shared bicycles have great advantages in connecting to public transportation and solving the problem of the city’s "last mile" by virtue of its convenience,flexibility,low carbon,environmental protection,and strong accessibility.However,as shared bicycles are developing rapidly,they are also facing problems such as "difficulty borrowing" and "disorderly parking and occupying public space",especially in the morning and evening peak hours.The lack of overall planning of dispatching personnel and vehicles and the lack of reasonable and effective dispatching strategies are the "culprits" that cause similar problems.It is necessary to meet the demand of shared bicycle borrowing and repayment during peak periods through scientific and reasonable dispatching strategies or plans.Therefore,this article mainly studies the demand forecasting and scheduling optimization of shared bicycles during peak periods.First,on the basis of a brief overview of the definition,development,functional positioning and existing problems of shared bicycles,this article introduces the data sources and preprocesses the data.Combining the preprocessed data,this article analyzes the travel characteristics of shared bicycles.The characteristics of the author,the characteristics of space-time dimensions and the influence of external conditions are analyzed and summarized.According to the different degree of influence of the factors that affect the borrowing and returning demand of shared bicycles,the research object and the research scope are determined,and the research scope is based on the spatial characteristics based on the consideration of the maximum walking distance that users can accept to find a car.Based on cluster analysis,it is divided into 31 travel areas.After comparing the commonly used demand forecasting models,based on the advantages and disadvantages of each model and the scope of application,a method for forecasting the borrowing and repayment demand during peak hours of shared bicycles based on the LSTM neural network considering the site relationship is proposed.City Mobike conducted an example analysis for the data.First,the demand for borrowing cars in the evening peak in the No.6 travel area was predicted,and the mean square error was only 9.8,which proved the validity of the model.It was compared with the previous models that participated in the comparison.The demand for returning cars at night in the No.4 travel area is predicted.The results show that the prediction error of the model proposed in this paper is the lowest,which proves the superiority of the model.Secondly,the content,form,and problem classification of shared bicycle scheduling are briefly described,a multi-objective optimal scheduling path model with the smallest total scheduling cost and the highest user satisfaction is constructed,and a genetic algorithm-based solution algorithm is designed.Finally,based on the scheduling demand predicted in the previous article,and the central point of each travel area as the scheduling station,the example analysis is carried out,and the optimal scheduling plan is solved by the genetic algorithm,which proves the feasibility and effectiveness of the constructed model and design algorithm.Sex.