Research On Optimization Algorithm Of Shared Bicycle Delivery Scheduling

Nowadays,with the rapid development of the sharing economy,shared bicycles gradually enter people’s vision.The emergence of shared bicycles is mainly to solve the problem of people’s travel.It facilitates people’s lives and reduces the degree of air pollution from automobile exhaust.But what comes with it is the problem that the supply of shared bicycles is in short supply and the supply exceeds demand.In order to solve the problem of path scheduling optimization,this paper takes 20 rental points near the White House in Washington area as research objects.Firstly,this paper analyzes the relevant factors that affect the number of shared bicycles in the rental points and obtains the geographic location of the rental points in the region.Then,the main influencing factors are used in the prediction model for training to predict the real-time number of shared bicycles required by each rental point in the future period,and then the number of shared bicycles to be distributed is calculated based on this value.Finally,the improved ant colony algorithm is used for optimal scheduling to obtain the distribution route with shorter distribution distance.The main research work is as follows:1)The acquisition of the number of shared bicycles at each rental point and the geographical location of the rental point is completed.This paper mainly obtains the number of shared bicycles owned from some rental points in Washington,D.C.,at the same time,obtains the corresponding factors affecting the rental number from Kaggle and sofasofa.The distance between rental points is obtained through the longitude and latitude on Baidu map.2)A model for predicting the number of shared bicycles to be shared in the future is established.Through thermodynamic correlation analysis,the key factors that affect the number of bicycles in the rental point are known,and the batch data obtained are simply processed according to the key factors to obtain the data format required for distribution scheduling.In this paper,the prediction model based on XGBoost is adopted,and the related parameters affecting the prediction results are adjusted.Compared with the gradient lifting decision tree and the neural network model,the prediction error is reduced and the result is more accurate.3)An ant colony algorithm based on pheromone attenuation method is designed.Considering the pheromone concentration that affects the ants’ judgment in the travel path,the initial pheromone concentration is set to a larger value to avoid all volatilization in a short time,and certain attenuation operation is carried out on pheromones released by ants in the path of searching food.The experiment shortens the distribution distance of the dispatching vehicle,saves the distribution time,and can quickly and effectively complete the distribution of shared bicycles required by each rental point.The experimental results show that the calculation method of ant residual pheromone concentration between lease points proposed in this paper can schedule shared bicycles more efficiently,and at the same time obtain shorter scheduling routes to complete the scheduling of shared bicycles.The problems existing in the research work of this article and the future work to be studied are explained at the end of the article.