Research On Demand Forecasting And Intelligent Scheduling Method Based On Shared Bicycle Track Data

With the acceleration of the modernization process,congestion and environmental pollution caused by the surge in urban road traffic has become one of the important factors affecting the people's happiness index.With the continuous development of green,sharing and other development concepts,the shared bicycle travel mode has emerged.It not only reduces the pressure and congestion of the urban road network,but also provides convenience for people to travel.However,due to the lack of a well-established shared bicycle operation management system,the unreasonable operation and unscheduled operation caused problems such as chaos and chaos,and difficulty in finding a car,which made the shared bicycle play a greater role in the urban road transportation system.Therefore,accurately predicting user travel demand and arranging reasonable scheduling schemes play an important role in bringing the efficiency and dvantages of shared bicycle system.This paper takes 3 million shared bicycle travel records in a certain area of Beijing as a sample,analyzes the demand forecasting and demand allocation optimization of the shared bicycles in the area,and studies the intelligent scheduling strategy to provide intelligent solutions for shared bicycle operation management.Firstly,the bicycle trajectory data is used to obtain position and distance information through Geohash transform.The sample destination is constructed by sample filtering,rule pre-filtering and starting drift.Statistics and mining are performed on the data,and data is constructed from two aspects:feature group and feature type.Then,based on the historical travel trajectory information and travel rules of each region,the bicycle traffic between different regions is obtained.According to the data type,the secondary K-means algorithm is used to realize the division of the scheduling region,and the regional cluster directly related to the prediction region is obtained..At the same time,based on the self-flow and correlation of travel between different travel locations,an LSTM-based linear network prediction model is established,which can extract the shared bicycle travel flow for a long time on the basis of solving the gradient explosion and gradient dissipation of the traditional neural network model.The variation law of the interval,and the distance feature extracted from the trajectory data is included in the prediction model,and the correlation between each travel area is described from two dimensions of time and space,and the demand is accurately predicted.Finally,based on the scheduling area division,the static scheduling mathematical model with the optimal scheduling co st is established to meet the regional scheduling requirements.The scheduling model solving method based on genetic algorithm and scheduling route construction scheme are proposed.The analysis results show that the route construction scheme based on genetic algorithm can obtain high-quality solutions in a limited time;increasing the dispatching vehicle within a certain range can not only reduce the scheduling cost,but also speed up the algorithm.