A Data-driven Scheduling Scheme for Shared Bicycles based on Spatio-temporal Nested Line Segment Tree Modeling

Abstract

The scheduling of shared bicycles often relies on manual experience, but frequent scheduling still cannot address the coexistence of "overcrowding" and "vacancies"; If artificial intelligence algorithms and big data frameworks are used, the cost is high and the operation is complex, making it difficult for small and medium-sized scheduling operators to undertake. The project constructs a mathematical model using a spatio-temporal dual dimensional nested line segment tree, breaking through the limitations of traditional single dimensional line segment trees and enabling rapid acquisition of bicycle supply and demand data at any time period and region. The designed data-driven bidirectional scheduling scheme can combine the stock and supply and demand data of parking points with the optimal quantity of bicycles across hourly intervals, introduce quantitative supply-demand differences and dynamic thresholds, and calculate and filter the scheduling volume of "small supply-demand fluctuations" based on spatio-temporal dimensions, thereby avoiding ineffective scheduling and achieving the transformation of scheduling operations from experience driven to data-driven. The implementation of the solution does not require complex big data frameworks or databases, is lightweight and easy to deploy, and adapts to the technical capabilities of small and medium-sized shared bicycle dispatch operators. It can also provide reference for span data processing in areas such as library partition book allocation and scarce water and electricity supply allocation.