基于Markov过程天气预测的共享单车调度优化研究

Optimization of Bike-Sharing Scheduling Based on Markov Process for Weather Prediction

针对某大学城共享单车资源分配不平衡问题,建立了与调度成本和用户满意度相关的单目标数学规划模型,对现有共享单车调度策略进行分析,并制定改善方案.通过调查问卷分析单车调度过程中存在的问题,利用Python获取单车的相关数据.在此基础上,基于齐次Markov过程对天气状况进行预测,采用LSTM时间序列对区域内未来单车数量的变化情况进行预测,利用遗传算法求解最优调度路径和调度数量,提出优化方案.通过敏感性分析验证本文所提方案的可行性.结果表明,优化后的调度方案缩短了单次调度时间并提高了用户的满意度,对共享单车管理部门具有指导意义.

To address the imbalanced resource allocation of shared bikes in a university town,a single-objective mathematical planning model related to scheduling cost and user satisfaction is established to analyze the existing bike-sharing scheduling strategy and make improvement solutions.The problems in the process of bike-sharing scheduling are analyzed through questionnaires,and Python is used to obtain the relevant bike data.Based on this,weather conditions are predicted using the homogeneous Markov process,future changes in the number of bikes in the region are predicted using LSTM,optimal scheduling paths and scheduling quantities are solved using genetic algorithms,and the optimization solutions are proposed.The feasibility of the solutions proposed in this paper is verified by a sensitivity analysis.The results show that the optimized scheduling solutions shorten the single scheduling time and improve the user satisfaction,which is instructive for the management of shared bikes.