Computer Engineering and Applications ›› 2023, Vol. 59 ›› Issue (4): 280-289.DOI: 10.3778/j.issn.1002-8331.2108-0495

• Engineering and Applications • Previous Articles     Next Articles

Study on Multi-Period Robust Tracking Shelter Hospital Location Based on Data-Driven

XIAO Yingchen, CHEN Gang   

  1. School of Management, Guizhou University, Guiyang 550025, China
  • Online:2023-02-15 Published:2023-02-15



  1. 贵州大学 管理学院,贵阳 550025

Abstract: The existing medical services cannot meet the current treatment needs when major infectious diseases break out on a large scale. Whether the Fangcang shelter hospital can be built in time and reasonably affects directly the safety of people’s lives and the effect of disease control. However, the number of disease infections fluctuates with time and is highly uncertain, which increases the difficulty of decision-making. To solve this problem, a robust tracking location model of multi-period Fangcang shelter hospital with the goal that minimum service capacity meets the maximum expected treatment demand and minimum expected total weighted distance based on data-driven is established. Firstly, the sigmoid function is used to learn the cumulative diagnosis data and predict the number and trend of infections. Secondly, the box uncertainty set is introduced to describe the uncertainty of the number of infected people, and the robust optimization is used to control the accuracy of prediction data. Thirdly, the model is transformed into mixed integer programming by robust counterpart theory, and a solution method is designed based on Python GUROBIPY and SCIPY modules. Finally, a case study on the location of Wuhan emergency Fangcang shelter hospital is proposed to verify the feasibility and effectiveness of the model. The results show that the robust tracking location has strong robustness. When resources are insufficient, treatment needs can be met in most periods, and when resources are sufficient, treatment needs in all cycles can be met without wasting resources.

Key words: emergency facilities location, multi-period, robust optimization, data-driven, uncertainty

摘要: 重大传染疾病大规模爆发时,现有医疗服务不能满足当前治疗需求,方舱医院能否及时合理地建成,直接影响人民群众的生命安全及疾病控制效果,但疾病感染人数随时间波动且高度不确定增加了决策的难度。针对该问题,从数据驱动的视角,构建以预期最小的服务能力满足最大的预期治疗需求和预期总加权距离最小为目标的多周期方舱医院鲁棒追踪选址模型。首先利用sigmoid函数对累计诊断数据进行学习,预测感染人数及趋势。其次引入box不确定集合刻画预测感染人数的不确定性,运用鲁棒优化控制预测数据的精准度。然后通过鲁棒对等理论将模型转化为混合整数规划,并基于Python的SCIPY和GUROBIPY模块设计求解方法。最后以武汉应急方舱医院选址为例,验证模型的可行性和有效性。结果表明鲁棒追踪选址具有较强的鲁棒性,在资源稀缺的情况下,能够满足大部分周期治疗需求,在资源充足的情况下,能够满足所有周期的治疗需求且未造成资源浪费。

关键词: 应急设施选址, 多周期, 鲁棒优化, 数据驱动, 不确定性