Massive COVID-19 vaccination can significantly reduce both mild and severe infection rates. Some governments have adopted mobile vaccination vehicles, offering a more convenient and flexible service compared to static walk-in sites. This paper addresses a new scheduling problem arising from the mobile COVID-19 vaccination planning practice. Given a set of communities, each with a specific number of residents to vaccinate, the objective is to assign mobile vaccination vehicles to communities and determine each vehicle’s service capacity and routes, attempting to minimize the total operational cost. To our knowledge, this is the first attempt to tackle the joint challenge of mass vaccination scheduling and routing. We formulate the problem as a mixed-integer nonlinear program model, which we linearize by treating each vehicle with multiple stations as separate units. Given that the problem is NP-hard, we then developed a tailored adaptive large neighborhood search (ALNS) approach that effectively solves practical-sized instances by utilizing the intrinsic structure of the problem. To illustrate the efficiency of the suggested model and solution methodologies, we conduct numerical experiments on instances of varying sizes. The results demonstrate the effectiveness of the developed ALNS algorithm in solving instances with realistic sizes, efficiently handling up to 100 communities and 14 vaccination vehicles. In addition, a case study shows that our method significantly reduces operational expenses compared to some experience-based greedy methods.

中文翻译：

---

具有容量选择的移动 COVID-19 免疫接种计划


大规模 COVID-19 疫苗接种可以显着降低轻度和重度感染率。一些政府已经采用了移动疫苗接种车，与静态的步入式接种点相比，提供了更方便、更灵活的服务。本文解决了移动 COVID-19 疫苗接种计划实践引起的新调度问题。给定一组社区，每个社区都有特定数量的居民需要接种疫苗，目标是为社区分配移动疫苗接种车并确定每辆车的服务容量和路线，以尽量减少总运营成本。据我们所知，这是应对大规模疫苗接种计划和路线规划的联合挑战的首次尝试。我们将问题表述为混合整数非线性规划模型，通过将具有多个站点的每辆车视为单独的单元来对其进行线性化。鉴于问题是 NP 困难的，我们随后开发了一种量身定制的自适应大邻域搜索 （ALNS） 方法，该方法通过利用问题的内在结构有效地解决了实际大小的实例。为了说明建议的模型和求解方法的效率，我们对不同大小的实例进行了数值实验。结果表明，开发的 ALNS 算法在求解具有实际大小的实例方面的有效性，可有效处理多达 100 个社区和 14 辆疫苗接种车。此外，案例研究表明，与一些基于经验的贪婪方法相比，我们的方法显著降低了运营费用。