题目：Stochastic Mixed-Integer Programming for Natural Disaster Relief

时间：2023年5月19日 14:00-15:30

地点：必赢线路检测中心 振华会议室

邀请人：刘冉 副教授（工业工程与管理系）

报告人简介

杨浩翔，香港中文大学（深圳）数据科学学院副研究员。研究领域：不确定性下的优化、大规模优化和分解方法、混合整数规划、非线性优化，应用于能源系统、救灾和援助、公共卫生、航空运输、供应链与体育数据分析。曾任Operations Research、Mathematical Programming等期刊编辑，多篇论文在Mathematical Programming、SIAM Journal on Optimization以及Nature Communications等期刊发表。

报告摘要

Stochastic optimization is an effective approach for modeling decision-making in natural disaster relief operations, particularly when faced with stochastic disruptions. In this talk, we discuss two specific application examples of infrastructure operations under natural disasters. The first example addresses the integrated protection and restoration of interdependent infrastructure networks. In the formulated two-stage stochastic program under disruption, the first stage focuses on the decisions regarding fortification of components, while the second stage considers repair crew schedules and network flow. The second example considers the preemptive shutoffs and power flow operations of an electric power system under wildfire uncertainty. The occurrence of the wildfire is also uncertainty, which motivates us to construct a two-stage stochastic mixed-integer program, where the recourse problem may have a random length. We show that in spite of non-convex recourse problems, both problems can be solved exactly by cutting-plane algorithms, featuring novel variants of Lagrangian cuts, and we will demonstrate the efficacy of such algorithms.