[4Xin2] Poster session 2

[4Xin2-83] Search System for Production Backup Policy Using Agent-Based Infection Simulator and Bayesian Optimization

〇Masaki Kitazawa^1,3, Satoshi Takahashi², Atsushi Yoshikawa² (1.Kitazawa Tech, 2.Kanto Gakuin University, 3.Rikkyo University)

Keywords:Agent-based simulation, Infectious disease, Production backup, Bayesian optimization

This study develops a search system combining an agent-based infection simulator with Bayesian optimization to identify optimal backup policies that minimize infections and backup workers at production sites during outbreaks. The backup policy sets the scope of absence for the infected worker and surrounding workers. Appropriate adjustment of the backup scope is crucial; a scope too narrow won't contain the infection, while one too wide requires unnecessary personnel. We experimented with the developed system in 12 different scenarios and obtained three findings: 1) The developed system could provide optimal backup policies similar to public health strategies in certain scenarios, such as when influenza was prevalent and masks were not used, 2) A single policy was optimal throughout the entire simulation in certain scenarios, 3) The appropriate backup implementation policy varied depending on the factory size, mask usage rate, and virus type. It was also important to quickly identify the causative virus and utilize this information for countermeasures.