Bees, Bats and Glowworms: Swarm Algorithms for Optimizing Industrial Plants from the Bottom-Up

The scheduling problem in production plants with diverse product ranges is NP-hard, posing challenges in finding optimal solutions. To overcome this, we propose a novel approach utilizing three bio-inspired optimization algorithms: artificial bee colony (ABC), bat, and glowworm algorithm. We apply these algorithms to address the job-shop scheduling problem and overcome computational barriers associated with traditional methods. Unlike previous studies, our approach departs from using these algorithms for global optimization within the solution space. Instead, we adopt a bottom-up strategy, directly applying them as verbatim swarm intelligence algorithms. Focusing on a semiconductor production plant, we employ agent-based modeling in the NetLogo simulation platform. By mapping bees, bats, and glowworms to plant entities like lots and machines, we establish direct correspondences. Agents interact with each other and the environment based on local rules, resulting in the emergence of desired global behavior—the industrial plant optimization. To evaluate performance, we compare our approach to a baseline algorithm employing engineered heuristics like First-In-First-Out (FIFO) and filling fullest batches first. Through this comparison, we assess the effectiveness of the bottom-up algorithms. Our results show promising performance improvements achieved with these algorithms, which rely on low-effort local calculations.