Scheduling Under Uncertainty For Industry 4.0 Applications: A Simulation-Based Genetic Algorithm Approach
Year: 2025
Type: conference-paper
Abstract: In highly uncertain and complex environments like industrial production systems, accounting for system variability is essential for informed and resilient decision-making. Industry 4.0 envisions intelligent systems that autonomously make decisions based on comprehensive analyses of current and future states. Incorporating uncertainty into production planning models, rather than relying solely on deterministic assumptions, is key to achieving these goals. Simulation enables realistic modeling of complex systems with minimal assumptions, making it essential to Industry 4.0 technologies like digital twins. Though effective in capturing stochastic behavior, simulation struggles to explore large solution spaces efficiently. Combining it with metaheuristics addresses this by leveraging simulation's feedback and uncertainty management with the search power of metaheuristics, leading to more robust and comprehensive solutions. Most production planning studies rely on deterministic models, often neglecting real-world uncertainities. This study presents a simulation-optimization approach for the single machine scheduling problem with stochastic processing times and sequence-dependent setup times. A genetic Algorithm (GA) solves a simplified deterministic version of the problem in each iteration, generating promising solutions, which are then evaluated through multiple replications in a discrete event simulation. The simulation provides feedback based on the average and variance of fitness values, which is used to update the pool of promising solutions fed back to the GA. After completing all generations, the best solution is simulated over a longer horizon to refine its confidence interval. This process assesses Schedule performance under uncertainty and enhances optimization with realistic feedback. To validate the method, it is tested on problems of varying sizes and compared with the traditional GA+Simulation approach, demonstrating superior scheduling performance.
Authors: beyza günesen akansu, Muzaffer Kapanoğlu
Institution: Eskişehir Osmangazi University
Language: English
Subfield: Industrial and Manufacturing Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7104014499
FWCI:
Cites:
Cited by:
Related to: