Automatic Algorithm Design for Hybrid Flowshop Scheduling Problems

Industrial production scheduling problems are challenges that researchers have been trying to solve for decades. Many practical scheduling problems such as the hybrid flowshop are NP-hard. As a result, researchers resort to metaheuristics to obtain effective and efficient solutions. The traditional design process of metaheuristics is mainly manual, often metaphor-based, biased by previous experience and prone to producing overly tailored methods that only work well on the tested problems and objectives. In this paper, we use an Automatic Algorithm Design (AAD) methodology to eliminate these limitations. AAD is capable of composing algorithms from components with minimal human intervention. We test the proposed AAD for three different optimization objectives in the hybrid flowshop. Comprehensive computational and statistical testing demonstrates that automatically designed algorithms outperform specifically tailored state-of-the-art methods for the tested objectives in most cases.