Enabling the Evaluation of Production Scheduling Algorithms in Complex Production Environments Using Individually Deployable Scheduling Services

Abstract

Changes in customer demands and technological advances increase the complexity of production scheduling. Hence, current production scheduling algorithms are not sufficiently good. Additionally, advances in the research of Machine Learning algorithms drive the development of new scheduling algorithms. Each algorithm’s quality is problem-dependent, making it challenging to find the best algorithm for a given production scenario. Benchmark problems only provide guidance as they do not reflect real-world situations. To address this issue, in this article, we develop the software artifact Simfia that allows researchers and practitioners to evaluate production scheduling algorithms in highly customizable experiments. To create the solution, we follow a design science research approach. We identify 30 functional requirements, develop the solution prototypically and demonstrate it in an exemplary production scenario. The artifact is developed in a service-oriented architecture where scheduling algorithms are provided as individually deployable scheduling services that Simfia manages. This solution enables researchers and practitioners to experiment with production scheduling algorithms in highly configurable production scenarios.