Application of Machine Learning Within Hybrid Systems Modelling

Due to the easy and affordable access to more and more computing power, as well as the continuous improvement and distribution of easy-to-understand and easy-to-use open-source libraries and packages, the popularity of machine learningMachine learning algorithms has increased significantly in the last years. This has led to a debate in the simulationSimulation research community as to whether the discipline of modellingModelling and simulationSimulation will eventually be made obsolete by machine learningMachine learning. However, this apprehension could not be further from the actual reality. In fact, both disciplines can complement each other very well to form a whole that is greater than the sum of its parts, provided that they are combined in a hybrid system in a reasonable way. In this paper, an overview of the existing possibilities is given for each of the three main categories of machine learningMachine learning, i.e., supervised, unsupervised, and reinforcement learningsReinforcement learning. This includes guidelines, application potentials, and use case examples on how to combine simulationSimulation and machine learningMachine learning in a hybrid system.