Krenn and Wotawa \cite{Krenn2009} proposed a similar approach. Instead of just using the first rule of the list, the rules' selection frequency could be dynamically updated during operation. The rules' selection frequency was in this approach based on biological processes of DNA transcription. Furthermore, rules do not only have precondition, but preconditions and postconditions.
Our approach builds upon this line of research. Our approach extends this further with two main contributions. First, we created an interface to PDDL, enabling the approach to apply to a wide range of already existing models. Second, instead of the biological inspiration, we use SFL, which was already proven to be successful at detecting faulty components in software testing, which is more akin to the problem of distinguishing faulty from non-faulty actions.