Imperative vs. Declarative Modeling of Industrial Process. The Case Study of the Longwall Shearer Operation.

Process modeling is an important and necessary step for further analysis and monitoring of industrial processes. In the process modeling two main paradigms exist, namely imperative and declarative ones. In our work, we analyzed information potential of these model paradigms regarding to conformance checking task of real life industrial process - longwall shearer operation carried out in an underground coal mine. The objective of our work was an analysis of selected imperative and declarative models to discover which approach is more appropriate from a practical point of view, taking into consideration criteria formulated by the domain expert. The first novelty of our work rely on real life industrial sensor data analysis and creation of event log with heuristic approach for case ID identification and labeling with expert rules. In parallel, we created prescribed process models. As representatives of imperative and declarative languages, we have selected the Petri nets and Declare models, respectively. We created two Petri nets (with Inductive and Heuristic Miner) and seven declarative models differ in restriction power. Due to the better description of the ideal cycle, to the further analysis and conformance checking task, we selected the Petri net created by Heuristic Miner. After the process model creation, we compared selected Petri net with Declare models using the natural language approach and constraints hierarchy. Based on created similarity measures, we choose one declarative model to conformance checking task and comparison with Petri net due to formulated quantitative and qualitative criteria. As main artifact in the conformance checking task, we used obtained real event log. Evaluation of the created models indicates that in the case of the longwall shearer operation monitoring, the declarative model better captures the necessary information to decision-makers than the Petri net, thus being more appropriate for practical use.