Model Goodness of Fit for Virtual Commissioning Purposes Based on Fuzzy-inference System

This paper is concerned with goodness of fit evaluation for virtual commissioning modelling purposes. Our goal is to propose a coefficient that could take into consideration several commonly used methods and expert knowledge referring to model quality evaluation. In this paper we try to find an answer to the question whether the model is good from the virtual commissioning point of view and if it can be used in virtual commissioning. The aim of virtual commissioning is to create a simulation model of a plant. It is very useful and crucial from the modern automation point of view (especially Industry 4.0) owing to the fact that potential changes and upgrades can be tested before they are implemented to an existing process. It is a formidable challenge to introduce a method which allows to unambiguously decide whether the model could be used in virtual commissioning. In order to evaluate the goodness of fit, commonly used performance indices are NRMSE (Normalized Root Mean Square Error) and ME (Maximum Error). In this work the unique way of combining NRMSE and ME with fuzzy logic has been introduced. For evaluating the goodness of fit of a model we propose a coefficient that is based on Takagi-Sugeno-Kang fuzzy-inference system. The suggested method is flexible and well-suited for all kind of models and processes because of taking into consideration all aspects of a process. What is more, it also gives an easy way of applying expert knowledge into it.