Comparison of Data-Driven (Fuzzy) Modelling Methods tested on NOx Data

This report is an application-experiment paper based on experimenting with real (NOx) data provided by Fuzzy Logic Laboratorium Linz-Hagenberg (FLLL) Johannes Kepler University in Linz. The NOx data which are described below have been studied and functional dependencies between them successfully modelled by several fuzzy models identified by data-driven methods by the FLLL institute; some particular result can be found e.g. in [2]. Bilateral project Aktion 41p19 between IRAFM-OU and FLLL-JKU made possible to realize a deeper cooperation between both institutes. One of key issues of the proposed cooperation was (for IRAFM) to benefit from experiences based on many applications and industrial projects solved by FLLL. Second part of the issue was (for FLLL) to benefit from theoretical research and techniques developed in IRAFM and implemented in the software package LFLC2000. Based on the cited project, it was possible to obtain a real data and to make lots of experiments which enriched IRAFM by experiences and prompted several improvements in the techniques developed in the institute as well as changes and implementations and the software package. FLLL institute which cooperated on the project will be provided with all methods used in the experiments and experiment results.