DOI: 10.5176/2251-211X_WebTech25
Authors: András Rövid, Péter Várlaki
Abstract: It is known that most real-life systems are mainly non-linear. Supply chains fall into this category, as well. The main aim of the paper is to show how these systems may be modelled by using heuristic approaches in combination with analytical methods. The proposed concepts are based on fuzzy reasoning and neural network like models. When modelling parameter varying systems local models corresponding to predefined regions in the parameter space are considered and blended to obtain the system output. Concerning the blending procedure fuzzy clusters and the so called tensor-product based representation and the related higher order singular value decomposition will be described.
Keywords: System Modelling, Identification, Local Models, Model Approximation
Updating...