Metamodelling of Queuing Systems using Fuzzy Graphs

Abstract

The dynamic behavior of queuing systems under sophisticated traffic can be analyzed using simulation models. Unfortunately, often the results are only available in a form of large datasets, which makes it hard to extract the underlying regularities. One of the interesting applications is the approximation of the behavior of simulation models, called metamodelling. The goal of this paper is to approximate the behavior of queuing systems as well as to extract some understandable knowledge about the simulation model. In this paper we present the knowledge extraction from trained Neural Networks. The underlying knowledge can be extracted from the Network in form of a Fuzzy Graph. The Fuzzy Graphs are generated using Rectangular Basis Functions of Neural Networks. The research results are illustrated with a range of experiments performed. Ill. 1, bibl. 8 (in English; summaries in English, Russian and Lithuanian).