Joining Analytic Network Process and Bayesian Network Model for Fault Spreading Problem
Gábor Szücs, Gyula Sallai · BiblioBoard Library Catalog (Open Research Library) · 2010
The ANP has been applied to a large variety of decisions: marketing, medical, political, military, social, prediction and many others. ANP is able to take analysis of benefits, opportunities, costs, and risks (BOCR) (Wijnmalen, 2007). Bayesian Networks are applied in query languages in scientific area of information retrieval (Cheng & Yang, 1999)0, in environmental modelling (Uusitalo, 2007). There are some improvements or combined versions of BN, e.g. combination with fuzzy (Li & Kao, 2005), and many authors deal with further development. There are some complex (decisional and engineering) problems, where neither ANP nor BN could help to solve alone. Some of these problems are usually such sophisticated, which involve human opinions with uncertainty, causal relationships, and uncertainties in the occurrence of events. These tasks can be solved by the proposed method constructed by combination of ANP and extended version of Bayesian Networks, i.e. by joining these two methods in cascade. Vector Bayesian Networks (VBN) is a generalized BN, which able to handle not only the probabilities, but any numerical value attached to nodes. This extension is able to calculate spreading of effects in any network. The combined method is particularly useful to investigate fault spreading problem in infocommunication networks.