Model of MRDTAN and Its Reasoning Mechanism System
Xu Ma · Industrial Engineering and Management · 2014
Aiming at problem on false decision losses in fault diagnosis procedures,a new model named MRDTAN(minimal risk decision based on TAN)was presented,which integrates tree augmented naive Bayes(TAN)with minimum risk decision.Decision loss measure function was introduced to construct TAN based reasoning model with minimal risk criteria,based on which,a three layered reasoning mechanism was constructed so that the fault diagnosis could be minimized.The effectiveness of the model was validated with the operator's physiological parameters data set of aerospace system in practical diagnostic applications.Comparing with the diagnosis results of other relative intelligent inference models,the experimental results shows that Bayesian network classifiers with minimal risk criteria is lower total decision loss in diagnostic reasoning.