Factoring Dynamic Bayesian Networks using Possible Conflicts

Abstract

Dynamic Bayesian Networks (DBNs) are temporal probabilistic graphical models that represent in a very compact way dynamic systems. They have been used for model based diagnosis of complex systems because they naturally cope with uncertainties in the diagnosis process, particularly sensor uncertainty in noisy environments. A caveat of DBN is the complexity of the inference procedure which is usually performed with Particle Filtering algorithms. Recently, factoring has been proposed to decompose a DBN into subsystems, distributing the diagnosis process and reducing the computational burden. This paper proposes decomposing a system with Possible Conflicts (PCs) and, afterwards, building a DBN factor from each resultant PC. The method can be systematically applied to a state space representation of a dynamic system to obtain minimal observable subsystems with analytical redundancy. Assuming single fault hypothesis and known fault modes, the method allows performing consistency based fault detection, isolation and identification with the unifying formalism of DBN. The three tank system benchmark has been used to illustrate the approach. Two fault scenarios are discussed and a comparison of the behaviors of a DBN of the complete system with the DBN factors is also included.