Model based diagnosis of large continuous dynamic systems requiring quantitative simulation has a high computational cost, which can be reduced by distributing the computation. Distribution can be obtained partitioning the original diagnosis problem into the analysis of simpler subproblems. In this work, Possible Conflicts are used to partition a system because they provide a systematic way to decompose a system. How- ever, a requirement of any decomposition method is that the resulting subsystems are observable. This paper focuses on structural observability, a powerful concept because it allows analyzing the observability of a system in terms of its configuration, i.e., independently of system parameter values. However, the literature provides different definitions of structural observability, adapted to different modeling formalisms: equations, bipartite graphs and bond graphs. This paper shows that definitions for these formalisms are equivalent. The three tank system benchmark and a spring-mass system are used to illustrate the definitions and their equivalence. Then, it will be applied through Possible Conflicts to build independent subsystems that can be used for monitoring and diagnosis.
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.
Hybrid systems diagnosis requires different sets of equations for each operation mode in order to estimate the continuous system behaviour. In this work we rely upon Hybrid Possible Conflicts (HPCs), which are an extension of Possible Conflicts (PCs) for hybrid systems, that introduce the information about potential system modes as control specifications that activate/deactivate different sets of equations. We also introduce the concept of Hybrid Minimal Evaluation Models (H-MEMs) to represent the set of globally consistent causal assignments in an HPC for any potential mode.
H-MEMs can be explored for a specific operation mode, and its computational model automatically generated. In this work, the selected computational models are minimal Dynamic Bayesian Networks (DBNs). Since DBNs can be directly generated from PCs, and can be used for fault detection and isolation, we propose to efficiently generate Mini- mal DBNs models on-line using the H-MEM structure. By introducing fault parameters in the DBN model, we can also perform fault identification, providing an unifying framework for fault diagnosis, under single fault assumption. We test the approach in a simulation four-tank system.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.