Consensus Based Fault Detection and Isolation for Systems of Systems

In this paper a new algorithm is proposed for distributed fault detection and isolation (FDI), applicable to systems of systems. The algorithm is composed of two stages. The first stage is proposed in the form of a consensus based network of interconnected optimal stochastic FDI observers connected to the constituent systems. At this stage residuals are generated, which provide high efficiency, scalability and robustness under general conditions concerning locally available models and the topology of the overall system. The second stage is aimed at fault detection based on the generated residuals. A new distributed detection algorithm based on consensus is proposed, enabling overall decision at the constituent system level, without fusion center. Simulation results illustrate the efficiency of the proposed algorithm.