AbstractIn this article, a novel distributed fault detection and isolation method is proposed for large‐scale power system affected by additive type of faults and energy bounded disturbances. First, in order to design the local residual generators, a bank of Luenberger observers are designed to estimate the local states of the system, and the parameters of each local residual generator are determined by solving the corresponding optimization problem, which guarantees that the residual generator is sensitive to faults while robust to energy bounded disturbances. When the local residual exceeds the corresponding threshold, the occurrence of a fault in each agent is recognized. Then, for the case that power system with both process noises and measurement noises, the specific fault detection thresholds are designed with the aid of the Chebyshev inequality. Finally, the effectiveness of the proposed scheme is validated via illustrative example.