A Cooperative Fault Detection Approach for Stochastic Multi-Agent Systems Using Maximum Likelihood Estimation Method

This study addresses the fault detection problem in multi-agent systems (MASs) with additive faults and stochastic uncertainties. The main focus is on enhancing the fault detection capability of each agent through a cooperative fault detection scheme, fostering cooperation between agents in two scenarios. For Gaussian uncertainties, one scheme is developed using the maximum likelihood estimation (MLE) matching expectation maximization (EM) algorithm. Additionally, a novel cooperative fault detection scheme is introduced to handle non-Gaussian uncertainties, where the cooperation mechanism among agents is determined by approximating non-Gaussian uncertainties using the Gaussian mixture model (GMM). The effectiveness and improvements of the proposed cooperative fault detection method are validated through numerical simulations.