An algorithm for fault detection in stochastic non-linear state-space models using particle filters

We propose a novel model-based algorithm for fault detection in nonlinear and non-Gaussian systems. The algorithm utilizes particle filters to generate a sequence of hidden states, which are then used in a log-likelihood ratio test to detect faults. The state-space models considered in this article are not easily amenable to standard log-likelihood ratio test, hence, a novel test statistic based on the joint likelihood function of hidden states and measurements is proposed. The proposed scheme is illustrated through an implementation on a highly non-linear multi-unit chemical reactor system.