Alarm design for nonlinear stochastic systems

Abstract

The nonlinear stochastic systems pose two important challenges in designing alarms: 1) measurements are not necessarily Gaussian distributed and 2) measurements are correlated - in particular for closed loop systems. We present an algorithm for designing alarms for such systems with unknown and known models. In the case of unknown models our approach is based on Monte Carlo simulations. In the case of known models, it makes use of a probability density function approximation algorithm called particle filtering. The alarm design algorithm is illustrated through two simulation examples. It was shown that the proposed alarm design was effective in detecting the fault even though the measurements were non-Gaussian.