Improved approach to KDI-based fault detection for non-linear black-box systems

This paper is concerned with an application study of model-based fault detection method to a ship propulsion system which is the object system for benchmark test of fault diagnosis. When modeling the object, quasi-ARMAX model with multi-model form is used. In this model, the system non-linearity is incorporated into model parameters by using non-linear non-parametric models (NNMs). Kullback discrimination information (KDI) is introduced as fault detection index to evaluate the distortion in identified model, which is caused by a fault. Although the fundamental effectiveness of the method has been verified through simulation studies, there have been still remained problems as to the detection performance. In this paper, some improvement schemes to solve these problems are proposed together with simulation results for various fault modes.