On the application of a fast identification algorithm to fault detection and diagnosis in car engines

This paper presents the results of an investigation into the applicability of a fast identification algorithm for fault detection and diagnosis in a multicylinder car engine. A nonlinear ARMA model is established, where the coefficients of the model are nonlinear functions of the shaft angle. This approach is compared with a previously proposed parameter estimation based approach. It is shown through simulations that the new algorithm produces the same results with only ten per cent of the computational load therefore making it more suitable for online implementation.