Sensor fault detection in a class of nonlinear systems using modal Kalman filter.

Kalman filter and its different variants are commonly used as optimal methods for fault detection in various types of system components. In this paper, a newly introduced type of aforementioned filters, called modal Kalman filter, is extended and utilized in order to estimate the states of nonlinear systems, for sensor fault detection purposes, in a class of nonlinear certain systems. This method, in contrast to the extended Kalman filter, which employs only the linear term of Taylor expansion, retains higher-order terms; as a result, the estimation error will reduce accordingly. Practicality and effectivity of this method, and its superiority over Kalman filter, in terms of accuracy and promptness of sensor fault detection, are also verified with simulation results. • A novel model-based fault detection approach, based on a new modification of Kalman filter is proposed. • This approach, called Modal Kalman filter, is based on modal series. • Applied to nonlinear systems for sensor fault detection. • Both bias fault and loss of effectiveness fault are taken into consideration. • The results approve the efficiency of the proposed filter. [ABSTRACT FROM AUTHOR]