Merging fuzzy observer-based state estimation and database classification for fault detection and diagnosis of an actuated seat

This paper proposes an approach for Fault Detection and Diagnosis (FDD) of an actuation system for passengers seats in commercial aircrafts. The FDD is performed using classification algorithms. The supervised classification algorithms are usually based on data collected from the different sensors installed on a real system. Thus, to reduce the number of embedded sensors and so the costs of seat components in commercial aircrafts, a fuzzy Takagi-Sugeno (T-S) state observer is considered to estimate non-measured state variables in order to enrich the database used for the supervised classification process. From experimental measurements on a prototype of the actuated seat, the benefit of adding T-S observer-based estimations is illustrated through a comparison of the classification results obtained using databases without then with estimated data.