Fuzzy parity equation for fault detection and identification of nonlinear systems

A fault detection and identification method for nonlinear systems based on fuzzy parity equations is presented. The proposed technique involves approximating first the nonlinear system by a T-S model, and then fully-decoupled parity equations are derived for the local linear models. Residuals generated from the fuzzy parity equations in the T-S model are fused by another T-S model. As the residual generated by the fuzzy parity equation is sensitive to a specific actuator or sensor fault, and insensitive to other faults, system states and disturbance inputs, faults can be detected and/or identified. A simulation example involving a nonlinear airplane model is presented to illustrate the performance of the proposed technique. © 2007