DATA DRIVEN METHOD FOR FAULT DETECTION AND ISOLATION IN NONLINEAR UNCERTAIN SYSTEMS

The problem of fault detection and isolation is studied for critical purpose systems whose features are nonlinearities and uncertainties. Solution of the problem assumes the checking redundancy relations existing among measured system inputs and outputs. A novel method is proposed for constructing redundancy relations on the base of system model described by differential equations whose right-hand side is polynomial. The method involves nonlinear transformation of the initial system model into strict feedback form. Algebraic and geometric approaches are used for this transformation. Application of the method is demonstrated for motion model of autonomous underwater vehicle.