This thesis mainly focuses on Fuzzy Fault Tolerant Predictive Control for a class of nonlinear systems. The Takagi-Sugeno (T-S) fuzzy approach is introduced as a modelling technique in order to consider the active control methods adapted to linear models. To obtain the convex form, two approches are applied in this work, the proposed global non stationary linearization method and the sector nonlinearity approach. The contributions of this thesis and novelties with respect to other works are based on a combination between Parallel Distributed Compensation control law and Model Predictive Control where the T-S fuzzy aspect uses measured and unmeasured premise variables. The optimization problem is formulated as a quadratic programming problem. A nonlinear observer and A T-S fuzzy observer are designed for the proposed strategies, in order to estimate faults and system state variables. The controller and observer gains are obtained by solving Linear Matrix Inequalities (LMIs) derived from the Lyapunov theory. Convergences are performed by using Lyapunov asymptotic stability and L2 optimization. Actually, the use of the sector nonlinearity approach has reduced the conservatism related to the number of LMIs to solve. On top of that, the chosen form of the candidate function of Lyapunov and the T-S fuzzy structure have significantly decreased the pessimism of sufficient stability conditions derived from Lyapunov theories. The proposed Fuzzy model based predictive control is designed to achieve desired set points and control objectives in the the healthy operating and to accommodate and tolerate unexpected faults. Furthermore, the uncertain case and robustness with respect to constraints are investigated. The effectiveness and the validity of the proposed Fault Tolerant Control (FTC) strategies is illustrated through an application to an academic example and to a Diesel Engine Air Path (DEAP) system.