This paper proposes an innovative fault diagnosis system for a turbocharged diesel engine with variable geometry turbocharger control. Numerous and diversified actuator faults are identified and analyzed such as air leakage in the admission collector, compressor malfunctioning, intake valves fault, intercooler fault, deterioration in the turbine-compressor coupling, defect in the variable geometry of the turbine. Furthermore, a complete non-linear engine model with four state variables is adopted. The proposed strategy consists in developing Fault Detection and Estimation algorithm (FDE) based on the theory of mu analysis control, operated on a linearization LTI model of the diesel, permits a fault detection and estimation while attenuating the effect of uncertainties, disturbances and noise. Simulations with LTI Diesel model in the presence of noise and uncertainties were carried out which demonstrated the effectiveness of the proposed algorithm. Moreover, the robustness properties of the used H-infinity/mu FDE filters exhibited significant rejection of disturbances and noise-effects attenuation as well as robustness against uncertainties which make the proposed strategy adequately close to the conditions of the real physical system.