Abstarct: The goal of this thesis is to design the robust model-free observer for detecting the fault existed in the suspension and brake railway (Diesel Electric) system in the presence of external disturbances. In this thesis, the fault detection unit is developed baxsed on TSK fuzzy observer to estimate and detect the fault occurred in the system. Moreover, residual signal is robust against the fault and disturbance signal. Satbility of the overall system is derived to guarantee the convergence of the error estimation to zero. In order to show the capability of the proposed method, the fuzzy observer is designed for both suspension and brake systems. In addition, a neural observer baxsed on Leunberger method is designed for brake system of the train.