This paper proposes a new approach to designing the observer for the nonlinear Large-scale system. The nonlinear large-scale system is modeled under the T-S fuzzy system framework that allows us to apply powerful linear methodologies to synthesize the observer. The interconnection terms of considered large-scale T-S fuzzy systems are unknown and do not need to satisfy any constraint. In addition, both the system and sensor of the large-scale T-S fuzzy system are impacted by the fault. A decentralized fuzzy observer is designed to estimate the unknown states as well as reconstruct the fault, simultaneously. With the support of the augment technique, Lyapunov methodology, and Linear Matrix Inequality (LMIs) technique, the conditions for observer design are derived in the main theorems. Finally, a numerical example is provided to show the effectiveness and merit of the proposed method.