A sensor fault estimation method is proposed for a general class of uncertain nonlinear systems. Taking the sensor faults as auxiliary states, an augmented descriptor system is constructed. For this augmented descriptor system, an adaptive fuzzy proportional‐derivative sliding mode observer is designed to estimate the sensor faults and states, simultaneously. Adaptive fuzzy systems based on the universal approximation theorem are used to approximate the nonlinearities. Thus, the nonlinearities can be unknown and the Lipschitz assumption can be utterly relieved. To relax the requirement of knowing the bound of uncertainties, an adaptive gain for the sliding mode term is used. The semi‐global asymptotic convergence of the proposed fault estimation observer is proved using the Lyapunov approach. The proposed method is applied to a highly nonlinear mathematical model with both bounded and unbounded sensor faults. Simulation results show the satisfactory performance of the proposed method.