In this paper, parameter-varying technique is firstly addressed for modelling a 4.8 MW wind turbine system which is nonlinear in essence. It is worthy to point out that the proposed parameter-varying model is capable of describing a nonlinear real-time process by using real-time system parameter updating. Secondly, fault reconstruction approach is proposed to reconstruct system component fault and actuator fault by utilizing augmented adaptive observer technique with parameter-varying. Different from the offline tuning adaptive scheme, the proposed adaptive observer includes adaptive tuning ability to online adjust the observer based on varying parameter. The effectiveness of the proposed parameter-varying modelling and fault reconstruction methods is demonstrated by using a widely-recognized 4.8 MW wind turbine benchmark system.