This paper proposes a new approach for designing a robust polynomial observer to estimate the unmeasurable states and reconstruct the faults of the Wind Energy Conversion Systems (WECSs). The WECSs in this study are affected by the uncertainties and faults. In this paper, firstly, the WECSs are modeled under the framework of the polynomial systems instead of linearizing at operating points to reduce the modeling errors. Secondly, the WECSs are reformulated in a new form then a polynomial observer is synthesized for this system to eliminate the influences of uncertainties and faults, estimate the unknown states, and reconstruct the faults. On the basis of Lyapunov theory and Sum-Of-Square approach, the conditions for observer design are derived in the main theorems. Finally, the simulation results are provided to prove the effectiveness of the proposed method.