In this paper, a distributed consensus-based algorithm is introduced for social welfare maximization problems in the presence of responsive loads and Wind Turbines (WTs) in smart grids. This distributed algorithm eliminates the need for a central controller in the system. Consequently, it prevents the curse of dimensionality, the computational burden and the communication congestion which are some drawbacks of using centralized algorithms for controlling smart grids. The effect of uncertainty of output power of WTs generation is another discussed issue. Modeling this uncertainty is investigated using Unscented Transformation (UT) method. In MATLAB simulation, the performance of this distributed algorithm is illustrated in the presence of demand response in some multi-agent systems. Additionally, the impact of WTs uncertainty is inspected in a 29-node system.