With the integration of large-scale distributed renewable energy, the model of active distribution network (ADN) becomes more complicated. This paper proposes an equivalent ADN model considering the uncertainties of wind turbines, photovoltaic (PV) systems and loads. The model output is composed of deterministic and uncertain components. First, the model of wind turbines, PV systems and loads are developed to describe the deterministic components and the Gaussian probability density model is used to describe the uncertain components. Then, model parameters are optimized using a reinforcement learning algorithm to track the time-varying equivalent wind turbines, PV systems and loads. Finally, case studies are carried out on a 63-bus distribution network. Simulation results demonstrate that errors of power flow calculation of the equivalent ADN model are reduced in the case that the uncertainties of wind turbines, PV systems and loads are considered.