The world is moving towards renewable sources of energy for its energy needs and solar and wind power are one of the most promising sources of ‘clean’ energy. There is a high level of uncertainty and variability associated with these sources of energy due to their dependence on weather conditions. For large scale deployment of Renewable Energy Sources (RES), it is important to develop strategies to integrate RES with the grid. The aim of this study is to develop wind and solar power forecasting models to improve the accuracy of RES using advanced forecasting techniques. A model using a closed-loop non-linear autoregressive artificial neural networks (CL-NAR-ANN) has been implemented to forecast solar power without the use of numerical weather prediction (NWP) as input. This method is compared with its exogenous variant with solar irradiance as input as well as other data driven models. The results suggest that the proposed model outperforms other models and can serve as a low-cost backup solution for situations when NWP data is not available. A probabilistic forecast is also implemented. A neural network based one hour ahead to one day ahead wind power forecasting model using NWP data as input is presented. The inputs for the model are chosen after performing a sensitivity analysis of the variables. A multi-model ensemble forecast approach with NWP members is presented. A probabilistic wind power forecast is also presented using the multiple NWP members as input.