Data Driven Approach for Long Term Forecasting of Renewable Energy Generation

With the recent issues of the fossil fuels extinction for the electricity generation there is a paradigm shift towards renewable energy sources (RES) based generation. Even though the RES are clean source of energy, it suffers from the issue of intermittency due to the dependency on environmental condition. To address this the paper focuses on the forecasting of solar and wind power generation data using the available data to mitigate the effect of electricity demand-supply mismatch and facilitate the demand-side management program. For achieving this claims the paper implements the Dynamic Mode Decomposition (DMD) algorithm for forecasting the solar and wind generation. The DMD is a strategy for predicting system states by reducing data down into its primary modes, which are derived from a series of training data. The primary modes are useful for determining the system's behaviour and projecting future states even in a noisy environment. Using the DMD algorithm the RES such as solar and wind energy data is predicted in different test scenarios. To test the effectiveness of prediction algorithm the evaluation metrics are calculated and from the prediction results and the metrics data available it can be claimed that the DMD algorithm performs better in different case scenarios.