Solar Radiation Forecasting using Customized Artificial Neural Network

The current electricity situation demands the adoption of renewable resources such as solar energy in order to contribute to the advancement of sustainable development, as excessive use of fossil fuels and their catastrophic impacts on the environment. Unlike fossil fuels, the sun has more energy than is required to power the entire planet, but due to the time-varying nature of solar energy, converting it into electricity economically and effectively is a major challenge. For this, forecasting solar energy a few days in advance for the estimation of the likely energy generation for those few days can be proposed as an effective solution. With the same objective, the present research offers a realistic solution to this problem known as an artificial neural network composed of 6 input layers, 15 hidden layers, and one response layer for solar power generation forecasting. The proposed ANN model has been simulated using solar power irradiance data from Panchkula region of Haryana to validate its performance. The experiment results show that the proposed ANN model is an effective approach to solar power generation forecasting and can also be useful with integration to photovoltaic systems.