Artificial neural network based simplified one day ahead forecasting of solar photovoltaic power generation

Abstract

The intermittency of solar energy resources possesses a serious challenge in balancing the power generation and load demand. To enhance the consistency of the system, it is crucial to forecast solar photovoltaic power. Among numerous techniques, Artificial Neural Network (ANN) is an efficient tool that may help simplify this problem. In this study, all 63 combinations of six input parameters, i.e., temperature, dew point, wind speed, cloud cover, relative humidity, and pressure, were applied one by one to ANN to forecast 24 hours ahead PV generation. The power forecast results were obtained based on weather forecast data of 21 days sampled from the recorded forecasted data of 180 days. To quantify the error between predicted and measured solar PV generation, Root Mean Squared Error (RMSE) was used, and the results of different input combinations were compared on basis of this statistical matrix. The analysis showed that the generation is best predicted on two combinations: the first is comprising of temperature, dew point, relative humidity, and cloud cover, while the second consists of all six parameters. And some of the combinations consisting of three parameters also resulted in RMSEs in close proximity of the least error value.