Application of Artificial Intelligence Technique in Predicting 7-Days Solar Photovoltaic Electrical Power

Abstract

To be able to dispatch electrical power effectively to consumers using solar photovoltaic (SPV) cells, there is a need to have information about the SPV power generation. This information is best derived from predicting the SPV power ahead of any supply. Artificial neural network intelligence technique is employed in this study with the aim of predicting SPV electrical power for a period of 7 days. The maximum power produced on a daily basis is been identified as well as the daily average power that is produced and predicted. With this information, the short-term availability of daily solar irradiation can be maximized. A statistical regression analysis has been used to establish the relationship between the produced and predicted power, using statistical functions like the mean bias error (MBE), the mean square error (MSE), root mean square error (RMSE), mean absolute deviation (MAD), mean absolute percentage error (MAPE), and the correlation coefficient (CC). The algorithm used in training the network is the backpropagation algorithm with feed-forward neural network. A total of 14,300 datasets have been used to establish this study with the application of artificial neural network (ANN) for prediction analysis. The result indicates that, the uncertainty in SPV power generation can be mitigated using ANN to predict its performance, thereby creating visibility as to what the SPV system can generate. This enables load balancing, efficient power dispatch and accurate scheduling.