Solar Power Monitoring and Forecasting System using Artificial Intelligence

Solar photovoltaic (PV) systems have been improving over the years resulting in an increased installation at the residential sites. This paper proposes an effective solar power monitoring and forecasting system to help residential users to monitor their solar PV systems more efficiently. Most of the current solar monitoring systems are either too expensive, or not smart enough to predict the future profile of the power generation. A ready app called ‘Geo-Solar Power Monitoring and Forecasting App’ for local monitoring and a ThingSpeak graphical user interface for remote monitoring has been developed. Compared to the existing apps, this app is enhanced to self-train when a new data set is received. Feed Forward Back Propagation Neural Network (FF-BP-NN) has higher accuracy of prediction than Nonlinear-Autoregressive neural network with external exogenous input (NARX).