Day-ahead prediction of solar power output for grid-connected solar photovoltaic installations using Artificial Neural Networks

Abstract

Solar Photovoltaic (PV) systems are gaining popularity as a form of alternative energy with increased environmental awareness, renewable energy usage and concern for energy security. Lack of area-specific forecasts for the power output of gridconnected photovoltaic system hinders tapping solar power on a large scale. The objective of this paper is to estimate the profile of produced power of a grid-connected 20 kWp solar power plant in a reputed manufacturing industry located in Tiruchirappalli, India [10° 44' 42.3816'' N, 78° 47' 9.4524'' E]. An Artificial Neural Network (ANN)-based model is proposed in this paper. An experimental database of solar power output (from 7 th January 2014 to 10 th February 2014) has been used for training the ANN. Simulations were carried out with the Neural Network Fitting Toolbox of MATLAB software. Day-Ahead Forecasting results indicate that the proposed model performs well with great accuracy and efficiency.Statistical error analysis in terms of Mean Absolute Percentage Error (MAPE) was conducted and the best result was found to be 0.2887%. Reliable area-specific solar power production map can provide better utilization of solar energy resource and help in power system management.

Energy Storage Control for Dispatching Photovoltaic Power

“…Different types of neural networks have been analyzed by changing the number of layers and neurons and their activation functions, and by training the networks using the Levenberg-Marquardt backpropagation algorithm. In contrast with [19], where a twolayer feed-forward network is adopted, this research employs a three-layer feed-forward network because it is identified as the best compromise. The creation of a good input set, as well as a good network architecture, is of great importance in limiting the forecasting errors.…”

Influence of Artificial and Natural Cooling on Performance Parameters of a Solar PV System: A Case Study

“…During the training, weights of neural network are updated to compute the forecast. With the inclusion of different climate dependent parameters and hybrid models, forecast error is continuously reduced [16][17][18][19][20]. Although several researchers have enlightened us regarding the performance of solar PV system but since its performance is highly site dependent, therefore true potential of system can be analyzed at local level only [21,22].…”