Title :
Sensitivity of artificial neural network based model for photovoltaic system actual performance
Author :
Ameen, Ammar Mohammed ; Pasupuleti, Jagadeesh ; Khatib, Tamer
Author_Institution :
Babylon Electr. Distrib. Directorate, Minist. of Electr., Iraq
Abstract :
A novel prediction model for the output current of PV module is proposed in this paper. The proposed model is based on cascade-forward back propagation artificial neural network with two inputs and one output. Solar radiation and ambient temperature are the inputs and the predicted current is the output. Experiment data for a 1.4 kWp PV systems installed in Sohar city, Oman are utilized in developing the proposed model. These data has an interval of 2 seconds in order to consider the uncertainty of the system´s output current. In order to evaluate the accuracy of the neural network, three statistical values are used namely mean absolute percentage error (MAPE), mean bias error (MBE) and root mean square error (RMSE). Moreover, the ability of the proposed model to predict performance with high uncertainty rate is validated. The results show that the MAPE, MBE and RMSE of the proposed model are 7.08%, -4.98% and 7.8%, respectively.
Keywords :
backpropagation; cascade networks; mean square error methods; neural nets; power engineering computing; prediction theory; solar cells; solar radiation; MAPE; MBE; Oman; PV module; RMSE; Sohar city; ambient temperature; cascade-forward back propagation artificial neural network sensitivity; mean absolute percentage error; mean bias error; photovoltaic system actual performance; prediction model; root mean square error; solar radiation; Artificial neural networks; Mathematical model; Neurons; Power generation; Predictive models; Solar radiation; Uncertainty; CFNN; Modeling of PV system; PV system; field performance;
Conference_Titel :
Power and Energy (PECon), 2014 IEEE International Conference on
Print_ISBN :
978-1-4799-7296-8
DOI :
10.1109/PECON.2014.7062449
