Comparison and Assessment of PV Module Power Prediction Based on ANN for Iraq Weather

Abstract

An artificial neural network (ANN) with backward-propagation technique was used to predict the power generation of PV module in sunny and cloudy weathers of Baghdad city-Iraq. Experiment tests were investigated in winter and summer days to get the best sunny and cloudy days. Three weather parameters were measured including: solar irradiance, ambient temperature and wind speed. In addition, the output electrical characteristics of PV module (voltage, current, power) and module temperature were measured. Therefore, the dataset of ANN system consists of four input and one output. Furthermore, the structure of ANN includes single and double hidden layers with backward propagation technique. Besides, number of neurons were optimized in training process. The evaluation of the ANN model was depended on determination coefficient (R) and Mean Squared Error (MSE). The obtained results show that the architecture of ANNs is appropriated for predicting the power generated from PV module. The two developed ANN models have good accuracy and the sunny model is relatively more accurate than the cloudy model. Where, the MSE is 0.002062 at epoch 6 in sunny model and 0.0087085 at epoch 9 in cloudy model. Furthermore, the R is recorded 0.993 and 0.982 in validation process for sunny and cloudy model respectively. In addition, the optimization number of neurons in hidden layer gave sufficient accuracy without referring to choose the neurons by trial and error.