Comparison of Feedforward Perceptron Network with LSTM for Solar Cell Radiation Prediction

Intermittency of electrical power in developing countries, as well some European countries such Turkey, can be eluded by taking advantage solar energy. Correct prediction radiation constitutes a very important step to take PV panels. We propose an experimental study predict the amount using classical artificial neural network (ANN) and deep learning methods. panel data were collected at Duzce University Turkey. Moreover, we included meteorological from Meteorological Ministry Turkey Duzce. Data on daily basis with 5-min interval. cleaned preprocessed train long-short-term memory (LSTM) ANN models one day ahead. Models evaluated coefficient determination (R2), mean square error (MSE), root squared (RMSE), absolute (MAE), biased (MBE). LSTM outperformed R2, MSE, RMSE, MAE, MBE 0.93, 0.008, 0.089, 0.17, 0.09, respectively. compared our results two similar studies literature. The proposed paves way for utilizing renewable energy leveraging usage