Applying Multi-layer Perceptron Neural Network to Predict Wind Speed in Lebanon

The objective of the present study is to identify the suitable variables for accurate wind speed prediction using Multi-Layer Perceptron Neural Network (MLPNN) model. To achieve this, monthly data consisting of average temperature, relative humidity, surface pressure, solar radiation, and wind speed were collected from meteorological services for different locations. In this study, 30 models are developed to find the relevant parameters. The results demonstrated that Model#30 and Model#25 with the combination of [Lat, Long, Alt, M, P, T, RH, GSR] and [Lat, Long, Alt, T, RH, GSR] give the best performance compared to the other models. Consequently, the geographical coordinates, temperature, relative humidity, and solar radiation are considered the most relevant that affect wind speed reduction.