Prevailing Wind Direction Forecasting for Natural Ventilation djustment in Greenhouses Based on LE-SVM

It is necessary to predict wind speed and direction accurately for realizing optimal adjustment of natural ventilation in greenhouse climate control. Due to strong randomness of wind direction least squares support vector machine (LS-SVM) is applied to built a prediction model for a defined prevailing direction of wind for natural ventilation of greenhouse. Experiment shows that in 10 min. ahead prediction is satisfying with MAPE is 6.77% and the number of data points with APE over 15% is only 5% of total number of predicted samples. For prediction in 20 min. ahead the MAPE is 11.83% which is acceptable but the number of the predicted with APE over 15% is 19 .6% of total number of test sample. The proposed model is also compared with the model based on artificial neural network, and the comparison results show that the proposed LE-SVM model is better than ANN model in forecasting of prevailing wind direction.