Wind Speed Interval Prediction Based on the Hybrid Ensemble Model With Biased Convex Cost Function

This study proposes a combination interval prediction based hybrid ensemble (CIPE) model for short-term wind speed prediction. The (CIP) employs the extreme learning machine (ELM) as predictor with biased convex cost function. To relieve heavy burden of hyper-parameter selection function, technique is developed by combining bagging and stacking methods. Multiple CIP models random hyper-parameters are first trained on sub-datasets generated bootstrap resampling. linear regression (LR) utilized meta to aggregate models. By introducing binary variables, LR can be formulated mixed integer programming (MIP) problem. With benefit function technique, high computational efficiency stable performance proposed guaranteed simultaneously. Multi-step ahead 10-min conducted actual farm data. Comprehensive experiments carried out verify superiority model.