Ultra-short-term Interval Prediction of Wind Farm Cluster Power Based on LASSO

Abstract

Efficient and accurate power prediction of wind farm cluster is an effective method to improve the safety and reliability of power system for large-scale wind power. In this paper, the probabilistic prediction model of regional wind power is studied. The nonparametric method based on least absolute shrinkage and selection operator (LASSO) is used for the ultra-short-term probabilistic prediction. In this paper, the prediction model of nonlinear quantile regression (NQR) model based on quantile regression (QR) and extreme learning machine (ELM) is studied. Then, LASSO is utilized to shrink the output weights for the sparsity. The penalty of LASSO can prevent the overfitting and improve the performance of prediction intervals (PIs), without the reduction of computational efficiency. With the actual dataset of the wind farms in northeast China, the PIs performance is verified, compared with other well-established benchmarks.