Performance Improvement of Very Short-term Prediction Intervals for Regional Wind Power Based on Composite Conditional Nonlinear Quantile Regression

Accurate regional wind power prediction plays an important role in the security and reliability of systems. For performance improvement very short-term intervals (PIs), a novel probabilistic method based on composite conditional nonlinear quantile regression (CCNQR) is proposed. First, hierarchical clustering weighted multivariate time series motifs (WMTSM) studied to consider static difference, dynamic meteorological difference series. Then, correlations are used as sample weights for linear programming (CLP) CCNQR. To optimize PIs, evaluation including accuracy PI coverage probability (PICP), average width (AW), offsets points outside PIs (OPOPI) quantify appropriate upper lower bounds. Moreover, adaptive boundary quantiles (ABQs) quantified optimal PIs. Finally, real farm data, superiority proposed verified by adequate comparisons with conventional methods.