A method and system for predicting hub height wind speed for an offshore wind farm

Abstract

The present application belongs to the technical field of offshore wind farm wind speed prediction, and specifically discloses a kind of offshore wind farm hub height wind speed prediction method and system.The method of the present application constructs WRF- ROMS- SWAN sea- air wave full coupling simulation framework, and extracts wind speed, wind direction, temperature, humidity and air pressure data at hub height from the coupling mode output, and constructs wind direction period coding feature, time period feature, historical lag feature and rolling statistical feature.Then further build data-driven error correction model to correct wind speed error.Compared with the traditional single machine learning model, the method of the present application is heterogeneous integrated by XGBoost and BiLSTM-Attention, which takes into account nonlinear feature analysis and time series dynamic dependence modeling.In addition, compared with the full-sample unified error correction model, the wind speed partition strategy proposed in the present application can adapt to the error distribution difference of different wind speed intervals, and improve the deviation control ability, error compression ability and trend fitting ability in low, medium and high wind speed intervals.

Images