78 / 2021-07-20 20:29:39

A novel hybrid models bsaed on Gaussian Process Regression for short-term wind power forecasting

Special Session > SS7 - Data-driven Engineering Practices in The Fields of Energy, Environment, and Biomedicine

Accurate and stable short-term wind power prediction is highly valuable for the scheduling plan and operational security of the wind power system. However, it is relatively difficult to obtain satisfactory forecasting results in the wind power system due to the complexity and nonlinearity of the wind speed series. This paper is proposing highly accurate hybrid model for wind power forecasting using different artificial intelligent systems for optimal performance. A least squares support vector machines (LSSVM) optimized by the improved whale optimization algorithm (IWOA) and a long short-term memory (LSTM) optimized by Bayesian optimization (BO) algorithm are used as the two basis models to predict wind power time series, respectively. A Gaussian process regression (GPR) model is utilized to combine independent forecasts generated by the two forecasting models. Comparing the proposed prediction model with two basis models based on the public wind farm datasets of the National Renewable Energy Laboratory (NREL), the proposed model significantly improves the prediction accuracy.