As wind speed being a vital indicator of wind energy, realizing accurate wind speed prediction plays an essential role in wind resource assessment, wind load analysis, wind damage precaution, etc. The multiple internal and external forces interfering with wind farm determine the highly non-linear and randomness of wind speed of a site, thus many researchers choose Artificial Neural Network (ANN) to analyze and forecast wind speed for its ability of self-learning and nonlinear fitting. However, lots of previous models only utilize one single model to directly predict wind speed, neglecting the importance of feature extraction. In this paper, a hybrid wind speed prediction model based on empirical mode decomposition (EEMD), Long short-term memory (LSTM) and Bayesian Optimization (BO) is proposed. The original wind speed data are first decomposed into several components by the method of EEMD. Then, LSTM networks are built for preliminary predictions of each components. Final predictions are reconstructed after optimize each LSTM models using BO. To evaluate the performance of the proposed model, several experiments are conducted on a wind dataset collected in a site within northeastern United States by a research team in Cornell University (quote). The results of the experiments illustrate that the proposed model outperforms traditional BO-LSTM prediction models, showing an improvement in prediction accuracy from multiple criteria.


A Hybrid Wind Speed Prediction Approach Based on Ensemble Empirical Mode Decomposition and BO-LSTM Neural Networks for Digital Twin

Division / Sponsors: 
Power Division & Nuclear Engineering Division (NED)

Power Conference
Paper Proceedings: https://asmedigitalcollection.asme.org/POWER/POWER2020/volume/83747

International Conference on Nuclear Engineering
Paper Proceedings: https://asmedigitalcollection.asme.org/ICONE/ICONE2020/volume/83761

POWER-ICONE - Virtual Conference

Supporting the need among researchers worldwide for additional online and multimedia content, ASME now offers the ASME Conference Video Collection. This platform will collect and provide long-term access to conference presentation videos, encompassing all technology areas within ASME's technical conference and events program. About 30,000-40,000 engineers across all mechanical engineering and related disciplines participate in ASME conferences each year. This includes authors of approximately 6,000 to 8,000 technical papers published by ASME, the majority of whom also submit video presentations for our virtual conferences. The video collection will be available as an add-on to existing product packages or as a separate subscription.