1. Multiperiod Wind Speed Forecasting Approach Based on ELM and Association Rules
- Author
-
Hou Ke, Cheng Yi-Fan, Jin Lijun, and Jiang Tao
- Subjects
Mathematical optimization ,Wind power ,Association rule learning ,business.industry ,Computer science ,020209 energy ,02 engineering and technology ,Wind speed ,Physics::Space Physics ,0202 electrical engineering, electronic engineering, information engineering ,Astrophysics::Solar and Stellar Astrophysics ,business ,Physics::Atmospheric and Oceanic Physics ,Membership function ,Extreme learning machine - Abstract
The wind speed forecasting is one of the key technologies in the operation of wind power plants and micro grids. However, the wind speed prediction is extremely complex and difficult due to the wind speed fluctuation and coupled weather conditions. This paper presents a multiperiod wind speed forecast approach based on data mining technique, which is the multi-objective optimal extreme learning machine with the association rules (AR-MOELM). On one hand, in order to improve the performance and efficiency of the extreme learning machine (ELM), the number of the hidden layer nodes and the penalty factor are optimized through multi-objective particle swarm algorithm (MOPSO)and membership function. On the other hand, the association rules(AR)are adopted to extract the preferable combination of the meteorological factors into the input matrix of the ELM for each season. The evaluation of the AR-MOELM is depicted in the Python environment by ultra-short-term, short-term and medium-term prediction experiments based on real wind speed datasets. The research results reveal that the proposed method improves the multiperiod wind speed prediction accuracy and outperforms existing methods.
- Published
- 2018