Research Area:  Metaheuristic Computing
As a classification model, a broad learning system is widely used in wind turbine fault diagnosis. However, the setting of hyperparameters for the models directly affects the classification accuracy of the models and it generally relies on practical experience and prior knowledge. In order to effectively solve the problem, the parameters of the broad learning system such as the number of feature nodes, the number of enhancement nodes, and the number of mapped features layer were optimized by the improved pelican optimization algorithm, and a classification model was built based on the broad learning system optimized by the improved pelican optimization algorithm. The classification accuracy of the proposed model was the highest and reached 98.75%. It is further shown that compared with the support vector machine, deep belief networks, and broad learning system models optimized by particle swarm optimization algorithm, the proposed model effectively improves the accuracy of wind turbine fault diagnosing.
Keywords:  
fault classification
wind turbine
broad learning system (BLS)
pelican optimization algorithm (POA)
parameter optimization
Author(s) Name:  Wumaier Tuerxun , Chang Xu , Muhaxi Haderbieke , Lei Guo and Zhiming Cheng
Journal name:   Machines
Conferrence name:  
Publisher name:  MDPI
DOI:  10.3390/machines10050407
Volume Information:  Volume 10 Issue 5
Paper Link:   https://www.mdpi.com/2075-1702/10/5/407