Relevance vector machine based fault classification in wind energy conversion system

Rekha S. N. (Sapthagiri College of Engineering)
P. Aruna Jeyanthy (Kalasalingam Academy of Research and Education (KARE))
D. Devaraj (Kalasalingam Academy of Research and Education (KARE))

International Journal of Electrical and Computer Engineering
Vol 9, No 3: June 2019

This Paper is an attempt to develop the multiclass classification in the Benchmark fault model applied on wind energy conversion system using the relevance vector machine (RVM). The RVM could apply a structural risk minimization by introducing a proper kernel for training and testing. The Gaussian Kernel is used for this purpose. The classification of sensor, process and actuators faults are observed and classified in the implementation. Training different fault condition and testing is carried out using the RVM implementation carried out using Matlab on the Wind Energy Conversion System (WECS). The training time becomes important while the training is carried out in a bigger WECS, and the hardware feasibility is prime while the testing is carried out on an online fault detection scenario. Matlab based implementation is carried out on the benchmark model for the fault detection in the WECS. The results are compared with the previously implemented fault detection technique and found to be performing better in terms of training time and hardware feasibility.