Fault diagnosis of Wind Turbine Pitch Systems based on Kohonen network
Xinli Li, Wanye Yao, Qingjie Zhou, Jianming Wang · Advances in computer science research · 2015
The wind turbines has a lot of operational failure parameters and some isolated sample;what's more , so direct use of neural networks for fault diagnosis easily lead to performance decreased .For this situation, we propose use of similarity function combined with Kohonen neural network for fault diagnosis:first use similarity function method to eliminated the redundant information for samples optimized; then due to the vagueness and uncertainty that exists between fault symptoms and causes of failure,it needs fuzzy clustering based on Kohonen neural network to solve,so the optimized samples input Kohonen network to obtain various type of standard fault model,then put the test samples in the model ,its results were compared with the standard fault sample can get the type of fault .Simulation results show that: in the wind turbine pitch system use the fault diagnostic method, establish the relationship model accuracy is relatively high, able to make quick and accurate diagnosis of the turbine pitch systems operational status and fault type.