A deep boosted transfer learning method for wind turbine gearbox fault detection

Deep learning methods have become popular among researchers in the field of fault detection. However, their performance depends on availability big datasets. To overcome this problem started applying transfer to achieve good from small available datasets, by leveraging multiple prediction models over similar machines and working conditions. influence negative limits application. Negative increases when environment conditions are changing continuously. effect transfer, we propose a novel deep method, coined boosted learning, for wind turbine gearbox detection that prevents only focuses relevant information source machine. The proposed method is an instance-based updates weights target machine training samples separately. different gradually decreased reduce impact final model. verified Case Western Reserve University bearing real farm results show ignores achieves higher accuracy compared standard methods.