Open AccessTransfer Learning with Pretrained Neural Network Between Unrelated Tasks for Machine Health Diagnosis
Author(s) -
Youssef Maher*,
Boujemaa Danouj
Publication year - 2020
Publication title -
international journal of recent technology and engineering
Language(s) - English
Resource type - Journals
ISSN - 2277-3878
DOI - 10.35940/ijrte.f8489.038620
Subject(s) - computer science , transfer of learning , retraining , artificial intelligence , robustness (evolution) , machine learning , artificial neural network , task (project management) , field (mathematics) , predictive maintenance , deep learning , recurrent neural network , reliability engineering , engineering , biochemistry , chemistry , mathematics , systems engineering , international trade , pure mathematics , business , gene
Deep Learning (DL) has contributed a lot in the field of industrial maintenance, in particular predictive maintenance by detecting potential failures and breakdowns before their appearance. Unfortunately, the DL has some limitations like the need for a large amount of data to produce an effective prediction model and also the fragility of the model in the face of changes in operating conditions. Another approach, the Transfer Learning (TL), had demonstrated in the literature that he can overcome these weaknesses. In this article, we will be using this technique with the pretrained neural network, AlexNet, which had been previously trained with the ImageNet database. Our method doesn’t require a high amount of input data and thus saves a lot of time in retraining the network in another task, which can be related or unrelated to the source task. In fact, the prediction model was successfully adapted to the bearings diagnosis case. It showed also high degree of robustness against changes of functioning conditions.