PROJECT TITLE :

An Intelligent Fault Diagnosis Method Using Unsupervised Feature Learning Towards Mechanical Big Data

ABSTRACT:

Intelligent fault diagnosis may be a promising tool to accommodate mechanical huge knowledge due to its ability in rapidly and efficiently processing collected signals and providing accurate diagnosis results. In ancient intelligent diagnosis methods, but, the options are manually extracted depending on prior knowledge and diagnostic experience. Such processes use human ingenuity however are time-consuming and labor-intensive. Galvanized by the thought of unsupervised feature learning that uses artificial intelligence techniques to learn options from raw information, a two-stage learning methodology is proposed for intelligent diagnosis of machines. In the first learning stage of the strategy, sparse filtering, an unsupervised 2-layer neural network, is used to directly learn features from mechanical vibration signals. Within the second stage, softmax regression is employed to classify the health conditions primarily based on the learned options. The proposed methodology is validated by a motor bearing dataset and a locomotive bearing dataset, respectively. The results show that the proposed methodology obtains fairly high diagnosis accuracies and is superior to the existing methods for the motor bearing dataset. Because of learning options adaptively, the proposed technique reduces the need of human labor and makes intelligent fault diagnosis handle massive information additional easily.