Deep Focus Parallel Convolutional Neural Network for Imbalanced Classification of Machinery Fault Diagnostics

Abstract

Artificial intelligence-based machinery fault diagnosis techniques have been increasingly considered in many industrial fields. The convolutional neural network (CNN) is able to learn features from raw signals because of its filter structure. Thus, several studies have applied CNN-based methods for machinery fault recognition and classification. However, most of these studies are based on a balanced data set, while ignoring that normal data and fault data tend to be highly imbalanced in real-world applications. Conventional CNNs do not work well for highly imbalanced fault diagnostics tasks and often lead to the degradation of performance. Therefore, in this article, a learning framework called deep focus parallel convolutional neural network (DFPCN) is proposed to overcome the weakness. It has powerful feature learning capabilities due to its parallel convolutional architecture. A new loss function named adaptive cross entropy loss (ACE loss) is designed for the DFPCN to focus training on minority health condition samples which are hard to classify. The effectiveness and superiority of the proposed DFPCN are validated by a highly imbalanced data set constructed from bearing vibration signals. The diagnostics results demonstrate that DFPCN outperforms the state-of-the-art CNN-based methods in terms of accuracy and stability, and avoids adding computational burden with the redundant samples when compared with oversampling methods.