Abstract
Motors, which are one of the most widely used machines in the manufacturing field, take charge of a key role in precision machining. Therefore, it is important to accurately estimate the health state of the motor that affects the quality of the product. The research outlined in this paper aims to improve motor fault severity estimation by suggesting a novel deep learning method, specifically, feature inherited hierarchical convolutional neural network (FI-HCNN). FI-HCNN consists of a fault diagnosis part and a severity estimation part, arranged hierarchically. The main novelty of the proposed FI-HCNN is the special inherited structure between the hierarchy; the severity estimation part utilizes the latent features to exploit the fault-related representations in the fault diagnosis task. FI-HCNN can improve the accuracy of the fault severity estimation because the level-specific abstraction is supported by the latent features. Also, FI-HCNN has ease in practical application because it is developed based on stator current signals which are usually acquired for a control purpose. Experimental studies of mechanical motor faults, including eccentricity, broken rotor bars, and unbalanced conditions, are used to corroborate the high performance of FI-HCNN, as compared to both conventional methods and other hierarchical deep learning methods.
Highlights
Motors are widely used in manufacturing applications that require a rotating force due to their low cost and high reliability
While existing studies on DLbased motor fault diagnosis (FD) use vibration signals, we propose a new deep learning (DL) method for FD and Severity Estimation (SE) of induction motors based on the stator current signal
The structure of feature inherited hierarchical convolutional neural network (FI-hierarchical CNN (HCNN)) was hierarchically composed to lead to an FD module that can learn the types of faults and an SE module that is able to estimate their severity
Summary
Motors are widely used in manufacturing applications that require a rotating force due to their low cost and high reliability. The degradation of motors can lead to deterioration in product quality, it is crucial to diagnose the motor state and evaluate the fault severity [1] To cope with these problems, motor current signature analysis (MCSA) has been studied for fault diagnosis (FD). Mathematical models have been formulated to investigate inter-turn shorts in stator windings [6]; further, the stator current spectrum was analyzed for SE of unbalance, eccentricity, and bearing faults [7]. These methods can be applied to generic motor systems; real-world applications are limited because specific motor expertise—which is not known—is necessary. Studies on data-driven MCSA, on the other hand, make an effort to extract fault-sensitive features and apply the proper
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
