A Fault Diagnosis Technique with the Combined DNN and CNN Using Motor Current Data

With the global demand for energy efficiency and safety increasing, the need to monitor the condition of electric motors and diagnose faults is being emphasized. Motor failures have a significant impact on operational downtime, economy, and social trust, thereby making effective diagnostic methods crucial. Non-contact methods have been primarily used for fault diagnosis, while Motor Current Signal Analysis (MCSA) is a widely used fault detection method today because it can easily detect common mechanical defects such as rotor shorts, bar cracks/damages, and bearing degradation. In this study, we propose a new type of fault diagnosis method using an architecture that combines DNN and CNN. This technique deeply analyzes complex patterns in current data, extracts sophisticated features, and accurately determines faults by considering nonlinear and temporal characteristics. The experimental results show that our proposed method achieved the improved performance for motor faults compared to existing methods.