Leveraging systems' non-linearity to tackle the scarcity of data in the design of Intelligent Fault Diagnosis Systems
Researchers have developed a novel approach to create Intelligent Fault Diagnosis Systems (IFDS) that can function effectively even with limited labeled data. This method utilizes Deep Transfer Learning (DTL) combined with a unique multi-excitation procedure that exploits the inherent non-linearities of real-world systems. The technique generates visual data suitable for analysis by pre-trained Convolutional Neural Networks (CNNs), addressing the common challenge of data scarcity in IFDS design. Experimental results on a railway pantograph structure have demonstrated the efficacy of this proposed method. AI
IMPACT This research offers a potential solution for developing more robust AI-driven diagnostic systems in industries where data collection is challenging.