Leveraging systems' non-linearity to tackle the scarcity of data in the design of Intelligent Fault Diagnosis Systems

The article presents a novel approach for designing Intelligent Fault Diagnosis Systems (IFDS) using Deep Transfer Learning (DTL) under conditions of data scarcity. It introduces a periodic multi-excitation level procedure that exploits the intrinsic non-linearities of systems to generate images for analysis by pre-trained Convolutional Neural Networks (CNNs). This method, along with a new data visualization and augmentation technique, addresses the challenge of limited labeled data, demonstrating effective results through experimental validation on a railway pantograph structure.