Integrated Multimodal Fault Diagnosis of Industrial Gearboxes Using Vibration Signals and Infrared Thermal Imaging

With the increasing demand for precision in fault diagnosis of gearboxes in complex systems, diagnostic methods based solely on vibration signal characteristics have shown limitations. Therefore, multimodal information fusion technology has become a new trend in the field of gearbox fault diagnosis. This article innovatively proposes a multimodal diagnostic method that combines vibration signals with thermal images, aiming to use thermal imaging technology to intuitively capture the thermal changes during gearbox operation and integrate vibration signals to comprehensively reflect fault characteristics. By introducing Convolutional Neural Networks (CNN) for feature depth extraction, this method effectively improves the parsing ability of fault information, thereby enhancing the accuracy and reliability of fault diagnosis. Experimental data confirms that compared to diagnostic methods using a single signal source, the method proposed in this paper performs well in fault detection of helical gearboxes, significantly improving diagnostic accuracy and precision.