E-jet printed sensors and deep learning for advanced EMS cart condition monitoring

This study presents an intelligent condition monitoring framework for electric monorail system cart systems, integrating conformal printed sensing, machine vision, and deep learning. Overcoming sensor integration challenges, thermal-field assisted electrohydrodynamic jet printing is employed to fabricate high-resolution zinc oxide sensors on complex surfaces, guided by a multiphysics model elucidating laser–jet interactions. Key process parameters (Zn²⁺ concentration, temperature, and voltage) are optimized for jet stability and microstructure. For operational diagnostics, a hyperparameter-optimized neural network is developed for track strain signal analysis, and a transfer learning-enhanced convolutional neural network is implemented for visual detection of track cable cracks. Experimental validation confirms the framework’s efficacy in achieving precise state identification, significantly boosting fault detection accuracy, reducing labor costs, and enabling intelligent operation and maintenance.