Towards soft real-time fault diagnosis for edge devices in industrial IoT using deep domain adaptation training strategy

•A soft real-time bearing fault detection for motors is proposed in this research.•Deep domain adaptation training methods are used for edge devices.•Deep learning models are employed for effective condition monitoring of motors.•Condition of the motor is monitored through an internet of things (IoT) plateform.•Benchmarked the tri-axial vibration dataset through applying deep learning methods on publically available datasets.