Vibration Sensor-Based Fall Detection in Smart Factory Shop Floor

Safety, become one of the most important aspects for workers especially for those who worked in the factory. There are some areas in factory that required protection of someone's privacy while keep monitoring them in case an emergency situation happened. These issues have led the authors to propose a vibration sensor-based fall detection in a smart factory shop. In this paper, G-Link 200 is used to collect vibration datasets wirelessly with WSDA-200-USB as a gateway to the internet. The dataset collection is gathered with private room scenarios where it also can be applicable for some areas in smart factory which impossible to use anything that can make the discomforts. Comparative deep learning approaches are used to detect the malicious vibration between normal and abnormal activities. The simulation shows that CNN is outperformed other algorithms by achieving 99.67% accuracy while Bi-LSTM is 92.47% and LSTM is 89.38%.