AI-Enabled Sensing Wristband for Student Behavior Detection

With the advancement of the Internet of Things (IoT) and artificial intelligence (AI) technologies, wearable sensors will play a significant role in smart healthcare and behavior detection. Based on multiscale convolutional channel attention residual network (MCRnet), this research proposes an AI-enabled sensing wristband that integrates three core modules: the electromagnetic generator (EMG) module, the triboelectric nanogenerator (TENG) module, and the user behavior detection (UBD) module. The electromagnetic generator module harnesses wrist motion to drive a magnetic ball rolling within a nylon hollow tube, inducing variations in the magnetic flux of the coil and generating current. Simultaneously, as a PTFE ball rolls inside the nylon tube, copper electrodes arranged on the tube's exterior undergo charge transfer, producing triboelectric signals. The electromagnetic generator module was simulated using the finite element method magnetics (FEMM) software, while vibration tests and human motion test experiments were also conducted simultaneously. The experimental results indicate that the electromagnetic generator module produces an output power of 2.42 mW when stimulated by wrist swinging, providing sufficient electrical energy for the system. Then, we formulated eight types of student behaviors and collected the activity signals of 20 individuals as a data set. By optimizing the network parameters, the success rate of student behavior detection in the MCRnet network ultimately reached 98.75%, demonstrating the system's excellent sensing performance. Finally, we demonstrated a smart classroom application by integrating digital twin and 5G communication technologies, highlighting the considerable potential of the sensing wristband in shaping future intelligent living environments.