Key points are not available for this paper at this time.
Portable wearable sensors are pivotal in advancing long-term human rehabilitation exercise training, yet challenges remain in achieving multi-functionality, intelligence, and sustainable energy supply. To address these limitations, we introduce a novel machine learning-enhanced triboelectric-electromagnetic-piezoelectric wearable sensor (TEP-WS) for comprehensive and prolonged monitoring of human motion. This device integrates three energy harvesting mechanisms-a triboelectric nanogenerator (TENG), a piezoelectric generator (PEG), and an electromagnetic generator (EMG)-into a compact, space-efficient design. The TENG generates triboelectric charge via copper ball reciprocation, while the PEG uses magnetic interactions to drive piezoelectric materials. The EMG achieves contactless power generation through relative motion between magnets and coils. These components share a rotating shaft, enhancing energy harvesting efficiency. Multi-physical sensing signals from these components are processed through machine learning algorithms, improving motion recognition accuracy. Experimental results demonstrate exceptional performance: at 4 Hz, the TENG, PEG, and EMG modules achieve peak power outputs of 13.28 nW, 0.63 µW, and 1.31 µW, respectively. A 10 µF capacitor can be charged within 14 s, showcasing the system's ability to provide continuous power for wearable devices. For gesture recognition and gait impairment detection, the TEP-WS achieves accuracy rates exceeding 93.3% and 95.0%, validated by the Random Forest (RF) algorithm. These findings highlight the TEP-WS's potential as a valuable tool in health monitoring and rehabilitation applications.
Zhou et al. (Sat,) studied this question.