.png){kind=icon}
.png){kind=icon}
Return
High performance Pd/g-C3N4/PDMS triboelectric nanogenerator for energy harvesting and machine learning-based human activity recognition
.png){kind=icon}
PRE
AI
S
A
P
K
G
S
B
S
DOI：10.1088/1361-6463/ae6120.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The growing use of wearable internet of devices and smart sensors demands sustainable, self-powered systems for real-time human motion monitoring. Current sensing technologies rely on batteries, which have short lifespans, require frequent maintenance and pose environmental risks. In this work, we present a battery-free motion sensing system using a triboelectric nanogenerator (TENG) embedded in footwear, with a Pd/g-C3N4/PDMS composite as the active layer. Two different sized Pd nanoparticles were synthesized and simultaneously decorated on g-C3N4 nanosheets by the synchrotron x-ray irradiation. The characterization results indicate the formation of strong Pd-N and Pd-O bonds at the metal-semiconductor interface, which enhances charge transfer and storage by introducing additional trapping sites. High-performance and stable-output TENGs were fabricated using two types of Pd nanoparticle-decorated g-C3N4 nanosheets incorporated into PDMS insulating matrix. Compared to pristine PDMS and g-C3N4/PDMS-based TENGs, the Pd/g-C3N4/PDMS composite showed significantly better performance due to its higher contact surface area, surface charge density and charge trapping centres. The optimized TENG achieved a peak power density of 9.6 W m(-2) and demonstrated practical utility by charging a 33 & micro;F capacitor in 310 s and powering 102 LEDs through finger tapping. Additionally, we developed a machine learning (ML) based human activity recognition system utilizing the electrical signals generated by the TENG. The TENG data was collected through an Arduino microcontroller and processed using a random forest classifier to distinguish between four human activity states, no movement, walking, running and jumping, enabling real-time monitoring. Experimental tests confirmed the 92% accuracy in classifying each activity. This work combines TENG technology with ML to create an efficient, self-powered human motion detection system, offering potential applications in health monitoring, sports analytics and smart wearable systems.
Keywords:
triboelectric nanogenerator
synchrotron x-ray irradiation
Pd/g-C3N4 nanocomposite
charge trapping
human activity recognition
random forest classifier
self-powered sensor
Journal
IF:
3.2
Papers:
2.6W
Citations:
4.9W
Organization
Cited Papers