Abstract
Powering small biomedical and Internet-of-Things devices with piezoelectric harvesters is challenging because the available power is low, intermittent, and difficult to condition, which can actually overstress storage elements if charging is not adapted. This study evaluates lithium-ion battery longevity under such constraints using NASA PCoE datasets and designs charge policies matched to limited source power. Firstly, we examine temperature – voltage behavior, and identify a moderate operating window, which motivates a slow-charge baseline that reduces electrochemical stress. Secondly, we introduce two control layers: a dynamic, rule-based controller that derates current as terminal voltage approaches 4.2 V or temperature exceeds 24°C within a defined safe window, and intelligent controllers (Random Forest, XGBoost, Gradient Boosting) that predict incremental degradation from routinely measured signals and select the current that minimizes expected damage under the same constraints. Using the number of cycles to SoH = 0.7 as the endpoint, dynamic control extends life from 30 to 43 cycles, while the intelligent controllers reach 45, 47, and 48 cycles. Reduced voltage peaks and ripple, together with lower thermal exposure, support the mechanism. Overall, aligning slow, condition-aware, and predictive charging with piezoelectric availability robustly extends service life and improves energy-capture efficiency, enabling more reliable, lower-maintenance biomedical and Internet-of-Things systems.
Go to article
Authors and Affiliations
Chaymae Amri
1
ORCID:
Khawla Gaouzi
1
ORCID:
Abdelilah Jilbab
1
ORCID:
My Hachem El Yousfi Alaoui
1
ORCID:
- UM5 ENSAM Rabat, Morocco