Adaptive Estimation of Near-Optimal Electrostatic Force in Micro Energy-Harvesters

Recent advancements in micro-electronics have led to the development of miniature-sized wearable sensors that can be used for a variety of health monitoring applications. These sensors are typically powered by small batteries which could require frequent recharge. Energy harvesting can reduce the charging frequency of these sensors. Longer operational lifetime can simplify the everyday use of these wearable sensors in many applications. Our objective in this paper is to maximize the output power of a kinetic-based micro energy-harvester. A hybrid machine learning and analytical approach is proposed to adaptively adjust the electrostatic force in a harvester with Coulomb-Force Parametric Generator (CFPG) architecture. The results show considerable improvement in the output power.