Self-Sensing Closed-Loop Control of Dielectric Elastomer Generators

Dielectric elastomer generators (DEGs) are electrostatic devices that convert mechanical work into electrical energy leveraging cyclic stretching and electrical loading of polymeric variable capacitors. Existing implementations of DEG harvesting cycles either rely on a priori knowledge of the deformation profile, limiting real-world applicability, or require external mechanical sensors, increasing system complexity. This work presents a closed-loop control architecture for DEGs that does not require external sensors. By injecting a high-frequency voltage signal and measuring the corresponding electrical current, we infer the DEG capacitance in real time and use it as a feedback for real-time control of the harvesting cycle. While similar self-sensing (SS) architectures have been explored for actuators, where energy efficiency is not a primary concern, their application to generator systems must explicitly account for the electrical losses introduced by the injected sensing signal. Using state-of-the-art SS algorithms, we systematically investigate the tradeoff between conversion efficiency and sensing accuracy, to maximize the energy yield. We then introduce a new SS method specifically tailored to DEG applications, which uses a modified high-frequency injection to further reduce dissipation. Experimental validation, conducted under sinusoidal and polychromatic deformation profiles, shows that SS control strategies allow converting 85%–95% of the energy achievable with ideal open-loop cycles.