Investigation of Peak Detection Strategies for Self-Sensing Dielectric Elastomer Generators

When controlling the charging cycle of a dielectric elastomer generator (DEG), accurate recognition of capacitance extrema is needed to correctly synchronize the charge and discharge events, enabling in turn energy conversion. Such recognition is commonly achieved via peak detection strategies based on dedicated mechanical sensors, which increase system size and complexity. The intrinsic self-sensing capability of dielectric elastomers can be exploited to reconstruct the capacitance from available electrical measurements, allowing to synchronize the DEG charging cycle without dedicated sensors. Since self-sensing requires a high-frequency signal injection, a fundamental trade-off emerges between robustness to measurement noise (demanding high current signals) and low resistive losses (demanding low current signals). In this paper, we investigate how this trade-off affects the energy yield of DEGs controlled via self-sensing. Using high-fidelity simulation models for both DEG and measurement noise, three real-time peak detection strategies are evaluated on the estimated capacitance over a wide range of self-sensing parameters and noise levels. The results show the existence of optimal self-sensing parameters for any given noise level. These findings provide practical tuning guidelines for self-sensing closed-loop control of DEGs.