Intrinsic Seebeck Feedback for Thermoelectric Temperature Regulation Using a Peltier Module.

Thermoelectric modules provide compact and solid-state thermal regulation capabilities; however, accurate temperature control commonly relies on external sensing elements, which increase system complexity and become impractical in spatially constrained or embedded applications. Although self-sensing approaches based on the Seebeck effect have emerged as an attractive alternative, reliable simultaneous sensing and actuation remain challenging because the driving voltage applied during thermoelectric operation masks the comparatively weak sensing signal. Furthermore, nonlinear thermal dynamics arising from Joule heating, heat conduction, and thermal saturation complicate state estimation under transient operating conditions. To address these limitations, this work proposes an observer-assisted thermoelectric framework integrating time-multiplexed self-sensing with physics-informed digital twin estimation for continuous thermal-state reconstruction and cold-side temperature regulation. The proposed method alternates between actuation and sensing phases, allowing Seebeck measurements to be acquired during non-actuated intervals, while a predictive thermal model reconstructs system behavior during periods in which direct observations are unavailable. An embedded implementation based on a microcontroller platform was developed and experimentally evaluated under multiple operating conditions and duty-cycle inputs. Experimental results demonstrated characteristic nonlinear thermoelectric behavior, including thermal saturation effects and variations in Seebeck sensitivity across operating regions. The findings indicate that Seebeck measurements provide meaningful thermal-state information at moderate and high temperature gradients, whereas reduced signal amplitude at lower duty cycles limits direct sensing reliability. The integrated digital twin and observer framework maintained continuous state awareness despite intermittent sensing availability and enabled improved interpretation of thermoelectric dynamics. The proposed approach establishes a practical pathway toward self-observable thermoelectric systems and demonstrates the potential for transforming thermoelectric devices from passive cooling components into intelligent cyber–physical elements capable of supporting predictive monitoring and advanced thermal regulation.