Physics-of-Failure-Based Degradation Prediction of Aluminium Electrolytic Capacitor Under Thermal Overstress Conditions

Aluminium electrolytic capacitors are indispensable in modern electronics, yet their long-term reliability is often compromised by electrolyte evaporation under thermal overstress. In this work, we propose a physics-of-failure framework that explicitly models electrolytic capacitor degradation by capturing the primary failure mechanism of electrolyte loss. Our approach integrates a non-linear, physics-based degradation model with an adaptive parameter estimation strategy that leverages both device-specific operational data and insights from comparable neighbouring units. To enhance predictive accuracy, the framework employs robust least-squares estimation, dynamic time warping for optimal neighbour selection, and Monte Carlo bootstrapping for rigorous uncertainty quantification and probabilistic calibration. Accelerated ageing experiments conducted under controlled isothermal conditions validate the model’s ability to reliably estimate the remaining useful life of AECs, thereby facilitating proactive maintenance, mitigating failures and bolstering overall system reliability. The seamless integration of physics-based modelling with advanced statistical techniques offers a transparent, data-efficient, and scalable solution for long-term prognostics of electronic components.