What if you could predict a circuit's failure before it happens? This project builds a PyTorch neural network that estimates the Remaining Useful Life (RUL) of a capacitor in an RC circuit from its voltage-decay behavior.
The idea: as an RC circuit runs, the capacitor's dielectric breaks down under stress and it slowly loses capacitance, changing how its voltage decays (T = RC). By learning the rate of change of that decay, the model reverse-engineers how close the capacitor is to end-of-life — defined here as a 20% drop in capacitance.
The video walks through the physics background, the theory, the model, and the results:
• Dataset — NASA's Capacitor Electrical Stress Dataset (capacitors stressed at 10V, 12V, 14V; EIS + charge/discharge signals)