What I'm working on
At the Ziyou Song Lab at UMich, I'm applying ML to predict the remaining useful life of lithium-ion batteries using electrochemical impedance spectroscopy (EIS) data.
EIS measures how a battery responds to small AC signals across a range of frequencies - the resulting impedance spectrum is a fingerprint of the battery's internal state.
Why this matters
Second-life batteries (repurposed from EVs into stationary storage) could unlock massive value for grid storage, but only if we can reliably assess their health. Today this is largely manual and expensive.
Our approach
We're combining physics-based equivalent circuit models with neural networks - encoding known electrochemical relationships as inductive biases. Early results look promising.
More updates as the research progresses.