A signal-processing framework for preemptive detection of soft shorts in lithium-metal batteries

<h2>Summary</h2> Lithium-ion batteries are central to sustainable energy systems, yet internal short circuits (ISCs) can trigger catastrophic failures, such as fires and explosions. Current detection methods are based on costly tests, large datasets, or computationally heavy models, which limit their scalability and practical application. Here, we report a physics-based framework for the early detection of soft ISCs by representing the battery as an equivalent resistance-capacitance (RC) circuit with time-varying resistances. Using signal processing, we track periodic resistance fluctuations in charging data to identify the onset of ISCs without any abuse conditions. Fast Fourier transform (FFT) highlights distinct frequency-domain signatures, while Welch's algorithm extracts characteristic power spectral density patterns that serve as detection criteria. Together, these methods provide robust, real-time monitoring of ISC formation. Our results establish a new diagnostic standard that links microscopic ISC dynamics with macroscopic signals, significantly improving battery safety for electric vehicles, consumer electronics, and grid-scale energy storage.