Real-time monitoring of mechanical damage in lithium-ion batteries using ultrasonic guided waves and cumulative energy feature analysis

Indentation damage in lithium-ion pouch cells poses safety risks and induces hidden internal degradation. Conventional electrical monitoring (voltage, temperature, and current) typically detects anomalies only after severe structural damage occurs, providing limited warning time. In this study, an in-situ monitoring experiment is conducted to track progressive mechanical damage from an intact state to failure in lithium-ion cells subjected to quasi-static indentation. Ultrasonic guided-wave (UGW) signals, surface temperature, and terminal voltage were continuously recorded until voltage drops and temperature spikes were observed. To process the acoustic responses during loading, a monitoring method is proposed, coupling waveform decomposition, baseline subtraction, and a cumulative energy-feature (CEF) scheme with threshold-controlled integration. The cumulative shift of signal centroid (CoC) is proposed as the primary indicator to convert wavefield fluctuations into a monotonic trend, addressing the non-monotonicity of standard energy metrics and the early saturation of shape coefficients. The experimental results indicate that the CoC exhibits a monotonic relationship with damage progression and provides an early warning as early as 15 min prior to the onset of electrical/thermal abnormalities in the tested cells, outperforming voltage and temperature monitoring. Moreover, combining complementary features further improves the prediction performance, providing a quantitative approach for the early detection of mechanically induced battery faults.