Session: Research Posters

Paper Number: 172989

Non-Destructive Characterization of Lithium-Ion Battery Health via Vibration Response Analysis and Principal Component Projection 

Lithium-ion batteries are integral components of modern electrical and electronic systems, particularly in electric vehicles, renewable energy storage systems, and aerospace platforms. High energy density, long cycle life, and compact size are the most important parameters that render them applicable to a wide range of applications. However, like any electrochemical system, lithium-ion batteries also degrade over time due to aging, cyclic charge-discharge, mechanical stress, or exposure to high temperatures. Unfortunately, these internal degradation processes do not manifest until they result in visible performance loss or, in the worst cases, catastrophic failure. Early detection of these issues is therefore crucial for maintaining safety and efficiency, as well as avoiding costly downtime or replacement.

This study presents a novel, non-invasive method for assessing the health state of lithium-ion batteries by analyzing their vibrational response under dynamic excitation. Instead of traditional electrical measurements, the proposed approach captures mechanical signatures associated with the internal condition of the battery. In this study, four distinct conditions of batteBatteryare examined, including fully charged and healthy, fully discharged, physically damaged, and aged. Moreover, surface vibrations were measured using a high-precision, non-contact Laser Doppler Vibrometer (LDV) on a shaker table subjected to controlled sinusoidal excitation. This technique enabled the capture of detailed dynamic response data without physically altering or affecting the internal electrochemical properties of the batteries.

The vibration signals recorded in the time domain were analyzed and predicted using Python-based signal processing functionality. A complete set of spectral, temporal, and statistical features was extracted from the signals, including properties such as dominant frequency, RMS amplitude, and signal energy. Principal Component Analysis (PCA) is used in this study to reduce the high-dimensional features and visualize relationships between battery conditions. The PCA projection revealed distinct clusters for each battery state, which indicates the method's sensitivity to internal structure alterations and the viability of vibration-based classification.

The proposed framework has several significant advantages: it is fast, completely non-contact, and compatible with various types and formats of batteries. With its prospects for detecting subtle mechanical changes related to electrochemical health, this approach has strong potential for integration into advanced battery management systems (BMS). This integration would offer the potential for real-time state-of-health monitoring, early fault detection, and predictive maintenance in high-consequence applications, such as electric mobility, aerospace systems, and grid-scale storage infrastructure. In conclusion, this research demonstrates the promise of vibration measurement and data-driven analytics as a promising method for battery diagnostics.

Presenting Author: Arefeh Salimi Beni Georgia Southern University

Presenting Author Biography: Arefeh Salimi Beni is a master's student in engineering at Georgia Southern University. Her research focuses on experimental analysis of lithium-ion batteries, with an emphasis on non-destructive testing (NDT) and non-destructive evaluation (NDE) techniques. She has experience applying machine learning methods for the prediction and assessment of battery health. Her current work involves developing data-driven approaches to evaluate the condition of lithium-ion batteries under dynamic excitation, aiming to improve early fault detection and safety monitoring.

Authors:

Arefeh Salimi Beni Georgia Southern University
Hossein Taheri Georgia Southern University