A Practical Analysis of Data Driven Lithium - Ion Battery Health Estimation

Lithium-Ion (Li-Ion) batteries are used as the main source of power for various applications, such as electric vehicles and portable devices. Monitoring the health of these batteries is crucial for ensuring safety, longevity, and optimal performance. Central to this monitoring is the concept of State of Health (SOH), which quantifies a battery’s remaining capacity in relation to its maximum when the battery was new, as well as battery functionality in energy storage and delivery. This work aims at accurate SOH estimation, which is critical for safe battery operation, preventing catastrophic failures, and minimizing safety risks. A practical analysis of various data-centric techniques like ML and Artificial Neural Network (ANN) and their effectiveness in estimating the maximum capacity available of Li-Ion batteries is discussed in this paper. The data-driven models’ performance is validated by the following error metrics: Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and R-squared(R 2 ). The study utilizes distinct Li-Ion battery datasets, and a validation dataset created by real-time Li-Ion battery parameters. It was found that supervised ML models were better at accurately estimating SOH for the existing dataset, but ANN did better when real-time Li-Ion battery parameters were used to test it. This proposed approach establishes a foundation for accurate and robust battery health monitoring, mitigating the challenges posed by uncertainty factors and further promoting the widespread adoption of Li-Ion batteries in diverse industries.