Research pick: Current affairs - "Design of an augmented unknown input estimator for the lithium-ion battery state of charge and sensor fault estimation"

A new approach to accurately determining the state of charge (SoC) of a lithium-ion battery could improve safety and longevity of electric vehicles, renewable energy systems, and portable electronics, according to research in the International Journal of Critical Infrastructures. The work builds on an approach that not only measures a battery’s remaining charge with exceptional precision but also detects and corrects sensor faults in real time.

The state of charge, or SoC, represents how much energy remains in a battery relative to its full capacity. However, the SoC cannot be measured directly as if it were some digital parameter because lithium batteries are complicated electrochemical systems. The SoC can only be estimated through models and data, and the results change as the battery ages through repeated charge-discharge cycles over its lifetime. Nevertheless, inaccurate estimation can lead to overcharging or deep discharging, conditions that cause degradation and ultimately failure, especially when repeated overheating has occurred.

The standard SoC estimation techniques all have their pros and cons. Empirical methods, such as Coulomb counting, which tracks charge entering and leaving a battery cell, are simple but prone to error from sensor drift and environmental changes. Voltage and impedance-based approaches work only when the battery is not in use. Model-based systems require extensive computation, making them difficult to apply in real-time conditions such as driving or rapid charging.

The new research closes the circuit by providing a hybrid strategy that combines nonlinear physical modelling and advanced estimation algorithms. The system treats sensor faults as part of the model itself, considering them as unknown inputs to be estimated alongside the SoC. This approach allows the battery management system to get an accurate reading even when its sensors are malfunctioning or operating under fluctuating temperature and ageing conditions.

In their laboratory tests, the researchers were able to make usable and accurate estimates of SoC even under stressful conditions. The model’s computational efficiency also makes it suitable for integration into embedded battery management systems used in vehicles and stationary storage units.

Fang, L. (2025) ‘Design of an augmented unknown input estimator for the lithium-ion battery state of charge and sensor fault estimation’, Int. J. Critical Infrastructures, Vol. 21, No. 11, pp.30–54.