Unknown Input Observer-Based Soc Estimation for Lithium-Ion Batteries with Current Sensor-Free and Voltage Bias

Due to space limitations, current sensors are often impractical in many portable devices, such as smartphones, smartwatches, and other wearable technologies that demand minimalistic hardware footprints, leading to the widespread adoption of current-sensorless methods for lithium-ion battery state-ofcharge (SOC) estimation. However, most of these methods, which typically leverage sophisticated mathematical models or precalibrated lookup tables to infer SOC from observable parameters, assume perfect voltage sensor measurements, an idealized condition that fails to account for the practical imperfections encountered in real-world applications. To address this critical challenge and improve estimation robustness, this paper presents an innovative unknown input observer (UIO) to estimate SOC with biased voltage measurements, eliminating the need for a current sensor entirely, byingeniously treating the fixed voltage bias as an additional state variable integrated into the the dynamic framework of the observer, thus enabling simultaneous estimation of both SOC and the bias term. Experimental results on a 2.45 Ah lithium-ion battery, conducted under rigorously controlled laboratory conditions with varying charge-discharge cycles to simulate realistic usage scenarios, demonstrate the root mean squared error (RMSE) of SOC estimation is $\mathbf{1. 0 2 \%}$.