A Variational Bayes Based State-of-Charge Estimation for Lithium-Ion Batteries Without Sensing Current

Hou, Jing; Yang, Yan; Gao, Tian · 2021 · DOAJ

DOI: 10.1109/ACCESS.2021.3086861

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the challenge of estimating the State-of-Charge (SOC) for lithium-ion batteries in portable devices where current sensors are absent due to cost, volume, or power constraints. Traditional SOC estimation methods rely on current measurements, while existing sensor-less approaches often employ simplified battery models or linear approximations that compromise accuracy. The authors propose a novel Variational Bayes-based Unscented Kalman Filter (VB-UKF) to simultaneously estimate SOC and unknown current input. This approach reformulates the problem as optimal filtering for a nonlinear system with unknown inputs, leveraging Variational Bayesian inference to approximate the joint posterior distribution of the state and input, thereby avoiding the computational complexity of exact Bayesian solutions. The study utilizes a first-order resistor-capacitor (RC) equivalent circuit model to balance accuracy and computational efficiency. Battery parameters, including ohmic resistance, polarization resistance, capacitance, and the Open Circuit Voltage (OCV)-SOC relationship, were identified using a pulsed-discharge test on a Samsung ICR18650 lithium-ion battery. The OCV-SOC curve was characterized by a seventh-order polynomial, while other parameters were fitted using exponential and polynomial functions derived from experimental data. The VB-UKF algorithm combines the Unscented Kalman Filter for nonlinear state estimation with standard Kalman filtering for input estimation, iteratively updating estimates by minimizing the Kullback-Leibler divergence between the approximate and true posterior distributions. Experimental validation was conducted under pulsed-discharge and Urban Dynamometer Driving Schedule (UDDS) profiles. The results demonstrate that the VB-UKF algorithm outperforms the previously proposed Unscented Recursive Three-Step Filter (URTSF) in both convergence rate and estimation accuracy. Specifically, the SOC root mean square errors for the VB-UKF remained bounded within ±3% after convergence. The model validation showed maximum modeling errors of less than 10 mV for pulsed-discharge and less than 3 mV for UDDS tests. The proposed method effectively handles the nonlinearities of the battery system and the uncertainty of the unknown current input without requiring current sensors. The significance of this work lies in providing a robust, accurate, and computationally feasible solution for SOC estimation in low-cost portable applications lacking current sensors. By integrating Variational Bayesian inference with the Unscented Kalman Filter, the method achieves superior performance compared to existing techniques that rely on weighted least squares or simplified models. This approach enhances the reliability of Battery Management Systems (BMS) in resource-constrained environments, ensuring safe and efficient battery operation through precise state monitoring.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• drowsiness detection algorithms