Adaptive optimal observer for real-time state of charge estimation of lithium-ion batteries in robotic systems

Abstract

Purpose

This study aims to address the challenge of the real-time state of charge (SOC) estimation for lithium-ion batteries in robotic systems, which is critical for monitoring remaining battery power, planning task execution, conserving energy and extending battery lifespan.

Design/methodology/approach

The authors introduced an optimal observer based on adaptive dynamic programming for online SOC estimation, leveraging a second-order resistor–capacitor model for the battery. The model parameters were determined by fitting an exponential function to the voltage response from pulse current discharges, and the observer's effectiveness was verified through extensive experimentation.

Findings

The proposed optimal observer demonstrated significant improvements in SOC estimation accuracy, robustness and real-time performance, outperforming traditional methods by minimizing estimation errors and eliminating the need for iterative steps in the adaptive critic and actor updates.

Originality/value

This study contributes a novel approach to SOC estimation using an optimal observer that optimizes the observer design by minimizing estimation errors. This method enhances the robustness of SOC estimation against observation errors and uncertainties in battery behavior, representing a significant advancement in battery management technology for robotic applications.

Citation

Zhao, J., Lu, Z. and Wang, G. (2024), "Adaptive optimal observer for real-time state of charge estimation of lithium-ion batteries in robotic systems", Robotic Intelligence and Automation, Vol. 44 No. 6, pp. 841-853. https://doi.org/10.1108/RIA-04-2024-0091