[Paper Review] Dynamic Cell Modeling of Li-Ion Polymer Batteries for Precise SOC Estimation in Power-Needy Autonomous Electric Vehicles

Qasim Ajao, Lanre Sadeeq|arXiv (Cornell University)|Jun 19, 2023

Advanced Battery Technologies ResearchEngineering3 citations

TL;DR

This paper proposes a high-accuracy dynamic cell model for Li-ion polymer batteries in power-needy autonomous electric vehicles (AEVs), using an advanced Kalman filtering approach with nonlinear autoregressive and radial basis function networks to estimate state-of-charge (SOC). The method achieves sub-quantization noise RMS error, enabling full utilization of the battery pack without overcharge or undercharge risks.

ABSTRACT

This paper presents findings on dynamic cell modeling for state-of-charge (SOC) estimation in an autonomous electric vehicle (AEV). The studied cells are Lithium-Ion Polymer-based with a nominal capacity of around 8Ah, optimized for power-needy applications. The AEV operates in a harsh environment with rate requirements up to +/-25C and highly dynamic rate profiles, unlike portable-electronic applications with constant power output and fractional C rates. SOC estimation methods effective in portable electronics may not suffice for the AEV. Accurate SOC estimation necessitates a precise cell model. The proposed SOC estimation method utilizes a detailed Kalman-filtering approach. The cell model must include SOC as a state in the model state vector. Multiple cell models are presented, starting with a simple one employing "Coulomb counting" as the state equation and Shepherd's rule as the output equation, lacking prediction of cell relaxation dynamics. An improved model incorporates filter states to account for relaxation and other dynamics in closed-circuit cell voltage, yielding better performance. The best overall results are achieved with a method combining nonlinear autoregressive filtering and dynamic radial basis function networks. The paper includes lab test results comparing physical cells with model predictions. The most accurate models obtained have an RMS estimation error lower than the quantization noise floor expected in the battery-management-system design. Importantly, these models enable precise SOC estimation, allowing the vehicle controller to utilize the battery pack's full operating range without overcharging or undercharging concerns.

Motivation & Objective

• Address the challenge of accurate SOC estimation in autonomous electric vehicles (AEVs) operating under high dynamic power demands exceeding 25C rate profiles.
• Overcome limitations of conventional SOC estimation methods used in portable electronics, which are inadequate for the harsh, rapidly varying conditions in AEVs.
• Develop a precise battery cell model that captures relaxation dynamics and nonlinear behaviors critical for reliable SOC estimation in power-needy applications.
• Enable full utilization of the battery pack’s operating range by minimizing SOC estimation error below the system’s quantization noise floor.
• Ensure robustness and accuracy under highly dynamic rate profiles typical of autonomous vehicle operation, avoiding overcharging or undercharging risks.

Proposed method

• Formulate a dynamic cell model with SOC as a state variable in the state vector, using Coulomb counting as the initial state equation.
• Incorporate Shepherd’s rule as the output equation to model open-circuit voltage, initially without relaxation dynamics.
• Enhance the model by introducing filter states to capture cell relaxation and transient voltage dynamics in closed-circuit voltage.
• Implement a nonlinear autoregressive filtering technique to improve prediction accuracy under varying load conditions.
• Integrate dynamic radial basis function (RBF) networks to model nonlinearities and time-varying behaviors in the battery system.
• Apply an extended Kalman filter (EKF) framework to recursively estimate SOC using the enhanced model, minimizing estimation error.

Experimental results

Research questions

• RQ1 Can a dynamic cell model that accounts for relaxation and nonlinear behavior improve SOC estimation accuracy in high-power AEVs compared to conventional methods?
• RQ2 How does the inclusion of filter states for relaxation dynamics affect the accuracy of SOC estimation under highly dynamic current profiles?
• RQ3 To what extent can nonlinear autoregressive filtering and dynamic RBF networks reduce SOC estimation error in Li-ion polymer cells under extreme C-rates?
• RQ4 Does the proposed model achieve estimation accuracy below the quantization noise floor of typical battery management systems?
• RQ5 Can the model enable safe and full utilization of the battery pack’s capacity without overcharge or undercharge concerns in autonomous vehicle applications?

Key findings

• The proposed dynamic cell model with filter states for relaxation dynamics significantly improves SOC estimation accuracy over basic Coulomb counting and Shepherd’s rule.
• The integration of nonlinear autoregressive filtering and dynamic radial basis function networks yields the best overall performance in SOC estimation.
• The most accurate model achieves an RMS estimation error lower than the expected quantization noise floor of the battery management system.
• The model enables precise SOC estimation across highly dynamic current profiles, including rates up to +/-25C, ensuring safe operation within the full battery capacity range.
• Laboratory test results confirm strong agreement between physical cell behavior and model predictions, validating the model’s robustness under real-world AEV conditions.
• The method effectively prevents overcharging and undercharging by maintaining SOC estimation accuracy below system-level noise limits, enhancing battery safety and longevity.