Scholarlink

<p>Battery–supercapacitor hybrid energy storage systems (HESSs) require coordinated state awareness because of the differing dynamic time scales of the two storage devices and their susceptibility to measurement uncertainty, parameter mismatch, and battery aging. A simulation-based dual-branch co-estimation framework that integrates complementary battery and supercapacitor (SC) estimation layers within a unified HESS environment was presented in this paper. Achieving an accurate system integration is the primary focus of this simulation, rather than developing a novel Kalman filter (KF) algorithm. In this framework, the battery branch employs a 2RC equivalent-circuit model (ECM) in conjunction with an adaptive innovation-based extended Kalman filter (AIEKF) to enable rapid estimation of state-of-charge (SOC) and terminal voltage (Vterm). While a slower effective-capacity update layer is also employed as Q-based state-of-health (SOH) tracking. The SC branch employs a nonlinear leakage-aware three-state ECM and compares AIEKF and UKF observers for internal-voltage, terminal-voltage, and energy-based SOC estimation. The underlying system is tested across a range of scenarios, including degraded Qbat due to aging, measurement noise, AIEKF/UKF parameter mismatch, initialization uncertainty, aggressive SC transients, and HESS excitation from the US06 driving cycle. During repeated US06 operations with scheduled rest-anchor intervals, the estimated Qbat is reduced from 180,000&#xa0;C to the aged value of about 162,000&#xa0;C (corresponding to SOH_Q ≈ 0.90), the battery SOC root-mean-square error (RMSE) is 0.2249%. On the SC side, the UKF estimator demonstrates a SOCsc RMSE of 0.1998% and a Vterm RMSE of 0.01028&#xa0;V, improving the performance compared to AIEKF. Using current-splitting sensitivity analysis for τ = 5,10,20,40&#xa0;s confirms stable co-estimation functionality across different time-constant values. Component-wise battery ablation shows that SOC re-anchoring and capacity feedback are the main contributors to long-term SOC consistency. At the same time, adaptive noise tuning plays a minor role under stationary-noise conditions. This indicates that leakage modeling and the third redistribution branch solely enrich internal dynamics and have minor impact on global SOCs and terminal-voltage accuracy under the tested condition based on US06. In summary, the result shows that coordinated battery–SC state and health co-estimation is feasible in a controlled simulation environment, while elucidating the contributions of each estimator and the limitations of each estimator mechanism.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Simulation-based state and health co-estimation for battery–supercapacitor hybrid energy storage systems

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年8月5日

查看全文
  • Jouhayna Bouanani,
  • Mourad Yessef,
  • Ahmed Lagrioui,
  • Mohammed F. Allehyani,
  • Husam S. Samkari

摘要

Battery–supercapacitor hybrid energy storage systems (HESSs) require coordinated state awareness because of the differing dynamic time scales of the two storage devices and their susceptibility to measurement uncertainty, parameter mismatch, and battery aging. A simulation-based dual-branch co-estimation framework that integrates complementary battery and supercapacitor (SC) estimation layers within a unified HESS environment was presented in this paper. Achieving an accurate system integration is the primary focus of this simulation, rather than developing a novel Kalman filter (KF) algorithm. In this framework, the battery branch employs a 2RC equivalent-circuit model (ECM) in conjunction with an adaptive innovation-based extended Kalman filter (AIEKF) to enable rapid estimation of state-of-charge (SOC) and terminal voltage (Vterm). While a slower effective-capacity update layer is also employed as Q-based state-of-health (SOH) tracking. The SC branch employs a nonlinear leakage-aware three-state ECM and compares AIEKF and UKF observers for internal-voltage, terminal-voltage, and energy-based SOC estimation. The underlying system is tested across a range of scenarios, including degraded Qbat due to aging, measurement noise, AIEKF/UKF parameter mismatch, initialization uncertainty, aggressive SC transients, and HESS excitation from the US06 driving cycle. During repeated US06 operations with scheduled rest-anchor intervals, the estimated Qbat is reduced from 180,000 C to the aged value of about 162,000 C (corresponding to SOH_Q ≈ 0.90), the battery SOC root-mean-square error (RMSE) is 0.2249%. On the SC side, the UKF estimator demonstrates a SOCsc RMSE of 0.1998% and a Vterm RMSE of 0.01028 V, improving the performance compared to AIEKF. Using current-splitting sensitivity analysis for τ = 5,10,20,40 s confirms stable co-estimation functionality across different time-constant values. Component-wise battery ablation shows that SOC re-anchoring and capacity feedback are the main contributors to long-term SOC consistency. At the same time, adaptive noise tuning plays a minor role under stationary-noise conditions. This indicates that leakage modeling and the third redistribution branch solely enrich internal dynamics and have minor impact on global SOCs and terminal-voltage accuracy under the tested condition based on US06. In summary, the result shows that coordinated battery–SC state and health co-estimation is feasible in a controlled simulation environment, while elucidating the contributions of each estimator and the limitations of each estimator mechanism.