| 融合参数识别双有源桥变换器鲁棒模型预测控制 |
投稿时间：2025-09-22 修订日期：2025-12-11  点此下载全文 |
| 引用本文： |
| 摘要点击次数: 17 |
| 全文下载次数: 0 |
|
| 基金项目:黑龙江省重点研发计划 |
|
| 中文摘要:传统双有源桥变换器的模型预测控制性能高度依赖于精确的数学模型，对电感等关键参数的失配尤为敏感，导致控制性能下降及鲁棒性不足。为此，本文提出一种融合在线参数辨识与鲁棒补偿机制的移动离散控制集鲁棒模型预测控制策略。该策略主要包括两方面：其一，采用基于梯度下降法的在线参数辨识，实时估计时变参数并动态修正预测模型；其二，设计鲁棒补偿项，以抑制参数估计误差、模型不确定性和外部扰动的综合影响。在线辨识与鲁棒补偿协同作用，显著降低了系统对参数失配的敏感度。仿真结果表明，所提策略在不同工况下均能有效提升双有源桥变换器的输出电压控制精度与整体鲁棒稳定性，验证了该方法的有效性。 |
| 中文关键词:双有源桥变换器 模型预测控制 移动离散控制集 参数识别 鲁棒项 |
| |
| Robust Model Predictive Control with Online Parameter Identification for Dual-Active-Bridge Converters |
|
|
| Abstract:The performance of conventional model predictive control (MPC) in dual-active-bridge (DAB) converters heavily relies on an accurate mathematical model and is highly sensitive to mismatches in key parameters such as inductance, which leads to degraded control performance and insufficient robustness. To address this issue, this paper proposes a moving discrete control set robust model predictive control (MDCS-RMPC) strategy that integrates online parameter identification and a robust compensation mechanism. The strategy consists of two key components: first, an online parameter identification method based on gradient descent is employed to estimate time-varying parameters in real time and dynamically update the prediction model; second, a robust compensation term is designed to suppress the combined effects of parameter estimation errors, model uncertainties, and external disturbances. The synergistic action of online identification and robust compensation significantly reduces the system's sensitivity to parameter mismatches. Simulation results demonstrate that the proposed strategy effectively enhances both the output voltage control accuracy and the overall robust stability of the DAB converter under various operating conditions, verifying the effectiveness of the method. |
| keywords:Dual active bridge converter Model predictive control Moving discrete control set Parameter identification Robust term |
| 查看全文 查看/发表评论 下载pdf阅读器 |
位访问者,今日访问
次