← 返回洞察首页
论文2026年7月1日待评估

TD3-Enhanced Adaptive Active Disturbance Rejection Control for Bidirectional DC--DC Converter in Photovoltaic-Battery Hybrid Microgrids面向光伏—电池混合微电网双向 DC-DC 变换器的 TD3 增强自适应自抗扰控制

英文 Abstract

This paper addresses the stability and reliability issues associated with the application of linear active disturbance rejection control (LADRC) in hybrid PV-battery microgrids. A hierarchical control strategy tailored for bidirectional DC-DC converter in the system is proposed. The strategy comprises an outer voltage regulation loop implemented with an enhanced LADRC and an inner current tracking loop driven by a proportional-integral (PI) controller. Specifically, the traditional LADRC is enhanced via an improved linear extended state observer, which achieves separate compensation for low-order and high-order disturbances. Additionally, the observer parameters are optimized in real time based on the Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm. Finally, simulations and experiments under multiple operating conditions are conducted to validate the proposed control strategy, demonstrating its superior performance over double closed-loop PI control and traditional LADRC.

中文翻译

本文针对混合光伏—电池微电网中采用线性自抗扰控制(LADRC)时的稳定性和可靠性问题,提出一种适用于系统双向 DC-DC 变换器的分层控制策略。该策略由采用改进型 LADRC 的外部电压调节环和由比例积分(PI)控制器驱动的内部电流跟踪环组成。具体而言,研究通过改进的线性扩张状态观测器增强传统 LADRC,使其能够分别补偿低阶和高阶扰动;同时基于双延迟深度确定性策略梯度(TD3)算法实时优化观测器参数。最后,作者在多种运行工况下开展仿真和实验,验证了所提控制策略,并表明其性能优于双闭环 PI 控制和传统 LADRC。

中文概述

本文面向光伏—电池混合微电网中双向 DC-DC 变换器的稳定控制,提出分层控制策略,并使用 TD3 强化学习实时优化扩张状态观测器参数。仿真和实验结果表明,该方法相较传统双闭环 PI 与 LADRC 具有更好的控制性能。

内容分类

应用场景
控制与自治
技术方法
优化与强化学习

原文与代码