Scholarlink

<p>The critical need for power management strategy enhancement exists now when grid modernization and renewable penetration lead energy innovation toward smart grids of resilient quality. Multiphase inverters that use flying capacitor constructions demonstrate restricted scalability and subpar real-time correction alongside restricted fault tolerance when working with high renewable power variations. The proposed research integrates a hybrid deep reinforcement learning (DRL) and extreme gradient boosting (XGBoost) control strategy with modular multilevel converters (MMC) to develop an advanced intelligent power management system for overcoming present challenges. The proposed modular multilevel converter system delivers advanced dynamic renewable integration solutions and operational reliability improvements because of its modular scalability and superior fault ride-through features and voltage balancing characteristics. The simulated model shows major performance advancement with a grid power factor reaching 0.992 and a THD value below 3.5% and an efficiency increase from 85 to 95%. Failure rate performance increased to (<i>λ</i> ≤ 0.05 failures/year) accompanied by MTTR reduction to less than 20&#xa0;min. When used in place of traditional FCMLI-based methods, the proposed hybrid DRL-XGBoost-driven MMC system presents real-time learning adaptability and higher modular flexibility and fault resilience. Both MATLAB/Simulink and Typhoon HIL real-time simulation platforms serve to perform the complete system validation process.</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

Energy-efficient modular power management in smart grids using DRL-XGBoost-driven multilevel converterss

  • 期刊论文

  • 发表日期：

    2026年3月2日

查看全文
  • Ravindran Kannan,
  • M. Malini,
  • S. Suganya,
  • Y. Lakshmi Prasanna,
  • Karthik Prabhu Durairaj,
  • Tamil Selvi Sakthivel

摘要

The critical need for power management strategy enhancement exists now when grid modernization and renewable penetration lead energy innovation toward smart grids of resilient quality. Multiphase inverters that use flying capacitor constructions demonstrate restricted scalability and subpar real-time correction alongside restricted fault tolerance when working with high renewable power variations. The proposed research integrates a hybrid deep reinforcement learning (DRL) and extreme gradient boosting (XGBoost) control strategy with modular multilevel converters (MMC) to develop an advanced intelligent power management system for overcoming present challenges. The proposed modular multilevel converter system delivers advanced dynamic renewable integration solutions and operational reliability improvements because of its modular scalability and superior fault ride-through features and voltage balancing characteristics. The simulated model shows major performance advancement with a grid power factor reaching 0.992 and a THD value below 3.5% and an efficiency increase from 85 to 95%. Failure rate performance increased to (λ ≤ 0.05 failures/year) accompanied by MTTR reduction to less than 20 min. When used in place of traditional FCMLI-based methods, the proposed hybrid DRL-XGBoost-driven MMC system presents real-time learning adaptability and higher modular flexibility and fault resilience. Both MATLAB/Simulink and Typhoon HIL real-time simulation platforms serve to perform the complete system validation process.