Scholarlink

<p>The rapid integration of intermittent renewable resources into power systems poses a significant challenge to reliable frequency regulation, a problem exacerbated by conventional studies that rely on idealized inputs and suboptimal controller tuning. This paper proposes a robust, forecast integrated frequency regulation framework for a two-area Virtual Power Plant (VPP) and Conventional Power Plant (CPP) integration that aggregates diverse distributed energy resources (DERs)—including wind, solar PV, solar thermal units with molten-salt based storage, and a fleet of plug-in hybrid electric vehicles (PHEVs)—coordinating with a conventional plant via tie-line control. Replacing simplistic step disturbances, day-ahead wind and regional load forecasts for Assam and Meghalaya, generated via machine learning, provide realistic external inputs. A novel Multiplayer Battle Game Inspired Optimizer (MBGO) is introduced to offline tune I, PI, and PID controllers, benchmarking against Particle Swarm Optimization (PSO), Salp Swarm Optimization (SSO), Green Leafhopper Flame Optimization (GLFOA), Hippopotamus Optimization Algorithm (HOA), and the PUMA optimizer. Results across multiple test cases demonstrate MBGO’s superior convergence speed and the consistent performance of its tuned controllers, which achieve a minimal Integral Square Error (ISE) of 0.000352 alongside reduced overshoot, faster settling, and enhanced tie-line stability under forecast driven variability. This paper shows that machine learning predictions combined with heterogeneous energy storage and those boosted by MBGO and classical control can significantly improve the dispatchability of Virtual Power Plants (VPPs) and add quantifiable frequency stabilisation properties to power systems with substantial renewable generation.</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

MBGO-PID Based Frequency Regulation of Virtual Power Plants with Machine Learning Forecasted Day-Ahead Profiles for Diverse DER-Based Storage

  • 期刊论文

  • 发表日期：

    2026年1月5日

查看全文
  • Subhajit Roy,
  • Dulal Chandra Das,
  • Nidul Sinha

摘要

The rapid integration of intermittent renewable resources into power systems poses a significant challenge to reliable frequency regulation, a problem exacerbated by conventional studies that rely on idealized inputs and suboptimal controller tuning. This paper proposes a robust, forecast integrated frequency regulation framework for a two-area Virtual Power Plant (VPP) and Conventional Power Plant (CPP) integration that aggregates diverse distributed energy resources (DERs)—including wind, solar PV, solar thermal units with molten-salt based storage, and a fleet of plug-in hybrid electric vehicles (PHEVs)—coordinating with a conventional plant via tie-line control. Replacing simplistic step disturbances, day-ahead wind and regional load forecasts for Assam and Meghalaya, generated via machine learning, provide realistic external inputs. A novel Multiplayer Battle Game Inspired Optimizer (MBGO) is introduced to offline tune I, PI, and PID controllers, benchmarking against Particle Swarm Optimization (PSO), Salp Swarm Optimization (SSO), Green Leafhopper Flame Optimization (GLFOA), Hippopotamus Optimization Algorithm (HOA), and the PUMA optimizer. Results across multiple test cases demonstrate MBGO’s superior convergence speed and the consistent performance of its tuned controllers, which achieve a minimal Integral Square Error (ISE) of 0.000352 alongside reduced overshoot, faster settling, and enhanced tie-line stability under forecast driven variability. This paper shows that machine learning predictions combined with heterogeneous energy storage and those boosted by MBGO and classical control can significantly improve the dispatchability of Virtual Power Plants (VPPs) and add quantifiable frequency stabilisation properties to power systems with substantial renewable generation.