Scholarlink

<p>In response to the distinctly different heating load characteristics within heterogeneous building complex, traditional heating load allocation strategies based on fixed weights can no longer meet the requirements for energy conservation and improving indoor temperature satisfaction rates. This study addresses this problem by proposing an adaptive-weighted multi-objective reinforcement learning (Adaptive-Weighted MORL) framework for a heterogeneous building complex comprising a training gym, office building, dormitory, and cafeteria. The framework achieves dynamic balance optimization between heating load and thermal comfort through an adaptive weight adjustment mechanism integrating proximal policy optimization (PPO) algorithm and non-dominated sorting genetic algorithm II (NSGA-II). PPO learns optimal heating load allocation strategies to adapt to environmental changes, while NSGA-II generates Pareto-optimal solution sets to guide PPO’s weight coefficient updates. This mechanism dynamically adjusts the heating load weight and thermal comfort weight, prioritizing thermal comfort weight under extreme weather conditions. Results demonstrate that, compared to the PPO method and traditional fixed-weight approach, the proposed framework achieves an overall energy saving rate of 22.1%, and a peak heating load reduction exceeding 40%, while maintaining indoor temperature satisfaction rates above 91% in most building types. Notably, under extreme conditions (such as the peak load day of March 17), the framework achieves a 39% peak reduction rate and a 22.2% daily energy saving rate. These findings thoroughly validate its effectiveness in complex dynamic environments. Overall, this framework provides an intelligent solution for optimizing heating load allocation across heterogeneous building types, effectively balancing the conflicting objectives of energy efficiency and thermal comfort while adapting to dynamic environmental conditions.</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

Adaptive method for dynamic collaborative allocation of heating load in heterogeneous building complex: An adaptive-weighted multi-objective reinforcement learning framework

  • 期刊论文

  • 发表日期：

    2025年10月20日

查看全文
  • Junfan An,
  • Yuechao Chao,
  • Yahui Du,
  • Jianjuan Yuan,
  • Zhihua Zhou,
  • Xuejing Zheng

摘要

In response to the distinctly different heating load characteristics within heterogeneous building complex, traditional heating load allocation strategies based on fixed weights can no longer meet the requirements for energy conservation and improving indoor temperature satisfaction rates. This study addresses this problem by proposing an adaptive-weighted multi-objective reinforcement learning (Adaptive-Weighted MORL) framework for a heterogeneous building complex comprising a training gym, office building, dormitory, and cafeteria. The framework achieves dynamic balance optimization between heating load and thermal comfort through an adaptive weight adjustment mechanism integrating proximal policy optimization (PPO) algorithm and non-dominated sorting genetic algorithm II (NSGA-II). PPO learns optimal heating load allocation strategies to adapt to environmental changes, while NSGA-II generates Pareto-optimal solution sets to guide PPO’s weight coefficient updates. This mechanism dynamically adjusts the heating load weight and thermal comfort weight, prioritizing thermal comfort weight under extreme weather conditions. Results demonstrate that, compared to the PPO method and traditional fixed-weight approach, the proposed framework achieves an overall energy saving rate of 22.1%, and a peak heating load reduction exceeding 40%, while maintaining indoor temperature satisfaction rates above 91% in most building types. Notably, under extreme conditions (such as the peak load day of March 17), the framework achieves a 39% peak reduction rate and a 22.2% daily energy saving rate. These findings thoroughly validate its effectiveness in complex dynamic environments. Overall, this framework provides an intelligent solution for optimizing heating load allocation across heterogeneous building types, effectively balancing the conflicting objectives of energy efficiency and thermal comfort while adapting to dynamic environmental conditions.