电力系统及其自动化学报2025，Vol.37Issue(8)：69-75,86,8.DOI:10.19635/j.cnki.csu-epsa.001624
考虑虚拟储能系统的建筑能源系统近端策略优化控制方法
Proximal Policy Optimization Control Method for Building Energy System Considering Virtual Energy Storage System
摘要
Abstract
Deep reinforcement learning(DRL)is an effective approach for realizing the optimal control of a building en-ergy system(BES).However,its practical applications face challenges such as a low convergence efficiency during model training and violations of room temperature constraints.To solve these problems,a proximal policy optimization(PPO)control method for BES which takes a virtual energy storage system(VESS)into account is proposed in this pa-per.First,based on the thermal inertia characteristics of the building envelope structure,a building VESS model is con-structed,and three VESS parameters including the virtual power,virtual capacity and virtual state-of-charge are formu-lated to quantify the BES-adjustable potential provided by thermal inertia.Second,the BES model with the consider-ation of a VESS is transformed into a Markov decision process,and the corresponding state variables,control actions,reward functions and transfer functions are set.Finally,the PPO algorithm is employed for the optimal control of BES.The calculation results of an example show that the proposed method effectively reduces the operating costs of BES and the proportion of room temperature violations while significantly increasing the generation speed of the optimal control strategy.关键词
近端策略优化/马尔可夫决策过程/虚拟储能系统/建筑能源系统/优化控制Key words
proximal policy optimization(PPO)/Markov decision process/virtual energy storage system(VESS)/building energy system(BES)/optimal control分类
信息技术与安全科学引用本文复制引用
庄重,段梅梅,黄艺璇,方凯杰,武泽清,徐延泽..考虑虚拟储能系统的建筑能源系统近端策略优化控制方法[J].电力系统及其自动化学报,2025,37(8):69-75,86,8.基金项目
国网江苏省电力有限公司科技项目资助(J2023126). （J2023126）
