基于DDPG模型的建筑能耗控制方法  

A BUILDING ENERGY CONSUMPTION CONTROL METHOD BASED ON DDPG MODEL

作　　者：周鑫 陈建平[1,2] 傅启明 Zhou Xin;Chen Jianping;Fu Qiming(School of Electronics and Information Engineering,Suzhou University of Science and Technology,Suzhou 215009,Jiangsu,China;Jiangsu Province Key Laboratory of Intelligent Building Energy Efficiency,Suzhou 215009,Jiangsu,China)

机构地区：[1]苏州科技大学电子与信息工程学院,江苏苏州215009 [2]江苏省建筑智慧节能重点实验室,江苏苏州215009

出　　处：《计算机应用与软件》2023年第2期40-47,共8页Computer Applications and Software

基　　金：国家自然科学基金项目(61772357,61672371);江苏省研究生科研创新计划项目(SJCX18-0881)。

摘　　要：针对居民建筑能耗逐渐增加、传统控制方法效率低下的问题,提出一种基于深度确定性策略梯度的建筑能耗控制方法。该方法利用深度强化学习模型,将建筑电力使用问题建模为强化学习的控制问题,解决负荷降低和成本最小化的问题。根据某开源数据库中居民的能耗使用数据,结合深度Q网络、确定性策略梯度和深度确定性策略梯度算法进行实验验证。实验结果表明,该方法能够有效降低负荷峰值与电力能源使用成本,实现建筑节能的目的。In order to solve the problem of the increasing energy consumption of residential buildings and low efficiency of traditional control methods,this paper proposes a building energy consumption control method based on deep deterministic policy gradient.This method used the deep reinforcement learning model to model the problem of building power use as the control problem of reinforcement learning to solve the problem of load reduction and cost minimization.Based on the energy consumption data of residents in an open source database,DQN,DPG and DDPG algorithms were combined for experimental verification.The experimental results show that the method can effectively reduce the peak load and power energy use cost and achieve the purpose of building energy conservation.

关 键 词：深度强化学习 深度确定性策略梯度 策略优化 建筑节能 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]