基于MADDPG算法协同预测的多智能体节能制热饮水机的能效提升  

Energy Efficiency Enhancement of a Multi-Agent Energy-Saving Heating Water Dispenser Through Collaborative Prediction Based on the MADDPG Algorithm

作　　者：严安 杨雨琪 蒋鑫阳 王佳玺 关志涛[2] YAN An;YANG Yuqi;JIANG Xinyang;WANG Jiaxi;GUAN Zhitao(School of Economics and Management,North China Electric Power University,Beijing 102200,China;School of Control and Computer Engineering,North China Electric Power University,Beijing 102200,China)

机构地区：[1]华北电力大学经济与管理学院,北京102200 [2]华北电力大学控制与计算机工程学院,北京102200

出　　处：《自动化应用》2024年第1期1-4,共4页Automation Application

摘　　要：为提升公共场所制热饮水机的群体性能,降低整体能耗,设计了多智能体协同预测的节能制热饮水机。利用季节性ARIMA模型预测饮水机使用人数,结合决策树及遗传算法设定单台饮水机的临界阈值,应用多智能体协同MADDPG算法建立多台饮水机协同控制。通过构建用电成本回归模型测试样机性能,实验结论表明,与市面的商用饮水机相比,新设计的饮水机可节约55%的电费,在提高饮水机能源利用效率方面具有潜在应用价值。To improve the group performance of heating drinking water systems in public places and reduce overall energy consumption,a multi-agent collaborative prediction energy-saving heating water dispenser is proposed.The system uses a seasonal ARIMA model to predict the number of users of the water dispenser,combines decision trees and genetic algorithms to set critical thresholds for individual water dispensers,and applies multi-agent collaborative MADDPG algorithms to establish collaborative control between multiple water dispensers.By constructing an electricity cost regression model to test the performance of the prototype,the experimental results show that compared with commercial water dispensers on the market,the newly designed water dispenser can save 55%of electricity costs,and has potential application value in improving the energy utilization efficiency of water dispensers..

关 键 词：制热饮水机 多智能体协同预测 MADDPG算法 季节性ARIMA 

分 类 号：TP29[自动化与计算机技术—检测技术与自动化装置]