Safe operation of online learning data driven model predictive control of building energy systems

作     者：Phillip Stoffel Patrick Henkel Martin Ratz Alexander Kumpel Dirk Muller 

作者机构：RWTH Aachen UniversityE.ON Energy Research CenterInstitute for Energy Efficient Buildings and Indoor ClimateMathieustraße 10Aachen52074Germany 

出 版 物：《Energy and AI》 (能源与人工智能（英文）)

年 卷 期：2023年第14卷第4期

页      面：536-549页

核心收录：

学科分类：12[管理学] 080702[工学-热能工程] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 081104[工学-模式识别与智能系统] 08[工学] 0807[工学-动力工程及工程热物理] 0835[工学-软件工程] 0811[工学-控制科学与工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：This project has received funding from the European Union’s Hori-zon 2020 research and innovation programme under grant agreement No.101023666 

主　　题：Data-driven model predictive control Online learning Novelty detection Artificial neural networks Building energy systems 

摘      要：Model predictive control is a promising approach to reduce the CO 2 emissions in the building ***,the vast modeling effort hampers the widescale practical ***,data-driven process models,like artificial neural networks,are well-suited to automatize the ***,the underlying data set strongly determines the quality and reliability of artificial neural *** general,the validity domain of a machine learning model is limited to the data that was used to train *** based on system states outside that domain,so-called extrapolations,are unreliable and can negatively influence the control *** present a safe operation approach combined with online learning to deal with extrapolation in data-driven model predictive ***,the k-nearest neighbor algorithm is used to detect extrapolation to switch to a robust fallback *** continuously retraining the artificial neural networks during operation,we successively increase the validity domain of the artificial neural networks and the control *** apply the approach to control a building energy system provided by the BOPTEST *** compare controllers based on two data sets,one with extensive system excitation and one with baseline *** system is controlled to a fixed temperature set point in baseline ***,the artificial neural networks trained on this data set tend to extrapolate in other operating *** show that safe operation in combination with online learning significantly improves performance.