Parameter estimation for building energy models using GRcGAN

作     者：Hansol Shin Cheol-Soo Park Hansol Shin;Cheol-Soo Park

作者机构：Department of Architecture and Architectural EngineeringCollege of EngineeringSeoul National UniversitySeoul08826R.O.Korea Institute of Construction and Environmental EngineeringInstitute of Engineering ResearchCollege of EngineeringSeoul National UniversitySeoul08826R.O.Korea 

出 版 物：《Building Simulation》 (建筑模拟（英文）)

年 卷 期：2023年第16卷第4期

页      面：629-639页

核心收录：

学科分类：08[工学] 0813[工学-建筑学] 081301[工学-建筑历史与理论] 

基　　金：This work was supported by the Korea Institute of Energy Technology Evaluation and Planning(KETEP)and the Ministry of Trade Industry&Energy(MOTIE)of the Republic of Korea(No.20202020800360) 

主　　题：generative adversarial networks generative model parameter estimation inverse problem model calibration parameter uncertainty 

摘      要：Parameter estimation methods can be classified into(1)manual(trial-and-error),(2)numerical optimization(optimization,sampling),(3)Bayesian inference(Bayes filter,Bayesian calibration),and(4)machine learning(generative model).Bayesian calibration has been widely used because it can capture stochastic nature of uncertain ***,the results of Bayesian calibration could be biased by(1)the prior distribution assumed by the expert’s subjective judgment;(2)the likelihood function that cannot always describe the true likelihood;and(3)the posterior distribution approximation method,such as the Markov Chain Monte Carlo,which requires significant computation *** overcome this,a new approach using a generator-regularized continuous conditional generative adversarial network(GRcGAN)is presented in this *** target parameters of the DOE reference building model were *** was trained to estimate uncertain parameters using simulated monthly electricity and gas *** can successfully estimate five uncertain parameters based on 1,000 training data *** proposed approach presents a potential for stochastic parameter estimation.