Open AccessMethod of System Identification for Air Conditioning Systems in Operation
Author(s) -
Tetsuyasu Yamada,
Hisao Ayame,
Shigeyuki Nagasaka,
Hiroo Hirose
Publication year - 2022
Publication title -
international journal emerging technology and advanced engineering
Language(s) - English
Resource type - Journals
ISSN - 2250-2459
DOI - 10.46338/ijetae0522_05
Subject(s) - autoregressive model , control theory (sociology) , air conditioning , computer science , system identification , bayesian probability , process (computing) , control engineering , control (management) , engineering , mathematics , statistics , artificial intelligence , data modeling , mechanical engineering , database , operating system
This study aims to optimize the operation of an air conditioning (AC) system by tracking situational changes due to outside temperature, number of people and computers, and other factors. Therefore, we studied the accurate estimation of system parameters of an AC unit during operation. We modeled the AC system using the first-order plus dead time model and discretized it using the autoregressive with exogenous input model. We developed a technique to estimate the system parameters using Bayesian optimization. Here, the system parameters are values that determine the physical characteristics of the combined air conditioning system and room. Therefore, we determined that there are cases where the characteristics deteriorate after repeated estimation. By solving this problem, we were able to establish a practical system. Keywords—ARX time-series model, Bayesian optimization method, First-order plus dead time, Gaussian process, PID control