METHOD FOR OPTIMALLY CONTROLLING CHILLER USING ONLINE/OFFLINE HYBRID MACHINE LEARNING MODEL

摘要

The present invention provides a method of controlling a building system (500) including a building energy management system (BEMS) with a plurality of sensors (S). According to the present invention, a method of optimally controlling a chiller using an online/offline machine learning model includes the steps of: (a) constructing, by a machine learning model generation unit (200), the online machine learning model by using input variables which are measured by the sensor (S) and automatically input to the machine learning model generation unit (200) through the input variable input unit (100); (b) constructing, by the machine learning model generation unit (200), the offline machine learning model by using input variables which are manually input to the machine learning model generation unit( 200) through an input manual input unit (110); (c) generating an offline/online hybrid machine learning model by using both the constructed online machine learning model and the offline machine learning model; (d) searching for a control set value for operating a refrigerator (510) at an optimum efficiency by using the generated offline/online hybrid machine learning model and a preset optimization algorithm; and (e) controlling the refrigerator (510) by using the control set value.;COPYRIGHT KIPO 2019