TY - JOUR
T1 - Development of optimization control system for chiller plant based on a predictive model with Multi Stack LSTM and Deep Learning Neural Network Multi Output
AU - Napitupulu, Haposan Yoga Pradika
AU - Nugraha, I. Gde Dharma
AU - Sari, Riri Fitri
N1 - Publisher Copyright:
© 2024 Elsevier Ltd
PY - 2024/12/1
Y1 - 2024/12/1
N2 - Net-zero emissions become necessary for every country, implementing energy efficiency is the way to achieve it. Released report by the International Energy Agency in September 2022 stated that energy efficiency and electrification are the main priorities for Indonesia in achieving NZE. Currently, more than 50 % of energy consumption in buildings is intended for cooling systems (chillers plants). Thus, energy efficiency in buildings cooling systems has great potential to help achieve NZE and support SDGs. This research is intended to find an optimization control system for chiller plant in order to reduce energy consumption of cooling system in building. The development of this optimization control system is based on predictive model with Multi Stack LSTM and Deep Learning Neural Network. This research developed four different optimization control system frameworks. Every developed framework then will be tested by carrying out simulation with Chiller Plant model. The important parameter was found by using correlation matrix. Based on the correlation matrix, it was found that the most influence parameter that affect chiller plant performance are Tcws and Twb. It leads developed Frameworks 1 become the best compared with others developed frameworks in this research, because it used Twb as a variable input in parallel with performance prediction. This developed framework has deficient error MAE, MSE, RMSE respectively 0.8079, 0.6527, 0.8079 and able to reduce energy consumption by 10.72 %.
AB - Net-zero emissions become necessary for every country, implementing energy efficiency is the way to achieve it. Released report by the International Energy Agency in September 2022 stated that energy efficiency and electrification are the main priorities for Indonesia in achieving NZE. Currently, more than 50 % of energy consumption in buildings is intended for cooling systems (chillers plants). Thus, energy efficiency in buildings cooling systems has great potential to help achieve NZE and support SDGs. This research is intended to find an optimization control system for chiller plant in order to reduce energy consumption of cooling system in building. The development of this optimization control system is based on predictive model with Multi Stack LSTM and Deep Learning Neural Network. This research developed four different optimization control system frameworks. Every developed framework then will be tested by carrying out simulation with Chiller Plant model. The important parameter was found by using correlation matrix. Based on the correlation matrix, it was found that the most influence parameter that affect chiller plant performance are Tcws and Twb. It leads developed Frameworks 1 become the best compared with others developed frameworks in this research, because it used Twb as a variable input in parallel with performance prediction. This developed framework has deficient error MAE, MSE, RMSE respectively 0.8079, 0.6527, 0.8079 and able to reduce energy consumption by 10.72 %.
KW - Algorithm
KW - Chiller plant
KW - Energy efficiency
KW - HVAC
KW - Machine learning
UR - http://www.scopus.com/inward/record.url?scp=85206467238&partnerID=8YFLogxK
U2 - 10.1016/j.jobe.2024.111029
DO - 10.1016/j.jobe.2024.111029
M3 - Article
AN - SCOPUS:85206467238
SN - 2352-7102
VL - 98
JO - Journal of Building Engineering
JF - Journal of Building Engineering
M1 - 111029
ER -