| Title |
Domestic Hot Water Load Prediction in Residential Communities using the Structured Probabilistic Statistical Models |
| Authors |
김철호(Kim, Chulho) ; 변지욱(Byun, Jiwook) ; 고재현(Go, Jaehyun) ; 허연숙(Heo, Yeonsook) |
| DOI |
https://doi.org/10.5659/JAIK.2022.38.3.211 |
| Keywords |
Probabilistic model; Domestic hot water load; Residential community; Household variation; Temporal variation |
| Abstract |
This study developed structured probabilistic statistical models to systematically reflect individual variations in the domestic hot water load
factoring household characteristics and temporal variations in the hourly pattern of consumption. The hourly domestic hot water data of 15
households derived from the Korea Energy Agency’s public data were used. Models 1 and 2 were based on bilinear regression models to
predict the daily average domestic hot water load based on the household characteristics and daily variations. Model 3 was based on the
multivariate normal distribution to generate the average hourly domestic hot water load profile which varied per household. Model 4 used the
beta distribution probability density function to randomly generate hourly variations from the average load profiles reflecting temporal
variation. As a result of applying these four models, individual and temporal variations were reflected in the whole year hourly load
prediction. The resulting probabilistic domestic hot water loads were compared with those determined using the deterministic method of the
American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE), ECO2 criteria and the single multivariate distribution
model derived during the entire set of hourly load data across the 15 households. This comparison reflected that the structured probabilistic
models predicted individual and temporal variations with sufficient accuracy. |