| Title |
A Study on Energy Consumption Prediction from Building Energy Management System Data with Missing Values Using SSIM and VLSW Algorithms |
| Authors |
(QUAN JUNLONG) ; 신지웅(Jee-Woong Shin) ; 고정림(Jeong-Lim Ko) ; 신승권(Seung-Kwon Shin) |
| DOI |
https://doi.org/10.5370/KIEE.2021.70.10.1540 |
| Keywords |
Building Energy Management Systems; Missing Values; Energy Prediction; Variable-Length Sliding Window; Sequence-to-Sequence Imputation Model |
| Abstract |
BEMS provide energy forecasting capabilities for effective energy management. Energy consumption prediction requires complete time series data, including missing data sets that may lead to biased results. However, it is a common problem that BEMS data is not measured due to power failure, communication barrier, maintenance and other reasons. Therefore, a solution is required for missing values. The existing methods will affect the results of the prediction model in the case of continuous missing interval or small amount of data. Therefore, this study uses VLSW algorithm which can generate fully learned data with only a small amount of data and SSIM model based on it to study the scheme to improve the prediction accuracy of building energy data set including missing values, and compares the difference with existing models. The results show that the SSIM model has better prediction accuracy than the existing LSTM technology. The concentration range of ASHRAE error level was satisfactory |