| Title |
A method of multi-model machine learning for electrical energy prediction accuracy improvement |
| Authors |
임정현(Jung Hyun Im) ; 성영락(Yeong Rak Seong) ; 오하령(Ha-Ryoung Oh) |
| DOI |
https://doi.org/10.5370/KIEE.2022.71.6.876 |
| Keywords |
Electrical energy prediction; machine learning; multi-model; RNN; Continuous time series |
| Abstract |
The rapid increase in power consumption is increasing the importance of predicting power consumption for stable power supply.
Accordingly, power consumption prediction methods using machine learning are being actively studied. RNN-based models are mainly used in systems with continuous data such as power consumption. Power consumption prediction using a single RNN-based model also shows high accuracy, but this paper proposes a multi-model-based power consumption prediction method for predicting with higher accuracy. In order to minimize noise data learning occurring in a single model, multiple models were used to learn only data that meets the learning conditions according to the conditions of the most important feature. Also, the power consumption prediction results were derived using the learned multiple models. In addition, it was verified that the prediction of power consumption using multiple models produced more accurate results compared to the prediction results of a single model. |