| Title |
A Study on Short-term Load Forecasting of Switch based on Graph Neural Network |
| Authors |
Jaein Kim ; Joo-Young Moon ; Jae-Hyun Lee ; Sung-Ho Park ; Sung-min Kim ; Dong-Sub Kim |
| DOI |
http://doi.org/10.5207/JIEIE.2022.36.1.037 |
| Keywords |
ARIMA; Graph neural networks; Load forecasting; LSTM; Switch |
| Abstract |
In this paper, the short-term load forecasting for each switch is performed by graph neural network-based ST-GCN(Spatio Temporal Graph Convolution Networks) model. The proposed model can predict the hourly load by using the connection information of the automatic switch belonging to the D/L(Distribution Line). To this end, the connection information between the switches belonging to the D/L is simplified, and the load information per hour is prep-processed. Next, our model is trained by constructing a training set with the features of the four D/L, which are the experimental subjects.
Finally, to verify the predictive model's performance, we compared the performance of ARIMA(Auto-Regressive Integrated Moving Average) and LSTM(Long-Term Short-Term Memory).
And its limitations are discussed. |