DOI QR코드

DOI QR Code

특수일 분리와 예측요소 확장을 이용한 전력수요 예측 딥 러닝 모델

Deep Learning Model for Electric Power Demand Prediction Using Special Day Separation and Prediction Elements Extention

  • 박준호 (가천대학교 에너지 IT학과) ;
  • 신동하 (가천대학교 에너지 IT학과) ;
  • 김창복 (가천대학교 에너지 IT학과)
  • 투고 : 2016.07.25
  • 심사 : 2017.08.25
  • 발행 : 2017.08.31

초록

본 연구는 전력수요 패턴이 다른 평일과 특수일 데이터가 가지는 상관관계를 분석하여, 별도의 데이터 셋을 구축하고, 각 데이터 셋에 적합한 딥 러닝 네트워크를 이용하여, 전력수요예측 오차를 감소하는 방안을 제시하였다. 또한, 기본적인 전력수요 예측요소인 기상요소에 환경요소, 구분요소 등 다양한 예측요소를 추가하여 예측율을 향상하는 방안을 제시하였다. 전체데이터는 시계열 데이터 학습에 적합한 LSTM을 이용하여 전력수요예측을 하였으며, 특수일 데이터는 DNN을 이용하여 전력수요예측을 하였다. 실험결과 기상요소 이외의 예측요소 추가를 통해 예측율이 향상되었다. 전체 데이터 셋의 평균 RMSE는 LSTM이 0.2597이며, DNN이 0.5474로 LSTM이 우수한 예측율을 보였다. 특수일 데이터 셋의 평균 RMSE는 0.2201로 DNN이 LSTM보다 우수한 예측율을 보였다. 또한, 전체 데이터 셋의 LSTM의 MAPE는 2.74 %이며, 특수 일의 MAPE는 3.07 %를 나타냈다.

This study analyze correlation between weekdays data and special days data of different power demand patterns, and builds a separate data set, and suggests ways to reduce power demand prediction error by using deep learning network suitable for each data set. In addition, we propose a method to improve the prediction rate by adding the environmental elements and the separating element to the meteorological element, which is a basic power demand prediction elements. The entire data predicted power demand using LSTM which is suitable for learning time series data, and the special day data predicted power demand using DNN. The experiment result show that the prediction rate is improved by adding prediction elements other than meteorological elements. The average RMSE of the entire dataset was 0.2597 for LSTM and 0.5474 for DNN, indicating that the LSTM showed a good prediction rate. The average RMSE of the special day data set was 0.2201 for DNN, indicating that the DNN had better prediction than LSTM. The MAPE of the LSTM of the whole data set was 2.74% and the MAPE of the special day was 3.07 %.

키워드

참고문헌

  1. B. G. Koo, H. S Kim, H. S. Lee, and J. H. Park, "Short-term electric load forecasting for summer season using ttemperature data," The Journal of the Korean Institute of Electrical Engineers, Vol. 64, No. 8, pp. 1137-1144, Aug. 2015. https://doi.org/10.5370/KIEE.2015.64.8.1137
  2. H. W. Jung, S. Y. Kim, K. B. Song, "Weekly maximum electric load forecasting method for 104 weeks using multiple regression models," The Journal of The Korean Institute of Electrical Engineers, Vol. 53. No. 9, pp. 1186-1191, Sep. 2014.
  3. S. Y. Kim, H. W. Jung, J. D. Park, S. M. Baek, W. S. Kim, K. H. Chon, and K. B. Song, "Weekly maximum electric load forecasting for 104 weeks by seasonal ARIMA model," Journal of the Korean Institute of IIIuminating and Electrical Installation Engineers, Vol. 28, No 1, pp. 50-56, Jan. 2014. https://doi.org/10.5207/JIEIE.2014.28.1.050
  4. M. K. Kim, and C. E. Hong, "The artificial neural network based electric power demand forecast using a season and weather informations," The Journal of the Korean Institute of Electrical Engineers, Vol. 53, No. 1, pp. 71-78, Jan. 2016.
  5. D. H. Shin, C. B. Kim, "A study on deep learning input pattern for summer power demand prediction," Journal of Korean Institute of Information Technology, Vol. 14, No. 11, pp. 127-134, Nov. 2016.
  6. H. S. Kim, H. C. Song, S. K. Ko, B. T. Lee, and J. W. Shin, "RNN-LSTM based short-term electricity demand forecasting using holiday information," in conference of the Institute of Eelectronics Engineers of Korea, Gwangju: Korea, pp. 552-555, Nov. 2016
  7. J. H. Lim, S. Y. Kim, J. D. Park, and K. B. Song, "Representative temperature assessment for improvement of short-term load forecasting accuracy," Journal of the Korean Institute of IIIuminating and Electrical Installation Engineers, Vol. 27, No. 6, pp. 37-43, May. 2013.
  8. F. A. Gers, N. N. Schraudolph, and J. Schmidhuber, "Learning precise timing with LSTM recurrent networks," Journal of Machine Learning Research 3, pp. 115-143, Mar. 2002.
  9. Christopher Olah, Understanding LSTM Networks, Github blog [Internet]. Available: http://colah.github.io/posts/2015-08-Understanding-LSTMs/