날씨 변수를 고려한 전력 수요량 예측구간 시계열 분석 연구 = Time series analysis for prediction intervals of electricity demand with weather variables

국문 초록 (Abstract)

연간 전력 수요량은 꾸준히 증가하는 추세이다. 전력 수요량이 늘어나는 만큼 국내 전력 의존도 또한 높아지고 있어 전력 수요 예측의 중요성이 부각되고 있다. 본 연구는 계절적 특성과 날씨를 고려하여 전력 수요량을 예측 분석하고자, 시계열 모형 연구와 이를 통한 예측값 및 예측구간에 대해 연구를 수행한다. ARIMA, SARIMA, ARIMAX, SARIMAX 모형에 각각 전력 수요량 월별 데이터를 적합하여 최적 차수를 찾고 표본외 예측분석(out-of-sample forecasting)을 수행하여 예측값의 RMSE(root mean squared error; 평균제곱근오차), MAE(mean absolute error; 평균절대오차), 평균절대백분율오차(MAPE)와 평균제곱로그오차(MSLE)를 계산하고 예측성능을 비교한다. 최적의 전력 수요량 예측 모형은 냉방도일(CDD), 난방도일(HDD), 일사량을 고려한 SARIMAX (2,1,1)(2,1,2,12) 모형으로 선정되었다. 예측구간의 엄밀한 분석을 위하여 정규분포, 라플라스분포 및 Bootstrap 방법을 이용한 80%, 95% 예측구간을 추정한다. 각 예측구간의 포함확률(coverage probability, CP), 평균길이(average length, AL), 평균구간점수(mean interval score, MIS)를 계산하여 비교 분석한다. 본 연구는 세 가지 방법으로 접근하여 예측구간 추정 결과를 도출하고 비교했다는 점에서 차별성이 있다. 본 연구에서의 전력 수요량 시계열 모형 및 예측구간 추정 방법은 효율적인 저비용 전력 공급 시스템 운영에 실질적 도움이 되기를 기대한다.

다국어 초록 (Multilingual Abstract)

Annual electricity demand is steadily increasing and the importance of forecasting the electricity demand is emerging since the dependence on domestic electricity is also increasing. The purpose of this study is analysis of the electricity demand considering seasonality and weather variables, for which we study time series modelling and forecasting analysis. ARIMA, SARIMA, ARIMAX, SARIMAX models are handled with optimally selected orders. Out-of-sample forecasting analysis is carried out to evaluate predicted values and their errors such as RMSE, MAE, MAPE and MSLE. The errors by four models are compared. As the optimal model, SARIMAX(2,1,1) (2,1,2,12) model with CDD, HDD and insolation has been chosen. For strict statistical analysis of prediction intervals, normal distribution, laplace distribution and bootstrap method are adopted to construct 80%, 95% prediction intervals. Coverage probability (CP), average length (AL) and mean interval score (MIS) of the prediction intervals by three approximations are evaluated. This study contributes with novelty in that it provides three approaches of prediction intervals along with their comparison. The results of this study are expected to be of practical help in the efficient operation of electricity supply system.

참고문헌 (Reference)

1 Box G. E. P., "Time series analysis: forecasting and control" Princeton-Hall International 1994

2 장성연, "Time series analysis of electricity demand and long-term forecast" 25 (25): 1363-1376, 2023

3 Durbin J., "Testing for serial correlation in least squares regression" 37 (37): 409-428, 1950

4 Gneiting T., "Strictly proper scoring rules, prediction, and estimation" 102 : 359-378, 2012

5 Amjady N., "Short-term hourly load forecasting using time-series modeling with peak load estimation capabaility" 16 (16): 498-505, 2001

6 Datta G., "Semi-parametric prediction intervals in small areas when auxiliary data are measured with error" 28 (28): 2309-2335, 2018

7 홍창곤 ; 이상진, "Robust interval estimation using density power divergence" 24 (24): 1-10, 2022

8 김재휘 ; 김재희, "Prediction of electricity consumption in a hotel using ensemble learning with temperature" 32 (32): 319-330, 2019

9 Hwang E., "Prediction intervals of the COVID-19 cases by HAR models with growth rates and vaccination rates in top eight affected countries: bootstrap improvement" 155 : 2022

10 Grillo M., "Prediction intervals for neural network models using weighted asymmetric loss functions"

11 Park J., "Poaer consumption prediction algorithm development based on the machine learning by considering the impact of COVID-19" 211-212, 2021

12 Jo Y., "Performance Comparison and Evaluation of Recurrent Neural Network based Learning for Electricity Demand Forecasting" Graduate School Kumoh National Institute of Technology 2020

13 문형빈 ; 안재균 ; 이철용, "Forecasting electricity demand of jeju insland using machine learning" 18 (18): 49-63, 2023

14 Chowell G., "Ensemble bootstrap methodology for forecasting dynamic growth processes using differential equations: application to epidemic outbreaks" 21 (21): 2021

15 정상욱 ; 김삼용, "Electricity demand forecasting for daily peak load with seasonality and temperature effects" 27 (27): 843-853, 2014

16 이지원 ; 김형준 ; 김문겸, "Design of short-term load forecasting based on ANN using bigdata" 69 (69): 792-799, 2020

17 권숙희 ; 김재훈 ; 손석만 ; 이성덕, "Comparison of time series predictions for maximum electric power demand" 34 (34): 623-632, 2021

18 Thombs LA., "Bootstrap prediction intervals for autoregression" 85 (85): 486-492, 1990

19 Efron B., "Bootstrap methods : another look at the jackknife" 7 (7): 1-26, 1979

20 이정현 ; 오상진 ; 윤여창 ; 안영호 ; 김재성 ; 조완섭 ; 이성덕, "Analysis of time series to support decision making on V2G using energy consumption data" 30 (30): 401-414, 2019

21 Zhixia Zou ; 이광봉, "An empirical test for temperature forecasting models" 14 (14): 91-100, 2012

22 Peters P., "A time series modelling approach to forecasting and interpolation" 2 (2): 1-22, 1980

23 Kim TJ., "A study on measurement method for solar heat gain coefficient of fenestration and shading device under solar simulator" Graduate School of Hanyang University 2023

24 양준모 ; 정민수 ; 조인숙, "A study on electricity demand prediction model by periodicity characteristics of time series data" 19 (19): 1991-2003, 2017

25 박종배 ; 심상우 ; 이다한 ; 노재형, "A machine learning based algorithm for short-term weekends load forecasting" 71 (71): 1578-1584, 2022