[논문]Short-term Load Forecasting of Buildings based on Artificial Neural Network and Clustering Technique

Ngo, Minh-Duc; Yun, Sang-Yun; Choi, Joon-Ho; Ahn, Seon-Ju

doi:10.7471/ikeee.2018.22.3.672

Short-term Load Forecasting of Buildings based on Artificial Neural Network and Clustering Technique 원문보기

전기전자학회논문지 = Journal of IKEEE, v.22 no.3, 2018년, pp.672 - 679

Ngo, Minh-Duc (Dept. of Electrical Engineering, Chonnam National University) , Yun, Sang-Yun (Dept. of Electrical Engineering, Chonnam National University) , Choi, Joon-Ho (Dept. of Electrical Engineering, Chonnam National University) , Ahn, Seon-Ju (Dept. of Electrical Engineering, Chonnam National University)

Abstract ▼ AI-Helper

Recently, microgrid (MG) has been proposed as one of the most critical solutions for various energy problems. For the optimal and economic operation of MGs, it is very important to forecast the load profile. However, it is not easy to predict the load accurately since the load in a MG is small and highly variable. In this paper, we propose an artificial neural network (ANN) based method to predict the energy use in campus buildings in short-term time series from one hour up to one week. The proposed method analyzes and extracts the features from the historical data of load and temperature to generate the prediction of future energy consumption in the building based on sparsified K-means. To evaluate the performance of the proposed approach, historical load data in hourly resolution collected from the campus buildings were used. The experimental results show that the proposed approach outperforms the conventional forecasting methods.

주제어

AI 본문요약
AI-Helper

* AI 자동 식별 결과로 적합하지 않은 문장이 있을 수 있으니, 이용에 유의하시기 바랍니다.

문제 정의

This paper described an artificial neural network approach combined with a clustering method and effective input feature selection to predict short-term energy consumption of buildings with highly nonlinear load profiles. The proposed method is computationally simple and also suitable for analyzing a large set of data whose pattern changes over time.

제안 방법

When there is knowledge of previous load values such as from the previous day and from the previous hour, it is easier to predict the load with good accuracy. In this paper, we generate five features from the past recorded load based on the availability of the recorded load dataset from our campus buildings. The features are previous hour load, previous 24 hours load, the average of three closest hour loads, the average of six closest hour load, and the average load of 24 hours before.
The past 24 hours data is extremely small data in comparison with other published researches which usually use past one week data. The mean absolute percentage error (MAPE) and root mean square error (RMSE) metrics are selected to evaluate the performance of the proposed method. The experimental results show that MAPE and RMSE of the proposed method are smaller than other conventional approaches.
This paper described an artificial neural network approach combined with a clustering method and effective input feature selection to predict short-term energy consumption of buildings with highly nonlinear load profiles. The proposed method is computationally simple and also suitable for analyzing a large set of data whose pattern changes over time. The forecasting model was applied to several campus building loads.
We evaluate the performance of different forecasting methods including MLR, SVM, RT, ANN without K-means, ANN combined with K-means for only load.

대상 데이터

The proposed methodology is tested on some campus building datasets. The evaluation metrics most frequently used to assess the performance of a model are MAPE and RMSE defined in equations (10) and (11), respectively.

이론/모형

In this paper, we proposed a novel approach based on sparsified K-means clustering and resilient back-propagation ANN (RBP ANN) algorithm. Firstly, all input variables including historical load, time, and weather variables are preprocessed and grouped based on the sparsified K-means clustering algorithm. Then, RBP ANN algorithm was used to forecast the load in each group.
In this paper, we propose to use two hidden layers neural networks with the resilient back propagation training algorithm. The inputs-outputs architecture of forecasting model employed in this study is shown in Figure 2.
Firstly, all input variables including historical load, time, and weather variables are preprocessed and grouped based on the sparsified K-means clustering algorithm. Then, RBP ANN algorithm was used to forecast the load in each group. The past 24 hours data including time calendar, temperature, and load are investigated to predict the electrical power at the current time.

성능/효과

Table 3 and Figure 6 show the performances and load profile of one day ahead forecast using various methods. It can be observed that the proposed method not only get the highest accuracy for a whole day but also reach the smallest error at every hour.

참고문헌 (12)

J. N. Zagrajek and R. Weron, "Modeling electricity loads in California: ARMA model with hyperbolic noise," Signal Processing, Vol. 82, No. 12, pp. 1903-1915, 2002. DOI:10.1016/S0165-1684(02)00318-3

상세보기
V. M. Bianco and S. Nardini, “Linear regression models to forecast electricity consumption in Italy,” Energy Sources, Part B: Economics, Planning and Policy, Vol. 8, No. 1, pp. 86-93, 2013. DOI:10.1080/15567240903289549

상세보기
G. Dudek, "Multivariate Regression Tree for Pattern-Based Forecasting Time Series with Multiple Seasonal Cycles," Proc. of 38th International Conference on Information Systems Architecture and Technology-ISAT 2017, pp. 85-94, 2017. DOI:10.1007/978-3-319-67220-5_8
R. K. Jain, K. M. Smith, P. J. Culligan and J. E. Taylor, "Forecasting energy consumption of multi-family residential buildings using support vector regression: Investigating the impact of temporal and spatial monitoring granularity on performance accuracy," Applied Energy, Vol. 123, pp. 168-178, 2014. DOI:10.1016/j.apenergy.2014.02.057

상세보기
M. A. Daut, M. Y. Hassan, H. Abdullah, H. A. Rahman, M. P. Abdullah and F. Hussin, "Building electrical energy consumption forecasting analysis using conventional and artificial intelligence methods: A review," Renewable and Sustainable Energy Reviews, 70, pp. 1108-1118, 2017. DOI:10.1016/j.rser.2016.12.015

상세보기
N. Ding, C. Benoit, G. Foggia, Y. Besanger and F. Wurtz, “Neural Network-Based Model Design for Short-Term Load Forecast in Distribution Systems,” IEEE Transactions on Power Systems, Vol. 31, No. 1, pp. 72-81, 2016. DOI:10.1109/TPWRS.2015.2390132

상세보기
F. Rodrigues, C. Cardeira and J. M. F Calado, "The Daily and Hourly Energy Consumption and Load Forecasting Using Artificial Neural Network Method: A Case Study Using a Set of 93 Households in Portugal," Energy Procedia, Vol. 62, pp. 220-229, 2014. DOI:10.1016/j.egypro.2014.12.383

상세보기
W. Mai, C. Y. Chung, T. Wu and H. Huang, "Electric load forecasting for large office building based on radial basis function neural network," In Proceedings of the 2014 IEEE PES General Meeting Conference & Exposition, pp. 1-5, 2014. DOI:10.1109/PESGM.2014.6939378
B. S. Kwon, R. J. Park and K. B. Song, “Analysis of Short-Term Load Forecasting Accuracy Based on Various Normalization Methods,” Journal of Korean Institute of Illuminating and Electrical Installation Engineers, Vol. 32, No. 06, pp. 30-33, 2018.
P. A. Farhad and B. Stephen, "Preconditioned Data Sparsification for Big Data with Applications to PCA and K-means," IEEE Transactions on Information Theory, Vol. 63, No. 5, pp. 2954-2974, 2017. DOI:10.1109/TIT.2017.2672725

상세보기
A. David and S. Vassilvitskii, "k-means++: The Advantages of Careful Seeding," Proceeding of the 18th Annual ACM-SIAM Symposium on Discrete Algorithms, 2007.
M. Riedmiller and H. Braun, "A direct adaptive method for faster backpropagation learning: The RPROP algorithm," In Proceedings of IEEE International Conference on Neural Networks, 1993. DOI:10.1109/ICNN.1993.298623

저자의 다른 논문 :

LOADING...

내보내기 구분	파일저장 인쇄 메일전송
구성항목	기본정보 상세정보 관리번호, 논문명, 저널/프로시딩명, 저자 , 발행년, 권, 호, 시작페이지, 끝페이지, 발행기관 관리번호, 논문명, 대등논문명, 저자 , 저널/프로시딩명, 발행기관, 발행년, 발행언어, 권, 호, 시작페이지, 끝페이지, ISBN, ISSN, 주제분야, 키워드, 초록(한글), 초록(영문), 저자(소속기관)
저장형식	Text(ASCII format) Excel format RefWorks Direct Export RIS format (for Reference Manager, ProCite, EndNote), Scholar's Aids, Mendeley
메일정보	받는사람 (필수) @ 보내는사람 (선택) @ 제목 내용 KISTI 검색결과 이메일 서비스
안내	총 건의 자료가 검색되었습니다. 다운받으실 자료의 인덱스를 입력하세요. (1-10,000) 검색결과의 순서대로 최대 10,000건 까지 다운로드가 가능합니다. 데이타가 많을 경우 속도가 느려질 수 있습니다.(최대 2~3분 소요) 다운로드 파일은 UTF-8 형태로 저장됩니다. 파일의 내용이 제대로 보이지 않을실 때는 웹브라우저 상단의 보기 -> 인코딩 -> 자동선택 여부를 확인하십시오. ~ Text(ASCII format) Excel format

연합인증