[논문]제조업 전력량 예측 정확성 향상을 위한 Double Encoder-Decoder 모델

조영창; 고병길; 성종훈; 조영식

doi:10.3745/ktsde.2020.9.12.419

제조업 전력량 예측 정확성 향상을 위한 Double Encoder-Decoder 모델
Double Encoder-Decoder Model for Improving the Accuracy of the Electricity Consumption Prediction in Manufacturing 원문보기

정보처리학회논문지. KIPS transactions on software and data engineering. 소프트웨어 및 데이터 공학, v.9 no.12, 2020년, pp.419 - 430

조영창 ((주)에스더블유엠 부설연구소) , 고병길 ((주)에스더블유엠 부설연구소) , 성종훈 ((주)에스더블유엠 부설연구소) , 조영식 ((주)AMEP 기술연구소)

초록
AI-Helper

본 연구는 기존 전력량 예측 모델의 구조를 변경하여 모델의 예측 능력을 향상 시킬 수 있는 방법에 관하여 연구하였다. 전기에 대한 수요는 그 어느 때보다 증가하고 있다. 산업 부문에서는 그 어느 부문 보다 전기 소모량이 많음으로, 더욱 정확한 공장 지역의 전력량 소모 예측 모델이 잉여 에너지 생산을 줄이기 위해 주목을 받고 있다. 우리는 2개의 개별 encoder와 한개의 decoder를 사용하여, 장기와 단기 데이터를 모두 사용하는 double encoder-decoder 모델을 제안한다. 우리는 제안된 모델을 세홍(주)의 생산 구역에서 2019년 1월 1일부터 2019년 6월 30일 까지 모집된 전력 소모량 데이터에서 평가 하였다. double encoder-decoder 모델은 기존의 encoder-decoder 모델을 사용했을 때와 비교하여 약 10 %의 평균 절대 비율 오차의 감소를 기록 하였다. 본 결과는 제안한 모델이 encoder-decoder 모델에 비해 생산 지역의 전력 사용량의 예측을 더 정확하게 하는 모델임을 보여준다.

Abstract ▼ AI-Helper

This paper investigated methods to improve the forecasting accuracy of the electricity consumption prediction model. Currently, the demand for electricity has continuously been rising more than ever. Since the industrial sector uses more electricity than any other sectors, the importance of a more precise forecasting model for manufacturing sites has been highlighted to lower the excess energy production. We propose a double encoder-decoder model, which uses two separate encoders and one decoder, in order to adapt both long-term and short-term data for better forecasts. We evaluated our proposed model on our electricity power consumption dataset, which was collected in a manufacturing site of Sehong from January 1st, 2019 to June 30th, 2019 with 1 minute time interval. From the experiment, the double encoder-decoder model marked about 10% reduction in mean absolute error percentage compared to a conventional encoder-decoder model. This result indicates that the proposed model forecasts electricity consumption more accurately on manufacturing sites compared to an encoder-decoder model.

주제어

표/그림 (5)

그림 Fig. 1. Stacked RNN Computation Graph. Each Arrow Represents the Computational Flow. Dashed Arrows Represent Skipped Computation in the Graph. Inputs are used in RNN Computation to Produce hidden States. Superscript in hidden States Notation Represent the Order of the Layer, and Subscript Represents the Time Step
그림 Fig. 2. Flow Diagram of Encoder-decoder Model, Blue Squares Represent Inputs to the Encoder, and Red Squares Represent the Output of the Decoder. With the Stacked RNN Computation, Inputs to the Encoder are Transformed to Hidden States, which is then used as Initial Hidden State of the Decoder. Decoder Forecasts Current Output Given Previous Hidden States and Previous Output Value
그림 Fig. 3. Computational Graph of Double Encoder-decoder. It is the Same as Fig. 2, Except that it has One Additional Encoder Unit. The Last Hidden States from both Encoders are used as Initial Hidden States in the Decoder after Concatenation
표 Table 1. Descriptions of Sehong Dataset
표 Table 2. Comparison between Double Encoder-decoder and Double Encoder-decoder with Attention

AI 본문요약
AI-Helper

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

문제 정의

In this paper, we examine the effectiveness of recent advancements in deep learning approaches on electricity demand forecasting tasks. We firstly adapt the encoder-decoder model with LSTM layers as proposed in [11, 12], and then make some modifications on encoder units for better forecasting.
We used the electricity usage in the manufacturing room as a target variable, since our main object of the research was to forecast electricity usage on the manufacturing site based on previous historical observations. As shown in Table.

제안 방법

Our proposed model is expected to incorporate with both long term and short term measurements in order to predict accurate forecasts based on both long-term trend and short-term momentum of the dataset. Secondly, we apply the attention mechanism [22, 23] on our proposed model to align two context vectors from encoders for decoder by assigning appropriate weights.
. We fitted our parameters with MAE loss, but also tracked other metric, MAPE, for convenient interpretation between models.

대상 데이터

When training the model, several hyperparameters needed to be tuned. For the hidden unit size of each LSTM layer, we examined the hidden unit size between 64, 128, and 256 units, with two layers for each encoder and decoder unit. In order to optimize the parameters in the LSTM layers, we chose the ADAM optimizer [29] with the learning rate exponentially decaying from 0.

이론/모형

Our results were calculated using only the test dataset. For the baseline model, we used the conventional encoderdecoder model [11, 12] as a baseline model.

성능/효과

In contrast to [2], where the study was performed from the perspective of overall electricity usage in a certain area, the study of electricity consumption of individual households was conducted in [3]. Empirical mean, k-nearest neighbor, neural networks, and LSTM were benchmarked, and LSTM yielded the best accuracy in terms of MAPE. One of the interesting results from [3] was that forecasting the aggregated electricity usage yielded higher MAPE than summing results from forecasting individuals' usage.
Our DED model achieved about 10 percent reduction on percentage error compared to the baseline model’s MAPE, which scored 50.47% in their best hyperparameter settings. Also, DED scored lower MAE compared to the baseline model’s MAE, 0.
In [2], LSTM was used to calculate a short-term load for the smart grid. Their results showed that LSTM achieved far better accuracy compared to both conventional machine learning approaches, such as neural network, SVR, and traditional time series forecasting methods like seasonal ARIMA. In contrast to [2], where the study was performed from the perspective of overall electricity usage in a certain area, the study of electricity consumption of individual households was conducted in [3].

참고문헌 (37)

Y. Cho, Energy info. Korea. Kyonggi-do, Korea: Korea Energy Economics Institute, 2018.
J. Zheng, C. Xu, Z. Zhang, and X. Li, "Electric load forecasting in smart grids using Long-Short-Term-Memory based Recurrent Neural Network," 2017 51st Annual Conference on Information Sciences and Systems (CISS), 2017.
W. Kong, Z. Y. Dong, Y. Jia, D. J. Hill, Y. Xu, and Y. Zhang, "Short-term residential load forecasting based on LSTM recurrent neural network," IEEE Transactions on Smart Grid, Vol.10, No.1, pp.841-851, 2019.

상세보기
J. Bedi and D. Toshniwal, "Deep learning framework to forecast electricity demand," Applied Energy, Vol.238, pp. 1312-1326, 2019.

상세보기
R. K. Agrawal, F. Muchahary, and M. M. Tripathi, "Long term load forecasting with hourly predictions based on long-short-term-memory networks," 2018 IEEE Texas Power and Energy Conference (TPEC), 2018.
G. E. P. Box and G. M. Jenkins, Time series analysis: forecasting and control. Oakland: Holden-Day, 1976.
E. Erdogdu, "Electricity demand analysis using cointegration and ARIMA modelling: A case study of Turkey," Energy Policy, Vol.35, No.2, pp.1129-1146, 2007.

상세보기
K. He, X. Zhang, S. Ren, and J. Sun, "Deep residual learning for image recognition," 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016.
J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, "You only look once: Unified, real-time object detection," 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016.
G. Huang, Z. Liu, L. V. D. Maaten, and K. Q. Weinberger, "Densely connected convolutional networks," 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017.
Sutskever, Ilya, Oriol Vinyals, and Quoc V. Le, "Sequence to sequence learning with neural networks," In Advances In Neural Information Processing Systems, pp.3104-3112. 2014.
K. Cho, B. V. Merrienboer, C. Gulcehre, D. Bahdanau, F. Bougares, H. Schwenk, and Y. Bengio, "Learning phrase representations using RNN encoder-decoder for statistical machine translation," Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2014.
I. Goodfellow, Y. Bengio, and A. Courville, Deep learning. Cambridge, MA: MIT Press, 2017.
E. E. Elattar, J. Goulermas, and Q. H. Wu, "Electric Load Forecasting Based on Locally Weighted Support Vector Regression," IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews), Vol.40, No.4, pp.438-447, 2010.

상세보기
G. Dudek, "Short-Term Load Forecasting Using Random Forests," Advances in Intelligent Systems and Computing Intelligent Systems 2014, pp.821-828, 2015.
T. He, Z. Dong, K. Meng, H. Wang, and Y. Oh, "Accelerating Multi-layer Perceptron based short term demand forecasting using Graphics Processing Units," 2009 Transmission & Distribution Conference & Exposition: Asia and Pacific, 2009.
A. Graves, Supervised sequence labelling with recurrent neural networks. Berlin: Springer, 2012.
Hochreiter, Sepp, and Jurgen Schmidhuber, "Long short-term memory," Neural Computation, Vol.9, No.8, pp.1735-1780, 1997.

상세보기
F. Rosenblatt, "The perceptron: A probabilistic model for information storage and organization in the brain," Psychological Review, Vol.65, No.6, pp.386-408, 1958.

상세보기
M. Leshno, V. Y. Lin, A. Pinkus, and S. Schocken, "Multilayer feedforward networks with a nonpolynomial activation function can approximate any function," Neural Networks, Vol.6, No.6, pp.861-867, 1993.

상세보기
Y. Bengio, A. Courville, and P. Vincent, "Representation Learning: A Review and New Perspectives," IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol.35, No.8, pp.1798-1828, 2013.

상세보기
T. Luong, H. Pham, and C. D. Manning, "Effective Approaches to Attention-based Neural Machine Translation," Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, 2015.
D. Bahdanau, K. Cho, and Y. Bengio, "Neural Machine Translation by Jointly Learning to Align and Translate," International Conference on Learning Representations, 2015.
J. Cheng, L. Dong, and M. Lapata, "Long Short-Term Memory-Networks for Machine Reading," Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, 2016.
A. Parikh, O. Tackstrom, D. Das, and J. Uszkoreit, "A Decomposable Attention Model for Natural Language Inference," Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, 2016.
L. Breiman, "Random Forests," Machine learning, Vol.45, No. 1, pp. 5-32, 2001.
Drucker, Harris, Christopher JC Burges, Linda Kaufman, Alex J. Smola, and Vladimir Vapnik, "Support vector regression machines," In Advances in Neural Information Processing Systems, pp. 155-161. 1997.
El Hihi, Salah, and Yoshua Bengio, "Hierarchical recurrent neural networks for long-term dependencies," In Advances in Neural Information Processing Systems, pp.493-499. 1996.
D. P. Kingma and J. Ba, "Adam: A method for stochastic optimization," 3rd International Conference for Learning Representations, 2015.
L. Bottou, "On-line Learning and Stochastic Approximations," On-Line Learning in Neural Networks, pp. 9-42, 1999.
T. Hastie, J. Friedman, and R. Tisbshirani, The Elements of statistical learning: data mining, inference, and prediction. New York: Springer, 2017.
D. Barber, Bayesian reasoning and machine learning. Cambridge: Cambridge University Press, 2018.
S.-Y. Shih, F.-K. Sun, and H.-Y. Lee, "Temporal pattern attention for multivariate time series forecasting," Machine Learning, Vol.108, No.8-9, pp.1421-1441, 2019.

상세보기
G. Lai, W.-C. Chang, Y. Yang, and H. Liu, "Modeling Long- and Short-Term Temporal Patterns with Deep Neural Networks," The 41st International ACM SIGIR Conference on Research & Development in Information Retrieval, 2018.
M. Abadi, "TensorFlow: learning functions at scale," Proceedings of the 21st ACM SIGPLAN International Conference on Functional Programming - ICFP 2016, 2016.
Hall, Mark A. "Correlation-based Feature Selection for Machine Learning," PhD diss., The University of Waikato, 1999.
Song, Fengxi, Zhongwei Guo, and Dayong Mei. "Feature selection using principal component analysis," In 2010 International Conference on System Science, Engineering Design and Manufacturing Informatization, Vol.1, pp.27-30. IEEE, 2010.

표제어: PCR

동의어: Packet Collision Rate

용어 설명 출처 목록 (6)

용어 설명: PCR은 세균 특이성이 있는 primer를 이용하여 적은 수의 세균이 있을지라도 쉽게 검출할 수 있는 유용한 방법이며, 이를 이용하여 구강 내 치면세균막이나 타액에서 직접 세균을 검출할 수 있게 되었다[8].

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