🗞️

News and Events

Conferences

Participation in 2023 SAREK Winter Conference

Date : 2023.11.24

Location : The Korea Science and Technology Center, Seoul

외피 단열재 위치에 따른 동적 정상상태 엑서지 분석법의 적절성 presented by @Habin Jo

복사와 대류 엑서지 전달을 고려한 건물 외피에 대한 계절별 비정상상태 엑서지 분석 presented by

Presented papers

by Sieun Kim

by Habin Jo

2023 SAREK Winter Conference

2023/11/27

Conferences

Participation in 2023 Autumn KIAEBS Conference

Date : 2023.11.10

Location : Pusan National University, Busan

지열교환기 설계 파라미터 불확실성에 기인한 지열교환기 출구 유체 온도 변동성 분석 presented by

균일 물질의 열확산계수 및 시스템 길이에 따른 동적 엑서지 분석법의 적절성 검토 presented by @Habin Jo

Presented papers

2023 KIAEBS Autumn Conference

2023/11/10

Conferences

Participation in 2023 KSES Autumn Conference

Date : 2023.11.6~2023.11.08

Location : Sono Calm Convention Center, Yeosu

Wonjun, Euntak, Sieun and Suhyun

전역 민감도 분석 기반의 지열교환기 설계 길이 불확실성 분석 presented by @Euntak Shin

복사와 대류를 고려한 건물 외피에 대한 비정상상태 엑서지 분석 presented by

Presented papers

by Euntak Shin

by Sieun Kim

2023 KSES Autumn Conference

2023/11/07

Conferences

Participation in 2023 Autumn KSGHE Conference

Date: 2023.10.12

Location: Chosun University, Gwangju

ASHRAE 지열교환기 설계법에 대한 전역 민감도 분석 @Euntak Shin

토양 열응답 함수 불확실성에 의한 지열히트펌프 입구 유체 온도 변화의 확률론적 분석

The latest research trends and technological advancements in the geothermal and hydrothermal fields were actively shared at the 2023 KSGHE Autumn Conference. There were a total of 33 oral presentations and 43 poster presentations.

Me and Yoonseong presented in the Ground Source Heat Pump session. It was a unique experience because we had previously presented in the same session with other researchers in same field at another conferences like SAREK.

In the past, I couldn't fully concentrate on the presentations of other researchers in the same field because of nervousness about my upcoming presentation. However, meeting them again in the same session made me more focused on their research methodologies and directions.

Moreover, I felt that I could learn more about the aspects I didn't know through the presentations of others.

I also felt that it was essential to pay close attention to the feedback that follows the presentations. Knowing what questions other researchers have about my research can provide valuable hints for preparing future presentations and is an excellent opportunity to critically evaluate what I might have missed. Therefore, even if busy, active participation in conferences is considered essential for expanding one's research capabilities.

After the conference, thanks to Professor Wonjun Choi's consideration, we had the opportunity to participate in a dinner. Although it felt a bit awkward, I was delighted to receive advice and words of encouragement from senior researchers in similar fields during the meal.

Additionally, it was a time to realize the importance of networking through interactions with various professors and industry professionals.

Written by: Euntak Shin

Presented papers

2023 KSGHE Autumn Conference

2023/10/12

Conferences

 by Wonjun Choi

•

 by Euntak Shin

2022 KIAEBS Autumn conference

2022/11/25

Conferences

Participation in 2022 Autumn KSES

Researchers at the University of Seoul and members of the betlab

We attended the Autumn Conference of Korea Solar Energy Society event on Jeju Island, which took place from October 19 to 21, 2022

Presented paper

•

 by Wonjun Choi

2022 KSES Autumn conference

2022/10/19

Conferences

Participation in 2022 Summer SAREK

왼쪽부터 차례대로 김윤성, 신은탁, 최원준교수님, 김윤선, 조하빈

From June 22nd to 24th 2022, we attended the Summer Conference of the Korean Society of HVAC (Heating, Ventilation and Air Conditioning) held at Phoenix Park in Pyeongchang, Gangwon Province."

Presented paper

•

 by Wonjun Choi

2022 SAREK Summer conference

2022/06/22 → 2022/06/24

Conferences

Papers

A new paper titled, has been published in Energy and Buildings.

Paper link

Interpretable deep learning model for load and temperature forecasting: Depending on encoding length, models may be cheating on wrong answers

Data-driven time-series forecasting models for model predictive control of building energy systems have an autoregressive structure that uses historic…

https://www.sciencedirect.com/science/article/pii/S0378778823006400?via=ihub

Summary

Data-driven time-series forecasting models for model predictive control of building energy systems have an autoregressive structure that accepts historical data as input. One customary practice is to encode the past 24 h of data to extract daily periodicity information when forecasting 24 h ahead. However, deep learning models have different forecasting mechanisms than traditional time-series models that decompose periodicity. This study focuses on the impact of encoding length in deep learning models used for building energy demand forecasting.

To investigate this, we developed an interpretable deep learning model using an attention mechanism and gated recurrent units. We tested three different encoding lengths: 8 hours, 24 hours, and 168 hours for 24-h-ahead forecasting of zone temperatures and loads.

The findings revealed that the choice of encoding length significantly impacts the model's forecasting behavior and performance. The 24-hour and 168-hour models rely heavily on daily periodicity information, while the 8-hour model depends more on recent state information. This makes the latter more adaptable to sudden changes, providing better accuracy when boundary conditions deviate from typical daily patterns.

The study suggests treating encoding length as a crucial hyperparameter to be optimized based on target characteristics, environmental conditions, and model architecture. It also highlights that an attention weight map can explain a model's inference mechanism which can help improve its performance by adjusting factors like encoding length.

Featured figures

Attention maps of three different encoding length models; (a) 24-h model, (b) 8-h model, and (c) 168-h model

Interpretable deep learning model for load and temperature forecasting: Depending on encoding length, models may be cheating on wrong answers

2023/08/11

Papers

A new paper titled has been published in Case Studies in Thermal Engineering.

Paper link

Real-time probabilistic backfill thermal property estimation method enabling estimation convergence judgment

This study proposes a real-time probabilistic estimation method for determining the thermal properties of backfill used in underground power transmiss…

https://www.sciencedirect.com/science/article/pii/S2214157X23004148?via=ihub

Summary

This paper presents a novel real-time probabilistic estimation method for determining the thermal properties of backfill material used in underground power transmission lines and ground heat exchangers. The proposed method combines a fast linear analytical model, Bayesian inference, and Jensen-Shannon divergence to enable real-time sequential estimations during field experiments, quantify the estimation uncertainty, and determine the estimation convergence. The method can capture the contextual information affecting the estimation uncertainty, such as the quality of the experiment and the construction state of the backfill. Thus the application of the developed estimation method can lead to significant cost savings by avoiding unnecessary and prolonged experiments.

Featured figures

Illustration of real-time sequential estimation of the probability density function for the backfill thermal conductivity, along with the temporal change in Jensen-Shannon divergence.

Real-time probabilistic backfill thermal property estimation method enabling estimation convergence judgment

2023/06/27

Papers

A new paper titled, has been published.

Paper link

Performance evaluation of deep learning architectures for load and temperature forecasting under dataset size constraints and seasonality

Buildings and their energy systems have unique characteristics and temporally changing dynamics. Additionally, building data have strong seasonality. …

https://www.sciencedirect.com/science/article/pii/S0378778823002578?via=ihub

Summary

Buildings and their energy systems are characterized by unique and complex features that can evolve over time. Moreover, building data is highly seasonal, which means that it is subject to variations according to the time of the year. Therefore, an effective deep learning model with exceptional adaptability (i.e., high performance with little data) is required to accurately forecast and control the energy systems of buildings.

Considering the intricate nature of the building-energy field, we conducted a benchmark study to compare the performance of six different deep learning architectures. These included the multilayer perceptron (MLP), simple recurrent neural network (RNN), long short-term memory, gated recurrent unit (GRU), dilated convolutional neural network (DCNN), and transformer. To analyze the effect of data seasonality on forecasting performance, we also developed a data similarity analysis method.

To ensure the reproducibility and accessibility of our benchmark, we utilized a publicly accessible data generator and the open-source Python library DeepTimeSeries. Our forecasting targets were the zone temperatures and thermal loads over a future 24-hour period. The benchmark results, with varying training dataset sizes ranging from 0.3 to 0.9 y, showed that the transformer architecture performed the best, especially on small training datasets. The GRU and RNN came in second and third place, respectively, while the rankings of other architectures varied depending on the training dataset size and forecasting targets.

Additionally, the data similarity analysis revealed that simply increasing the training dataset size does not necessarily improve the model performance. This highlights the importance of model training with highly similar data for the forecasting period.