受賞国：日本国

本奨学金は、文部科学省（MEXT）により、九州大学における大学院修学のために授与されたものです。
研究生課程（2019年4月～2019年9月）、修士課程（2019年10月～2021年9月）、および博士課程（2021年10月～2024年9月）を支援しました。
在ヨルダン日本国大使館による推薦（大使館推薦）により採用されました。

受賞国：日本国

Since 2004, IGSES has honored their graduate students annually with the IGSES Awards for their outstanding achievement.

Winners are nominated based on the comprehensive evaluation of academic scores, research achievements, attitude toward research, and their master’s thesis and presentation.

受賞国：日本国

IEICES国際会議2022において、優れた研究発表に対して授与された賞。

担当区分：筆頭著者,　責任著者   記述言語：英語   掲載種別：研究論文（学術雑誌）   出版者・発行元：Energy and Buildings  

Smart meter data offers numerous opportunities for smart grid applications and insights, yet access to high-quality, large-scale data remains challenging due to issues with accessibility, data privacy, and the difficulties of setting up long-term data collection experiments. Generative adversarial network (GAN) models recently gained attention due to their generalizability and strong performance, often surpassing conventional models that require case-specific tuning and optimization. The DoppelGANger (DGAN), a recent multi-generator, multi-discriminator time series GAN model, has shown very promising performance in time series data synthesis but remains not well explored for electrical time series data. This study introduces a strategy to synthesize a 2-year hourly demand profile for a dwelling by training DGAN sub-models on monthly data to capture short-term patterns and aggregating their results to reproduce seasonal demand trends. The model was evaluated for dwellings with four different types of demand behaviors using various similarity measures between the real and synthesized data such as autocorrelation, hourly distributions, weekday distributions, top 2.5 % occurrences, load duration curves, as well as a thorough hourly demand percentile analysis. The results showed high performance in reproducing fully synthesized demand, especially capturing the autocorrelation as well as the hourly distribution. However, it was less accurate in learning the weekday distributions. A sensitivity analysis of the hourly percentile deviation for 5 different percentile ranges (<10 %, 10–25 %, 25–75 %, and > 90 %) showed that the model struggles more with matching the lower percentiles, especially (<10 %) in comparison to the highest and middle percentiles ranges.

DOI： 10.1016/j.enbuild.2025.116722

Scopus

担当区分：筆頭著者,　責任著者   記述言語：英語   掲載種別：研究論文（学術雑誌）   出版者・発行元：Energy and Buildings  

Decoding and unveiling the complex behaviors of electrical demand are vital steps toward achieving optimal power planning and operation of the power grid. Residential demand is stochastic and unique; hence, discovering unique demand behavior patterns and characteristics is essential for achieving sustainability targets in the residential energy sector and zero-energy housing through improved renewable optimization, building energy management, and policy making. While such endeavors have already been made in recent research, little to no research has extended beyond the total demand of a dwelling at the main meter level to unveil behind-the-meter energy demand behavior correlations of many appliances and dwelling space demands, primarily owing to the difficulty in acquiring such data for long periods and for a large number of dwellings. In this study, the 2-year minutely sampled electrical demand data of 479 dwellings in a large residential complex in Osaka, Japan, with 18 behind-the-meter energy demand attributes (BTMEDAs) of appliances, dwelling space energy demand, and water and gas consumption, are analyzed to (a) investigate the seasonal diversity of demand behaviors at the main meter level, namely, total household demand (THD) and BTMEDAs. (b) Quantify the correlation between the THD of each dwelling and various BTMEDAs through a multi-method approach to determine the most impactful appliances/spaces on global electrical demand behaviors across different seasons. (c) Finally, we demonstrate how the cluster encoding method can enhance user data privacy while preserving the key association of various BTMEDAs characteristics with THD. The results depicted that 6% of the users have Very High (VH) energy demand behavior, 26% High (H), 35% Medium (M), and 33% Low (L), although behavior proportions change based on seasons. Furthermore, across all seasons, the living room's outlet was a significant contributor to THD behaviors with a weight of 20.1%, followed by its AC with 15 % weight, and the Hallway and WC lighting outlets with 9.15%, despite having low average demand.

DOI： 10.1016/j.enbuild.2025.116147

Scopus

記述言語：日本語  

DOI： https://doi.org/10.18948/shase.49.330_1

担当区分：筆頭著者,　最終著者,　責任著者   出版者・発行元：E3s Web of Conferences  

The outbreak of the SARS-CoV-2 virus forced people to work from home. This study aimed to examine the relationship between residents' mental stress, indoor environment quality (IEQ), preventive behaviors, and socioeconomic status (SES) in Indonesia by using a cross-sectional study with a questionnaire survey in Indonesia. A total of 1004 valid responses were obtained during the survey during the COVID-19 pandemic period (November-December 2021). Logistic regression and odds ratio (OR) was used to evaluate the association between the possibility of mental stress and sleeplessness relying on the classified group of income, education, and age. In addition, a structural equation model (SEM) was used to analyze the inter-relationship between these characteristics and their effects on mental stress and sleeplessness as a crisis variable. The results indicate that mental stress was more inclined among low-income households during the COVID-19 pandemic than middle-up and high-income groups, with OR = 0.48 and 0.50, respectively. Moreover, the SEM suggested that SES also had significant direct effects (p-value < 0.05) on preventive behaviors (ω = 0.105), IEQ (ω =0.102), and crisis (ω = -0.237). It evidenced that the higher socioeconomic levels could have less possibility of experiencing a crisis. The findings of this study could add to practical implications that support the researchers and public policy stakeholders in mitigating the long-term effect of COVID-19 in Indonesia related to mental health and indoor environments.

DOI： 10.1051/e3sconf/202339601020

Scopus

担当区分：筆頭著者   記述言語：英語   掲載種別：研究論文（学術雑誌）   出版者・発行元：Energies  

Owing to the high energy demand of buildings, which accounted for 36% of the global share in 2020, they are one of the core targets for energy-efficiency research and regulations. Hence, coupled with the increasing complexity of decentralized power grids and high renewable energy penetration, the inception of smart buildings is becoming increasingly urgent. Data-driven building energy management systems (BEMS) based on deep reinforcement learning (DRL) have attracted significant research interest, particularly in recent years, primarily owing to their ability to overcome many of the challenges faced by conventional control methods related to real-time building modelling, multi-objective optimization, and the generalization of BEMS for efficient wide deployment. A PRISMA-based systematic assessment of a large database of 470 papers was conducted to review recent advancements in DRL-based BEMS for different building types, their research directions, and knowledge gaps. Five building types were identified: residential, offices, educational, data centres, and other commercial buildings. Their comparative analysis was conducted based on the types of appliances and systems controlled by the BEMS, renewable energy integration, DR, and unique system objectives other than energy, such as cost, and comfort. Moreover, it is worth considering that only approximately 11% of the recent research considers real system implementations.

DOI： 10.3390/en15228663

Scopus

記述言語：英語   著書種別：学術書

DOI： https://doi.org/10.1007/978-981-16-0135-4_2