国際学会発表のお知らせ(M2大曽根） Notice of International Conference Presentation（M2 Osone)
7月3日～6日にフランス・パリで開催された ICECET 2025 (International Conference on Electrical, Computer and Energy Technologies) において、当研究室修士二年の大曽根佑紀が研究成果を発表しました。
発表タイトルは
「A Deep Reinforcement Learning-based Approach for Revenue Optimization in PV-Battery Storage Systems」
（太陽光発電＋蓄電池システムの収益最適化に向けた深層強化学習手法）です。
本研究では、再生可能エネルギーの普及に伴う課題の一つである「インバランスペナルティ（需給計画と実績の差に対する罰金）」に着目し、蓄電池の充放電スケジュールを深層強化学習により最適化しました。その結果、従来の手法に比べて 収益性を最大化しつつ、インバランスペナルティを大幅に削減できること を実証しました。
（なお、発表当日は大学院博士後期課程の入試と日程が重なってしまったため、特別にオンラインでの発表に切り替えて参加しました。この経験を通じて、「学会選定の段階からイベントや重要日程との重複に配慮することの大切さを学んだ」としています。）
これからの研究活動にも大いに期待です。
From July 3rd to 6th, Yuki Osone, a second-year master's student in our laboratory, presented his research findings at ICECET 2025 (International Conference on Electrical, Computer and Energy Technologies) held in Paris, France.
The title of the presentation is “A Deep Reinforcement Learning-based Approach for Revenue Optimization in PV-Battery Storage Systems”
In this study, he focused on “imbalance penalties” (penalties for discrepancies between supply and demand plans and actual results), which are one of the issues associated with the spread of renewable energy, and optimized the charging and discharging schedules of storage batteries using deep reinforcement learning. As a result, he demonstrated that it is possible to maximize profitability while significantly reducing imbalance penalties compared to conventional methods.
(On the day of the presentation, there was a conflict with the graduate school doctoral program entrance exam, so he switched to an online presentation to participate. Through this experience, he said, “I learned the importance of considering conflicts with events and important dates from the stage of selecting a conference.”)
Let's look forward to his future research activities!
The title of the presentation is “A Deep Reinforcement Learning-based Approach for Revenue Optimization in PV-Battery Storage Systems”
In this study, he focused on “imbalance penalties” (penalties for discrepancies between supply and demand plans and actual results), which are one of the issues associated with the spread of renewable energy, and optimized the charging and discharging schedules of storage batteries using deep reinforcement learning. As a result, he demonstrated that it is possible to maximize profitability while significantly reducing imbalance penalties compared to conventional methods.
(On the day of the presentation, there was a conflict with the graduate school doctoral program entrance exam, so he switched to an online presentation to participate. Through this experience, he said, “I learned the importance of considering conflicts with events and important dates from the stage of selecting a conference.”)
Let's look forward to his future research activities!