About Me
Dye the Sky.
I am Zhixu Duan (段之续), an undergraduate at SMEE (School of Mechanical and Electrical Engineering) of UESTC (University of Electronic Science and Technology, China) 
, major in mechanical engineering. Now as RA in Center for System Reliability and Safety, China (ReliaLab), advised by Zuoyi Chen and Hong-Zhong Huang. Also RA in EPICLab (Efficient and Precision Intelligent Computing Lab), SAI of SJTU (School of Artificial Intelligence, Shanghai Jiao Tong University) 
, advised by Linfeng Zhang.
Feel free to Ask me anything! Always keep a passion for exploring strange fields . Research interests focus on AI4Reliability including but are not limited to zero/few-shot learning, transfer learning, and llm.
🎓Education & Background
● 2025.11–Present, RA at EPICLab, SAI of SJTU
● 2023.12–Present, RA at ReliaLab, UESTC
● 2023.9–2027.9, B.S. at SMEE of UESTC.
● 2020.9–2023.9, In No.1 Middle School of Huairen, Shanxi Province.
🏆Achievements
● National Scholarship in 2025 Top 1.8%
● National Scholarship in 2024 Top 1.8%
● Ranked 1/83 in my major🥇 Top 1
● Research/Papers📃
🔥𝐏𝐚𝐫𝐚𝐥𝐥𝐞𝐥 𝐑𝐞𝐥𝐚𝐭𝐢𝐨𝐧 𝐍𝐞𝐭𝐰𝐨𝐫𝐤 𝐟𝐨𝐫 𝐈𝐧𝐭𝐞𝐥𝐥𝐢𝐠𝐞𝐧𝐭 𝐅𝐚𝐮𝐥𝐭 𝐃𝐞𝐭𝐞𝐜𝐭𝐢𝐨𝐧 𝐚𝐧𝐝 𝐋𝐨𝐜𝐚𝐥𝐢𝐳𝐚𝐭𝐢𝐨𝐧 𝐨𝐟 𝐓𝐫𝐚𝐢𝐧 𝐓𝐫𝐚𝐧𝐬𝐦𝐢𝐬𝐬𝐢𝐨𝐧 𝐒𝐲𝐬𝐭𝐞𝐦𝐬 𝐰𝐢𝐭𝐡 𝐙𝐞𝐫𝐨-𝐟𝐚𝐮𝐥𝐭 𝐒𝐚𝐦𝐩𝐥𝐞
⭐First Author and delivered an oral presentation to IEEE PHM 2024
TL;DR: A PRN model combining RSN and KAN is proposed for train transmission fault detection with few samples, achieving over 98% accuracy and strong generalization.
🔥𝐏𝐬𝐞𝐮𝐝𝐨-𝐟𝐚𝐮𝐥𝐭 𝐝𝐚𝐭𝐚 𝐞𝐧𝐡𝐚𝐧𝐜𝐞𝐝 𝐫𝐞𝐥𝐚𝐭𝐢𝐨𝐧 𝐧𝐞𝐭𝐰𝐨𝐫𝐤 𝐟𝐨𝐫 𝐟𝐚𝐮𝐥𝐭 𝐝𝐞𝐭𝐞𝐜𝐭𝐢𝐨𝐧 𝐚𝐧𝐝 𝐥𝐨𝐜𝐚𝐥𝐢𝐳𝐚𝐭𝐢𝐨𝐧 𝐢𝐧 𝐭𝐫𝐚𝐢𝐧 𝐭𝐫𝐚𝐧𝐬𝐦𝐢𝐬𝐬𝐢𝐨𝐧 𝐬𝐲𝐬𝐭𝐞𝐦𝐬
⭐First Author published in EAAI (SCI Q1)
TL;DR: A relation network using pseudo-fault data improves train transmission fault detection and localization performance.
🔥𝐃𝐞𝐜𝐨𝐮𝐩𝐥𝐢𝐧𝐠 𝐈𝐧𝐭𝐫𝐢𝐧𝐬𝐢𝐜 𝐅𝐚𝐮𝐥𝐭 𝐅𝐞𝐚𝐭𝐮𝐫𝐞𝐬 𝐟𝐫𝐨𝐦 𝐃𝐨𝐦𝐚𝐢𝐧 𝐕𝐚𝐫𝐢𝐚𝐭𝐢𝐨𝐧𝐬 𝐯𝐢𝐚 𝐃𝐨𝐦𝐚𝐢𝐧-𝐀𝐭𝐭𝐫𝐢𝐛𝐮𝐭𝐞 𝐅𝐮𝐬𝐢𝐨𝐧 𝐟𝐨𝐫 𝐔𝐧𝐬𝐞𝐞𝐧-𝐃𝐨𝐦𝐚𝐢𝐧 𝐅𝐚𝐮𝐥𝐭 𝐃𝐢𝐚𝐠𝐧𝐨𝐬𝐢𝐬
⭐First Author submitted manuscript to MSSP (SCI Q1)
TL;DR: A domain-attribute fusion model decouples intrinsic fault features, improving unseen-domain diagnosis robustness.
🔥𝐔𝐧𝐢𝐟𝐢𝐞𝐝 𝐇𝐞𝐚𝐥𝐭𝐡 𝐃𝐨𝐦𝐚𝐢𝐧 𝐑𝐞𝐥𝐚𝐭𝐢𝐨𝐧 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠 𝐟𝐨𝐫 𝐓𝐫𝐚𝐢𝐧 𝐓𝐫𝐚𝐧𝐬𝐦𝐢𝐬𝐬𝐢𝐨𝐧 𝐒𝐲𝐬𝐭𝐞𝐦𝐬 𝐅𝐚𝐮𝐥𝐭 𝐃𝐞𝐭𝐞𝐜𝐭𝐢𝐨𝐧 𝐮𝐧𝐝𝐞𝐫 𝐂𝐨𝐦𝐩𝐥𝐞𝐱 𝐎𝐩𝐞𝐫𝐚𝐭𝐢𝐧𝐠 𝐂𝐨𝐧𝐝𝐢𝐭𝐢𝐨𝐧𝐬
⭐Second Author (first author of adviser) submitted manuscript to IEEE TII (SCI Q1)
TL;DR: A unified health-domain relation learning approach enhances fault detection under complex operating conditions.
🔥𝐎𝐩𝐞𝐧-𝐒𝐞𝐭 𝐅𝐚𝐮𝐥𝐭 𝐃𝐢𝐚𝐠𝐧𝐨𝐬𝐢𝐬 𝐔𝐬𝐢𝐧𝐠 𝐂𝐋𝐈𝐏 𝐰𝐢𝐭𝐡 𝐅𝐨𝐫𝐰𝐚𝐫𝐝-𝐑𝐞𝐯𝐞𝐫𝐬𝐞 𝐑𝐞𝐚𝐬𝐨𝐧𝐢𝐧𝐠
⭐Second Author (first author of adviser) minor revision of COMPUT IND (SCI Q1)
TL;DR: A CLIP-based forward–reverse reasoning model enable for fault diagnosis
● Second author of an invention patent (first author of adviser), accepted for filing.
● Principal leader of a National Innovation and Entrepreneurship Program (College Student Project)
Leadership
● Vice President and founding core member of UESTC Interdisciplinary Association (IA).
1.Invited Professor Dezhong Yao (Founding Dean of UESTC’s School of Life Science and Technology) and brain scientist Professor Pedro Antonio Valdes Sosa as advisors
2.To promote the society’s outreach and research activities in fields such as bioinformatics, bioimaging, bioelectronics, and AI for Life Science.
3.Responsible for building and maintaining the association’s official website.
● Student Mentor for SMEE 2025 Freshmen
Mentored freshmen under the guidance of Counselor Keyu Chen
Talks & Reports
● 【SMEE idol】Zhixu Duan:”Take Each Step Steadily, and the Distance Will Unfold” –Interviewed by SMEE of UESTC
● 【UESTC-IA×DP Technology】Uni-Lab Developers’ Offline Workshop –Interviewed by DP Technology
Academic Positions
● 😏A key reader of Nature and a potential contributor to Science.
● 😎Attending major international conferences, and leading the field in tea break self-service acquisition and multimodal collection.