Fault Detection and Diagnosis Software of LHAASO
H. Zhang, M. Gu and S. Fan, "Fault Detection and Diagnosis Software of LHAASO," in IEEE Transactions on Nuclear Science, doi: 10.1109/TNS.2024.3454806.
======
Education
- B.Eng. in Computer Science, Lanzhou University, 2016
- Ph.D. student in Institute of Hign Energy Physics / University of Chineses Academy of Science
Research Interests
- High Performance (TB/s) Computing Architecture
- Memory Disaggregation
- Fault Diagnosis
- AI for Science, Computing for Science.
- Web Technology
Research Highlights
- Fault monitoring and diagnosis in large-scale distributed clusters: H. Zhang, M. Gu and S. Fan, “Fault Detection and Diagnosis Software of LHAASO,” in IEEE Transactions on Nuclear Science, doi: 10.1109/TNS.2024.3454806.
- Open-source large-scale collaboration member management system, with over 3000 users: Link
- High-performance, efficient, and reliable distributed memory cache pool: in progress.
Professional experience
- 2021.5-2021.8 LHAASO Electronic Control System Development
Mainly responsible for JAVA Spring Cloud development, responsible for writing RESTful services to be provided to the front end.
- 2022.09-2023.04 Member Management System
Developed the member management system for the science experiment collaborations. The system uses NextJS for the front-end and SpringBoot for the back-end, and I am responsible for writing all the code for both the front-end and the back-end. Currently used in three large-scale international collaborations, with more than 3,000 users.
- 2022.12-2023.12 Fault Monitoring and Diagnosis System
Developed a fault detection and diagnosis system for large-scale scientific experiments. The system uses Python and provides users with a common monitoring tool, unified fault monitoring and DAG-based root cause analysis of faults. I am responsible for all aspects of the system. It has been used in LHAASO and achieved good results.
- 2023.10-NOW Distributed Memory Buffer Pool
Dedicated to improving the reliability and stability of large-scale online distributed data systems. Modified based on MinIO so that it can run on memory and manage distributed memory. Currently, it can manage a cluster of 50 computers with over 25TB of memory, achieving more than 10GB/s of mixed reads and writes, and running stably.
- 2024.01-NOW Application of Large Language Models to Online Systems for High Energy Physics
Deploying and fine-tuning ChatGLM3-6B to explore the use of large langugage models to improve the user experience of high-energy physics online systems.
Publications
Talks
LHAASO 2023 1st Collaboration Meeting
Talk at Sichuan, Sichuan, China
Languages
Mandarin(Native), English(Fluent)
Skills & Expertise
- Programming Skills: C++, Java, Python, Go, JavaScript.
- The abilities of a full-stack engineer have been fully exercised, with extremely strong system architecture skills.
- Strong interpersonal skills with a good sense of teamwork.