Age
25
Birthdate
1998/02/11
Nationality
Chinese
Chengyuan Li
Master's Degree Student
Personal Information
Research interests
- Robust diagnostic methods using monitoring data representations
- System state prognosis method considering uncertainty
- Autonomous accident mitigation
Education
Nuclear Power Institute of China, Chengdu, China
Nuclear Energy Science and Engineering - Master of Engineering
2020 - 2023
- Working on a project on system health management for abnormal transients in nuclear reactors.
- Focusing on the implementation of diagnosis, prognosis and decision making methods for operational events after anomalies have occurred.
- Received the title of "Outstanding Graduate Student" from the Graduate School of China Nuclear Industry
- Received the Best Paper award at the annual meeting of the Science and Technology on Reactor System Design Technology Laboratory, a national key laboratory of China.
Xi'an Jiaotong University, Shaanxi, China
Nuclear Engineering and Technology - Bachelor of Engineering
2016 - 2020
- Received "National Scholarship" from the Chinese government in 2017** (rank 2/73)**
- Received "First-class scholarship" from Xi'an Jiaotong University in 2018** (rank 4/73)**
- Awarded "Outstanding Graduates Student Cadre" in 2020** (rank 6/73)**
- Awarded "Excellent Student Cadre" in 2017-2019** (3 years)**
- Obtained the qualification to pursue a Master's degree without entrance examination
Projects
Reactor Accident Disposal Method based on Deep Learning
Team Leader
August 2021 - Now
- I am responsible for investigating methods to diagnose, prognosis and mitigate decisions under complex system anomalies in the nuclear reactor
- I have used the existing reactor system steady-state numerical twin model, added anomalous initiation events with different parameters, and finally built the reactor transient data set
- I have used open source frameworks such as Pytorch to build deep learning models for diagnostic, prognostic, and decision-making tasks, and to build a process for model training schemes and validation experiments
- I am responsible for writing the first draft of the scientific paper after the experimental results were obtained, and for the PowerPoint presentation at the international conference
Publications Available Online
Representation Learning based and Interpretable Reactor System Diagnosis Using Denoising Padded Autoencoder
https://doi.org/10.48550/arXiv.2208.14319
- Submitted to the Journal of Reliability Engineering & System Safety (JRESS）
- Under Review
A Long-term Dependent and Trustworthy Approach to Reactor Accident Prognosis based on Temporal Fusion Transformer
https://doi.org/10.48550/arXiv.2210.17298
- Accepted by the International Conference on Nuclear Engineering (ICONE30)
An Unsupervised Learning-based Framework for Effective Representation Extraction of Reactor Accidents
https://doi.org/10.48550/arXiv.2208.13147
- Accepted by the Pacific Basin Nuclear Conference (PBNC2022)
- Two reviewers rated it as 90 and 100 out of 100
Post-hoc Interpretability based Parameter Selection for Data Oriented Nuclear Reactor Accident Diagnosis System
https://doi.org/10.1115/ICONE29-92505
- Accepted by the International Conference on Nuclear Engineering (ICONE29)
- One out of two reviewers rated the article as journal quality and recommended it as a candidate for best paper
Skills
Python Programming
Data Analysis: Numpy, Pandas, Scipy, Matplotlib
Machine Learning: Scikit-learn, XGBoost
Deep Learning: Pytorch
Black Box Optimization: Pymoo, OpenBox
High Performance Computing: Numba, Taichi, Ray
Domain Specific Skills
Self-supervised learning: to extract useful representation
Pattern recognition: to distinguish the root cause
Deep learning interpretability: to enable trustworthy decision making
Multi-horizon forecasting: to prognose future system status
Funding
My funding comes from two sources. On one hand, it is supported by the research project funds from my current institute, the Nuclear Power Institute of China. On the other hand, upon receiving a formal invitation, I will also obtain financial support from the China Scholarship Council (CSC).