About me

I am currently pursuing a Master of Science (MSc.) in Electrical and Electronic Engineering with a specialization in Communications Engineering at the University of Hong Kong, where I am developing advanced machine learning techniques for signal processing and communication systems.

Prior to this, I obtained a B.Eng. in Communication Engineering from the University of Electronic Science and Technology of China (UESTC), graduating in the top 25% of my cohort with a weighted GPA of 83.67, and a B.Eng. with Honours of First Class in Electronics & Electrical Engineering from the University of Glasgow through a joint-degree program.

My current research focuses on applying machine learning and image processing methods to challenges in engineering and industrial applications. My specific interests include developing deep neural networks for electromagnetic field reconstruction in power systems, utilizing computer vision techniques for autonomous vehicle navigation, and implementing gradient boosting algorithms for predictive modeling in geotechnical engineering. I have authored papers on transformer magnetic field prediction using deep learning approaches and landslide mitigation optimization through XGBoost algorithms, with publications presented at IEEE conferences.

Skills

Programmings

Python PYTHON
C/C++ C/C++
LaTeX LATEX

Embedded

ARM ASSEMBLY ARM ASSEMBLY
C/C++ (HAL) C/C++ (HAL)
MBED OS MBED OS

Office

Word WORD
PowerPoint POWERPOINT
Excel EXCEL

Languages

100%
中文
100%
English

Education

Education

  1. University of Electronic Science and Technology of China Sept. 2021 to Jun. 2025

    Bachelor of Engineering in Communication Engineering
    Weighted Grade: 83.67; Weighted GPA: 3.6 / 4; Rank (by Jun. 2024): Top 25%
    Microelectronic Systems (99); Circuit Analysis and Design (99); Communication Principles and Systems (95); Signals and Systems (94); Team Design Project and Skills (92); Embedded Processors (92); Wireless Sensor Network (92); Calculus I (92); Linear Algebra and Geometry (91); Probability Theory and Mathematical Statistics (91)
    Language Proficiency: English
    Dec. 2021, CET-4 553 pt.; Jun. 2022, CET-6 553 pt.; Oct. 2024, IELTS Band 6.5
    Scholarships
    Oct. 2022, Second Class of Excellent Student Scholarship, UESTC
    Oct. 2023, Second Class of Excellent Student Scholarship, UESTC
    Jun. 2025, Progress Award, UESTC
  2. University of Glasgow Sept. 2021 to Jun. 2025

    Bachelor of Engineering (Honours) in Electronics and Electrical Engineering with Communications Engineering
    GPA: 17.66 / 22.00
    Graduation with Honours of First Class
  3. University of Hong Kong Sept. 2025 to Now

    Master of Science (MSc.) in Electrical and Electronic Engineering with Communications Engineering (In Progress)

Contact

Contact Form

Researches

A Deep Neural Network Model for Transformer Magnetic Field Reconstruction

November 2022
A deep neural network (DNN) model for reconstructing transformer magnetic field distributions was proposed. Sampling of transformer magnetic field data simulated by finite element analysis (FEA) software was performed, followed by PCA scaling and normalisation. Responsible for neural network construction and result verification, and participated in the writing of the paper A Deep Neural Network Model for Transformer Magnetic Field Reconstruction, currently submitted to Networks.

Fast Prediction of Transformer Magnetic Fields under Defect Conditions Based on ML

November 2023
Addressing the issue that finite element analysis (FEA) is time-consuming for calculating transformers under four defect states and is not suitable for scenarios requiring rapid response, Random Forest, XGBoost, DNN, and CNN were used to establish a three-phase transformer magnetic field simulation model under defect states. The results showed that CNN had the best simulation effect and successfully improved the simulation time to the second or even millisecond level. Responsible for the rapid prediction of magnetic fields under overvoltage defect states and improving simulation accuracy.

Recommendation for Landslide Treatment Measures Based on XGBoost, First Author

January 2024
Collected a large amount of data on landslide mitigation measures and used five methods to calculate the number, length, and spacing of anti-slide piles for landslide bodies, demonstrating that the gradient boosting method (XGBoost) is superior to other methods in terms of quantitative recommendations for landslide mitigation. As the first author leading the writing of the paper Recommendation for landslide treatment measures based on XGBoost, which was submitted to the international conference IEEE IGARSS 2024 and presented as a Poster.

Intelligent Line-Following Obstacle-Avoiding Car Based on STM32H743ZI2

February 2024
Using the STM32H743ZI2 as the main controller and OpenMV as the vision module, the system implements functions such as line following, obstacle avoidance, traffic light recognition, pedestrian detection, and peripheral communication. Responsibilities include software algorithm development for the main controller and part of the vision system, RTOS porting and setup, NearLink communication establishment and testing, and TOF module development. The project performed exceptionally well in real-world environments, achieving a high score of 92 pt. (Top 10%).

Unified Image Restoration Model Based on Big Model Priors

September 2024
Lead the development of a unified image restoration framework aimed at addressing multiple image degradation issues (such as rain, fog, noise, and low resolution) using a single model, overcoming the inefficiency of traditional task-specific methods. Leveraging the prior knowledge of large models and the multimodal integration of visual-language models like CLIP to design a non-unified transformer-based architecture with a dynamic prompt adaptation mechanism. The model achieved outstanding performance on benchmark datasets such as Rain100L, RESIDE SOTS, BSD400, and DIV2K. This project was awarded the Outstanding Undergraduate Final Year Project by Glasgow College, UESTC (Top 40).