Characterizing Residual Stress on Pipework Weld Repairs Using Machine Learning.

点击翻译按钮获取中文摘要

As I rounded the curve while maintaining my pace within the green area of the floor, I was shocked to see a massive red-hot ingot being lowered by rails onto a platform after emerging from a building-sized furnace. I closed in on the delegation as we continued our factory tour at Doosan Heavy Industries' nuclear reactor production division in Changwon, South Korea, attempting not to be distracted by the surrounding welding activities. Participating in that two-week intensive IAEA-KHNP training programme on the successful launch of nuclear power programmes for newcomer countries while I was employed as a technologist at the Ghana nuclear power institute prompted a period of intense self-reflection regarding my purpose, something I had struggled to determine up until that point. Engaging with like-minded engineers and researchers from around the world as they shared knowledge in the spirit of cooperation to advance nuclear energy, I quickly realized that this was a field to which I was more suited and to which I would be content devoting myself to making significant contributions. Consequently, despite having a bachelor's degree in biomedical engineering, I decided to pursue master's studies in nuclear engineering. Having conducted research during both my undergraduate and graduate studies, as well as spending two years analyzing data in a nuclear research institution and completing certification courses on Udemy, I have acquired skills in data science, machine learning with Pytorch, molecular dynamics, Monte Carlo simulation, and scientific writing, among others. All of these inform my approach to problem-solving and allow me to construct research projects, conduct experiments and simulations, and derive meaningful insights from output data-all of which are crucial to research. For example, my master's thesis required me to develop custom data modules with Python scripts to automate the process of reading specific numerical values from simulation logfiles and processing, computing, and graphing the results. Consequently, by reducing repetitive workflows, I had more time to focus on result interpretation.