Keeping solar minigrids reliable
Through the Harvey Mudd Engineering Clinic, I worked with Mee Panyar on solar minigrids serving rural villages in Myanmar. Our question was a practical one: how do you spot a developing problem before the electricity stops working?
From measurements to earlier warnings
My part was to design and implement time-series anomaly detection algorithms in Python. These look for unusual behavior in measurements collected over time, with the goal of giving local maintenance teams earlier warning of potential problems.
The interesting challenge was connecting the data to the system it came from. An unusual pattern is useful if it helps someone recognize that a minigrid needs attention. That kept the machine learning work tied to the practical problem of maintaining electricity service.
Engineering in context
The Clinic pairs student teams with industry and community partners to work on real engineering problems. This project took me into energy systems and data analysis; my other clinic work took a rather different direction, recovering Halon 1301 through distillation.