Date of Award:

8-2026

Document Type:

Thesis

Degree Name:

Master of Science (MS)

Department:

Electrical and Computer Engineering

Committee Chair(s)

Regan A. Zane

Committee

Regan A. Zane

Committee

Hongjie Wang

Committee

Don Cripps

Abstract

Electric vehicles can be charged while driving through wireless charging coils buried in the road surface. To avoid running expensive cables from roadside equipment to each charging pad, the electronics that control the power transfer can also be buried in the road. However, burying these circuit boards removes the ability to cool them with fans or flowing air, so the heat generated during operation must escape naturally into the surrounding road material. If the boards get too hot, components fail—a problem that has already been observed in prototype systems.

Predicting how hot each component on a circuit board will get when buried in a road is difficult. Existing methods such as computer simulations and hand calculations are accurate but take days to weeks of engineering work for each new board design. When multiple boards are being developed at the same time and going through frequent design changes, this process cannot keep pace. Current designs are also oversized because engineers assume worst-case heat loads that rarely occur during actual roadway operation, adding unnecessary cost and bulk.

This thesis develops a faster approach. First, a computer simulation estimates how much heat each part of the system produces at different vehicle speeds, showing that actual heat loads under pulsed roadway use are substantially lower than continuous worst-case assumptions. An automated imaging pipeline then uses a thermal camera and contact sensors to measure actual component temperatures on prototype boards tested in both open air and sand—the sand simulating conditions inside a road. Finally, a machine learning model trained on these thermal measurements learns to predict component temperatures for new boards without repeating the full simulation or measurement process.

Testing shows that the trained model recognizes general heat-flow patterns rather than memorizing the layout of any single board, producing predictions within a few degrees of measured temperatures even on boards it has not seen before. This framework gives engineers a practical tool for checking whether a buried circuit board will overheat before it is installed in a road, reducing thermal evaluation time from weeks of engineering effort to minutes per design revision.

Download

Share

COinS

Copyright for this work is retained by the student. If you have any questions regarding the inclusion of this work in the Digital Commons, please email us at .