The solar array on Solar Miner III consisted of about 800 solar cells. To be placed on the car and produce power, these cells had to be wired into some combination of series and parallel strings. My project was to produce an optimum electrical configuration of these cells.
Solution
To solve this problem I first started out by collecting published route data
to determine driving direction and location at different times during the
race. Then I wrote a program to calculate how much total energy one
configuration would generate over the course of a race. Then by using a
combination of simluated annealing and genetic algorithms variations of the
configuration were tested automatically, examing the search space for an
optimum solution. Several numerically intensive computers were utilized
across campus simultaneously to help look.
The final solution as pictured below used a total of 6 segments, and produced approximately 6% more energy than a Solar Miner II style configuration and 4% more energy than new configurations generated by hand. Purchasing solar cells that produce 6% more power would have cost 1.5-2 times as much, in the neighboorhood of $10k-$15k as opposed to the $7k we paid.
Here is Solar Miner III somewhere in Arizona.
SMIII at the EDS building in St. Louis
SMIII at that same place in Arizona