I've never built one before, too.
I originally planned to use a BeagleBoard, but the application isn't all that demanding - and the little Mega2560 combines a better mix of the I/O capabilities I need with adequate speed and memory to do the job. I'd have liked it better if it had an FPU, but I can manage without.
I'm monitoring and logging pressure, temperature, and radiation level at 1Hz in a dedicated task. Each time samples are taken the code examines all three values, and calculates both first and second derivatives of pressure and temperature with respect to time. I think these will be enough for early detection of behavior instability.
Even though my startup() function includes subjecting the reaction chamber to a vacuum (twice) to remove oxygen, I'm concerned that water vapor could be produced by the H2
reduction of oxidized Ni (fuel) or Cu (reaction chamber inner wall). Water vapor pressure is described by the Antoine Equation up to its critical point, above which it behaves like an ideal gas with a very
large R value. I have a graph at http://www.iedu.com/Solar/Engines/Hydrodyne/AntoineEq.html
that contrasts the behavior of water with an ideal gas. It's an eye-opener.
A large part of the code I'm writing is targeted at finding any possible excuse to scram the reactor. That's going to be an inconvenience, but I intend to understand everything I can before I "soften" the code's "twitchiness".