They should plot frequency/direction/intensity = analyze, it might be possible to assess any resonances and glue weights(or other peak suppressors) to reduce this. This changes the frequency, but excursion magnitude needs a dissipative mechanism of some sort so vibrations are damped. A small good conductor with a magnet, sort of a voice coil like item at the back could extract energy by shorting like mag brakes. You could also counter drive the wind in a balanced way with multiplexed audio frequencies - needs some filtering, but the concept is valid. Each mirror would run it's own feedback loop - not sure if these mirrors are aligned and tightened, or if there is something in the way of a feedback loop. A fiddly add-on, but if it saves 15-20% energy efficiency.
I wonder how much it will increase the heliostat costs by doing something like this? (Or will offset on lifetime costs of the heliostat by making it cheaper)?