Strain in fiberglass wing panel in various temperatures

[KtAero]

I have several full bridge bending bridges set up on a fiberglass airplane wing to measure wing bending along the span. Should be thermal compensated with full bridge, but I'm getting significant (equivalent to hundreds of in-lb) thermal impact (roll from hanger into sun and trace drifts, or cloudy day drifts for about 15 min following takeoff as wing finds cooler equilibrium). I think this is mechanical- that the wing is actually physically straining, but I need a way to subtract off the drift and would like to do it automatically. I can do this by 1) waiting for temps to equalize 2) look only at short time (several seconds) where the drift is insignificant. However, this doens't allow me to correlate a known wing load in the hanger to a particular wing strain that can be used to determine strains in flight since the temp changes the strain/load relationship. Ideas? Thanks