You can always try a shorter approach by just building up the bilayer from non-minimized molecules and see if it works. I would definitely give it a try.

If it doesn't work then this should be a safe although quite laburious way:

1) Take your single molecule, minimize it in vacuum.

2) Build the bilayer out of the minimized molecules but make them enough space not to overlap. The bilayer will at first have higher area per lipid then expected at the end.

3) Add solvent and minimize this large bilayer.

4) Apply lateral pressure in xy plane to squeeze the bilayer. This can be done by 1000 bars at low temperature (like 50K for example).

5) Let it heat under slow thermostat.

6) Equilibrate the resulting structure at your target temperature for some tens of nanoseconds.

I think its better when you create bilayer then you can do minimization and short equilibration. Then you insert protein in to bilayer following gromacs tutorial.

http://www.mdtutorials.com/gmx/

first you minimize the single molecule in vacuum.Then build the bilayer with the minimized molecules but you have to make enough space so that molecules do not overlap. then add solvent and minimize this large bilayer. then perform an equilibration of short time span. after this you can compare the rmsd values and energy values of the initial structure with the equilibrated structure to conclude on that if there is any changes of conformation happened or not.

best of luck.