The invisible surface should be trim surface(s) with problematic trim curve(s) so it cannot shade surface(s) properly. Try untrim those invisible surfaces and check the curves.
The floral elements for the left ring (at least in your last upload) are not forming a closed polysurface. I don’t know if this has deteriorated from the point where you tried to boolean, but it looks like it needs a lot of cleanup to get to something that can be booleaned. Maybe hide everything else and take a fresh look, or rebuild from whatever your primitive parts were?
yes thats the result after trying to clean up with untrim. ive started back from the base but it seems almost at random the union boolean will cause this problem. insted of joining the solids it will remove one of the surfaces and i need to untrim every time.
The partial inner surfaces to the foliage part aren’t going to help. Lose them and remake a surface at the end by extracting a copy of the outer face of the band and trimming it to the foliage. Watch out for small parts left in the veins of the leaves…
Internal surfaces where the stems cross or enter the leaves - create intersections and use them to split out the unwanted pieces for deletion.
Small surfaces in corners - its easy to leave them out of joins.
Duplicate surfaces - not sure if you have any, I did after messing with your stuff.
Naked edges - you’ll need to find these and deal with them as you work through the issues. I found a couple of places where I needed to use Refit Trimmed Edge to tidy up the surfaces.
Vanishing surfaces - when I rebuilt the inner face, which involved lots of trimming or splitting, I hit a point where the face I wanted disappeared, just leaving edges and isocurves. I used Refit to Tolerance on the face to bring this back.
Once the foliage part is closed you want to make a small adjustment so that its inner face overlaps the outer face of the band - if the surfaces are just coincident then, with the complexity involved, they probably won’t boolean. I used Scale 2D in the Front View, originating from the centre, with a factor of 0.995.
This is going to take a lot of time to resolve but, provided you are methodical, it is definitely achievable (and a good learning experience).