You need to step back and take a wider view of things.
You have stated:
" we know from our old process that the force has a large window" How large? What happens if the force is too low? What happens if the force is too high?
" The old pneumatic cylinder has the gauge calibrated each year." But does anyone ever look at the gauge? If they do, do they make any adjustments? The gauge, in and of itself, has no effect on your process.
" I was going to use a load cell to verify the force once a year." and "we hope to perform a check regularly to check it" What will you do if you check and it's out of tolerance? What if it's out of tolerance, but your process is just fine? More importantly, what if your process goes bad 3 months after you do the check? Will anyone know?
The key is to establish the actually force limits that the process can tolerate. Then notice that "tolerate" is the root of "tolerance".
It's possible to take roydm's excellent suggestion of a driving to a fixed position against a spring and design a simple and robust device that also almost no chance of ever "going out of tolerance" as long as the actuator gets to position.
Failure to get to position can be easily detected and trigger an alarm.