Modicon Quantum hardware clock glitch

I'm trying to set up an annual hour meter on a piece of equipment that will only reset at midnight on January 1.

(picture 1) I've set up a sub block taking 41004 [value of 14 (current year)] and subtracting register 41755 from it.
(picture 2) 41755 is just taking a snap shot of 41004 and subtracting zero from it so that I can compare any changes of 41004's value and execute accordingly.

I am then using both the greater than and the less than output to energize coil 01755, which in turn I'm using a normally closed 01755 contact to reset my accumulated value. So in theory the middle output (equal to) of my first sub block will always be energized except for one scan when the year changes. Which also means that both the greater than (top rung) and less than (bottom rung) will be de-energized except for one scan when the year changes.

However, it seems that my accumulated value has already reset itself (more than once). Is there a better way to achieve an annual reset? We're using Proworx 32.

41754 seems to be self resetting when equal to #9999 rather than when 41004 is NEQ to 41755 as you described.
What does N184.1 look like? It might be helpful to see how you are conditioning your timer. It's a little strange that 41753 has no accumulated time.
Have you confirmed that the registers that you have chosen are not being written to somewhere else?

I just realized I had uploaded an outdated picture. The picture in my original post is how it's written now. I eliminated my counter reset just as a temporary measure until I could find out why it was randomly resetting itself. The picture on this post is how the program should be written.

UCTR 41754 is representative of the total annual elapsed time for the boiler (in hours). Since there are only approximately 8760 hours in a year, this should never reset itself.

Timer 41754 (N184.1) is a timer for 3600 seconds (1 hour) that is only enabled when the input for the boiler is active. After the boiler output has been active for 3600 seconds, it will increment UCTR 41754 by one, then reset itself.

My first thought was that I may have duplicated a register, or that it was being written to from another location. However, I've cross referenced our HMI (Cimplicity 8.0) and the address used table and I can't find any conflict.

Here are the pictures