Now it is very clear to me.
Thank you very much, sir!

Fermi level is not well-defined for molecule systems, since their HOMO-LUMO gap is generally not small, making the position of Fermi-level has a strong arbitrariness.

For ease of comparison of DOS/PDOS maps of molecular systems, you can simply shift energy levels so that HOMO occurs at E=0 position. In the DOS module of Multiwfn (latest version), you can find an option "-6 Set shift of energy levels", after choosing it and inputting H, all MO levels will be shifted by -E(HOMO), and thus HOMO will exactly appear at E=0 in the DOS map.

Sir, you answer helped me greatly. Thank you very much!
Just one more question, if you could answer me. When i make a DOS/PDOS graph to three similar systems, i get graphs with the fermi energy in different ranges of temperature. I see many articles that present this fermi energy in zero.
The correct procedure is to normalize all my graphs to present a fermi energy=0?

The choice of the fragments has somewhat arbitrariness, I suggest:
1 COC
2 Only OH
3 COOH
Namely defining the whole functional group in common sense as fragment. In the paper, this definition should be clearly mentioned.

I am doing a PDOS graph for a graphene quantum dot (GQDs) molecule. This GQDs has hydroxyl, epoxy and carboxyl groups.
I don't understand how to define the fragments in my case:
1) For epoxy groups, should i include the carbons attached to the oxygens (C-O-C) or only the oxygens?
2) For the hydroxyl group, should i include the hydrogen of the O-H, and the carbon attached to this group, or only the oxygen?
3) For the carboxyl, should i include the whole COOH group, or only the oxygens?

This is important because it affects the PDOS graph a lot.

Thank you in advance.