Dear Tian Lu,
Thank you for your help!! I change the Multiwfn version from 3.7 to 3.8. The problem has been solved.
I don't find this problem in using subfunction 2 of main function 15.
This is my modified .molden file
I am using latest version of Multiwfn and ORCA 5.0.1. Please make sure you are using latest version of Multiwfn.
Recently, I do some test by ORCA. The system is an I2(iodine) molecule. I transfer the .gbw to .molden file. And I also change the nuclear charge of I from 53 to 25. But when I use Fuzzy atomic analysis (function 15 in Multiwfn), wrong dipole moment is obtained. And the net charge is -56.0000. If I change the nuclear charge of I to 53, the net charge is +56.0000. I don't know why?