If you are certain the databases are in synch after a restart, you might try this command to get the standby in roll forward pending state. See the Command Reference manual:
db2rfpen - Reset rollforward pending state
However, if you can restart the standby while HADR is still running, then that would be best since it will catch up even if it is behind when restarted. I think it "may be" OK to stop HADR on the primary (but not the standby) when restarting the instances. "May be" means that I have not tested this and there are no guarantees, since this is based on my recollection of having worked with HADR last year.