Same was in my case
the data I was collecting using portable analyser and the casing mounted sensors both were showing similar “ghost peaks” and were fluctuating with power frequency.
But in your case the issue has started after a maintenance activity.
I would suggest you to see a waterfall plot and locate when this peak has started appearing. Was it present before the maintenance activity or not. Then list out ALL the activities you did during the maintenance and then strike one by one.
Hope you have checked for bearing fault frequencies
confirmed that this peak is not from the driven component through motor solo