This will find all objects whose text content includes at least four numerical characters in a row, which seems like a reliable criterion for automating selection:
(ssget "_X" '((1 . "*####*")))
I would bet that there are already routines available for the Searching in this Forum and elsewhere, probably mostly to add commas every third character, but easily adaptable for spaces instead if there aren't any already written that way.
But questions arise: I note the non-numerical letter at the beginning of some of your example objects. It might be easy enough to work around that, but would there ever be more than that? Suffix content beyond the end of the numerical part? More wording before and/or after? Multiple numerical portions within the same object? Would the numerical part(s) you want processed always be separated from other content by a space? Would there ever be more than three decimal places, and if so, should they also be space-separated into groups of three? Etc.
Kent Cooper, AIA