Seems like the most straight-forward way...and to do it one continuous pass, would be to:

   1. Get Filename
   2. Extract unique part of file to a $uniqueId variable
   3. Keep unique $uniqueId as key to hash
   4. Get date-time stamp of file
   5. If $uniqueId key doesn't ye exist in hash then
       store filename and date-time stamp in hash
       (perhaps as 'filename':'date-time stamp' to
       make it easy and quick to use 
         split(":",$uniqueIds{$uniqueId})  
       to recover the parts) at that $uniqueId
       key, 
      ELSE
       Get the filename and date-time stamp already in
         hash at that $uniqueId; 
       Compare date-time (probably using CPAN module to
         make this easy and simple) already in hash at
         $uniqueId key;
       Store new filename and date-time stamp into
         hash if new date-time is later than old one and
         Delete the old file that was idetified at the
         old entry in the hash,
       OTHERWISE,
         Delete this file.
   6. Repeat #5 until all files processed.
[download]

This will leave the names and date-time stamps of the 'latest' (or 'newest') filenames in the hash at the end of the run and will also leave only the newest files in the directory.