ogrinfo command line utility tells me its coordinate extent is:
Extent: (627.649021, 4665.315550) - (652.503142, 4698.809243)
And its coordinate reference system is:
Layer SRS WKT:
The "unknown" means anything we do now will be guesswork. Assuming these numbers are derived from UTM they are possibly in metres - does that make sense? Is your data from an area approximately 25 metres by 33 metres (the difference in the extent X and Y limits)? Or is it more like 25 kilometres by 33 kilometres? These numbers might make more sense if they are kilometres, otherwise they are very close to the (0,0) reference point.
Once that's established, the problem is going to be finding out where the reference (0,0) point is. This is related to the choice of UTM zone. If you know which UTM zones the points might come from you can try several until you hit the right spot. Maps of UTM zone codes are available if you search.
For your second example,
survlimits, the coordinates are in metres, but there's still no reference system associated:
Extent: (399497.662923, 4666343.184640) - (415616.488522, 4679517.726046)
Layer SRS WKT:
If I load it into QGIS it appears but it has no coordinate system, so I need to use the menus to give it one. But which one? There are several UTM 50N zones, but the WGS84 (EPSG code 32650) locates it here:
which for wider context is the red dot here:
If the UTM zone is correct then that should be in roughly the right place. If the longitude is wildly out then maybe its UTM49N or UTM51N.
I'm not sure what effect the Beijing 1954 datum will have at the moment though. I did also try WGS72 UTM zone 50N which moved the polygon slightly.