First, to identify children aged 12-24 months, I would use the birth recode file (BR file). This file has one line per birth, so there are multiple lines for female respondents with multiple births in their lifetime. Note that this file includes a line for each child whether he/she is currently alive or not (b5) and whether he/she is lives with the mother or not (b9).

To calculate the age of children (in months), take the difference between the mother's date of birth (v008) and the date of the child's birth (b3). For example, you could use the STATA code below to do this:

generate agemnth=.
replace agemnth=v008-b3 

Next you can create a variable to identify children who are between 12-24 months on the day of interview:

 generate oneyr = 0
replace oneyr= 1 if  agemnth>11 & agemnth<24 

There should be very few respondents with more than one child between 12-24 months on the day of interview. If you wish to restrict your analysis to just one line per mother, you can keep just one child in this age group per woman. For example, you could use something like the STATA code below:

sort oneyr caseid
by oneyr caseid  : gen dup = cond(_N==1,0,_n)
drop if dup>1