There is, of course, “more than one way to do it,™” but I think that the way that I would do it is to use the /g modifier as discussed in perldoc perlretut.

Something like ... (caution... extemporaneous code; your syntax may vary)

while (my $line = <FH>) {
  next unless $line =~ /^\~\|TOTAL/;  
  my @percents = ( $line =~ /([\d\.]+)/g );
  .. do something with @percents ..
}
[download]

First, we ignore any lines outright which do not begin with the proper string ... notice the use of the "^" symbol to anchor to start-of-line, and the backslash-escaping of special symbols that otherwise would be taken as part of (ill-formed) regular expression syntax.

Then, “the interesting bits” in the string are groupings of digits-and-decimal-points, so we gather up as many of them as are present anywhere in the line.   In so-called “array context,” Perl will return an array containing all of the values found, without using a loop to do so, although we certainly could have done so using so-called “scalar context.”   Notice the use of parentheses to indicate a substring that we wish to extract.