The first rule for any problem like this is: define a signature such that elements which should be seen as the same have the same signature, then use a hash with signatures as keys to group things.

I think by "repeated lines" you mean lines in which all three values are the same, so the signature should include all three values:

sub signature {
  my $arrayref = shift;
  # the examples imply that the inputs are integers
  # if not, use a different character to join them
  return join '.', @$arrayref;
}
[download]

Once you have a signature, you can use a hash to accumulate any duplicates; in this case we just need a count:

my %seen;
for my $arrayref (@all_inputs) {
  if ($arrayref->[0] && $arrayref->[1]) {
    # not interesting, just pass it through
    print_result($arrayref);
  } else {
    # get a signature, increase the count for that signature by 1
    ++$seen{ signature($arrayref) };
  }
}
# now construct combined results for the saved inputs
for my $sig (keys %seen) {
  my $count = $seen{$sig};
  my $arrayref = [ split /\./, $sig ];
  # keep it easy: multiplying the zero value is a noop
  $arrayref->[$_] *= $count for (0, 1);
  print_result($arrayref);
}
[download]

For your example data, this ends up with combined lines [ 0, 10, 1 ], [ -5, 0, 1 ], [ 30, 0, 1 ], [ 0, -45, 1 ].