use strict;
use warnings;

# Set up arrays, use hash to find elements of
# @arrB not duplicated in @arrA.
#
my @arrA = (1, 2, 3, 4);
my @arrB = (1, 5, 3, 6, 8, 2, 4);

my %hashA;
@hashA{@arrA} = ();
my @nonDup = grep { not exists $hashA{$_} } @arrB;

# Show arrays.
#
print
   qq{Before\n},
   qq{@arrA\n},
   qq{@arrB\n},
   qq{@nonDup\n};

# Look to replace value 3 in @arrA. Loop over @arrA
# to find the element we want.    
#
my $toReplace = 3;
foreach my $idx ( 0 .. $#arrA )
{
    next unless $arrA[$idx] == $toReplace;

    # Use splice to take the found element out of
    # @arrA and replace it with one randomly splice'ed
    # out of @nonDup.
    #
    splice
       @arrA,
       $idx,
       1,
       splice
          @nonDup,
          int rand @nonDup,
          1;
    last;
}

# Show what we have now.
#
print
   qq{After\n},
   qq{@arrA\n},
   qq{@arrB\n},
   qq{@nonDup\n};
[download]

The output is

Before
1 2 3 4
1 5 3 6 8 2 4
5 6 8
After
1 2 6 4
1 5 3 6 8 2 4
5 8
[download]

Note that I use only three arguments for the second splice as we do not need to replace the element taken out of @nonDup.

I hope this guess is something close to what you need and will be helpful.

Cheers,

JohnGG