Try accessing like this

foreach my $hash (@a){
    foreach my $key (keys %{$hash}){
        print "key >$key< value>",$hash->{$key},"<\n";
    }

}
[download]

use 'Data::Dumper' module to print your data structure @a, you can visualize what you have created.
try using

use strict;
use warnings;
use diagnostics;
[download]

show the code of routine 'create_hash_for_filename'
UPDATE

$a[$j]{$key} 
and 
$a[$j]->{$key}
both will work
[download]

you are using @a for merging, but finally expecting output in @octopus_plid, how it is possible?.

UPDATE
nothing wrong in using like

for my $key (keys %{$a[0]})
according to your above presented logic
[download]

before calling the second for loop just try
use Data::Dumper;
print Dumper \@a;
[download]

run your script under perl -d
debugger will help you definitely. see 'man perldebug' for usage of debugger