Teradata does not natively support regular expressions. That is why the '%[0-9.-]%' doesn't work.
Thanks Rluebke for your comment.
Do you know what could be alternative solution?
When I use the Oracle UDF's for Teradata in a Unicode system then teradata return an error.
Because UDFs are for the latin character set.
Any possibility to get UDFs for the unicode character set. I tried to do some changes on UDFs,
but I couldn't complete the functions.