Taguchi-based and intelligent optimisation of a multi-response process using historical data