SERIA REAL-TIME OPTIMIZATION O LIMITE SUPERIOR PARA SELF-OPTIMIZING CONTROL?