This paper proposed an effective system about micro-expression cognition and emotional regulation. First, the micro-expressional face was mapped into Arousal-Valence-Stance 3D emotion space. The micro-expression recognition is based on 3D-Gradient projection descriptor and the gradient magnitude weighted Nearest Neighbor Algorithm (NNA) in facial feature regions. Second, Gross cognitive reappraisal strategy was introduced to emotion analysis. In this research the emotional regulation process operated in the continuous emotional space enabling a wide range of intermediary emotions to be obtained. Finally, the micro-expression recognition algorithm was tested with the Yale University's facial database. The cognitive emotional system was applied to the human-robot interaction. The experiment results show that this micro-expression cognitive emotional model is generally consistent with human brain emotional regulation mechanisms and efficiently improve robot's emotion humanoid.