Aiming at the psychological impact on human and personalized matching problem, which are caused by the long-term coexistence between human and humanoid robot in the future, a more universal artificial emotion calculation model in the PAD (Pleasure, Arousal, Dominance) emotional space is proposed. The overall framework of the artificial emotion model consists of 3 sections: external event detection and processing, generation and correction of emotional response vectors and emotional discretization. Firstly, an effective external stimulus 3D annotation method is put forward, after that, according to the external stimuli, a computational model of emotion which can generate emotional response is proposed, and the personality and mood model are introduced to realize the personalized emotional response. Then, an emotional response vector discretization method is carried out, and the emotional intensity computational formula is proposed on basis of specific emotion types. Finally, the expression synthesis and voice response model is proposed according to the emotion generated by the emotion model. The speech dialogue and expression state migration experiment on the H&F robot-III humanoid head robot platform is carried out, and the results show that the artificial emotion model can generate personalized emotional response to external stimuli and simulate the process of human emotion changes effectively.