The target of Emotion modeling is to establish an system that can perceive, recognize, and express emotions with concurrency which humanity have by proper mathematical models. A big challenge in Emotion modeling is to establish the complex personal system of an intelligent agent or machine in a quasi‐physical and quasi‐sociological way, so that it rationally responds to emotions and behaviors by different internal and external stimuli with concurrency. For this purpose, an emotion classification model is presented for the audio‐visual external stimuli based on an improved long short‐term memory network. Then, based on Gross's emotional regulation theory, a hidden Markov model is constructed to imitate the process framework of human emotional cognition and personalized emotional regulation and expression, so as to realize the machine's intuitive and reasonable response to stimuli with concurrency. In this article, a framework of machine personalized artificial emotion simulation is preliminarily constructed, which provides a new model for human–computer interaction, in order to provide a reference solution for machine emotion understanding and expression.