Abstract: Integrating different disciplines is very important in today's society. Each discipline will encounter bottlenecks at a certain stage of development, and interdisciplinary integration will create new things with more meaning. Artificial intelligence technology has effectively promoted the development of experience economy. The establishment of intelligent systems that meet the needs of human experience has become the trend of the current research. A human-centered intelligent system will pay more attention to users’ psychologies and provides users with personalized services in the human-computer interaction. Since emotional experience plays an important role in human activities, emotional interactions is better than cold mechanical interactions in the human-computer interaction. With the aging population in China and the rapid development of society, the number of people suffering from various psychological problems has increased dramatically. It is time to use advanced artificial intelligence technology to develop systems that can regulate emotions. Those systems can alleviate the loneliness of the elderly, and provide an aid for the treatment of psychological problems. They will have a wide range of application values. In order to enhance the interactive experience, introduce agents in human-computer interaction is the trend of future development. However, without emotion, a virtual human will have a poor sense of reality and is difficult to attract users to join the interaction. How to make virtual humans have natural emotional performance is a very important research topic. The research should include both how to recognize emotions from users and how to generate various emotional expressions to users. As the forefront of artificial intelligence technology, Affective computing has developed rapidly in recent years. Meanwhile, virtual reality technology has attracted widespread attention in the industry. Modeling emotions for virtual humans is one of the core issues to be solved in the field of emotional computing. It is also a research hotspot and a very challenging scientific issue in the field of virtual reality. To find out how psychology theories can be applied in the virtual human modeling, this paper reviews the existing methods of emotional modeling for virtual human based on the artificial intelligence and virtual reality. The review is carried out from the interdisciplinary perspectives such as cognitive appraisal theory of emotions, emotional model and emotional expression algorithm. After summarizing the current status of the emotional models that based on the cognitive appraisal theory of emotions, the paper analyzes some classical emotional models and discusses how to generate human-like emotions for virtual humans, including how to generate virtual human expressions, how to use the relevant body movements to express the emotion, how to test and evaluate the emotions in human-computer interaction, etc. Based on these existing studies, the paper points out some shortcomings in the current stage and analyses the application prospect of deep learning in this field. Besides that, it gives some technical suggestions for the future research. Therefore, the findings generated from this paper can be a reference for further research on emotional models. The paper takes into account both theory and application. The content of this paper belongs to the intersection of liberal arts and science. It systematically summarizes the technical methods of establishing emotional virtual humans and explores a path to apply psychology to information science.

Key words: affective computing, emotion, emotional model, virtual agent

摘要： 情感计算是人工智能的前沿领域之一，建立具有情绪表现力的人机智能系统具有广泛的社会需求。本文分析了基于情绪认知评价理论的情绪建模方法，总结了典型的虚拟人情绪计算模型，归纳了情绪和感知的相关成果，梳理了虚拟人情绪表现方法。针对情绪建模所要深入研究的问题，分析现有研究中存在的不足，对虚拟人情绪模型研究的未来发展方向提出了建议，为进一步深入研究虚拟人情绪模型提供了参考。

关键词: 情感计算, 情绪, 情绪模型, 虚拟人