人机情感交互智能体及其在控制性能评价中的应用

Affective interactive agents with applications in control performance assessment

在智能体的信念、愿望、意图模型基础上,提出了一个人机情感交互模型。根据隐马氏链模型的情感强度空间表达和个性OCC模型的结构框架,融合了一种人工心理应激模型的内涵,建立起一个刺激转移情感计算模型。详细给出了智能体的情感计算方法,并进一步将其加入到智能体的人机交互机制中,构造了人机交互学习模型,提出了人机交互目标完成程度指标及其计算和情感学习算法。以过程控制性能评价为应用实例,讨论了智能体人机情感交互模型的构建和相关算法,并针对具体的控制回路进行了仿真和分析,验证了所提出方法的可行性和有效性。

Based on Belief,Desire,Intention（BDI） model of intelligent agents,a kind of human-computer Affective Interaction model（AI-BDI） was proposed.According to affect intensity space representations of Hidden Markov Models（HMM） and frameworks of personal Ortony,Clore,Collins（OCC） models,as well as combining implications of a kind of human psychological stimulation model,a stimulated and transferring affect computing model was set up,and agent affect computation method was presented in detail.The model was further introduced to construct a human-computer interactive learning model.Human-computer interactive objectives achievement degree and the computation as well as learning algorithms were proposed.By taking the process control performance assessments as example,construction and relevant algorithms of human-computer interactive model were discussed.Two specified control loops were simulated and analyzedt to demonstrated the feasibility and effectiveness of proposed method.