To overcome the weak local search ability and human subjective preference dependence of traditional Inter- active Evolutionary Computing（IEC）, a kind of affective learning Agent oriented to implement IEC was proposed by extending the typical Belief, Desire, Intention（BDI）structure. This affective computing and learning behavior based on human-computer interaction could study human＇s decision preference actively, thus the subjective fatigue of hu- man in human-computer interaction was reduced. The agent＇s structure models and functional modules were conduc- ted, and the realization of IEC algorithm was discussed in detail. Test functions as well as PID parameters of process controllers were employed to verify the effectiveness and feasibility of proposed method in IEC applications%为克服普通交互式进化计算方法局域搜索能力不强、过度依赖于人员主观偏好的问题，对智能体传统的信念、愿望、意图结构进行了扩展，提出了一类面向交互式进化计算的情感学习智能体，这种基于人机交互机制下的情感计算和学习行为，能够主动学习人员的决策偏好，从而减轻人员在交互式进化计算中的疲劳程度。给出了智能体的结构模型和功能模块，讨论了实现交互式进化计算的算法。通过数值实例以及工业过程比例、积分、微分控制参数整定问题，说明了在一类交互式进化计算问题中的应用，验证了所提技术方法的可行性和有效性。
展开▼
