Queueing Models with Behavioural Consideration

在世界上几乎到处都可以找到这种系统。 许多人工智能系统都涉及到人类。人的行为对系统的性能有很大的影响。 如何捕捉和量化这种影响是一个感兴趣的问题，设计和操作这样的嵌入式系统。 本研究的目的是探讨具有行为考量的决策系统。 研究分为三个阶段。 在第一阶段，对人类（客户和/或服务器）在不同的环境下（例如，长/短排队长度或等待时间）进行。 这些研究将主要在中国北京清华大学与中国同事合作进行。 在第二阶段，将开发和研究分析模型。 分析中将使用矩阵分析法、随机比较法和模拟法等。 有效和高效的算法将被开发用于计算性能指标和寻找最佳或启发式解决方案。 最后（第三）阶段将涉及在系统设计中应用结果，例如服务组织（例如，呼叫中心、银行、机场、餐厅等） 在文献中已经有关于考虑行为因素的决策模型的研究。 然而，人类行为被假定为理性的，而它通常是有限理性的。 本研究将考虑人类的有限理性，并在引进和研究相关模型时加以考虑。 研究结果将更适用于涉及人类活动和决策的决策系统的设计。 这些结果还有望提高这些焊接系统的效率、生产率和可靠性。

Queueing systems can be found almost everywhere in the world. Many of the queueing systems involve humans. The behaviour of humans has significant impact on the performance of the queueing systems. How to capture and quantify such impact is an issue of interest to the design and operations of such queueing systems. The objective of this research is to investigate queueing systems with behavioural consideration. The research consists of three stages. At the first stage, empirical studies on how humans (customers and/or servers) behave under different queueing circumstances (e.g., long/short queue length or waiting time) will be conducted. The studies will be carried out mainly, in collaboration with Chinese colleagues, at Tsinghua University, Beijing, China. At the second stage, analytic models will be developed and investigated. Matrix-analytic methods, stochastic comparison, and simulation, among others, will be utilized in the analysis. Effective and efficient algorithms will be developed for computing performance measures and for finding optimal or heuristic solutions. The last (third) stage will involve the applications of results in the design of systems such as service organizations (e.g., call centre, bank, airport, restaurant, etc.) There have been studies on queueing models with behavioural consideration in the literature. However, human behaviour is assumed to be rational, while it is usually bounded rational. The proposed research will take consideration of the bounded rationality of humans in the introduction and investigation of queueing models. The results are expected to be more suitable for the design of queueing systems where human activities and decisions are involved. The results are also expected to increase the efficiency, productivity, and reliability of those queueing systems.