人员能力与任务不确定环境下的一种任务分配方法研究

发布时间：2018-02-24 11:17

本文关键词： 不确定性任务-人员分配马尔科夫决策过程(MDP)模型 MATB　出处：《北京交通大学》2017年硕士论文　论文类型：学位论文

【摘要】：动态任务分配就是将合适的任务实时地分配给合适的成员,以充分利用系统的资源,提高任务的完成绩效。在一个不确定性系统中,任务到达的时间是随机的、不确定的,到达的任务类型是多样的、变化的,外界环境的变化会增加系统的不确定性,同时人员受外界以及自身状态的影响其处理任务的能力也是波动的。随着人工智能的发展,动态任务分配理论成果越来越成熟,其应用领域也更加广泛,但仍缺乏对不确定性人机系统中的任务-人员动态分配的研究。现有的任务-人员动态分配研究中,多集中于对外界不确定性的研究,而忽略人员能力波动性也会导致人机系统的不确定性。借鉴多Agent系统在不确定环境下智能体决策理论,建立马尔科夫决策过程(Markov Decision Processes,MDP)模型解决不确定性人机系统中任务-人员动态分配问题,丰富任务-人员动态分配理论,解决实际不确定性任务-人员动态分配问题、人员合作模式选择问题,从而降低不确定事件对系统的不良影响,提高系统的运作效率。论文完成的主要工作如下:(1)针对人机系统外界环境不确定和作业者能力波动,建立了基于MDP的人员能力与任务不确定性分配模型,并对MDP模型中的每个参数给出具体定义和说明,论证了其解的存在性,使用改进的策略迭代算法对模型进行了求解。(2)以MATB(Multi-Attribute Task Battery,多属性任务组)动态任务环境为应用背景,对MATB中任务-人员分配以及人员合作方式的选择问题进行了深入地分析,通过构建合理的人因实验设计,有效解决了 MDP模型求解所需的脑力负荷值、处理速度概率值、人员差异性等参数的获取问题。(3)采用MATLAB仿真,对随机任务-人员分配策略和基于MDP的分配策略进行了比较,验证了基于MDP分配策略的有效性和优越性,并对两种不同人员合作模式的分配策略进行了对比分析,证明了能力均等的合作方式优于能力差异的组合方式。
[Abstract]:Dynamic task allocation is to assign the right task to the right member in real time to make full use of the system's resources and improve the performance of the task. In an uncertain system, the time of task arrival is random and uncertain. The types of tasks that arrive are diverse and varied, and changes in the external environment increase the uncertainty of the system, and the ability of people to handle tasks under the influence of the outside world and their own states is also fluctuating. The theory of dynamic task assignment is becoming more and more mature, and its application field is more extensive. However, there is still a lack of research on dynamic assignment of task-personnel in uncertain man-machine system. Most of the researches focus on the uncertainty of the outside world, but ignoring the volatility of human ability will lead to the uncertainty of the man-machine system. This paper draws lessons from the theory of agent decision-making in the uncertain environment of multi-#en0# system. A Markov Decision process model is established to solve the problem of dynamic assignment of task-personnel in uncertain man-machine system, which enriches the theory of dynamic assignment of task-person, and solves the problem of dynamic assignment of uncertain task-person in practice. In order to reduce the adverse effects of uncertain events on the system and improve the operational efficiency of the system, the main work accomplished in this paper is as follows: 1) aiming at the uncertainty of the external environment of the man-machine system and the fluctuation of the operator's ability, In this paper, the model of personnel ability and task uncertainty assignment based on MDP is established, and each parameter in the MDP model is defined and explained in detail, and the existence of the solution is proved. Using the improved strategy iterative algorithm to solve the model. (2) taking the MATB(Multi-Attribute Task batch (multi-attribute task group) dynamic task environment as the application background, the problem of the task-person assignment and the choice of the cooperation mode in the MATB is deeply analyzed. By constructing a reasonable human factor experimental design, the problem of obtaining parameters such as mental load value, probability value of processing speed, personnel difference and so on for solving MDP model is effectively solved by MATLAB simulation. This paper compares the random task-personnel allocation strategy with the allocation strategy based on MDP, validates the effectiveness and superiority of the allocation strategy based on MDP, and makes a comparative analysis of the allocation strategies of two different personnel cooperation modes. It is proved that the cooperative mode of ability equality is superior to the combination mode of ability difference.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F224

【参考文献】