Sojourn Time Control for Assembly Type Queueing Networks

9713974 Ayhan本研究的重点是装配型排队网络中作业停留时间的实时控制。这种控制的目的是确保以及时和可预测的方式生产完整的作业。实时控制策略将基于对排队网络状态的观察来执行。这些观察将提供有关网络中当前所有作业的位置和位置时间的完整信息。研究内容包括以下几个方面。装配型排队网络将被表示为单输入自由选择网，它可以用一组随机差分方程组来刻画。这些差分方程式将捕捉装配型排队网络的动态，并将作为支持实时控制的数学模型。控制程序将被描述为优化问题，并将开发有效的求解程序。最后，将与现有的在制品(Work in Process，在制品)限制控制策略进行对比，测试新控制策略的性能。如果成功，这项研究将导致装配操作性能的改善。研究的实时控制策略将减少资源争用，从而提高吞吐量，减少作业停留时间。这项研究的一个主要贡献是开发了复杂装配操作的分析模型。由于这些模型将是制定实时控制范例的主要工具，它们将导致在装配型排队网络中建立实时控制所需的基础科学基础的发展。

9713974 Ayhan The focus of this research is on the real time control of job sojourn time in assembly-type queueing networks. The purpose of this control is to ensure that complete jobs are produced in a timely and predictable manner. The real time control strategy will be executed based on the observation of the state of the queueing network. These observations will give complete information about the location and the time in location of all jobs currently in the network. The research will involve the following tasks. Assembly type queueing networks will be represented as Single Input Free Choice Nets which can be characterized by a set of stochastic difference equations. These difference equation will capture the dynamics of assembly type queueing networks and will serve as the mathematical model underlying real time control. The control procedure will be formulated as an optimization problem and efficient solution procedures will be developed. Finally, the performance of this new control strategy will be tested against the existing WIP (Work in Process) limiting control strategies. If successful, this research will lead to improvements in the performance of assembly operations. The research real time control strategy will decrease the resource contention which will result in an increase in throughput and decrease in job sojourn time. A major contribution anticipated from this research is the development of analytical models of complicated assembly operations. Since these models will be the main tool in the formulation of real time control paradigm, they will lead to the development of the fundamental science base required to establish real time control in assembly-type queueing networks.