الگوریتم های تصمیم گیری در دو سطح سیستم عامل های پیچیده

A two-level manufacturing operation system composed of two interconnected subproblems, i.e. scheduling of independent, non-preemptive tasks on unrelated moving executors as well as motion control of a group of moving executors performing the tasks is investigated. As the performance index of the two-level system, the makespan is assumed. Three heuristic solution algorithms for the two-level system are presented. The first algorithm ensures the current modification of solutions for the scheduling subproblem during the decision procedure of the two-level system. In the second one, an iterative approach is applied, which consists in successive implementation of the solution algorithms for both subproblems. The third algorithm uses on-line procedure, which enables determination of the best solution in the current step of the decision procedure. Comparisons of the algorithms as well as a numerical example are also presented.

For complex operation systems composed of operations (tasks, jobs), which are mainly characterized by execution times, different decision-making problems are solved including scheduling and allocation. It is assumed that the execution times for tasks are not given a priori but they are direct or indirect results of solving the decision-making problems, which consist in controlling the execution of tasks under consideration. Taking into account the control of task execution in operation system makes it more complex on the one hand and closer to real-life applications on the other hand. Complex manufacturing systems like flexible manufacturing systems are the main application area for systems under consideration. In such systems, the movement of different elements should be taken into account and, roughly speaking, the movement is the subject of control. The movement of plants to be produced as well as the movement of executors being the performers of technological tasks can be distinguished as the most important cases. Further investigation will be confined to the second case. Then, each task consists of two parts: driving-up of the executor towards the plant to perform the job (transportation task) and the execution of the job to forward the performance of the plant (technological task). Both parts should be controlled. One can easily notice that two decision-making subproblems, i.e. solving the selected subproblem for operation system and control of the execution of tasks are interconnected. It is convenient to describe them in the form of a two-level decision-making system. The first subproblem being the more strategic one is placed at the upper level. For such a general and complex problem, particular decision-making subproblems can be generated and investigated as well as different solution algorithms can be derived and applied. The idea of the two-level decision-making system for complex manufacturing operation system has been first introduced in Ref. [2]. It corresponds to the development of the area of complex operation systems, e.g. [1], [11], [12] and [13].