Abstract of doctoral thesis entitled GA-BASED CONTROL OF MULTIOBJECTIVE FLEXIBLE JOB SHOP SCHEDULING PROCESSES

of doctoral thesis entitled GA-BASED CONTROL OF MULTIOBJECTIVE FLEXIBLE JOB SHOP SCHEDULING PROCESSES Doctoral student: math. ec. ELENA SIMONA NICOARĂ Doctoral advisor: prof. NICOLAE PARASCHIV In the beginning of this work, the manufacturing systems are hierarchically approached on three levels: the first one corresponds to the global enterprise, the second one to the production planning sector and the third level corresponds to the scheduling sector. A Job Shop Scheduling (JSS) problem tries to allocate a finite set of resources over time to a finite set of different jobs in a minimum time, while satisfying temporal and resource constraints. Each job has a fixed order of operations, these are to be processed on a specific machine for a specified duration and the solution of the problem is an optimal schedule (an optimal table of starting times for each operation of all jobs). The thesis focuses on Multiobjective Flexible Job Shop Scheduling (MOFJSS) problems, which are tackled as scheduling processes; the flexibility considered here is a type II flexibility (the routings of the jobs on the resources are flexible) and the processes objectives are three: minimizing the makespan, reducing the number of late jobs and minimizing the average ratio of the idleness time of the jobs. The author extends the mathematical framework associated to the JSS problems to a formal definition of the MOFJSS processes within the systems theory context, by specifying the process hypothesis, the input, the state and the output variables, the constraints and the objective. The hall-mark of these processes consists in their high level of complexity, mainly conditioned by the huge dimension of the search space, by the difficulty in the representation, by the hard constraints and the enormous effort to evaluate the candidatesolutions. The primary contributions of the thesis address a poorly approached direction in the literature, namely the systemic modeling of the optimization processes control based on genetic algorithms. These contributions consist in the following aspects: providing a procedural model for the optimal control of MOFJSS processes based on genetic algorithms; designing a particular genetic algorithm enriched with a control strategy able to avoid the premature convergence of the genetic algorithm to suboptimal regions of the search space (imperative for the complex processes); proving the model viability by experiments on two deterministic scheduling processes in the manufacturing area, approached as uniobjective and multiobjective