Abstract
The classic job shop scheduling aims at the optimization of goals such as lead time. With resource prices rising dramatically, more focus is put on not only pursuing lead time, but also energy and other consumables. This causes a dramatic rise in the complexity of the planning algorithms. A potential method of implementation is here presented and tested considering a simplified industrial task. In this work an extensive research on job shop scheduling methodology is presented and analyzed. A process chain is then simulated through a complex matlab executable. The program is first tested to find regions of high efficiency production. As a sufficiently accurate first approach the process is rated in terms of energy consumption, makespan and final material roughness. It is indeed of main interest to reduce machine energy demand to avoid high energy peaks consumption and therefore costs. Secondly a neighborhood genetic algorithm is developed and tested. The simulations are carried out in Matlab/Simulink and are intended to mimic the production of a semplified mechanical manufacture. As a final outcome the analysis has shown a good stability of the investigated algorithm. With the presented method the optimized makespan, output from the simulation ensures higher production efficiency compared to a random or hand programmed scheme. In a last step, the potentials of self-optimization to achieve overall optimization in better calculation time than neighborhood or genetic approaches will be analyzed.
|