Development and analysis of dispatching rules for dynamic multi-objective scheduling problems

In this work, a new method that uses Simulation and Gene Expression Programming is proposed for composite dispatching rules extraction for scheduling problems. In the suggested method, new dispatching rules that take into account the features of the production system have been developed for dynamic scheduling problems. Two different methods of Gene Expression Programming have been applied in the development of these rules. One of these methods uses multi-genic chromosomes based on production parameters whereas the other uses chromosomes with automatical defined function (ADF) that combine rules in the literature in a way linear and nonlinear. Although the simulated model is designed for multi objective dynamic flexible job shop scheduling problems, it has a structure which can easily implement for the other scheduling problems. According to the multi-objective view point, the dispatching rules obtained from the proposed methods are compared with the selected rules from the literature, and it is seen that the performance of extracting rules is significantly better than all other rules in literature. Also, when it is compared two kind of extracting rules, it was observed that ADF rules are better than multi-genic rules. Because the presented method and the achieved dispatching rules are robust, they can be applied for all real, big and dynamic scheduling problems which their optimization models have high complexity. Keywords: Dynamic job shop scheduling, Dispatching rules, Simulation, Gene expression programming, Multi-Objective optimization