Efficient Adaptive Algorithms for Dynamic Job Shop Scheduling

Many recent algorithmic approaches suffer from their limited applicability to dynamic scheduling scenarios. Genetic Algorithms have been shown to overcome this problem at the expense of searching less efficient. We propose a flexible dispatching procedure which is controlled by a Genetic Algorithm. For this algorithm we investigate the impact of scaling the search space heuristically. Then we apply a temporal decompostition to the dynamic scheduling problem and discuss efficiency issues of the resulting adaptive algorithm. Finally we compare the outcome with priority-rule based dispatching and with other adaptive approaches reported in literature.