Genetic Algorithm-based Adaptive Rescheduling in Flexible Job Shops via Double-level Encoding

Abstract Within flexible job shops, the rescheduling approach should have the capability of self adaption when the type of disturbance changes. Considering three types of uncertain disturbances, including rush orders, task delays and machine failures, self-adaptive framework under multiple uncertainties has been achieved by defining affected tasks and other system parameters in a uniform pattern, and executing in the mechanism with the capability of rescheduling. Then, based on genetic algorithm, genetic chromosome with double-level code is designed for describing the task assignment of all jobs and the production sequence on each machine concurrently. Subsequently, the crossover and mutation operators have been modified to be accordance with such double-level code chromosome. Computational experiments prove that the solution framework functions effectively and efficiently in large-scale industrial application under uncertainties, and the decoding speed and computational efficiency has been improved by the proposed double-level code genetic algorithm.