Flexible Job Shop Schedule generation in Evolution Algorithm with Differential Evolution hybridisation

Flexible Job Shop Scheduling becomes an emerging scheduling problem due to its nature to model constraints in holistic manufacturing systems. Its flexibility during sequencing and assigning tasks is typical for most smart factories, cyber-physical systems, new systems in distribution and procurement. There are many ways to deal with these scheduling problems, and population-based heuristics are the most common and thriving. Evolution Algorithms are the most popular as most general and practically used in many optimisation areas, while Differential Evolution principles are considered the most successful. This paper addresses the problem representation by chromosome and schedule generation to be suitable for hybridising Evolution Algorithm optimisation with Differential Evolution principles. Semi-active, Active and Non-Delay schedules are experimentally compared on benchmark models to find their suitability to be represented by one or two chromatids chromosome. Subsequently, several Differential Evolution strategies are tested and discussed to find their suitability to be implemented as a mutation operator in the Random key-based Evolution Algorithm.