A holonic approach to flexible flow shop scheduling under stochastic processing times

Flexible flow shop scheduling problems are NP-hard and tend to become more complex when stochastic uncertainties are taken into consideration. This paper presents a novel decomposition-based holonic approach (DBHA) for minimising the makespan of a flexible flow shop (FFS) with stochastic processing times. The proposed DBHA employs autonomous and cooperative holons to construct solutions. When jobs are released to an FFS, the machines of the FFS are firstly grouped by a neighbouring K-means clustering algorithm into an appropriate number of cluster holons, based on their stochastic nature. A scheduling policy, determined by the back propagation networks (BPNs), is then assigned to each cluster holon for schedule generation. For cluster holons of a low stochastic nature, the Genetic Algorithm Control (GAC) is adopted to generate local schedules in a centralised manner; on the other hand, for cluster holons of a high stochastic nature, the Shortest Processing Time Based Contract Net Protocol (SPT-CNP) is applied to conduct negotiations for scheduling in a decentralised manner. The combination of these two scheduling policies enables the DBHA to achieve globally good solutions, with considerable adaptability in dynamic environments. Computation results indicate that the DBHA outperforms either GAC or SPT-CNP alone for FFS scheduling with stochastic processing times.