A cluster-based scheduling model using SPT and SA for dynamic hybrid flow shop problems

Abstract

The hybrid flow shop (HFS) scheduling, typically found in a variety of real-world industries, is an NP-hard combinatorial optimisation problem. Consideration of uncertainties hugely aggravates its complexity. This paper considers makespan minimisation of dynamic HFS scheduling problems under machine breakdown and stochastic processing times. It presents a novel cluster-based scheduling model (CBSM) that combines the good features of the shortest processing time (SPT) algorithm and the simulated annealing (SA) heuristic to synergise HFS scheduling under uncertainties. In this model, a neighbouring agglomerative hierarchical clustering algorithm is firstly developed. This algorithm decomposes an HFS into an appropriate number of machine clusters with different stochastic natures. The CBSM then performs a decision-tree-based assignment procedure using the classification and regression trees to determine an appropriate approach, either SPT or SA, for each machine cluster. Finally, the machine clusters are scheduled by their assigned approaches. To validate the effectiveness of the CBSM, a discrete-event simulator is conducted to evaluate its performance. The simulation results show that the CBSM outperforms all the compared algorithms in solving the dynamic HFS scheduling problems.