Abstract:
Mass personalisation is now a reality. Driven by a continuous desire for product variability and individualised products, manufacturing systems need to adapt to produce personalised products at dynamic batch sizes. Production scheduling plays a vital role in the manufacturing system to address the challenges of mass personalisation. Effective production scheduling can meet customer demands, improve production efficiency, and save costs.
The Flexible Job-Shop Scheduling Problem (FJSP) can meet these demands. However, two assumptions are often seen in the shop scheduling literature. First, the processing order of operations in a job is assumed to be sequential; in practice, operational precedences are not always linear, and multiple dependencies may exist between operations in a job. Second, transportation time between machining centres are often neglected; in modern smart factories, transportation constraints need to be considered due to the introduction of complex transportation systems. Therefore, the Extended Flexible Job-Shop Scheduling Problem (EFJSP) with transportation constraints consisting of these two assumptions is proposed. The objective function is the minimisation of makespan.
To this end, a hybrid algorithm integrating multi-start, genetic algorithm and tabu search meta-heuristics, MS-GATS, is developed to optimise the problem. The hybrid algorithm utilises graph structures, dispatching rules and an ϵ-greedy policy to aid the search.
Computational experiments of widely-accepted benchmarking datasets are used to benchmark against current state-of-the-art algorithms. The benchmarking datasets are then adapted to accommodate transportation constraints to evaluate the feasibility of the proposed algorithm on the extended problem. Results from the standard dataset benchmarking illustrate MS-GATS is highly competitive against current state-of-the-art algorithms for the FJSP and EFJSP. Similarly, results from the extended dataset also confirm the feasibility and efficiency of the algorithm on the extended problem – EFJSP with transportation constraints.