| Popis: |
The upsurge of technological opportunities has brought about the speedy growth of the industry’s processes and machineries. The increased size and complexity of these systems, followed by high dependence on them, have necessitated the need to intensify the maintenance processes, requiring more effective maintenance scheduling approach to minimise the number of failure occurrences; which could be realised through a well-articulated perfect preventive maintenance with component replacement (PPMR) schedules from infancy stage through to completion stage. Then, using Strength Pareto Evolutionary Algorithm 2 (SPEA2), we devise a multi-objective optimisation approach that uses dependability and cost as objective functions. Typically, due to large scale problems, SPEA2 implementation on a single node proves to be computationally challenging, taking up to 1 h 20 min to run an instance of SPEA2 PPMR scheduling optimisation, the search time, the quality of the solution and its accuracy suffer much. We address this limitation by proposing a distributed architecture based on MapReduce, which we superimpose on SPEA2. The evaluation of our approach in a case study presents the following results: (1) that our approach offers an effective optimisation modelling mechanism for PPMR; (2) that the distributed implementation tremendously improves the performance by 79% computational time of the optimisation among 4 nodes (1 master-node and 3 worker-nodes) and the quality and accuracy of the solution without introducing much overhead. |
