Multi-Task Optimization in Reliability Redundancy Allocation Problem: A Multifactorial Evolutionary-Based Approach
Journal article, Peer reviewed
Accepted version
![Thumbnail](/bora-xmlui/bitstream/handle/11250/3128866/Multi-Task%2bOptimization%2bin%2bReliability%2bRedundancy%2bAllocation%2bProblem%2bA%2bMultifactorial%2bEvolutionary-Based%2bApproach.pdf.jpg?sequence=6&isAllowed=y)
View/ Open
Date
2024Metadata
Show full item recordCollections
- Department of Informatics [978]
- Registrations from Cristin [10237]
Original version
Reliability Engineering & System Safety. 2024, 244, 109807. 10.1016/j.ress.2023.109807Abstract
Evolutionary multi-task optimization attempts to solve multiple optimization problems simultaneously by modeling the solution structures of two or more problems within a single encoding. In this paper, we report a novel way for evolutionary multi-task optimization in the reliability redundancy allocation problem exploiting the concepts of the popular multifactorial evolutionary algorithm (MFEA). We demonstrate the working of the proposed method considering two test sets and show how they can be concurrently solved using the MFEA. In the first test set, we consider two optimization tasks (case studies): the complex (bridge) system and the series-parallel system. In the second test set, there are two optimization tasks: the over-speed protection system for the gas turbine and the life support system in a space capsule. The common attributes between the two systems, within a set, complement each other to enhance the evolution process through implicit knowledge transfer. We present the comparative results considering existing evolutionary methods such as particle swarm optimization, genetic algorithm, simulated annealing, differential evolution, and ant colony optimization. Results are analyzed and compared using the average reliability, best reliability, computation time, performance ranking, and the popular statistical significance test of analysis of variance. The outcome shows that our proposed approach can solve the multiple case studies of RRAP simultaneously without compromising the solution quality. Moreover, our MFEA based solution method tops the rank among all approaches and provides significant improvement in computation time where it gains 28.02% and 14.43% of improvement in computation time for first and second test set, respectively, when compared with genetic algorithm. The percentage improvements in the computational time of the MFEA significantly increases when it is compared with other approaches.