A hybrid approach to solve a bi-objective optimization problem of a capacitated-flow network with a time factor

• Published in: Reliability engineering & system safety Vol. 204; p. 107191
• Main Author: Yeh, Cheng-Ta
• Format: Journal Article
• Language: English
• Published: Barking Elsevier Ltd 01.12.2020; Elsevier BV
• Subjects: Capacitated-flow network; Communications systems; Component assignment; Component reliability; Computer applications; Entropy (Information theory); Genetic algorithms; Information entropy; Minimal path; Network reliability; Non-dominated sorting genetic algorithm II; Optimization; Quickest path problem; Reliability analysis; Reliability aspects; Reliability engineering; Simple additive weighting; Sorting algorithms; System reliability; Time factor; Time factors; Simple additive weighting; Non-dominated sorting genetic algorithm II; Quickest path problem; Capacitated-flow network; System reliability; Time factor; Information entropy; Component assignment; Minimal path
• ISSN: 0951-8320; 1879-0836
• DOI: 10.1016/j.ress.2020.107191

•Discuss a system reliability multi-objective optimization problem with a time factor.•Adopt a MP based algorithm to evaluate system reliability under a component assignment.•Apply NSGA-II to determine the non-dominated solutions.•Utilize information entropy and simple additive weighting to determine an objective compromise solution. The problem of system reliability evaluation for a capacitated-flow network has been extended by involving the characteristics of the quickest path problem. System reliability is thus defined as the probability that d units of demand can be successfully delivered from a source node to a sink node within time t. System reliability optimization considering the time factor is important for demand transmission, especially for communication systems. However, this factor has not been taken into account in previous research. To this end, in this study, the time factor is employed to adopt a component assignment strategy to maximize the system reliability and minimize the assignment cost. Because the considered problem is bi-objective, a hybrid approach integrating the non-dominated sorting genetic algorithm II, a concept of minimal path, information entropy, and simple additive weighting is developed to determine a set of non-dominated solutions and to subsequently determine the compromise solution from an objective viewpoint. Numerical examples are utilized to demonstrate the computational efficiency of the proposed method.