| dc.description.abstract | Currently, the world is experiencing scenarios in which the presence of technology grows and its complexity increases in several areas: industry, domestic life, education, government, and others. Technological evolution should bring an increase in quality, in such a way that it guarantees the correct functioning of the processes and offers more reliable products. In this context, the study of systems reliability is essential in order to understand how this reliability can be optimized. Several components may be involved, such as microchips, hardware, software, networks and infrastructure. Understanding the mathematical model that defines the reliability of a given system can be a challenging task, mainly due to the dependence of physical parameters, factors such as time, materials used, quantities and weights.
This dissertation work carries out a scientific survey of previous works on various system configurations and variations of the reliability and redundancy allocation problem. A particular case study is chosen for analysis, with the aim of proposing a new method for its resolution. The problem is originally formulated with a continuous, non-linear model, and then it is transformed into a linear and integer model. In the proposed approach, a finite set of component options is suggested, seeking to portray real market situations in which manufacturers' catalogs are pre-defined, each with a specific reliability value. Furthermore, the originally mono-objective problem is reformulated as multi-objective, and solved using linear programming, IBM's CPLEX tool, together with the epsilon-restricted method. Once the problem was solved, comparisons were made with previous works that explored the same problem by applying meta-heuristics. | |
