Optimizacion de redes con arcos de baja confiabilidad adicionando arcos redundantes.

摘要

This thesis develops a methodology to optimize the reliability of a complex network whose nodes are not necessarily in series or in parallel and their arcs have low known reliability. Network reliability is improved by using redundant arcs with a budgetary restriction.;Two heuristic optimization algorithms are used and compared to solve this redundancy allocation problem: a genetic algorithm (AG) and an algorithm proposed by the doctor Noel Artiles and developed by the author based on sequential integer linear programming (PLES). Since the exact evaluation of the objective problem is very difficult, it is evaluated by Monte Carlo simulation.;Two factorial designs were developed to compare the maximum network reliability (RRT) and the execution time (te ) of the heuristics optimization, AG and PLES. In conclusion, the maximum network reliability only was affected by the network size ( n), therefore both heuristics maximize the network reliability ( RRT) with the same efficiency. Furthermore, the experiment results shown that the proposed heuristic PLES is a faster algorithm than AG for all the network sizes.