The optimal design of a computer communication network belongs to NP-complete problem. It's hard to get the global solution using the traditional algorithm. Genetic algorithms are a natural evolution-based heurist...The optimal design of a computer communication network belongs to NP-complete problem. It's hard to get the global solution using the traditional algorithm. Genetic algorithms are a natural evolution-based heuristic search method, which have been successfully applied to a variety of problems. The difficulties in using the algorithm are how a particular problem is to be modeled to fit into the genetic algorithm framework, and how the operators (selection, crossover, mutation ) work due to the code strings. In this paper, authors establish a model for optimal design of networks, which is maximization of network reliability subject to a given cost constraint, and offer a corresponding modified genetic algorithms. Two examples are provided. The numerical results show the algorithm given in this paper has an idea solution speed and can get the optimal solution easily, and is also feasible to large scale problems.展开更多
文摘The optimal design of a computer communication network belongs to NP-complete problem. It's hard to get the global solution using the traditional algorithm. Genetic algorithms are a natural evolution-based heuristic search method, which have been successfully applied to a variety of problems. The difficulties in using the algorithm are how a particular problem is to be modeled to fit into the genetic algorithm framework, and how the operators (selection, crossover, mutation ) work due to the code strings. In this paper, authors establish a model for optimal design of networks, which is maximization of network reliability subject to a given cost constraint, and offer a corresponding modified genetic algorithms. Two examples are provided. The numerical results show the algorithm given in this paper has an idea solution speed and can get the optimal solution easily, and is also feasible to large scale problems.