This paper studies the three-state series-parallel (S-P) devices network reliability redundancy allocation problem (RAP) whose different components is paralleled, and the objective function
is maximized with cost and weight constraints. A new model is built. A new algorithm is constructed by using an objective function, that is, the discrete particle swarm optimization algorithm with compression coefficient to solve the RAP problem.
The algorithm is tested by two instances of the problem with MATLAB programming. As shown by the results, under the usually initial solution conditions, the particle swarm optimization algorithm can converge in each run and solve the system; at the same time, the algorithm is compared with the traditional simulated annealing algorithm, ant colony algorithm and genetic algorithm. The particle swarm optimization algorithm features advantages of easy programming and efficient convergence to the system optimal solution.