The main purpose of the current research is to develop a new mechanism that makes PSO to work better. Optimization is a tool to find the best solution of a multi-modal problem. Nowadays engineers and designers are looking for optimal designs due to restriction of resources. Recently many optimization algorithms have been developed to takle complex problems, such as Particle Swarm Optimization (PSO). This algorithm is known as a strong explorer but weak exploiter. The main problem is many internal parameters to be tuned. So that users should achive a set of sensitivity analyses to adjust them for the problem at hand, but in the mean time the obtainded values may not proper for other problems.
In the present work, in order to overcome this shortcoming, a mutation mechanism as an important element of genetic algorithm is implemented in PSO. Each particle in the optimization proccess is assigned with a random number, then, as the procedure goes on, the random number of each individual is compared with a treshhold and if it is smaller than the treshhold, the particle gets mutated. This new algorithm entitled as modified particle swarm optimization (PSO). Three reinforced concrete MRF optimum design problems are solved by MPSO, PSO and results compared together with another research results based on HS algorithm. Studying the results show that MPSO compared with adjusted PSO by sensitivity analysis and HS, not only brings to better solutions, but it also does not need to any manual adjustment by sensitivity analysis.
Manuscript profile
