Abstract
Three variations of the particle swarm optimization (PSO) method are used to solve the boundary inverse heat conduction problem, in one, two, and three dimensions. Both steady and transient problems are studied. It is shown that PSO can be successfully applied to inverse heat conduction problems, and can alleviate some of the stability problems of the classical approaches. The computational costs of the three variations of PSO (basic, repulsive, and complete repulsive) are compared with each other, and with an implementation of the genetic algorithm. For these problems, some variants of PSO are proven to be more efficient than other algorithms. Also, the effectiveness of PSO in dealing with noisy domains is investigated.