The Application of Modified Parallel Particle Swarm Optimization to Back Analysis of Rock Masses Parameters in Underground Engineering

摘要

As back analysis of rock masses parameters in geotechnical engineering involves enormous amounts of computation, the traditional serial implementation of the optimization algorithm is time-consuming and even unrealistic. To obtain enhanced computational throughput and global search capability, a parallel implementation of Particle Swarm Optimization (PSO)based on Message Passing Interface (MPI) on a computer cluster is presented in this paper. The method is further illustrated with its application to a large scale underground cavern of a hydropower station. The back analysis results are favorable and prove that the proposed method is effective and applicable.