Parameter Identifications of Elliptic Differential Equation by Hybrid Particle Swarm Optimization

摘要

Inverse problem which arise widely in scientific and engineering areas is to find out unknown properties of the objects. The ill-posed and nonlinear nature of inverse problem causes the difficulties of solving such problems. An improved particle swarm optimization with new velocity updating equation and without velocity limit is proposed to solve the parameter identification of elliptic differential equation problem. When the evolution process falls in the stagnation state, new promising particles are generated randomly by multi-parents crossover operator. The numerical results show that hybrid PSO algorithm is effective to solve parameter identification problems of elliptic differential equation and is not very sensitive to noise.