Parameter Identification Of Creep Model Of Rock Based On Modified Pso-Lm Algorithm

Identification of the creep parameter is one of the important research issues in the field of rock creep mechanics. Considered the defect of parameter identification in creep model which the particle swarm optimization (PSO) is convergence slowly, and prone to fall into local optimum solution, the Levenberg-Marquardt nonlinear least squares method (LM) is relied heavily on the initial value. Therefore, a new parameter identified method is proposed that it combined with linear decreasing weight particle swarm optimization(modified PSO) and LM method, which proceeds as: 1) the model parameters are identified by means of modified PSO firstly; 2) LM method is used to identify the model parameters by initial values from step 1. A case showed that the modified PSO-LM method could be effectively identified the parameters of rock creep model.