FEASIBILITY OF PSO-ANFIS-PSO AND GA-ANFIS-GA MODELS IN PREDICTION OF PEAK GROUND ACCELERATION

In the present study, two new hybrid approaches are proposed for predicting peak ground acceleration (PGA) parameter. The proposed approaches are based on the combinations of Adaptive Neuro-Fuzzy System (ANFIS) with Genetic Algorithm (GA), and with Particle Swarm Optimization (PSO). In these approaches, the PSO and GA algorithms are employed to enhance the accuracy of ANFIS model. To develop hybrid models, a comprehensive database from Pacific Earthquake Engineering Research Center (PEER) are used to train and test the proposed models. Earthquake magnitude, earthquake source to site distance, average shear-wave velocity, and faulting mechanisms are used as predictive parameters. The performances of developed hybrid models (PSO-ANFIS-PSO and GA-ANFIS-GA) are compared with the ANFIS model and also the most common soft computing approaches available in the literature. According to the obtained results, three developed models can be effectively used to predict the PGA parameter, but the comparison of models shows that the PSO-ANFIS–PSO model provides better results.

FEASIBILITY OF PSO-ANFIS-PSO AND GA-ANFIS-GA MODELS IN PREDICTION OF PEAK GROUND ACCELERATION Keywords:

ANFIS , metaheuristics , PSO , GA , peak ground acceleration.