PREDICTION OF MODAL STRAIN ENERGY USING MULTI-OBJECTIVE METAHEURISTIC-TRAINED ARTIFICIAL NEURAL NETWORKS

This paper uses multi-objective methods to improve the exploration for single-objective problems. This method involved splitting the objective function into two segments and these enhanced using specialized algorithms designed for handling multiple objective functions. Three MO algorithms include Colliding Bodies Optimization (MOCBO), Particle Swarm Optimization (MOPSO), and non-dominated sorting genetic algorithm (NSGA-II), which are used to get the best prediction of structural modal strain energy. The indicated method is then implemented to two spatial truss structures. The recently developed method has superior performance over the previous approach that relied on single-objective optimization (SO) algorithms.

کلیدواژه ها:

Neural networks ، Strain energy ، Optimization algorithms ، Multi-objective optimizations ، Truss structures