Fast Reliability-Based Optimization Using Parallel Genetic Algorithm, Developed Force Method and Artificial Intelligence

Abstract The aim of this paper is to present four strategies to increase the accuracy and speed of optimization of truss structures under the constraint of Structural System Failure Probability (SSFP). In the first strategy based on the probability rules, a criterion is defined to avoid producing many correlated paths and obtain more accurate upper bound of SSFP. In the second strategy, the force method formulation is improved and employed to analyze trusses with different topologies. In the third strategy two intelligent agents are utilized to identify the repeated paths and determine the fitness of the best chromosomes in each generation. Using the fourth strategy, the chromosomes, whose SSFP is much larger or smaller than the allowable value, are identified during the analysis and the analysis is terminated at that stage.