Uniformly sampled genetic algorithm with gradient search for structural identification - Part I: Global search

摘要

This paper is the first of two-part series on a uniformly sampled genetic algorithm with gradient search devised to efficiently solve the optimization-based structural identification. The strategy involves multi-species exploration, adaptive search space reduction and quasi-random sequence sampling. The use of a small number of uniform samples enables preliminary exploration in the solution space so as to shorten the "learning curve" considerably. The proposed strategy is shown by numerical study to give much better identification accuracy than the original search space reduction method, while using much less computational time for identification of known-mass and unknown-mass systems.