On Bayesian Approach to Grillage Optimization

Abstract

In this paper, a new simplified version of the Bayesian Approach to coordinate global optimization (BAcoor) is compared with the well-known algorithms. BAcoor is some method of multi-dimensional optimization by applying a sequence of one-dimensional global optimizers starting from the best points obtained by previous one-dimensional optimization. The globality of one-dimension search is controlled by the only parameter ε. The new element is that observation points are defined by explicit formulas. In other similar methods, such as [8, 15, 16, 10, 11], this is performed by some numerical techniques that minimize the risk functions. The efficiency of suggested method is investigated and compared with other methods by solving a real-life civil engineering global optimization problem of pile placement schemes in grillage-type foundations. This problem is a good benchmark, because the minimal value of the objective function is known so the optimization error can be defined exactly.

DOI: http://dx.doi.org/10.5755/j01.itc.41.4.1670