A hierarchical evolutionary algorithm with noisy fitness in structural optimization problems

  • Authors:

  • Ferrante Neri;Anna V. Kononova;Giuseppe Delvecchio;Marcello Sylos Labini;Alexey V. Uglanov

  • Affiliations:

  • Department of Electrotechnics and Electronics, Polytechnic of Bari, Bari, Italy;Laboratory of Theoretical-Probabilistic Methods, Yaroslavl' State University, Yaroslavl', Russia;Technical Area, University of Bari, Bari, Italy;Department of Electrotechnics and Electronics, Polytechnic of Bari, Bari, Italy;Laboratory of Theoretical-Probabilistic Methods, Yaroslavl' State University, Yaroslavl', Russia

  • Venue:
  • EC'05 Proceedings of the 3rd European conference on Applications of Evolutionary Computing
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

The authors propose a hierarchical evolutionary algorithm (HEA) to solve structural optimization problems. The HEA is composed by a lower level evolutionary algorithm (LLEA) and a higher level evolutionary algorithm (HLEA). The HEA has been applied to the design of grounding grids for electrical safety. A compact representation to describe the topology of the grounding grid is proposed. An analysis of the decision space is carried out and its restriction is obtained according to some considerations on the physical meaning of the individuals. Due to the algorithmic structure and the specific class of problems under study, the fitness function of the HLEA is noisy. A statistical approach to analyze the behavior and the reliability of the fitness function is done by applying the limit theorems of the probability theory. The comparison with the other method of grounding grid design shows the validity and the efficiency of the HEA.