A meta-heuristics algorithm for the global optimization of expensive, simulator evaluated, objective function

• Autorzy: Miazga P.
• Konferencja: Evolutionary Computation and Global Optimization 2006 / National Conference (9 ; 31.05-2.06.2006 ; Murzasichle, Poland)
• Języki publikacji: EN

Abstrakty

EN

This paper presents the idea of a meta-heuristics algorithm called Evolutionary Controlled Clustering Algorithm (ECCA) designed for implementation in simulation optimization. The method focuses on localization of function optima neighborhoods. A Evolutionary Algorithm (EA) with soft selection and gene injection is used for finding basin of attraction. It operates over nodes of a grid created in a continuous parameter space. ECCA manipulates the grid density as well as the simulation accuracy. Clustered data is used for identification of the basin of attraction. Later, surrogate optimization is applied for local optima search. ECCA was optimized for operation in an uncertain and dynamically changing environment of simulation data. It was tested on the design of the shape of waveguide transition. The computer program can be executed concurrently on a multi-processor machine or on a grid of computers.

Słowa kluczowe

EN

global optimization; evolutionary computation; evolutionary algorithm; EA; meta-heuristics algorytm; Evolutionary Controled Clustering Algorithm; ECCA

PL

optymalizacja globalna; obliczenia ewolucyjne; algorytm ewolucyjny; algorytm metaheurystyczny

• Wydawca: Oficyna Wydawnicza Politechniki Warszawskiej
• Czasopismo: Prace Naukowe Politechniki Warszawskiej. Elektronika
• Rocznik: 2006
• Tom: z. 156
• Strony: 273--280
• Opis fizyczny: Bibliogr. 10 poz., tab., rys., wykr.

Twórcy

autor

Miazga P.

• Warsaw University of Technology, Faculty of Electronics and Information Technology Institute of Radio Electronics, Warsaw, Poland,

Bibliografia

• [1] Miazga P., Gwarek W., Improved Design of Passive Coaxial Components Using Electromagnetic 2-D Solver in an Optimization Loop. IEEE Trans on MTT, vol. 45, May, 1997, pp. 858-860.
• [2] Bandler J.W. et al. Electromagnetic optimization of 3D structures. IEEE Trans on MTT, vol. 45, May 1997, pp. 770-779.
• [3] Carson Y., Maria A. Simulation Optimization: Methods and Applications. In proc. of 1997 Winter Simulation Conference, 118-126.
• [4] Currin C., Mitchell T., Morrin M., Yvisaker D. A Bayesian Approach to the Design and Analysis of Computer Experiments. Report : Oak Ridge National Laboratory, ORNL-6498, Sept. 1988.
• [5] Jones D.R. A Taxonomy of Global Optimization Methods Based on Response Surfaces. Journal of Global Optimization 21, 2001, 345-383.
• [6] Kohler J.B., Puchalski A.A., Simon B. Estimating Functions Evaluated by Simulation:A Bayesian/Analytic Approach. In The Annals of Applied Probability, 1998, Vol. 8, No. 4, 1184-1215.
• [7] Miazga P. Global Optimization of Expensive Simulator-Evaluated Function Based on Response Surface Interpolation. In Proc. of VIII KKAEiOG, Korbielów, May 2006, pp.151-156.
• [8] Mathias K.,Whitley D., Kushma A., Stork C. An Empirical Evaluation of Genetic Algorithms on Noisy Objective Functions, Genetic Algorithms for Pattern Recognition, CRC Press, 1996, pp. 65-86.
• [9] QuickWave 3D electromagnetic simulator users manual. "Qwed" sp. z o. o. Warszawa.
• [10] Knowles J. ParEGO: A Hybrid Algorithm With On-Line Landscape Approximation for Expensive Multi-objective Optimization Problems. IEEE Trans. on EC, vol.10, no. 1, Feb 2005, pp. 50-66.