Application of genetic algorithm in numerical multi-objective optimization of ceramic capacitors

• Published in: 2008 2nd Electronics System-Integration Technology Conference pp. 377 - 382
• Main Authors: Dowhan, L., Wymyslowski, A., Felba, J., Wiese, S., Wolter, K.-J.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.09.2008
• Subjects: Capacitors; Ceramics; Delamination; Evolution (biology); Genetic algorithms; Residual stresses; Resistors; Thermal expansion; Thermal factors; Thermal stresses
• ISBN: 9781424428137; 1424428130
• DOI: 10.1109/ESTC.2008.4684378

Passive electronic elements such as capacitors and resistors are the most numerous parts used in electronics. In this paper, the ceramic capacitors were taken in the consideration. The purpose was to improve the reliability of such elements by carrying out the numerical multiobjective optimization process. In such components the reliability is strictly related to thermal-mechanical integrity. The key factor in such structures is the residual stress which occurs due to the differences of thermal expansion coefficients between the layers. To minimize the risk of failure (i.e. cracking, delamination) the 3D numerical parametric model of CC structure was elaborated using the finite element method. Afterwards, in order to minimize the failure risk in the crucial areas of the capacitor, the multi-objective optimization process was designed and carried out. To obtain the global extrema, as a result of the multi- objective optimization process, the classic genetic algorithm was applied. The idea of these algorithms is taken from the biology. They base on the natural evolution process in which the best suited individuals are taken for creating the next population. As a result, the best individuals survive and represent the optimization's solutions. In the investigation the self-made optimization tool was used. The tool was made in Python scripting language and it has implemented the multi-objective algorithms and methods that allow to apply and to optimize the numerical models (i.e. made in Ansys or Abaqus).