Neural network approach to structural optimization: boundary conditions case

Article type: Research Article

Authors: Ratajewicz-Mikolajczak, Elzbieta^a | Sikora, Jan^{b; *}

Affiliations: [a] Department of Fundamental Electrical Engineering, Lublin Technical University, ul. Nadbystrzycka 38a, 20-618 Lublin, Poland E-mail: elarm@eltecol.pol.lublin.pl | [b] Department of Electrical Engineering, Warsaw University of Technology, ul. Koszykowa 75, 00-662 Warsaw, Poland

Correspondence: [*] Corresponding author.

Abstract: An electrical field model governing copper electrodeposition process on printed circuit board (PCB) aided by the presence an ultrasonic field is presented in this paper. The problems associated with plating of copper and the current distribution into through-holes and advantages gained by ultrasonic intensification of the electrodeposition process are well known and are the subject of many investigations. Our experiments indicate that a copper layer in a through-hole grows faster than on their flat surface of PCB in this process. On the basis of measurements, a numerical (FEM) model has been proposed. The inverse problem has been formulated for coupled (a boundary conditions) fields making it possible to control the current density and the layer thickness. An artificial neural network is applied to solve this problem.

DOI: 10.3233/JAE-1999-149

Journal: International Journal of Applied Electromagnetics and Mechanics, vol. 10, no. 4, pp. 293-301, 1999