Title :
Design of locally connected CMOS neural cells to solve the steady-state heat flow problem
Author :
Gobovic, D. ; Zaghloul, M.E.
Author_Institution :
Dept. of Electr. Eng. & Comput. Sci., George Washington Univ., Washington, DC, USA
Abstract :
A locally connected CMOS neural network is used to solve the steady-state heat flow problem. The proposed theory is based on the formulation of an energy function where the solution of the problem is obtained at the minimum of the function. An electrical circuit that tends to find the minimum of the energy function is designed. The circuit has a cellular neural structure where each cell is implemented as a neuron. The proposed method is illustrated by an example of heat flow in the plate. The circuit simulation results agree with the numerical results obtained by the software program Matlab
Keywords :
CMOS analogue integrated circuits; cellular arrays; circuit analysis computing; digital simulation; heat transfer; neural chips; Matlab program; cellular neural structure; circuit simulation; electrical circuit; energy function; locally connected CMOS neural cells; neural network; steady-state heat flow problem; Boundary conditions; Circuits; Difference equations; Differential equations; Heat engines; Neural networks; Partial differential equations; Resistance heating; Steady-state; Symmetric matrices;
Conference_Titel :
Circuits and Systems, 1993., Proceedings of the 36th Midwest Symposium on
Conference_Location :
Detroit, MI
Print_ISBN :
0-7803-1760-2
DOI :
10.1109/MWSCAS.1993.342937
