Application of time-varying cellular neural network for optimal solutions
N.K. Al-Ani, T. Kacprzak · 2002
A time-varying cellular neural network (TVCNN) with a new scheme of annealing is proposed for finding the global optimal solution of a multivariable cost function. The technique is an engineering annealing method, which is the advanced electronic version of mean-field annealing. The processing of finding the global minimum of the generalized energy function is implemented by first increasing the energy level by reducing the voltage gain of neurones. Then searching for the global minimum energy level by increasing the neurone gain. The process of the global optimization is explained by the system eigenvalues with two computer simulations.