Multi-Objective Optimization of the Field Distribution for an Electrostimulative Prosthesis

We describe methods for the automatic design optimization of a system of electrodes in a highly inhomogeneous domain to generate a field distribution that satisfies a set of objective functions. Due to the inherent complexity of the model we use numerical simulations to compute the field distribution within the domain of interest. We present additional algorithms to use some aspects of the optimization problem in order to reduce the numerical computation effort. Furthermore, we present an adapted optimization algorithm applicable to the problem presented here. Finally, we show first optimization results and make general remarks about the optimization process including the topological structure of the fitness landscape, measures to preserve diversity in the solution set and possible deteriorating effects to the solution candidates due to numerical inaccuracies.