Genetic Optimization for Pulsed-Power System Configuration

Abstract

Pulsed-power systems traditionally have been designed to provide a pulse that is non programmable or requires hardware modifications to adjust the output waveform shape. Advancements in pulsed-power technologies are enabling system designs that allow for greater flexibility such as programmable current shaping. Material science, which uses current pulse shaping to obtain data for the equation of state analysis, is driving much of this work. The programming of pulsed-power systems through the use of simulations and manual curve fitting techniques can work well for systems that only have a few controllable parameters and are generating waveforms with simple spectral content. Complex systems with many controllable parameters become unmanageable for manual trial and error to be effective. This paper discusses the characterization and modeling of a scaled down programmable current adder directed at investigating technical issues that will be encountered in full-scale drivers. A discussion of the procedure used to optimize the adder current output, using genetic algorithms, is presented. The approach to system programmability presented in this paper will allow for a more simplified user interface and system control, as the requirements for flexibility and complexity in future systems increase.