Rapid thruster-in-the-loop optimization for Hall thrusters

In order to expedite Hall thruster development, a method to rapidly and autonomously optimize thruster performance was created. This method combines rapid thrust measurement, thruster-in-the-loop control, and derivative-free optimization schemes to automate thruster optimization and rapidly find operational areas of interest, thus reducing the reliance on current-voltage-magnetic field (IVB) maps. Rapid thrust measurements are achieved by comparing test points to a known operating point, allowing fast and accurate thrust measurement while minimizing the effects of long-term thermal drift. Fast mapping of a Hall thruster’s operation was demonstrated at 72 test points per hour, with an average thrust measurement error of <1% compared to conventional thrust measurements. Two-dimensional (thruster discharge voltage and magnetic field strength) Nelder-Mead and Powell optimization schemes are shown to converge rapidly to maxima in total efficiency or specific impulse in fewer than 15 test points. The Powell optimization scheme remained effective in five dimensions, further increasing the peak thruster efficiency while adjusting three additional thruster dimensions (keeper current, cathode flow fraction, and magnetic field skew).