A three-objective Pareto genetic algorithm is applied to the design of practical impedance-modulated feeds for log-periodic monopole arrays (LPMAs) mounted on arbitrary structures. Traditional design techniques do not permit the synthesis of feed structures for arrays on complex grounding structures. The multimodality of the search space of the feed-network design parameters precludes the use of standard gradient-search methods. The use of Pareto genetic algorithms allows for the study of tradeoffs between competing performance goals, and permits the generation of a database of feed designs that represent optimal performance tradeoffs from which one can select a posteriori. Numerical results are presented for the design of an LPMA mounted on a realistic wing.