Design of Bean Pumping Units Based on Multi-objective Optimal Evolutionary Algorithm

General multi-objective optimal evolutionary algorithms (MOEA) can find as possible as optimal resolution set during solving multi-objective problems (MOP), however it cannot solve those problems which having strict constrained conditions. This paper proposed a novel method based on geometry character-geometrical pareto selection (GPS), which being used to optimize the two objective problems of beam pumping units (the maximum peak torque factor and acceleration during upstroke). This algorithm generated an initial population whose genes produced by complex method and coded the mechanism dimensions of beam pumping units with float, kept sufficient valid individuals throughout crossover and variation, selected those points which were more farther from the infinite far away point to form candidate set during every generation, and sifted valid Pareto frontier through the candidate set in the final. The experimental results proved that the algorithm proposed in this paper worked well for MOP with strict constrained conditions.