自动导引车, 模型辨识, 遗传算法;多目标优化

A novel model identification approach based on least square method and multi-objective genetic algorithm was presented for the driving system of AGV. The model identified by least square method was adopted by multi-objective genetic algorithm as an evolution tutor who selected these individuals having a balance among a set of objective performances as elitists. Lossless finite precision method was used to keep the population diversity, and a multi-objective combined function was designed for a quick search for Pareto optimal solutions. The model identification experiments of AGV driving system demonstrate that the response curve of the model identified by the novel approach is much consistent with the object response curve both in amplitude and phase, which will provide a precise system model for AGV servo control system design.