基于多目标优化的AGV驱动系统模型辨识

Model Identification Based on Multi-objective Optimization for AGV Driving Systems

以自动导引车(AGV)的驱动系统为对象,提出了一种基于最小二乘法和多目标遗传算法的新型模型辨识方法。以最小二乘法的辨识模型为多目标遗传算法的进化导师,选择具有多指标均衡优化性能的个体为精英,利用无损有限精度法保持种群多样性,设计多指标聚合函数快速搜索Pareto最优解。AGV驱动系统的模型辨识试验表明,该方法辨识的模型响应曲线在幅值和相位两方面皆与对象响应曲线较为吻合,研究结果可为AGV伺服控制系统的设计提供较为精确的系统模型。

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.