摘要
针对差速驱动AGV(automated guided vehicle)动力学的非线性、强耦合性、以及未建模问题导致无法建立精确模型进行轨迹跟踪控制任务,为此提出一种基于Koopman算子的数据驱动控制策略(koopman data-driven control,KDC)。首先,基于Koopman算子理论,利用扩展动态模态分解(EDMD)方法,得到了该AGV系统的近似高维线性动态显式表达式,即Koopman算子的近似表达。然后,基于导出的该高维线性Koopman动态表达式,设计线性模型预测控制器(MPC)实现AGV轨迹跟踪控制。最后,通过仿真测试对Koopman模型进行了对比验证,同时仿真实验验证轨迹跟踪效果,结果表明从数据中获取的模型信息精确,能高度拟合AGV的动力学模型,所设计的控制策略计算耗时少,在轨迹跟踪控制的精度和稳定性方面具有一定的有效性和先进性。
In view of the nonlinear,strong coupling and unmodeled dynamics of differential driven AGV(automated guided vehicle),it is impossible to establish an accurate model for trajectory tracking control task.This paper proposes a Koopman data-driven control(KDC)strategy based on Koopman operator.Firstly,based on the theory of Koopman operator and using the extended dynamic mode decomposition(EDMD)method,the approximate high-dimensional linear dynamic explicit expression of the AGV system,that is,the approximate expression of Koopman operator,is obtained.Then,based on the derived high-dimensional linear Koopman dynamic expression,a linear model predictive controller(MPC)is designed to realize the trajectory tracking control of AGV.Finally,a simulation test of Koopman model were compared to verify,at the same time,simulation results verify the trajectory tracking effect,the results show that the model from the data to obtain accurate information,can be highly fitting AGV dynamic model of the control strategy is designed to calculate less time consuming,in terms of the precision and stability of the trajectory tracking control has certain validity and advantage.
作者
丁承君
吴礼荣
朱雪宏
任超
张统
DING Cheng-jun;WU Li-rong;ZHU Xue-hong;REN Chao;ZHANG Tong(School of Mechanical Engineering,Hebei University of Technology,Tianjin 300401,China;School of Electrical and Information Engineering,Tianjin University,Tianjin 300072,China)
出处
《组合机床与自动化加工技术》
北大核心
2023年第3期109-112,117,共5页
Modular Machine Tool & Automatic Manufacturing Technique
基金
国家自然科学基金资助项目(U1913211)。