基于神经网络参数自学习的阻抗控制  

Impedance Control Based on Self-Learning Parameters of Neural Network

作　　者：党选举[1] 袁以坤 DANG Xuanju;YUAN Yikun(School of Electronic and Automation,Guilin University of Electronic Technology,Guilin 541004,China)

机构地区：[1]桂林电子科技大学电子工程与自动化学院,桂林541004

出　　处：《组合机床与自动化加工技术》2024年第1期123-126,130,共5页Modular Machine Tool & Automatic Manufacturing Technique

基　　金：国家自然科学基金项目(62263004,61863008)。

摘　　要：针对机器人在打磨过程中环境刚度和位置未知,传统的阻抗控制难以有效保持打磨质量的问题,提出了一种基于调节参数神经网络自学习的阻抗控制。由于基于李雅普诺夫稳定性理论设计的阻尼参数补偿方法中调节参数的选取直接影响系统的控制性能,根据阻尼补偿的数学描述,构建神经网络,用于其参数自适应调节,设计不同的激励函数用于反映阻尼在多种因素影响下变化的特征。通过所搭建的神经网络在线学习,实现参数的优化,以适应打磨过程环境变化。在斜面、平面及曲面等不同环境下,考虑其刚度突变、刚度动态变化、期望力动态变化等因素的仿真实验,结果表明所提出的控制方法与传统控制方法相比,具有更小的超调和稳态误差,并能够适应环境参数未知的情况,明显提高打磨质量和效率。Aiming at the problem that the robot′s environmental stiffness and position are unknown in the grinding process and the traditional impedance control is difficult to maintain the grinding quality effectively,an impedance control based on the self-learning of neural network with adjustment parameters is proposed.Since the selection of the regulation parameters in the damping parameter compensation method designed based on Lyapunov stability theory directly affects the control performance of the system,a neural network is constructed for its parameter adaptive adjustment based on the mathematical description of damping compensation,and different excitation functions are designed to reflect the characteristics of damping changes under the influence of multiple factors.Through the built neural network online learning,the optimization of parameters is realized to adapt to the environmental changes of the grinding process.Simulation experiments in different environments,such as inclined,flat and curved surfaces,considering their sudden changes in stiffness,dynamic changes in stiffness and dynamic changes in expected force,show that the proposed control method has smaller overshoot and steady-state error compared with the traditional control method,and can adapt to the situation where the environmental parameters are unknown,significantly improving the grinding quality and efficiency.

关 键 词：阻抗控制 工业机器人打磨 未知环境 变刚度 神经网络 

分 类 号：TH165[机械工程—机械制造及自动化] TG659[金属学及工艺—金属切削加工及机床]