基于非因果稳定逆的柔性机械臂学习控制

In this project, the learning control for trajectory tracking problems of flexible manipulator's endpoint is mainly studied. The non-minimum phase behaviour of flexible manipulator is exhibited by the flexible structures and underactuated character. Noncausal stable inversion method to cope with this will be mainly adopted to guarantee the running accuracy of control systems. Flexible manipulator dynamics equations based on Euler-Bernouli beam model will be developed through the assumed mode method and Lagrange equation fomulation. Under model uncertainties, the requirements of accurate tracking performance will be relaxed for the control system to ensure the system path to match the target path exactly. A learning method on parameter modification of the system model will be proposed based on noncausal stable inversion combined with model predictive method, which provided a technique to achieve precise path tracking control of flexible manipulator. Under the condition of the system is running repetitively, with a combination of iterative learning control (ILC) technology, using noncasusal iterative learning structure, an accurate trajectory tracking control algorithm will be given according to optimum performance index, which constructs the noncausal ILC law to improve the rate of convergence and the accuracy of tracking effect through the previous flexible manipulator running data. The performance of the proposed control technology will be verified with the actual experiment on a flexible arm system. This project's goal is to provide technical support for the flexible manipulator in the actual application.

本项目主要研究柔性机械臂末端轨迹精确跟踪的学习控制问题。针对柔性机械臂柔性结构和欠驱动造成的非最小相位特点，主要采用非因果稳定逆技术保证柔性臂末端的跟踪精度。期望以Euler-Bernouli梁模型为基础，结合Lagrange能量法与假设模态法，给出柔性臂动力学方程。针对模型的不确定性，放松对运行轨迹跟踪的要求，采用将非因果逆与模型预测相结合的学习方法，在线修正模型参数，实现学习、规划和控制一体的柔性臂路径精确跟踪控制，保证柔性臂末端路径与目标路径完全一致。在柔性臂重复运行的条件下，与迭代学习控制技术相结合，采用非因果的迭代学习律结构，根据最优化性能指标和稳定逆理论，采用以前系统运行的数据构造在实际应用中能够实现的系统稳定逆，实现柔性臂的轨迹精确跟踪控制。所提出的控制技术将在实际柔性臂系统上进行实验验证。旨在为柔性机械臂在实际生产中推广应用提供技术保障。

本项目针对柔性机械臂柔性结构和欠驱动造成的非最小相位特点，将非因果稳定逆技术与模型预测技术及迭代学习控制技术相结合，提出了一系列保证系统运行精度的柔性机械臂末端轨迹跟踪控制方法。给出了以Euler-Bernouli梁为基础模型，假设模态法与Lagrange公式法相结合的柔性臂动力学方程，建立了具有非最小相位特点的适合柔性机械臂末端跟踪控制的空间和平面柔性臂模型。在模型不确定情况下，采用反馈结合前馈的控制结构，结合非最小相位稳定逆技术，应用输入整形、模型预测、预作用优化和输出轨迹重定义等方法，设计了多种柔性机械臂的振动控制、轨迹跟踪控制和路径精确跟踪控制。在柔性臂重复运行的条件下，与迭代学习控制技术相结合，采用非因果的迭代学习律结构，在基函数空间逼近系统的稳定逆，并根据最优化性能指标，采用以前系统运行的数据构造在实际应用中能够实现的系统稳定逆，设计了多种实现柔性臂的轨迹精确跟踪控制的迭代学习控制方法。所提出的控制技术均在实际柔性臂系统上进行了实验，实验结果验证了所提方法的有效性。本项目取得的成果为柔性机械臂在实际生产中推广应用提供了技术保障。