首页> 外文期刊>Fortschritt-Berichte VDI, Reihe 8. Mess-, Steuerungs- und Regelungstechnik >Linear Parameter Varying Control Design for Industrial Manipulators
【24h】

Linear Parameter Varying Control Design for Industrial Manipulators

机译:工业机械手线性参数可变控制设计

获取原文
           

摘要

In this thesis, modern LPV gain scheduling techniques are used to improve the performance capabilities of industrial manipulators with notable elasticities. The goal is to increase the disturbance attenuation properties of the system. At the same time, the good tracking performance already possible with the classical combination of feedforward and simple robust feedback controllers must be maintained. The whole tool chain from the LPV controller synthesis algorithms over a dedicated plant modeling and identification to the actual controller design and tuning is covered. Two types of LPV controller synthesis algorithms are treated. One is based on a gridding of the parameter space while the other uses a linear fractional representation of the plant. New extensions of these synthesis algorithms are proposed which allow to avoid common numerical issues. The plant model structures are chosen with respect to the considered LPV controller synthesis methods. A reasonable tradeoff between model complexity and accuracy is obtained by using flexible joint manipulator models. Several scheduling signals are needed to capture the operating point dependency of the inertia matrix. Using a principle components method, the number of scheduling signals is reduced. In a further step, the reduced set of scheduling signals is modified in order to retain some physical insight into the scheduling signals. A closed loop frequency domain identification of unknown parameters in the synthesis models is performed. For the experimental determination of frequency response functions, important aspects concerning the closed loop excitation of industrial robots like the impact of nonlinear friction are considered. Two separate experiment designs are applied leading to consistent results. For the automation of the controller design and for insightful benchmarks of different control approaches, a controller optimization scheme is introduced. Several weighting structures for LPV controller designs are discussed. Amongst them is a novel weighting scheme which separates the output feedback problem into an observer and a state feedback problem. A benchmark with all LPV controller designs and classical controller designs is performed. In further benchmarks, the impact of important details in the LPV synthesis algorithms is studied. The performance of the best LPV controller is demonstrated in a hardware testbed. A proportional integral derivative controller, optimized in the same manner, can be clearly outperformed.
机译:本文采用现代的LPV增益调度技术来提高具有显着弹性的工业机械手的性能。目的是增加系统的干扰衰减特性。同时,必须保持前馈和简单鲁棒的反馈控制器的经典组合已经可能实现的良好跟踪性能。涵盖了从LPV控制器综合算法到专用工厂建模和标识,再到实际控制器设计和调试的整个工具链。处理了两种类型的LPV控制器综合算法。一种基于参数空间的网格化,而另一种则使用植物的线性分数表示。提出了这些合成算法的新扩展,它们可以避免常见的数值问题。根据考虑的LPV控制器综合方法选择工厂模型结构。通过使用柔性关节机械手模型,可以在模型复杂度和精度之间进行合理的权衡。需要几个调度信号来捕获惯性矩阵的工作点相关性。使用主成分法,减少了调度信号的数量。在另一步骤中,修改减少的调度信号集,以便保留对调度信号的一些物理了解。在综合模型中执行未知参数的闭环频域识别。为了实验确定频率响应函数,考虑了与工业机器人的闭环励磁有关的重要方面,例如非线性摩擦的影响。应用了两个独立的实验设计,以产生一致的结果。为了实现控制器设计的自动化以及各种控制方法的深入参考,引入了一种控制器优化方案。讨论了用于LPV控制器设计的几种加权结构。其中有一种新颖的加权方案,它将输出反馈问题分为观察者和状态反馈问题。执行所有LPV控制器设计和经典控制器设计的基准测试。在进一步的基准测试中,研究了LPV合成算法中重要细节的影响。最佳LPV控制器的性能在硬件测试平台上得到了证明。显然,以相同方式优化的比例积分微分控制器的性能优于。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号