Modeling and feed-forward control design of flexible multibody systems with environment contact

轻量化的机器设计变得越来越重要。这是由于对能源效率和更高工作速度的需求不断增加。然而，这通常会产生不希望的弹性振动。对于需要进行大运动的机械，必须采用柔性多体系统动力学方法对其进行建模。在许多实际应用中，机器与其环境之间的联系也变得越来越重要。因此，本课题针对具有环境接触的柔性多体系统的建模、基于逆的前馈控制和轨迹控制进行研究。在第一个项目阶段，开发了一种统一的方法，用于具有运动回路、永久闭合接触和轨迹跟踪的柔性多体系统的前馈控制。因此，基于所谓的伺服约束的模型反演方法被证明是最合适的。在第一个供资期取得这些令人鼓舞的成果的基础上，应在第二个供资期进一步发展这些模型和方法，以便能够在实际应用中有效使用。除了神话改进和数值测试外，还应进行选定的实验调查。从而证明所建立的模型和方法的实际适用性。第二个资助期的第一个目标是提高模型反演边值求解器的数值效率。因此，对于更大的问题，快速前馈控制计算应该成为可能。对于中等规模的问题，计算次数以预定轨迹的实时性为量级。第二个目标是提高模型精度，特别是关于摩擦建模以及扩展到接触的开启和关闭。这在两种情况下都会产生非光滑模型，因此必须对已开发的逆模型求解器进行修改。第三个重点是前馈控制与附加反馈控制的结合。因此，也可以控制弹性坐标中的外部干扰。这种扰动可以是由接触点的摩擦引起的振动，也可以是接触闭合时激发的振动。本项目旨在为控制任务建立一种输出反馈控制方法。因此，使用应变片测量柔性体的曲率。然后将这些与在逆模型中确定并反馈给关节执行器的所需曲率进行比较。此外，应研究将这种控制概念与接触力测量相结合的可能性。项目第二阶段的最后一点是利用现有的柔性并联运动学对开发方法进行数值和实验验证。

Light-weight machine designs are becoming more and more important. This is due to increasing demands for energy efficiency and higher working speeds. However, this often yields undesirable elastic vibrations. If these machines have to perform large working motion, they must be modeled using the method of flexible multibody system dynamics. Also in many practical applications the contact between machines and their environment is becoming increasingly more important. Therefore, this project is aimed at modeling, inversion-based feed-forward control and trajectory control of flexible multibody systems with environment contact. In the first project phase a unified approach was developed for feed-forward control of flexible multibody systems with kinematic loops, permanently closed contacts and trajectory tracking. Thereby, a model inversion method based on so-called servo-constraints has been proven to be most suitable. Based on these encouraging results of the first funding period, the models and methods should be further developed in the second funding period such that an efficient usage in real applications is possible. Besides mythological improvements and numerical tests, selected experimental investigations should also be performed. Therewith the practical applicability of the developed models and methods will be demonstrated. A first goal of the second funding period is the improvement of the numerical efficiency of the boundary value solver for the model inversion. Thus, also for larger problems, a fast feed-forward control computation should become possible. For midsize problems computation times in the magnitude of the real time of the predefined trajectory are sought. The second goal is the improvement of the model accuracy, especially concerning friction modeling as well as an extension to opening and closing of the contact. This yields in both cases non-smooth models, and thus the developed solver of the inverse models must be modified. A third focal point is the combination of the feed-forward control with additional feedback control. Thus, also external disturbances in the elastic coordinates can be controlled. Such disturbances can be vibrations which are caused by friction at the contact point or vibrations excited when the contact closes. This project seeks to establish an output feedback control approach for the control task. Thereby, using strain gauges the curvature of the flexible bodies is measured. These are then compared with desired curvatures which are determined in the inverse model and feed back to the joint actuators. Further, the possibility of combing such a control concept with contact force measurements should be investigated. Final points of the second project phase are numerical and experimental validation of the develop methods using an existing flexible parallel kinematics.