Abstract The mechanical dynamics of modern machines very often depend on the angular position of the driven axis. To obtain optimal control, such applications typically require an advanced control structure such as an adaptive controller. Moreover, the variation in the dynamics like changing inertia, load torque, and viscous friction limits the performance and reduces the energy efficiency. Energy savings can be obtained by using so-called trajectory optimization techniques combined with feedforward control. However, both optimization and adaptive control require the knowledge of the position dependency of the mechanical parameters. In the case of reciprocating mechanisms, for instance, this position dependency is significant. Consequently, the mechanical parameters change rapidly at high operating speed of the machine. This paper thus contributes towards fast and accurate estimation of rapidly varying mechanical parameters. A sliding discrete Fourier transform (SDFT) approach is proposed to track the inertia variation of a reciprocating mechanism online. The feasibility is verified with experiments on an industrial pick and place unit. Both the results on the real machine and its CAD equivalent, modelled in a multibody dynamics software package, are considered. In addition, the developed inertia tracking algorithm is proven to be implementable in standard commercial drive components.

中文翻译：

---

使用 SDFT 方法在线跟踪变化的惯性

摘要 现代机器的机械动力学通常取决于从动轴的角位置。为了获得最佳控制，此类应用通常需要先进的控制结构，例如自适应控制器。此外，诸如改变惯量、负载扭矩和粘性摩擦等动力学的变化限制了性能并降低了能效。通过使用结合前馈控制的所谓的轨迹优化技术可以实现节能。然而，优化和自适应控制都需要了解机械参数的位置相关性。例如，在往复机制的情况下，这种位置依赖性很重要。因此，机械参数在机器高速运转时迅速变化。因此，本文有助于快速准确地估计快速变化的机械参数。提出了一种滑动离散傅立叶变换 (SDFT) 方法来在线跟踪往复机构的惯性变化。通过在工业拾取和放置单元上的实验验证了可行性。真实机器上的结果及其在多体动力学软件包中建模的 CAD 等效结果都被考虑在内。此外，开发的惯性跟踪算法已被证明可在标准商用驱动组件中实现。通过在工业拾取和放置单元上的实验验证了可行性。真实机器上的结果及其在多体动力学软件包中建模的 CAD 等效结果都被考虑在内。此外，开发的惯性跟踪算法已被证明可在标准商用驱动组件中实现。通过在工业拾取和放置单元上的实验验证了可行性。真实机器上的结果及其在多体动力学软件包中建模的 CAD 等效结果都被考虑在内。此外，开发的惯性跟踪算法已被证明可在标准商用驱动组件中实现。