Enhancement of cutting force observer by identification of position and force-amplitude dependent model parameters
Enhancement of cutting force observer by identification of position and force-amplitude dependent model parameters
复制标题
通过识别位置和力幅相关模型参数来增强切削力观测器
DOI：
10.1007/s00170-019-04080-8
复制
发表时间：
2019
影响因子：
3.4
通讯作者：
Y. Kakinuma,
中科院分区：
文献类型：
--
作者：
S.Yamato;A. Sugiyama;N. Suzuki;N. Irino;Y. Imabeppu;Y. Kakinuma,
外部无传感器切削力估计在其可持续性方面具有良好的潜力。然而，其精度将恶化，由于变化的机器动力学取决于阶段的位置和切削力的振幅。在传统的方法中，物理模型参数如轴向刚度和粘性阻尼系数被视为在一定条件下识别的常值。结果，由于没有考虑上述参数变化，估计精度降低。为了解决这个问题，一个简单的参数识别方法在时域内采用最小二乘法（LSM）和切削力估计的负载侧扰动观测器（LDOB）提出了一个全封闭控制滚珠丝杠驱动的工作台。在不同的平台位置和不同的激励振幅下进行了一系列激励试验，以捕获位置和力-振幅相关的模型参数。研究了在工作台运动和停止状态下模型行为的差异。所提出的方法捕获的位置和力的振幅相关的模型参数安装到观测器。通过端铣试验验证了该方法的有效性。实验结果清楚地表明，通过考虑物理模型参数的位置和力-幅值依赖性，可以大大提高在进给和横向进给方向上的切削力的估计精度。
External sensor-less cutting force estimation has good potential in terms of its sustainability. However, its accuracy will deteriorate due to variation of machine dynamics depending on the stage position and cutting force amplitude. In the conventional methods, the physical model parameters such as the axial stiffness and viscous damping coefficient are regarded as constant values identified at a certain condition. As a result, the estimation accuracy decreases because the above parameter variation is not considered. To tackle this issue, a simple parameter identification method in time domain by employing the least-squares method (LSM) and a cutting force estimation by a load-side disturbance observer (LDOB) are proposed for a full-closed controlled ball-screw-driven stage. A series of excitation tests were conducted at different stage positions and various excitation amplitudes in order to capture the position and force-amplitude dependent model parameters. The difference of model behavior in the moving and stopped condition of the stage was also investigated. The position and force-amplitude dependent model parameters captured by the proposed method are installed into the observer. The validity of the proposed method was evaluated through end-milling tests. The experimental results clearly showed that the estimation accuracy of cutting force can be greatly improved in both feed and cross-feed directions by taking into account the position and force-amplitude dependency of physical model parameters.