Robust Interpolation and Iterative Learning Strategies for Modulated Tool Path Planning

1 online resource (68 pages) : PDF

2016

University of North Carolina at Charlotte

During single-point metal turning processes, a long, razor-sharp chip nest is often formed that can be a hazard to both operators and the part being machined. To combat this, a process called Modulated Tool Path (MTP) machining was developed by Barkman et al [1] that superimposes a sinusoidal motion on the feed direction of the tool path, causing the tool to remove itself from the cut and break chips. As determined by Berglind [2], the amplitude and frequency characteristics of this sinusoidal motion influence the material removal rate, chip length, and surface finish of the part, and thus, must be replicated by the machine tool controller as faithfully as possible. After the appropriate parameters for each MTP movement segment have been selected, the machine tool controller must attempt to reproduce these movements as closely as possible. In order to achieve this, a typical controller will rely on two components:1. An interpolation strategy, responsible for enforcing physical limitations of the tool, such as acceleration and jerk limits, and2. A closed-loop controller to ensure that the machine tool is reaching towards the setpoint determined by the interpolator at each loop closure. In this thesis, an alternative strategy for item 1 is proposed, as well as an iterative-domain technique to improve closed-loop control of repetitive processes. In order to tailor the standard interpolation strategy (known as jerk-limited linear interpolation) for MTP manufacturing, a technique called Exponential and Sigmoidal (E/S) interpolation has been developed which shows a marked improvement in both tracking error and acceleration required for a given trajectory. While item 2, the closed-loop controller (which is typically a PID type), has been researched in tremendous depth, it has no provisions for improving its performance during repeated tasks or for rejecting iteration-varying disturbances. To combat the first deficiency, an iterative-domain controller such as Proportional-Derivative Iterative Learning (PD-ILC) could be applied, however, the second criticism remains. This thesis proposes a variant of PD-ILC called Disturbance & Performance-Weighted ILC (DPW-ILC) which weights the relevancy of prior control inputs and errors based on criteria of merit, and determines an appropriate control input for the current iteration. As one of the main criteria for DPW-ILC is the measure of the disturbance, it is robust to iteration-varying disturbances. This thesis will show the vast improvement of DPW-ILC versus PID control only and a version of iterative control called PD-ILC in the face of these types of disturbances.

masters theses

Mechanical engineering
Artificial intelligence

M.S.

Controls
Disturbance Weighting
Iterative Learning
Modulated Tool Path
Performance Weighting
Trajectory Planning

Mechanical Engineering

Vermillion, Christopher
Ziegert, John

Kelly, Scott

Thesis (M.S.)--University of North Carolina at Charlotte, 2016.

This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s). For additional information, see http://rightsstatements.org/page/InC/1.0/.

Copyright is held by the author unless otherwise indicated.

DiMarco_uncc_0694N_11106

http://hdl.handle.net/20.500.13093/etd:1590