Repository logo
Log In(current)
  1. Home
  2. Colleges & Schools
  3. Graduate School
  4. Doctoral Dissertations
  5. Machine Learning-Enabled Modeling, Bifurcation and Stability Analysis, and Control for Machine Tool Dynamics
Details

Machine Learning-Enabled Modeling, Bifurcation and Stability Analysis, and Control for Machine Tool Dynamics

Date Issued
May 2026
Author(s)
Ma, Hang Mason
The University of Tennessee
Advisor(s)
Tony Shi
Additional Advisor(s)
Bradley Jared
Xueping Li
Bing Yao
Permanent URI
https://trace.tennessee.edu/handle/20.500.14382/57754
Abstract
This dissertation studies machine learning-enabled modeling, bifurcation and stability analysis, and control for machine tool dynamics. The scientific driver and also the major application area of this dissertation originate from the classical machine tool dynamics and chatter problem. I first developed an interpretable machine learning method, denoted by compressive sensing-assisted mixed integer optimization (CS-MIO) for general dynamical systems discovery as ordinary differential equations-type models from highly noisy data. The developed method combines various regularizations with discrete optimization for accurate model discovery from highly noisy data, with case studies on classical dynamical systems. This method is then extended for modeling thermal dynamics of additive friction stir deposition. Good alignment is achieved between the model prediction and experimental measurements. Next, bifurcation and stability analysis is conducted for the milling machine tool dynamics for rapid stability identification via numerical simulation. In particular, the nonlinear delay differential equations (DDEs) are numerically solved and accelerated by a newly developed loss-of-contact-based metric. This metric monitors the solution process of DDEs, enabling automated stability and early steady state identification. Numerical results show that the proposed metric can accelerate the numerical simulation by up to seven- and four-fold for stable and unstable cuts. This metric is integrated into state-of-the art method for accurate and fast stability map generation. Lastly, an interpretable machine learning based predictive control method is developed by extending CS-MIO to discover effective process models and achieve robust control performance for manufacturing processes. Case study on the turning dynamics is conducted. Compared with existing methods, the proposed method shows a two order-of magnitude anti-noise capability improvement for model discovery, and that enables stable process parameter selection with maximum material removal rate. The developed methods in this dissertation can be applicable for other machine tool dynamics in manufacturing processes, including additive friction stir deposition, and wire arc additive manufacturing.
Subjects

Engineering, Industri...

Engineering, Mechanic...

machine tool dynamics...

chatter

machine learning mode...

bifurcation and stabi...

control and automatio...

Disciplines
Engineering
Degree
Doctor of Philosophy
Major
Industrial Engineering
File(s)
Thumbnail Image
Name

MA-PRIMARY-2026.pdf

Size

10.93 MB

Format

Adobe PDF

Checksum (MD5)

24ae1f067df04a93951ba57f2b24413a


University Libraries

1015 Volunteer Boulevard
Knoxville, TN 37996
865-974-4351

Map & Directions
Donate to the Libraries
  • About
  • John C. Hodges Society
  • Speaking Volumes magazine
  • Outreach
  • Directory
  • Employment
  • Policies
  • Library Intranet
University of Tennessee power T logo

The University of Tennessee, Knoxville
Knoxville, Tennessee 37996
865-974-1000

Events
A-Z
Apply
Privacy
Map
Directory
Give to UT
Accessibility

Built with DSpace-CRIS software - Extension maintained and optimized by 4Science