Date of Award

Spring 2007

Project Type

Dissertation

Program or Major

Engineering: Systems Design

Degree Name

Doctor of Philosophy

First Advisor

Robert B Jerard

Abstract

Novel techniques and strategies are investigated for dynamically measuring the process capability of machine tools and using this information for Smart Machine System (SMS) research. Several aspects of the system are explored including system integration, data acquisition, force and power model calibration, feedrate scheduling and tool condition monitoring.

A key aspect of a SMS is its ability to provide synchronization between process measurements and model estimates. It permits real time feedback regarding the current machine tool process. This information can be used to accurately determine and keep track of model coefficients for the actual tooling and materials in use, providing both a continued improvement in model accuracy as well as a way to monitor the health of the machine and the machining process. A cutting power model is applied based on a linear tangential force model with edge effect. The robustness of the model is verified through experiments with a wide variety of cutting conditions. Results show good agreement between measured and estimated power.

A test platform has been implemented for performing research on Smart Machine Systems. It uses a commercially available OAC from MDSI, geometric modeling software from Predator along with a number of modules developed at UNH.

Test cases illustrate how models and sensors can be combined to select machining conditions that will produce a good part on the first try. On-line calibration allows the SMS to fine tune model coefficients, which can then be used to improve production efficiency as the machine "learns" its own capabilities.

With force measurements, the force model can be calibrated and resultant force predictions can be performed. A feedrate selection planner has been created to choose the fastest possible feedrates subject to constraints which are related to part quality, tool health and machine tool capabilities.

Monitoring tangential model coefficients is shown to be more useful than monitoring power ratio for tool condition monitoring. As the model coefficients are independent of the cutting geometry, their changes are more promising, in that KTC will increase with edge chipping and breakage, while KTE will increase as the flank wearland expands.

Download

Share

COinS