Adaptive Observer for Optimal Control of Lathe Tool Wear

Abstract

Lathe turning is a single point cutting process, thus the cutting edge of the tool is not accessible for direct tool wear measurement. A variety of indirect in-process measurements have been proposed and/or tested in laboratories, but none have found practical application. A new approach to in-process tool wear sensing is described in this paper. It employs an observable, control-oriented state space model having a six-variable state vector. Four state variables are measureable by practical means, but the other two are tool wear characteristics that cannot be directly measured. In past work, the operation of a linear observer for tool wear was verified by simulation. This paper describes an adaptive observer that reduces the dependence on knowledge of plant parameters. Its operation is being verified by simulation. The results of the simulation will permit the development of requirements for practical application in the adaptive control of lathes.