Monitoring of machine tools and optimization of manufacturing processes require accurate values of in-process measured mechanical quantities such as force, torque, or displacement. Commercial sensors available today do not provide the required performance, usually due to the very principle of their operation. Periodical tuning of these sensors to match the expected characteristics of their working environment is not a solution, since this environment changes rapidly. A system presented in this paper corrects dynamic characteristics of sensors in a continuous fashion. The implemented on-line algorithm comprises three steps performed repeatedly: (1) identification of parameters in a model of the machining process, in which the particular sensor is used; (2) estimation of parameters in the assumed model of the sensor—this estimation is based on results obtained in step 1; (3) evaluation of parameters of the adaptive filter that compensates the actual dynamic properties of the sensor. The developed algorithm has been implemented as a multi-processor, transputer based system integrated with digital signal processors.

This content is only available via PDF.
You do not currently have access to this content.