Output Only Functional Series Time Dependent AutoRegressive Moving Average (FS-TARMA) Modelling of Tool Acceleration Signals for Wear Estimation
33rd IMAC Conference and Exposition on Structural Dynamics, Florida, United States Of America, 2 - 05 February 2015, pp.111-122, (Full Text)
- Publication Type: Conference Paper / Full Text
- Doi Number: 10.1007/978-3-319-15230-1_11
- City: Florida
- Country: United States Of America
- Page Numbers: pp.111-122
- Keywords: Tool wear, Turning, FS-TARMA, Time series, ARTIFICIAL NEURAL-NETWORK, SENSOR FUSION, ACOUSTIC-EMISSION, ONLINE, MACHINE
- Middle East Technical University Affiliated: Yes
Abstract
In this paper, tool vibration signals obtained from a turning process are used for tool wear estimation purposes. During the cutting process, tool acceleration signals are recorded for different levels of wear. Due to non-stationarity of tool/holder system's response, Time dependent time series model of Functional Series Time dependent AutoRegressive Moving Average (FS-TARMA) type is used for modelling the signals and extraction of wear sensitive features that will be exploited in a wear estimation algorithm. Results of the analysis through FS-TARMA, reveals its higher accuracy with respect to stationary type models, since it captures time dependent properties as well, which can be used in an online tool wear estimation algorithm.