This paper deals with the condition monitoring of knives via the Empirical Mode Decomposition (EMD). The cutting process is basically transient, thus Fourier Analysis and similar signal processing tools aren’t optimal because they treat signals as they were periodic. EMD is a signal analysis technique which is particularly suited for non-stationary and/or non-linear data, since it adaptively decomposes the signal in a sum of Intrinsic Mode Functions (IMFs). The knives under analysis are used inside an automated packaging machine; they are hydraulically actuated and are mounted on a moving support, so it’s not possible to put sensors on them because of security reasons related to sensors wiring. Instead, the actuators control valve is hosted on a fixed machine part, so its pressure signal is the one analysed in this paper. The sum of two IMFs is used to estimate the knife state and to obtain a representation of the wearing process during a knife life.
Knife diagnostics with empirical mode decomposition / Cotogno, M.; Cocconcelli, M.; Rubini, R.. - 4:(2016), pp. 167-175. (Intervento presentato al convegno The Fourth International Conference Condition Monitoring of Machinery in Non-Stationary Operations tenutosi a Lyon, France nel 15-17 December 2014) [10.1007/978-3-319-20463-5_13].