In-process detection of tool breakages using time series monitoring of cutting forces

Abstract A time series model for on-line detection of tool failure has been developed for general milling operations. The deterministic components of milling forces are first separated by using fundamentals of milling mechanics, with the remaining force signal being tracked by a first order adaptive autoregressive time series filter. The filter allows separation of cutting transients from a tool breakage event. The efficient response of the algorithm allows the detection of tool failure within one revolution. The results of laboratory experiments performed confirm the practical applicability of the monitoring algorithm.