Tool-condition Monitoring from Degradation Signals Using Mahalanobis-taguchi System Analysis

Drilling is a widely used machining process. With every hole drilled, a drill-bit gradually degrades until it breaks. In certain special applications replacing a drill-bit after it's breakage can be costly. In such cases degradation signals are often used to make a decision whether or not to replace the tool. However, very often the requirement of making a decision online, leads to online collection of data on the degradation signals of interest. From such data appropriate features are identified that help to arrive at a decision for the tool replacement before it breaks. In this paper we study two degradation signals viz., thrust force and torque using Mahalanobis-Taguchi System analysis. The study includes ten features (five features per degradation signal) and obtains Mahalanobis distance values based on data from holes with 'normal' degradation levels. Data from the last drilled hole prior to the tool breakage, representing 'abnormal' degradation level, are used for the validation of the measurement scale. Subsequently, the useful features out of the ten under study are identified using the orthogonal arrays and signal-to-noise ratios. Robust Engineering ASI’s 20 Annual Symposium