Condition monitoring of micro tool with multi sensor using wavelet packet transform

In micro machining process, the tool condition monitoring is primary task for efficient process planning and high quality of processed product. Tool condition deteriorates with increase of machining distance due to tool wear. Therefore, the accurate relation between wear of tool and measured various signal indicates the machining condition in real time. Until now, energy of measured data is analyzed for estimation of tool condition, however, it could not give specific information of abnormal operation and subtle change due to the tool wear. In this research, single period signal is analyzed to reveal the component level symptom of fault in micro machining using laser displacement sensor. The multi sensors are applied to health monitoring of micro tool machining with aluminium printed circuit board (PCB). Vibration, acoustic emission (AE), force and laser displacement sensors are used to measure system condition according to increase of machining distance. To extract sensitive feature from each signal, we used the wavelet packet transform which extracts the time and frequency information simultaneously. A best base is selected by fisher linear discriminant which classifies the each feature according to the machining distance or system condition. Using this result, the system features are classified in the time-frequency domain, and the fault or abnormal peak position are revealed by best base.