Toward a process monitoring of CNC wood router. Sensor selection and surface roughness prediction

Abstract

The measurement of wood surface roughness is performed once the machining process is completed. It requires considerable time since the measurement is performed at slow speed. The objective of this study was to develop a method to evaluate the surface roughness of paper birch wood while routing. For this purpose, a number of transducers were mounted on the router spindle and also in the proximity of the workpiece and cutting zone. Signals were acquired during a wide range of cutting conditions and analyzed. Statistical regression and artificial neural networks were employed to establish relationships between the signals and the actual cutting depth and surface roughness. The sensor selection and the feasibility of the sensor placement were determined. The models were subjected to a validation procedure to confirm their performance. The placement of the microphone at constant distance from the cutting zone was determined to be the most useful one. A model able to predict the surface roughness of routed paper birch wood regardless of the depth of cut was produced. The performance of the model was valid independently of the length of the workpiece.