Prediction of dust emissions during CNC milling of spruce and pine with machine learning

Abstract

Wood dust generated by CNC machines during milling is hazardous to human health. This study aims to determine the wood dust emissions (PM2.5, PM10) according to the wood species, spindle speed (12000 rpm, 15000 rpm, and 18000 rpm), feed rate (3 m/min, 6 m/min, and 9 m/min), and cutting direction (contours, linear and spiral), and to predict them with machine learning algorithms. Oriental spruce (Picea orientalis L.) and Scots pine (Pinus sylvestris L.), known for their low and high dust emission values, respectively, were used as wood species. A blade with a diameter of 3 mm was preferred as a cutter for milling both wood species. The results of the analyses show that spindle speed, feed rate, and cutting direction parameters have a significant effect on PM. According to the PM2.5 and PM10 values, the highest wood dust emissions were measured at 121.42 µg/m³ and 173.02 µg/m³, respectively, in Scots pine wood material, with a spindle speed of 18,000 rpm, a feed rate of 3 m/min, and cutting direction being linear. The lowest wood dust emission was measured as 4.20 µg/m³ and 7.40 µg/m³ for PM2.5 and PM10 values, respectively, at a feed rate of 6 and 9 m/min, 15,000 rpm in Oriental spruce wood material under the conditions of cutting direction. However, the Cubist model performed best among the machine learning algorithms for predicting PM2.5 and PM10 levels. This study aims to provide data on wood dust emissions during CNC milling to inform the development of CNC parameter adjustments that minimize dust generation.