Neural network modelling of process parameters influence on tool wear for ball-nose end mills

Abstract

Ball-nose end mills are used extensively in the manufacture of dies and moulds, aerospace components and medical devices. The machining process optimisation of these types of operations are challenging due to the large amount of parameters involved, such as axial path of cut, radial depth of cut, variable cutting speed along the cutting edge, variable chip thickness along the cutting edge, among others. This paper focuses on the main process parameters that affect tool wear during ball-nose milling operations. The parameters analysed are cutting speed, tilt angle and tool wear. Tool wear experiments were conducted with an indexable ball-nose end mill, machining blocks of P-20 mould steel at 30 HRC, under conditions similar to those used to produce sculptured surfaces in dies and moulds. Under the abrasive wear mode, a relationship was established between the undeformed chip geometry and the tool wear behaviour. The relation between process parameters and tool wear characteristics has been modelled with linear regression approach and artificial neural networks. The results have been compared with experimental work, indicating that proposed models are suitable to model this type of tool wear.