Investigation on the machinability of a dual phase steel in milling and its estimation by artificial neural networks

In this study, the machinability of a low alloy steel in milling, which was inter-critically annealed and quenched at various temperatures, was examined. In addition, estimation of the experimental results was made with the use of artificial neural networks. The micro structures were observed and the martensite volume fractions were determined after the application of heat treatments which were carried out in the inter-critical region. Also, the mechanical properties of the steel with low, medium and high martensite volume fractions, which were obtained by quenching from 737ºC, 754ºC and 779ºC temperatures, were evaluated. Meantime, hardenability of the steel with various martensite volume fractions was detected by the application of Jominy end quench testing. Machinability was carried out by using the milling process at various cutting speeds. Besides, uncoated and TiN coated cemented carbide inserts used in order to make comparison. The effect of martensite volume fractions and the cutting parameters on tool wear and the surface roughness were determined. Along with the Elman?s recurrent neural networks, Levenberg-Marquardt algorithm was applied and verified to be successful.