Aspects regarding the Use of Genetic Programming in Modelling the Cutting Process

The paper deals with the milling force modelling concerning average and maximum forces. From the experience of the late decades, the modelling using numerical methods of approximation such as multiple regression algorithms has proved that the models are not enough accurate and the error could be significant. This research tries to replace the classical models using reliable ones, the milling process still remaining a challenge for researchers. The main purpose of the work is to obtain a function of three variables (cutting depth, feed per toot and cutting speed) using genetic algorithms. The achievement of the genetic relation is important in cutting force estimation and also in optimization of cutting parameters if the set of experiments are enough. The milling investigations were carried out on a CNC machining centre, FIRST MCV 300, with three axes. The measurements were made with a Kistler dynamometer fixed on the machine tool mass. The workpiece material used for the experimental tests was a 173 mm length, 85 mm width, and 35mm high piece of steel. The workpiece was initially tested in order to establish the material (spectrometric test and hardness measuring). The cutting tool used is CoroMill R 365-080Q27-S15M with cutting tool inserts R365-1505ZNE-PM 4230 (Sandvik Coromant). The experimental plan was chosen according to Taguchi’s Method and consist of 16 tests based on three variables. The fit between data and the model is very good, very good result being achieved after maximum 1 000 generations.