TY - JOUR AU - Sekulic, M. AU - Pejic, V. AU - Brezocnik, M. AU - Gostimirovic, M. AU - Hadzistevic, M. TI - Prediction of surface roughness in the ball-end milling process using response surface methodology, genetic algorithms, and grey wolf optimizer algorithm JO - Advances in Production Engineering & Management PY - 2018 VL - 13 IS - 1 SP - 018 EP - 030 DO - https://doi.org/10.14743/apem2018.1.270 UR - http://apem-journal.org/Archives/2018/Abstract-APEM13-1_018-030.html SN - 1854-6250 AB - In this research study proposed are a response surface methodology (RSM), genetic algorithm (GA) and a grey wolf optimizer (GWO) algorithm for prediction of surface roughness in ball-end milling of hardened steel. The RSM is a conventional predicting approach, GA is an evolutionary algorithm and GWO is a new swarm intelligence-based algorithm. Spindle speed, feed per tooth, axial depth and radial depth of cut were selected as input parameters. Experiments were performed on a CNC milling center and experimental data were collected based on a four-factor-five-level central composite design (CCD). RSM was applied for establishing the basic relationship between input parameters and surface roughness. After that analysis of variance (ANOVA) was conducted for the evaluation of the proposed mathematical model. A predefined reduced quadratic model was used as a reference model for a build-up of predictive models using GA and GWO algorithm. Predicted values of RSM, GA and GWO models are compared with experimental results. In the comparison of model performance for all the three models it was found that GWO model is the best solution. The model accuracy was found to be at 91.80 % and 89.58 % for training and testing data, respectively, which showed the effectiveness of the GWO algorithm for modeling machining processes. KW - Ball-end milling KW - Surface roughness KW - Response surface methodology (RSM) KW - Genetic algorithm (GA) KW - Grey wolf optimizer algorithm (GWO) ER -