Modeling and Prediction of Machining Parameters in Turning Operation using RSM, ANN and Particle Swarm Optimization

Abstract

This study presents modeling, analysis, optimization and prediction of turning process of low carbon steel. The input parameters used are cutting speed, feed rate and depth of cut. The experiments were carried out under various conditions, three levels of each input parameters and two different treatments for each level, and obtain the surface roughness and electric current consumption which are the output parameters. The experimental data was used to develop the mathematical model using response surface methodology (RSM), and the relationship between the input parameters and the output parameters is determined. Also the prediction of output parameters using artificial neural network with back propagation (NN_BP) and artificial neural network with particle swarm optimization (NN_PSO) is performed. The data for experiments is used to train the NN_BP and NN_PSO to update the weighs in neural network. After training, the neural network model are runned using test data and the results using NN_BP and NN_PSO are compared with each other. The result shows that for turning with lubricant at 300 iteration, the NN_PSO is better than NN_BP. At 1200 iteration, NN_BP is better than NN_PSO for the surface roughness, but NN_PSO is better than NN_BP for electric current consumption. For turning without lubricant at 300 iteration, NN_BP is better than NN_PSO. At 1200 iteration, for surface roughness the NN_PSO is better than NN_BP, but for electric current consumption NN_BP is better than NN_PSO.