multi objective optimization of end milling process parameter using ai techniques

Abstract

newline quotEnd milling is one of the most universal and widely accepted machining operation in many manufacturing industries such as automobile, aerospace, die making, biomedical instruments, etc. An appropriate selection of process variables influence on productivity, quality, cost and delivery time of the product therefore the right selection of process variables is very important for end milling operation. Prediction and optimization of process variables increase the productivity of the product in obviously. Multi objective optimization and prediction instead of single objective is a challenging task in manufacturing industries. The objectives are to use artificial neural network (ANN) tool of artificial intelligence (AI) techniques to predict the responses and multi objective genetic algorithm (MOGA) tool to optimize the process parameters of the end milling operation for selected materials before the practical work. newlineIn current research, ANN and multi objective genetic algorithm tools have been used for the prediction and optimization of multi-response parameters of the end milling process. Here input machining parameters such as cutting speed, feed rate, depth of cut and mechanical properties of the material such as density and hardness have been considered for various responses such as material removal rate, machining time, tool life, tangential cutting force, torque and power. The cutting tool material properties have been included as a constant and exponent that shows a relation between cutting tool material and work material. The mathematical model has been developed and established based on empirical equations. This mathematic model has been utilized for the development of ANN model for prediction and multi objective genetic algorithm tool for optimization of performance evaluation criteria of end milling process. MATLAB R2015a has been utilized for the training and testing of the ANN model. Feed forward back propagation algorithm has been used with 5 input neurons, 6 hidden layers, 10 neurons in ea