Forecasting of machining quality using predictive neural networks

Cylindrical grinding is one of the metal finishing process widely used in all manufacturing industries. Surface finish is the vital output measure in the grinding process with respect to productivity and quality. This paper proposes an efficient technique Artificial Neural Network to predict the process parameters (wheel speed, hardness of material and depth of cut) in the grinding process for a given set of machining parameters. Experiments are designed according to response surface method. Different sets of data from the response surface model are utilized to train the developed network. The trained network is applied to predict the quality grinding. The proposed ANN is developed using MATLAB platform. The proposed method is cost effective, reliable and better than existing models and scopes virtual automation.