Performance comparison of artificial neural network and multiple regression models for wire electrical discharge machining of haste alloy

Abstract

Haste Alloy C276 is a difficult to machine materials and exclusively employed to numerous engineering files such as gas, aerospace and nuclear applications. Since it has better strength and lower thermal conductivity, it has the tendency of reduced life of tool and poor machining performance by the use of traditional machining processes. Unconventional methods of material removal have been developed for overcoming the difficulties of traditional machining and claimed to be a suitable alternate approach for traditional method of material removal. Wire Electrical Discharge Machining (WEDM) is one among the available contemporary machining process that has been specifically adopted for machining of hard materials. Proper utilization of artificial decision making tools helps the manufacturer to attain benefits in manufacturing fields. This present exploration describes the development of neural network models and multiple regression models for WEDM process. The experimental runs are devised and analyzed by Taguchi’s approach. Grey Relational Analysis (GRA) is adopted for attaining the Grey Relational Grade (GRG) for representing a multi-performance index. An Artificial Neural Network (ANN) model and multiple regression models have been proposed to predict the Grey Relational Grade (GRG). Grey Relational Coefficient (GRC) values which are generated by the GRA method, considered as input for developing the Neural Network (NN) model to predict the multi-performance index (GRG). The closeness among the actual and predicted values are derived with the help of a comparative analysis. Finally, predicted values of GRG by NN models and regression models have been compared with actual values and the outcomes prove that neural network model offers better predictive control over regression models.