Modeling and prediction of WEDM performance parameters for Al/SiCp MMC using dimensional analysis and artificial neural network

Abstract

In the present work, The CNC wire electrical discharge machining (WEDM) of Al 2124 SiCp (0,15,20) Metal Matrix Composite (MMC) is analysed by using dimensional analysis approach (DA) & artificial neural network (ANN). The models are formulated to correlate the independent parameters such as pulse on time, pulse off time, wire feed rate, current, voltage, thermal conductivity of the work piece material, coefficient of thermal expansion, density and the wire tension with the dependent parameters surface roughness and the material removal rate through design of experiments (DOE) plan. From the experimental findings, it has been observed that the pulse on time, thermal conductivity, coefficient of thermal expansion, wire feed rate and the wire tension are the most influencing parameters. In order to find out the accuracy of the formulated DA and ANN models, correlation coefficient (R2) was calculated. From the R2 values, it was clear that both DA and ANN approaches are competent to predict the surface roughness and the material removal rate. In addition, the models formulated by using ANN approach were found to be more reliable than the DA approach. The higher values of R2 (99.9910%) and lower value of various error based parameters shows the adequacy and reliability of the DA and ANN models. Comparative study of DA and ANN models disclosed the accuracy of ANN models hence recommended.