In magnetic field-assisted finishing process, magnetorheological polishing fluid is used for precision polishing of freeform surfaces in the nanometer range. An efficient model is derived to accurately relate the input and output process parameters for better prediction of finishing performance. In this study, the relationship between the input and output process parameters of magnetic field-assisted finishing process is established using back-propagation neural network technique. Also, a close comparison between the regression analysis and neural network model has been carried out. The simulation results from neural network model better matches with the experimental data. Hence, this particular neural network model can be utilized to predict the response variables. A further optimization study using genetic algorithm and simulated annealing techniques is carried out to optimize the input process parameters for achieving maximum finishing performance. It is found that the results obtained from the genetic algorithm is more accurate and matches with the experimental results than the simulated annealing. Further, a characterization study of the finished workpiece surface is carried out which shows that MFAF process can achieve surface finish in the nanometer range having a minimum surface roughness value of 70 nm.
Using Lumped Element Equivalent Network Model to Derive Analytical Equations for Interpretation of Transformer Frequency Responses
