Academic Journals Database
Disseminating quality controlled scientific knowledge

Predictive Modelling & Analysis of AISI 1045 Tool Steel in Die Sinking E.D.M. using Neural Network Approach

ADD TO MY LIST
 
Author(s): Samaddar Prasenjeet 1 , B. B. Patel 2 , K. B. Rathod

Journal: International Journal of Engineering Trends and Technology
ISSN 2231-5381

Volume: 4;
Issue: 8;
Start page: 3383;
Date: 2013;
VIEW PDF   PDF DOWNLOAD PDF   Download PDF Original page

Keywords: Artificial Neural Network (ANN) | Back Propagation | Electro Discharge Machining | Material Removal Rate | Tool Wear Ratio.

ABSTRACT
Electrical Discharge Machining (EDM) is a non conventional machining process, where electrically conductive materials are machined by using a precisely controlled spark that occurs between an electrode and a work piece in the presence of a dielectric fluid. It has been a demanding research area to model and optimize the EDM process in the present scenario. In this work a neural network model is presented for predictions of material removal rate (MRR) & tool wear rate (TWR) in die sinking electrical discharge machining (EDM) process for American Iron and Steel Institute 1045 tool steel with copper electrode.Experimentation has been carried out on EDM of AISI 1045 tool Steel. The experimental results have been used to train ANN using Back-Propagation Algorithm which gives the optimum value of the performance parameters like Material Removal Rate (MRR) and Tool Wear Rate (TWR) based on the influence of various electrode materials and processing parameters such as Gap Voltage, Peak Current, Pulse on time and Pulse off time. According to the correlation coefficients diagram it was concluded that the ANN tool gives us the best possible predictions for the data we have trained.Also we are getting ANN MRR results very much closer to our experimental MRR values.This shows the values that are very much possible to correlate amongst one another apart from taking experiments and finally the values of experimental TWR and predicted TWR are nearly correlating with one another.This shows that the ANN can be trained enough to give us close results by predicting the values.
Affiliate Program     

Tango Jona
Tangokurs Rapperswil-Jona