Open Access Open Access  Restricted Access Subscription Access
Open Access Open Access Open Access  Restricted Access Restricted Access Subscription Access

A Neural Network (NN) and Response Surface Methodology (RSM) Based Surface Roughness Prediction Model for EDM


Affiliations
1 Production Engineering Division, Karunya Institute of Technology and Sciences (Deemed University), Coimbatore-641114, India
2 Information Technology Division, Karunya Institute of Technology and Sciences (Deemed University), Coimbatore-641114, India
     

   Subscribe/Renew Journal


Non-conventional machining processes such as the eiectro discharge machining process are used to machine hard tool and die materials. However, the process produces surfaces that have high tensile residual stresses, high surface roughness, cracks etc. The aim of this paper is to quantify the effect of some of the main EDM parameters on the surface roughness using Response Surface Methodology's Box-Behnken design and to develop a knowledge base using Neural Network. In this work, a mathematical model has been developed using the critical parameters namely the peak current, the pulse on time and the gap voltage using RSM. A response equation has been developed to predict the surface roughness values. The validity of this response equation has been proved by conducting additional experiments within the range of the experimentation. An Analysis of Variance (AVOVA) has been performed which revealed that the model developed is valid. The response equation has been used to generate training data, meant for training a Feed Forward Back Propagation Neural Network (NN). The critical parameters namely the peak current, the pulse on time and the gap voltage form the input parameters for the proposed NN model and the output being the surface roughness values. The predicted surface roughness values from the NN are found to be in reasonable agreement with the experimental values.
User
Subscription Login to verify subscription
Notifications
Font Size

Abstract Views: 299

PDF Views: 0




  • A Neural Network (NN) and Response Surface Methodology (RSM) Based Surface Roughness Prediction Model for EDM

Abstract Views: 299  |  PDF Views: 0

Authors

M. Joseph Davidson
Production Engineering Division, Karunya Institute of Technology and Sciences (Deemed University), Coimbatore-641114, India
S. Porpandi Selvi
Information Technology Division, Karunya Institute of Technology and Sciences (Deemed University), Coimbatore-641114, India
D. G. Harris Samuel
Production Engineering Division, Karunya Institute of Technology and Sciences (Deemed University), Coimbatore-641114, India
D. Alin Jeyadarlus Sofana
Production Engineering Division, Karunya Institute of Technology and Sciences (Deemed University), Coimbatore-641114, India
M. Jeeva Rose Shanthi
Production Engineering Division, Karunya Institute of Technology and Sciences (Deemed University), Coimbatore-641114, India

Abstract


Non-conventional machining processes such as the eiectro discharge machining process are used to machine hard tool and die materials. However, the process produces surfaces that have high tensile residual stresses, high surface roughness, cracks etc. The aim of this paper is to quantify the effect of some of the main EDM parameters on the surface roughness using Response Surface Methodology's Box-Behnken design and to develop a knowledge base using Neural Network. In this work, a mathematical model has been developed using the critical parameters namely the peak current, the pulse on time and the gap voltage using RSM. A response equation has been developed to predict the surface roughness values. The validity of this response equation has been proved by conducting additional experiments within the range of the experimentation. An Analysis of Variance (AVOVA) has been performed which revealed that the model developed is valid. The response equation has been used to generate training data, meant for training a Feed Forward Back Propagation Neural Network (NN). The critical parameters namely the peak current, the pulse on time and the gap voltage form the input parameters for the proposed NN model and the output being the surface roughness values. The predicted surface roughness values from the NN are found to be in reasonable agreement with the experimental values.