Anfis Based Modeling for Prediction of Surface Roughness during Machining of Glass Fiber Reinforced Epoxy Composites

Recently, the worldwide globalization and erotic development in the technological field has created immense market competition. Industries are now concerned on focusing towards refined product quality and increased productivity. In a mass production line, selection of appropriate parameter setting is indeed essential to avoid compromise in terms of quality as well as productivity. Product quality generally consists of multiple features which may be conflicting in nature depending on the requirements. Hence, achieving high quality product is definitely a challenging job. It is fact that various quality features are mutually correlated and assignment of priority importance of individual quality features is uncertain and vague due to subjective judgment of the decision-makers. Therefore, fuzziness arises and may adversely affect the solution. Surface roughness plays an important role in determining the interaction between the real object and surrounding environment. Decrease in surface roughness usually increases manufacturing costs exponentially, which results in a trade-off between the manufacturing cost of a component and its performance in application. Direct and on-line measurement of the surface roughness is very difficult. So there is indeed a need to develop a robust, autonomous and accurate predictive system. In this context, this paper highlights the application of integrated intelligent techniques i.e. neural network and the fuzzy inference system called Adaptive Neuro-Fuzzy Inference System (ANFIS) for prediction modeling of surface roughness during machining of GFRP composites. An experimental data set has been obtained by taking machining parameters like spindle speed, feed rate and depth of cut as input; and surface roughness of the machined composite IJMMD © Academic Research Journals (India), pp. 1-15 2 International Journal of Materials, Manufacturing and Design (IJMMD) product has been treated as output. Experimental data have been utilized for prediction-modeling of the surface roughness with an accuracy of 91%.