Development of ANFIS Model for Flow Forming of Solution Annealed H30 Aluminium Tubes

Abstract

Modeling of the cold flow forming process for manufacturing of tube shaped solution annealed H30 Aluminum alloy has been considered in this present study. Three inputs (feed-speed ratio, roller in-feed and roller axial stagger) and three outputs viz. spring back, ovality and internal diameter have been considered for the present study.. Adaptive network-based fuzzy inference system (ANFIS) in Matlab platform has used for modeling purposes and its performance is compared with regression model. ANFIS has completely outperformed the regression models. Percentage accuracy in predicting all the three responses are found to be very high with ANFIS models. Prediction of ovality against the test data using regression analysis is found to be extremely erroneous. It indicates that additional process parameters are involved in predicting ovality which are not captured during the experimentation.

Application of Adaptive Neuro-Fuzzy Inference System for Prediction of Mean Thickness in Backward Metal Flow Forming Process

“…Previously ANFIS approach has applied to various metal forming processes. Bikramjit et al [13] used ANFIS to model the dimensional accuracy like inside diameter, spring-back, and flow-formed tubes' ovality. The ANFIS model outperformed the regression model in the prediction of the inside diameter, springback and ovality.…”