An adaptive neuro-fuzzy inference system-based modelling to predict mechanical properties of hot-rolled TRIP steel

Abstract

A model based on adaptive neural network formalism coupled with fuzzy inference system has been developed to predict mechanical properties of hot-rolled TRIP steel. The developed model incorporates a wide range of data containing chemical compositions, thermo-mechanical processing parameters and mechanical properties of hot-rolled TRIP steel. A compact set of process variables has been selected as the model inputs for predicting tensile strength, yield strength, elongation and retained austenite under a given operating condition. The model predictions show that carbon, silicon and manganese content have a significant effect on the retained austenite which increases with the increased amount of these elements. The microalloying elements such as niobium and molybdenum have a little effect on the volume fraction of retained austenite. The present model provides a predictive platform for possible application of these artificial intelligence-based tools for automation, real-time process control and operator guidance in plant operation.