Prediction of deformation resistance for hot rolled strip based on ANFIS

A prediction model of deformation resistance for finishing stands during hot continuous rolling process based on ANFIS was proposed. The subtractive clustering algorithm was adopted to build an initial fuzzy inference system, so the initial structure and parameters of fuzzy system were determined adaptively. Then the original prediction model of deformation resistance was established based on ANFIS, the hybrid algorithm was applied to training the parameters of fuzzy rules and the final prediction model was obtained. The modeling experiments were carried out and the results show that the new built model is effective for carbon steel, micro-alloyed steel, alloy steel and other steel grades, and it has better prediction accuracy than other four models including BP neural networks, Zhou Jihua's model, Baosteel 1880HSM model and SMS model.