Application of Adaptive Neuro Fuzzy Inference System (ANFIS) for prediction the hardness of CK45 based on hot rolling parameters
نویسندگان
doi
10.71644/ admt.2025.1195749چکیده
Rolling stands as a crucial manufacturing technique that offers the dual benefit of enhancing steel's mechanical characteristics. Given the substantial time investment and financial burden associated with rolling experiment setups, implementing predictive models for mechanical properties can enhance precision while reducing both temporal and monetary costs. This study conducted hot rolling experiments on CK45 steel across two distinct environments. The specimens underwent rolling at five different temperature levels and five varying work-roll rotation speeds, maintaining consistent reduction percentages. Following the rolling process, the samples were rapidly cooled in ambient air and cold water conditions, with hardness measurements obtained using specialized testing equipment. The research employed the Adaptive Neuro-Fuzzy Inference approach to forecast hardness values based on operational parameters. The model utilized rolling temperature and rotational speed of the rollers as input variables, while the hardness measurements post-quenching in both air and water environments served as output data. The analysis yielded R² values exceeding 0.99 between measured and predicted results for both environments, demonstrating ANFIS's effectiveness in accurately predicting sample hardness across various rolling speeds and temperatures.