Robust optimisation of chemical composition of as cast chromium white cast iron using a green sand mould

Abstract

An L ₉ (3⁴) orthogonal experiment was performed for the optimisation of chemical composition of chromium white cast iron. The differences of the models by the orthogonal design and radial base function artificial neural network (RBFANN) were investigated. The results show that Si significantly influences the hardness, and Cu and Cr are main factors influencing the impact toughness. The predicted and simulated results indicate that the RBFANN can not only be used to establish robust model for the orthogonal experiment data but also the RBFANN model is rather better than the quadratic regression. Therefore, the RBFANN method could provide a more accurate model than orthogonal design method for the range of parameters investigated.

Keywords:

• CHEMICAL COMPOSITION
• CHROMIUM WHITE CAST IRON
• ORTHOGONAL DESIGN
• ARTIFICIAL NEURAL NETWORK