New method for rapid evaluation of spheroidisation and inoculation grade of hypereutectic cast iron

• Published in: International journal of cast metals research (Online) Vol. 17; no. 3; pp. 152 - 156
• Main Authors: Li, Dayong, Ma, Xuliang, Zhang, Yutong, Qiu, Xianbo
• Format: Journal Article
• Language: English
• Published: Taylor & Francis 01.01.2004
• Subjects: ARTIFICIAL NEURAL NETWORKS; INOCULATION GRADE; MULTI-PARAMETRIC MODELS; NODULAR CAST IRON; SPHEROIDISATION
• ISSN: 1364-0461; 1743-1336
• DOI: 10.1179/136404604225017492

Spheroidisation and inoculation are important steps in the production of qualified nodular cast iron, and technologies to evaluate spheroidisation and inoculation grade rapidly and accurately are highly desirable. Since the solidification of nodular cast iron is different from that of grey cast iron, either the cooling curve or the eutectic expansion curve can be used to determine the spheroidisation and inoculation grade. An artificial neural network multi-parametric approach to evaluate the effect of spheroidisation and inoculation is reported, using temperature and dimension characteristics on both cooling and eutectic expansion curves. With the new method, the spheroidisation and inoculation grade can be predicted within 3 min and the predicting network can be retrained when the production conditions change. The selection and measurement of the input parameters, the structure and training of the neural network model and evaluation of an example are discussed.