点击次数：

论文类型：期刊论文

发表时间：2008-10-01

发表刊物：4th International Conference on Continuous Casting of Steel in Developing Countries

收录刊物：SCIE、CPCI-S

卷号：15

页面范围：605-614

ISSN号：1006-706X

关键字：mould; transient heat transfer; inverse problem; neural network

摘要：Accurate and real-time detection of mould transient heat transfer is a strong assurance for producing good initial solidification strand and accurately controlling the solidification process in continuous casting. To solve the conflict between accuracy and real-time, based on the well used technologies of mould temperature measurement and heat transfer numerical simulation, the detection method for measuring the mould transient heat transfer corresponded to both stable and unsteady mould processes was developed. The inverse problem algorithm was used to resolve the problem of accuracy, and the neural network technology was introduced to improve the real-time issue. The results showed that the method of the measured temperature data + inverse algorithm + neural network + direct algorithm, could promote the detecting accuracy of mould transient heat transfer further more, meanwhile, satisfy the online measuring requirement. With the realization of this online method, the distributions of mould plates temperature and heat flux reflecting the actual working conditions will be obtained, which can provide the more accurate heat flux conditions between mould and strand than the methods used before. Furthermore we can obtain the 3D distribution of solidifying shell thickness and the mould flux thickness and lubrication status, and can provide the quantitative data support for predicting the strand cracks and optimizing and evaluating the thermal behavior of mould flux.