摘要
介绍了采用RBF神经元网络建立电炉冶炼的碳温终点工艺指标与其影响因素间的神经元网络模型,其基本思路为:分析电炉的工艺、原料和操作特点,从而确定模型,再对模型进行训练和学习,用电炉实际运行的结果参数验证模型的准确性。该模型对现场的数据进行了自学习和预报,预报结果逐渐与实际接近。实际运行结果表明,该模型有着更好的收敛特性和很强的自学习能力,运行准确可靠,预报的结果有很高的精度和实用性,能较好地指导生产实践。
The paper describes the application of RBF neural network to establishing the neural network model for the final carbon temperature parameters and influential factors of EAF melting. The basic concept is to analyze EAF process, raw material and operation characteristics to define the model, which is trained and studies, and the accuracy of which is verified with actual parameters in EAF running. The model studies and predicts field data and the results come to approach the reality. Actual results show the model, with better convergence characteristics and strong self-adaption capability, accurate and reliable, gives results of high precision and usefulness and provides suitable instructions for production.
出处
《天津冶金》
CAS
2013年第A01期79-82,共4页
Tianjin Metallurgy
关键词
电炉
终点预报
终点控制
RBF神经网络
EAF
end-point prediction
end-point control
RBF neural network
