摘要
以主元分析方法和新型ESN(回声状态网络)算法为核心,研究了转炉终点静态预测模型。通过对某钢厂转炉生产数据的主元分析,建立了ESN模型,同时将ESN模型与传统的BP和RBF神经网络模型进行了对比研究。结果表明,使用ESN建立的模型比传统的BP网络模型和RBF网络模型,在钢水温度预测方面精度分别提高了0.85%和0.45%,在钢水碳质量分数预测方面精度分别提高了0.45%和0.19%,能够有效的对转炉终点碳含量和温度进行预测,从而为转炉炼钢过程提供更准确的操作指导。
This paper is concerned with the BOF end-point static prediction model based on PCA algorithm and new ESN （echo state network）. The PCA processed is utlized to established the ESN model by anlasy- ing BOF production data of an steelwork. In addition, a comparative study was done with the traditional BP and RBF neural network model. The results show that the ESN model achieves encouraging 0.85 % and 0.45% in the aspect of the temperature prediction of molten steel,and 0.45% and 0.19% in the aspect of the carbon mass fraction prediction of molten steel compared to the tranditional BP and BRF neural network model. This indicate that the ESN model can effectively predict the carbon content and temperature of BOF end-point, and can guide the steelmaking process.
出处
《辽宁科技大学学报》
CAS
2009年第4期360-364,共5页
Journal of University of Science and Technology Liaoning
