摘要
针对高炉炼铁是一个动态过程,具有大延迟,工况复杂的特性。采用LSTM-RNN模型进行硅含量预测,充分发挥了其处理时间序列时挖掘前后关联信息的优势。首先根据时间序列趋势及相关系数选择自变量,并采用复杂工况的实际生产数据进行验证。然后用程序自动求解最优参数进行硅含量预测。最后将LSTM-RNN模型与PLS模型及RNN模型的结果进行对比,验证该方法的优势。研究发现LSTM-RNN模型预测误差稳定,预测精度较高,比传统的统计学及神经网络方法取得了更好的预测精度。
The ironmaking in blast furnace,with large delay and complex conditions,is a dynamic process.Thetraditional methods for prediction of silicon content in hot metal are mostly based on the statistics or the simpleneural networks,leading to lower accuracy.However,a model based on the long short-term memory-recurrent neuralnetwork(LSTM-RNN)is proposed to exploit the characteristics of the mutual information before and after the timeseries in this paper.The independent variables are selected according to the time series trend and the correlationcoefficient.After that,the silicon content is predicted according to the input variables by optimizing the parametersautomatically.In order to verify the constructed model,the extremely complex production data is used to comparethe LSTM-RNN and simple RNN models.Remarkably,the result shows that the prediction error of LSTM-RNNmodel is stable and the prediction accuracy is high.
作者
李泽龙
杨春节
刘文辉
周恒
李宇轩
LI Zelong;YANG Chunjie;LIU Wenhui;ZHOU Heng;LI Yuxuan(College of Control Science and Engineering, Zhejiang University, Hangzhou 310027, Zhejiang, China)
出处
《化工学报》
EI
CAS
CSCD
北大核心
2018年第3期992-997,共6页
CIESC Journal
基金
国家自然科学基金项目(61290321)~~
关键词
预测
动态建模
神经网络
高炉炼铁
硅含量
prediction
dynamic modelling
neural network
ironmaking
silicon content
