In order to solve the problems of traditional sinter quality prediction model, such as using single feature selection method and having no background of process mechanism, which results in low model prediction accuracy and lack of interpretability, a GA-BiLSTM prediction model with feature optimization is proposed. First, the optimal feature set is selected through various feature selection methods and combined with the sintering process mechanism, then GA is used to optimize BiLSTM, and finally the optimal feature set is used as the input of the GA-BiLSTM model to predict the FeO content in sinter. The GA-BiLSTM model with feature optimization was compared with other models. The results show that the prediction error of the established model is low, and the prediction accuracy for FeO mass fraction in sinter is as high as 94% within the allowable error range of ±0.5%, which may provide a new guiding direction for improving the quality of sinter.
最大互信息系数(maximal information coefficient,MIC)主要用来衡量多个变量之间的相关程度.MIC[14]主要思想是均匀划分变量,得到各自的划分区域,然后依次借助动态规划算法和近似最大互信息算法,求出各自区域的最大互信息值.MIC适合处理强非线性的烧结过程数据.因此,本文用该算法求出与烧结矿中FeO含量相关的特征集A.最大互信息系数MIC定义如下:
LSTM可以处理和预测时序特征长的数据,但它只能依靠历史数据进行预测,忽视了未来数据的信息.BiLSTM由2个信息传递方向相反的LSTM组成,相对于LSTM,双向长短期记忆网络(bidirectional long short-term memory,BiLSTM)能够结合未来数据的信息,实现对数据的全面分析,进而对数据进行更加准确的预测.BiLSTM的结构如图8所示.
ZhangB, ZhouJ M, LiM. Prediction of sinter yield and strength in iron ore sintering process by numerical simulation[J]. Applied Thermal Engineering, 2018, 131: 70-79.
[2]
LiuS, LyuQ, LiuX J, et al. Synthetically predicting the quality index of sinter using machine learning model[J]. Ironmaking & Steelmaking, 2020, 47(7): 828-836.
[3]
DuS, WuM, ChenX, et al. An intelligent control strategy for iron ore sintering ignition process based on the prediction of ignition temperature[J]. IEEE Transactions on Industrial Electronics, 2020, 67(2): 1233-1241.
[4]
NiuL L, LiuZ J, ZhangJ L, et al. Prediction of sinter chemical composition based on ensemble learning algorithms[J]. Journal of Sustainable Metallurgy, 2023, 9(3): 1168-1179.
[5]
XiaG L, WuZ X, LiuM Y, et al. Prediction interval estimation of sinter drum index based on light gradient boosting machine and kernel density estimation[J]. Ironmaking & Steelmaking, 2023, 50(8): 909-920.
DingCheng-yi, ChangRen-de, GuoSheng-lan, et al. Research status of intelligent control of sintering process and prediction of metallurgical properties of sinter[J].Iron & Steel, 2024,59(4):42-56.
[8]
HuJ, WuM, ZhangP, et al. Prediction performance improvement via anomaly detection and correction of actual production data in iron ore sintering process[J]. IEEE Transactions on Industrial Informatics, 2020, 16(12): 7602-7612.
[9]
LiuS, LyuQ, LiuX J, et al. A prediction system of burn through point based on gradient boosting decision tree and decision rules[J]. ISIJ International, 2019, 59(12): 2156-2164.
ZhangZhi-feng, LiuXiao-jie, LiXin, et al. Prediction of FeO content in sinter based on MIV-GA-BP mode[J].China Metallurgy, 2022,32(10):75-81.
[12]
LiuS, LiuX J, LyuQ, et al. Comprehensive system based on a DNN and LSTM for predicting sinter composition[J]. Applied Soft Computing, 2020, 95: 106574.
ZhangJun-hong, ShenFeng-man, XieAn-guo. The application of G-BP in FeO content prediction during sintering[J]. Journal of Northeastern University(Natural Science), 2002, 23(11):1073-1075.
JiXun-sheng, JingTian-tian, XiongNian-yun. Study on FeO content of sinter prediction[J]. Computer Simulation, 2015,32(10):318-322.
[17]
LiD C, HuangW T, ChenC C, et al. Employing box plots to build high-dimensional manufacturing models for new products in TFT-LCD plants[J]. Neurocomputing, 2014, 142: 73-85.
[18]
SimonN, TibshiraniR. Comment on “detecting novel associations in large data sets” by Reshef, et al, science dec 16, 2011[EB/OL]. (2014-01-29)[2024-01-10].