End sulfur content prediction method of molten iron in KR based on Kmeans–BP neural network
-
摘要: 針對KR工序終點鐵水硫含量預測問題，提出一種基于Kmeans聚類分析和BP神經網絡（BPNN）相結合的建模方法。首先，通過Kmeans聚類對KR工序生產數據進行模式識別和分類，構建不同工況特征的數據集；然后，基于BP神經網絡，針對不同數據集訓練預測模型；最后，將不同數據集的預測模型進行集成，形成最終的終點鐵水硫含量預測模型，實現對不同鐵水條件和工況條件的預測。利用某鋼鐵企業實際生產數據，分別用基于脫硫反應動力學、BP神經網絡和Kmeans–BPNN方法建立的預測模型，對KR工序終點鐵水硫含量進行預測。結果表明，Kmeans–BPNN的KR工序終點硫含量預測模型的精度顯著高于脫硫反應動力學和BP神經網絡的預測模型。Abstract: In the steel manufacturing process, an accurate prediction of end sulfur content in KR is crucial for steadily controlling sulfur content in molten iron and improving steel properties. Regarding the end sulfur content prediction in the KR process, an integrated modeling method based on Kmeans clustering analysis and the BP neural network (BPNN) is proposed in this paper. As an unsupervised learning method, Kmeans clustering analysis can complete data classification according to the similarity of influencing factors instead of depending on target values. The BPNN, as a supervised learning method, can effectively explore the correlation between influencing factors and target values. The integration of these two methods can realize information exploration of data from different dimensions. Based on this understanding and the actual production data in one steel plant, the prediction model of end sulfur content in KR based on Kmeans–BPNN is studied. First, datasets of different operating conditions are constructed according to the pattern recognition and classification of production data in the KR process through Kmeans clustering. By establishing the relation curve between the number of clustering centers and the mean error of clustering results and selecting the adjacent positions to 10% of the maximum mean error difference, the number of Kmeans clustering centers is confirmed as five. Then, the prediction model is trained by different datasets based on the BPNN. The input layer and hidden layer have five nodes, and the output layer has one node in the BPNN-based prediction model of end sulfur content in KR. A piecewise linear function is selected as the activation function, and the maximum number of training is fixed at 1,000. Finally, the prediction models of different datasets are integrated and formulated in the final prediction model of end sulfur content in molten iron, realizing the prediction of different molten iron conditions and operating conditions. To test and verify the effectiveness and accuracy of the prediction model based on the Kmeans–BPNN method, the end sulfur content prediction of molten iron in KR is performed by applying prediction models based on desulfurization reaction kinetics, routine BPNN, and Kmeans–BPNN using the same training and testing datasets. The prediction results indicate that the end sulfur content prediction in KR based on the Kmeans–BPNN method is significantly more accurate than that of the prediction model based on the desulfurization reaction kinetics and the routine BPNN model.
-
Key words:
- KR /
- sulfur content /
- prediction /
- Kmeans clustering /
- BP neural network /
- desulfurization reaction kinetics
-
圖 1 BP神經網絡構建KR工序終點硫含量預測模型示意圖
Figure 1. Diagram of the BP neural network for prediction model of sulfur content in KR
圖 2 聚類中心個數與聚類結果誤差的關系曲線.（a）聚類結果誤差均值；（b）聚類結果誤差均值差值
Figure 2. Relationship curve between the number of clustering centers and error of clustering results: (a) mean error of clustering results; (b) difference of mean error of clustering results
圖 3 基于Kmeans–BPNN的終點硫含量預測模型路線圖
Figure 3. Diagram of the sulfur content prediction model based on Kmeans–BPNN
圖 4 不同模型KR工序終點硫含量預測精度. (a)脫硫反應動力學模型; (b) BP神經網絡模型; (c) Kmeans–BPNN模型
Figure 4. Prediction hit rates of the end sulfur content in KR based on different models: (a) desulfurization reaction kinetics; (b) BP neural network; (c) Kmeans–BPNN
表 1 脫硫反應達到平衡時鐵水溫度與終點硫含量的關系
Table 1. Relationship between molten iron temperature and end sulfur content when desulfurization reaction reaches equilibrium
Temperature / K End sulfur content / 10–7 1673 2.22 1623 1.18 1573 0.60 
下載: 導出CSV
表 2 KR脫硫終點硫含量的影響因素的數據分布
Table 2. Data distribution of the influencing factors of the end sulfur content in KR desulfurization
Influencing factors Data range Average Standard deviation Incoming molten iron [S] content /10–5 10–106 36.24 12.38 Incoming molten iron temperature /℃ 1261–1470 1381.89 32.33 Molten iron weight / t 200–223.8 213.01 6.24 Desulfurizer consumption /kg 659–2978 1423.36 309.24 Mixing time /min 5–26 14.78 1.93 Notes: The [S] content in molten iron is measured as a mass percentage.
下載: 導出CSV
表 3 KR脫硫反應動力學、BP神經網絡和Kmeans–BPNN方法預測命中率匯總 %
Table 3. Summary of prediction hit rates based on desulfurization reaction kinetics, BPNN, and Kmeans–BPNN
Error range /10–5 Desulfurization
reaction kineticsBPNN Kmeans–BPNN [–1,1] 23.83 25.14 30.23 [–2,2] 44.08 51.24 55.51 [–3,3] 62.33 73.69 77.82 [–4,4] 76.86 91.60 95.18
下載: 導出CSV
www.77susu.com<span id="fpn9h"><noframes id="fpn9h"><span id="fpn9h"></span> <span id="fpn9h"><noframes id="fpn9h"> <th id="fpn9h"></th> <strike id="fpn9h"><noframes id="fpn9h"><strike id="fpn9h"></strike> <th id="fpn9h"><noframes id="fpn9h"> <span id="fpn9h"><video id="fpn9h"></video></span> <ruby id="fpn9h"></ruby> <strike id="fpn9h"><noframes id="fpn9h"><span id="fpn9h"></span> -
參考文獻
[1] Feng K, Xu A J, He D F, et al. Case-based reasoning method based on mechanistic model correction for predicting endpoint sulphur content of molten iron in KR desulphurization. Ironmaking Steelmaking, 2020, 47(7): 799 doi: 10.1080/03019233.2019.1615307 [2] Chen W, Wang B X, Chen Y. Prediction for the sulfur content in pig iron of blast furnace by combining artificial neural network with genetic algorithm. Adv Mater Res, 2010, 143-144: 1137 doi: 10.4028/www.scientific.net/AMR.143-144.1137 [3] Zhang J H, Xie A G, Shen F M. Optimization and analysis of sulfur content in hot metal based on neural network. J Mater Metall, 2006, 5(2): 86 doi: 10.3969/j.issn.1671-6620.2006.02.002張軍紅, 謝安國, 沈峰滿. 基于神經網絡對鐵水硫含量的優化和分析. 材料與冶金學報, 2006, 5(2):86 doi: 10.3969/j.issn.1671-6620.2006.02.002 [4] Zhang H S, Zhan D P, Jiang Z H. Final sulfur content prediction model based on improved BP artificial neural network for hot metal pretreatment. Iron Steel, 2007, 42(3): 30 doi: 10.3321/j.issn:0449-749X.2007.03.008張慧書, 戰東平, 姜周華. 基于改進BP神經網絡的鐵水預處理終點硫含量預報模型. 鋼鐵, 2007, 42(3):30 doi: 10.3321/j.issn:0449-749X.2007.03.008 [5] Zhou H, Tang Z Y, Wen B J, et al. Application of statistical analysis, Deng's relevancy, and BP neural network for predicting molten iron sulfur in COREX process. Int J Chem React Eng, 2020, 18(12): 20220122 [6] Wang W, Chen W L, Ye Y, et al. Application of neural network to predict sulphur content in hot metal. Iron Steel, 2006, 41(10): 19 doi: 10.3321/j.issn:0449-749X.2006.10.004王煒, 陳畏林, 葉勇, 等. 神經網絡在高爐鐵水硫含量預報中的應用. 鋼鐵, 2006, 41(10):19 doi: 10.3321/j.issn:0449-749X.2006.10.004 [7] Fang Z Q, Sun Y H. Desulphurization analysis of slag with high basicity and prediction model for sulfur distribution ratio. Steelmaking, 2014, 30(1): 46 doi: 10.3969/j.issn.1002-1043.2014.01.012方忠強, 孫彥輝. 高堿度精煉渣脫硫分析及硫分配比預測模型. 煉鋼, 2014, 30(1):46 doi: 10.3969/j.issn.1002-1043.2014.01.012 [8] Qiu D, Dai W J, Zhang N. Research on prediction model of end sulfur content for converter smelting. Adv Mater Res, 2014, 1037: 26 doi: 10.4028/www.scientific.net/AMR.1037.26 [9] Al-Jamimi H A. Prediction of sulfur content in desulfurization process using a fuzzy-logic based model. Solid State Phenom, 2019, 287: 80 doi: 10.4028/www.scientific.net/SSP.287.80 [10] Zhang G, Liu H C, Li P L, et al. Load prediction based on hybrid model of VMD–mRMR–BPNN–LSSVM. Complexity, 2020, 2020: 6940786 [11] Liu Y F, Wang Q, Zhang X D, et al. Using ANFIS and BPNN methods to predict the unfrozen water content of saline soil in Western Jilin, China. Symmetry, 2018, 11(1): 16 doi: 10.3390/sym11010016 [12] Ahmed W, Muhammad K, Siddiqui F I. Predicting calorific value of thar lignite deposit: A comparison between back-propagation neural networks (BPNN), gradient boosting trees (GBT), and multiple linear regression (MLR). Appl Artif Intell, 2020, 34(14): 1124 doi: 10.1080/08839514.2020.1824091 [13] Liu Y, Li B F. Bayesian hierarchical K-means clustering. Intell Data Anal, 2020, 24(5): 977 doi: 10.3233/IDA-194807 [14] Geng X Y, Mu Y K, Mao S L, et al. An improved K-means algorithm based on fuzzy metrics. IEEE Access, 8: 217416 [15] T?rn?uc? C, Gómez-Pérez D, Balcázar J L, et al. Global optimality in k-means clustering. Inf Sci, 2018, 439-440: 79 doi: 10.1016/j.ins.2018.02.001 [16] Ming Y W, Zhu E, Wang M, et al. Scalable k-means for large-scale clustering. Intell Data Anal, 2019, 23(4): 825 doi: 10.3233/IDA-173795 [17] Moodi F, Saadatfar H. An improved K-means algorithm for big data. IET Softw, 2022, 16(1): 48 doi: 10.1049/sfw2.12032 [18] Hu H, Zhang J F, Li T. A novel hybrid decompose-ensemble strategy with a VMD–BPNN approach for daily streamflow estimating. Water Resour Manag, 2021, 35: 15 [19] Jiang X P, Wang Z T, Zhu H, et al. Hydraulic turbine system identification and predictive control based on GASA–BPNN. Int J Miner Metall Mater, 2021, 28(7): 1240 doi: 10.1007/s12613-021-2290-6 [20] Lu Y, Xing Y, Li X, et al. A new approach of CMT seam welding deformation forecasting based on GA–BPNN. Frattura ed Integrità Strutturale, 2020, 14(53): 325 [21] Cai B, Pan G L, Fu F. Prediction of the postfire flexural capacity of RC beam using GA–BPNN machine learning. J Perform Constr Fac, 2020, 34(6): 04020105 doi: 10.1061/(ASCE)CF.1943-5509.0001514 [22] Sun Y B, Xiao J, Liu H, et al. Deformation prediction based on an adaptive GA–BPNN and the online compensation of a 5-DOF hybrid robot. Ind Robot, 2020, 47(6): 915 doi: 10.1108/IR-01-2020-0016 [23] Wang D H, Sun J Y, Dong A P, et al. Prediction of core deflection in wax injection for investment casting by using SVM and BPNN. Int J Adv Manuf Tech, 2019, 101(5-8): 2165 doi: 10.1007/s00170-018-3069-4 [24] Yang Z, Zhou Q, Wu X, et al. Detection of water content in transformer oil using multi frequency ultrasonic with PCA–GA–BPNN. Energies, 2019, 12(7): 1379 doi: 10.3390/en12071379 [25] Yang X, Zhou Q, Wang J, et al. Predictive control modeling of ADS's MEBT using BPNN to reduce the impact of noise on the control system. Ann Nucl Energy, 2019, 132(10): 576 [26] Cai B, Sun X, Wang J, et al. Fault detection and diagnostic method of diesel engine by combining rule-based algorithm and BNs/BPNNs. J Manuf Syst, 2020, 57(7): 148 -
點擊查看大圖
圖(4) / 表(3)
計量
- 文章訪問數: 371
- HTML全文瀏覽量: 180
- PDF下載量: 60
- 被引次數: 0
