To solve the bottleneck problem of difficult real-time prediction of the iron recovery during the hydrogen-based mineral phase transformation and magnetic separation of refractory iron ores, a soft measurement model of fuzzy dynamic stochastic configuration network was proposed. Non-convex membership functions were constructed using kernel density estimation, and the parameters of the fuzzy system were optimized by combining the max-min algorithm and the generalized projection gradient descent algorithm, which improved the smoothness of membership functions and the stability of inference. An adaptive weight search strategy based on error variation rate was proposed to dynamically adjust the parameter range, improve the node generation quality, and shorten the model training and prediction time. Experimental results indicate that in the high-dimensional complex function approximation task, the RMSE (0.003 94) of the proposed model is significantly lower than that of SCN (0.025 68) and SPSCN (0.013 34); in the industrial soft measurement task of magnetic separation recovery, compared with the traditional SCN model, the prediction accuracy of the proposed model has reached 99.3%, and the prediction time is shortened by 60.5%. This paper provides a high-precision and high-efficiency means for the real-time monitoring of the magnetic separation recovery in hydrogen-based mineral phase transformation and provides an innovative solution for the closed-loop optimization of process parameters and the modeling of complex industrial processes.
ZhaoLi-qun, WangChun-nyu, ZhangMin, et al. Current exploration status and supply-demand situation of iron ore resources in China mainland [J]. Geology and Exploration, 2020, 56(3): 635-643.
[3]
LiG R. The Chinese iron ore deposits and ore production [M]. London: Iron Ores and Iron Oxide Materials, 2018: 3-34.
GuoJuan, CuiRong-guo, ZhouQi-zhong, et al. Outlook and overview of mineral resources situation of China in 2023 [J]. China Mining Magazine, 2024, 32(1): 12-19.
[6]
SunY S, ZhuX R, HanY X, et al. Iron recovery from refractory limonite ore using suspension magnetization roasting: a pilot-scale study [J]. Journal of Cleaner Production, 2020, 261: 121221.
[7]
NingJ L, GaoP, YuanS, et al. Highly efficient and green separation of iron from complex low-grade polymetallic ore via hydrogen-based mineral phase transformation [J]. Powder Technology, 2024, 433: 119177.
[8]
LiW B, WangH Y, HanY X, et al. A zero-carbon emission approach for the reduction of refractory iron ores: mineral phase, magnetic property and surface transformation in hydrogen system [J]. International Journal of Hydrogen Energy, 2024, 89: 531-540.
[9]
TangZ D, LiP Y, GaoP, et al. Minerals phase transformation by hydrogen reduction technology: a new approach to recycle iron from refractory limonite for reducing carbon emissions [J]. Advanced Powder Technology, 2022, 33(12): 103870.
[10]
KrizhevskyA, SutskeverI, HintonG E. ImageNet classification with deep convolutional neural networks [J]. Communications of the ACM, 2017, 60(6): 84-90.
[11]
SuykensJ A K, VandewalleJ. Least squares support vector machine classifiers [J]. Neural Processing Letters, 1999, 9(3): 293-300.
[12]
LiawA, WienerM. Classification and regression by randomForest [J]. R News, 2002, 2(3):18-22.
[13]
ZhangQ, SunY S, GaoP, et al. Hydrogen-based mineral phase transformation of bastnaesite: detailed assessment of physicochemical properties and flotation behavior[J]. Chemical Engineering Journal, 2024, 500: 156992.
[14]
WangD G, LiM. Stochastic configuration networks: fundamentals and algorithms [J]. IEEE Transactions on Cybernetics, 2017, 47(10): 3466-3479.
[15]
LiK, QiaoJ F, WangD H. Fuzzy stochastic configuration networks for nonlinear system modeling [J]. IEEE Transactions on Fuzzy Systems, 2024, 32(3):948-957.
[16]
QiaoJ H, ChenY X. Stochastic configuration networks with chaotic maps and hierarchical learning strategy [J]. Information Sciences, 2023, 629: 96-108.
[17]
QiaoJ H, XiongN K, QiaoJ Y. Multivariable adaptive decoupling control based on stochastic configuration networks with serial-parallel switching distribution [J]. Information Sciences, 2025, 686: 121352.
GuoFeng-fang, HanYue-xin, ZhangQi, et al. Experimental study on hydrogen-based mineral phase transformation followed by magnetic separation of iron ore from Shilu, Hainan [J]. Metal Mine, 2024 (2): 152-156.
[20]
ZhangQ, SunY S, HanY X, et al. Pyrolysis behavior of a green and clean reductant for suspension magnetization roasting [J]. Journal of Cleaner Production, 2020, 268: 122173.
[21]
ZhuX R, HanY X, LiuP F, et al. Reduction mechanism of the porous hematite in limonite ore magnetization roasting[J]. Mineral Processing and Extractive Metallurgy Review, 2024, 45(2): 85-90.
YuJian-wen, HanYue-xin, LiYan-jun, et al. Experimental study on utilization of crude concentrate from an Anshan-type hematite ore by suspension magnetization roasting-magnetic separation [J]. Journal of Central South University (Science and Technology), 2018, 49(4): 771-778.
[24]
SunY, BabuP, PalomarD P. Majorization-minimization algorithms in signal processing, communications, and machine learning [J]. IEEE Transactions on Signal Processing, 2017, 65(3): 794-816.
[25]
JainP, KarP. Non-convex optimization for machine learning [J]. Foundations and Trends in Machine Learning,2017,10(3/4):142-336.
[26]
BotevZ I, GrotowskiJ F, KroeseD P. Kernel density estimation via diffusion [J]. The Annals of Statistics, 2010, 38(5): 2916-2957.
[27]
ZhangD Q, ZhangN, YeN, et al. Hybrid learning algorithm of radial basis function networks for reliability analysis [J]. IEEE Transactions on Reliability, 2021, 70(3): 887-900.
[28]
ZhangY, ZhouQ, SunC X, et al. RBF neural network and ANFIS-based short-term load forecasting approach in real-time price environment [J]. IEEE Transactions on Power Systems, 2008, 23(3): 853-858.
[29]
PodderD. CO₂ emission by vehicles [EB/OL]. (2020-08-05) [2024-10-01].