Element Detection in Scrap Steel Using Portable LIBS and Sparrow Search Algorithm-Kernel Extreme Learning Machine (SSA-KELM)
HUANG Xiao-hong1, 2, LIU Xiao-chen1, 2, LIU Yan-li3*, SONG Chao1, 2, SUN Yong-chang1, 2, ZHANG Qing-jun4
1. College of Artificial Intelligence, North China University of Science and Technology, Tangshan 063210, China
2. Hebei Key Laboratory of Industrial Intelligent Perception, Tangshan 063210, China
3. Physical and Chemical Testing Center of Material Technology Research Institute of HBIS Group Co., Ltd., Shijiazhuang 050000, China
4. Comprehensive Test and Analysis Center of North China University of Science and Technology, Tangshan 063210, China
Abstract：Elemental content detection is necessary for efficiently utilizing scrap steel, an important raw material for electric furnace steelmaking.In this study, a new method combining the sparrow search algorithm optimized kernel extreme learning machine (SSA-KELM) and laser-induced breakdown spectroscopy (LIBS) was proposed to analyze and model the element contents of 12 groups of steel samples, including medium-low alloy steel and low alloy steel. First, the portable LIBS spectrometer was used to collect laser-induced breakdown spectroscopy data of 12 different steel scrap samples in the range of 170~400 nm, and 28 different locations on the surface of each sample were selected for detection to reduce experimental fluctuations. The k-value check was used to eliminate gross errors, and the remaining data was averaged to obtain 336 groups of average spectrum data from 12 sample groups. Then, the obtained spectral data was subjected to baseline correction and normalization to reduce the baseline fluctuation. Multiple related spectral lines of the target elements were selected as the input features of the model, and the spectral data was divided into training and testing sets. A random sample from each steel type was selected as the model's testing set, and the remaining data was used as the model's training set. The sparrow search algorithm was used to optimize the parameters of the kernel extreme learning machine (KELM), and the model was established for the related elements. The final model for C, Cu, Mn, Cr, Ni, Si, V, Al, and Ti elements had an average correlation coefficient (R2) and root mean square error (RMSE) of 0.996 and 0.016, respectively, on the validation set. The quantitative analysis performance of the single variable calibration model and the genetic algorithm optimized KELM (GA-KELM) multivariate calibration model were compared, and the results showed that the SSA-KELM model had significant improvements in all indicators compared to the single variable calibration model and GA-KELM model. The combination of KELM and Sparrow search algorithm as a multivariate model can effectively reduce the interference of multiple factors on the target elements and enhance the performance of the quantitative analysis. It can rapidly and accurately detect various element contents in steel scrap on-site by combining it with the portable LIBS system.
黄晓红，刘晓辰，刘艳丽，宋 超，孙永长，张庆军. 便携式LIBS结合SSA-KELM的废钢成分定量分析方法[J]. 光谱学与光谱分析, 2024, 44(09): 2412-2419.
HUANG Xiao-hong, LIU Xiao-chen, LIU Yan-li, SONG Chao, SUN Yong-chang, ZHANG Qing-jun. Element Detection in Scrap Steel Using Portable LIBS and Sparrow Search Algorithm-Kernel Extreme Learning Machine (SSA-KELM). SPECTROSCOPY AND SPECTRAL ANALYSIS, 2024, 44(09): 2412-2419.
[1] Xuan Y, Yue Q. Resources, Conservation and Recycling, 2017, 120: 186.
[2] Dworak S, Fellner J. Resources, Conservation and Recycling, 2021, 173: 105692.
[3] Schroeder W W, Van Niekerk J J, Dicks L, et al. Spectrochimica Acta Part B: Atomic Spectroscopy, 1971, 26(6): 331.
[4] Prosr P, Polansky R, Pihera J. Proceedings of 2013 IEEE International Conference on Solid Dielectrics (ICSD), 2013: 740.
[5] Pereira G R, Rocha H S, Anjos M J, et al. Proceedings of 2006 IEEE Nuclear Science Symposium Conference Record, 2006, 6: 3932.
[6] Díaz D, Hahn D W, Molina A. Spectrochimica Acta Part B: Atomic Spectroscopy, 2017, 136: 106.
[7] Li X, Yang S, Fan R, et al. Optics & Laser Technology, 2018, 102: 233.
[8] Kim Y, Fabre C, Cauzid J. Spectrochimica Acta Part B: Atomic Spectroscopy, 2022, doi: 10.1016/j.sab.2022/106397.
[9] Bhatt C R, Goueguel C L, Jain J C, et al. Applied Optics, 2017, 56(28): 7789.
[10] Court A. Raman-LIBS, a Journey From Mars to Earth Via the Moon. 2019 IEEE Aerospace Conference, 2019, 1, doi: 10.1109/AERO.2019.8741619.
[11] Ruan F, Qi J, Yan C, et al. Journal of Analytical Atomic Spectrometry, 2017, 32(11): 2194.
[12] Sezer B, Unuvar A, Boyaci I H, et al. Journal of Cereal Science, 2022, 104: 103435.
[13] Afgan M S, Hou Z, Wang Z. Journal of Analytical Atomic Spectrometry, 2017, 32(10): 1905.
[14] Sheng L, Zhang T, Wang K, et al. Chemical Research in Chinese Universities, 2015, 31(1): 107.
[15] Zhu C, Lv J, Liu K, et al. Microchemical Journal, 2021, 168: 106408.
[16] Rifai K, Özcan L, Doucet F, et al. Spectrochimica Acta Part B: Atomic Spectroscopy, 2020, 165: 105766.
[17] GUO Zhi-wei, SUN Lan-xiang, ZHANG Peng, et al(郭志卫，孙兰香，张 鹏，等). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2019, 39(1): 278.
[18] Mei Y G, Cheng S S, Hao Z Q, et al. Plasma Science and Technology, 2019, 21(3): 171.
[19] Huang G B, Zhu Q Y, Siew C K. Neurocomputing, 2006, 70(1-3): 489.
[20] Huang G B, Ding X, Zhou H. Neurocomputing, 2010, 74(1-3): 155.
[21] Xue J, Shen B. Systems Science & Control Engineering, 2020, 8(1): 22.
[22] Sun L, Yu H. Spectrochimica Acta Part B: Atomic Spectroscopy, 2009, 64(3): 278.
