The  current  status  of  steel  scrap  utilization ,  classification ,  and  classification  standards  in china were analyzed , the current situation of steel scrap utilization and classification sorting were  ex- plored , and  the  existing  problems  were  put  forward.   On  this  basis ,  development  and application of steel scrap intelligent identification system and rapid detection technology were analyzed.  It is consid- ered that there is still a big gap between the development level of steel scrap classification and sorting technology and the actual demand of steel enterprises .  In the future , the coupling of the two technolo- gies must be  achieved  in  order  to  realize the  intelligent  smelting  of  electric  arc  furnace.  The  article points out that the classification and sorting of steel scrap in china is still relatively primary and exten- sive , it is not yet possible to achieve rapid and effective identification of the  composition of steel scrap.under the background of“ dual-carbon”, as china , s steel industry shifts from the stage of large-scale and high-speed development to the  new  stage  of green  low-carbon  and high-quality  development , the importance of steel scrap  quality  is  becoming  more  and  more  important .   The  utilization  level  of  steel scrap  should be improved as soon as possible , not only to improve the scrap ratio ofsteelmaking , more importantly , it is necessary to improve the technical level of steel scrap classification and sorting , so as to lay a foundation for the realization of intelligent smelting .

Although scrap bundles have many advantages , the diversity of scrap types has an impact on smelting .  considering the complexity of the steel plant environment and the  complexity of the  current scrap type recognition technology , the use of mobile devices to realiZe the accurate recognition of scrap bundles in complex scenes is of vital significance to improve the accuracy and productivity of smelting models .   The  dataset  was  enriched  by  adding  new  pictures  of  scrap  bundle  under  complex  lighting scenes in the original dataset , and the improved hybrid network model was applied to  study the  scrap bundle recognition algorithm.  The results of the study show that the improved Edge Next hybrid model has a better performance in the recognition scenarios .  on the experimental dataset , its test accuracy is improved by 2. 81%  compared to Mobilenetv3 ; one round of training time consumed is reduced by 16 seconds compared to  the  viT model ;  and  the  model  shows  better  convergence  speed  and oscillation amplitude during the training process .  In summary , the improved Edge Next model provides solid theo-retical support for improving the intelligent recognition of scrap bundles . 

The quality  disputes  during  inspection  of  scrap  steel  have  always  been  an important issue plaguing major steel enterprises .   In order to  solve the problems  such as the  great influence  of subjec - tive factors , the  difficulty  in  tracing  the  grading  process ,  and  quality  disputes  existing  in the  scrap steel quality  inspection  process ,  the  scrap  steel  intelligent  grading  system  based on  artificial  intelli- gence technology has emerged as the times require and has received great attention in the steel indus- try .  As a new thing , many enterprises have  doubts , incomplete  and unscientific  understandings , and even misunderstandings about the scrap steel intelligent grading system.  Based on the technical princi- ples  and  the  functions  of  the  technical  architecture ,  the  key  technological  breakthroughs  in  aspects such as the automatic collection of pictures , standard unification , the intelligent grading ofspecial ma- terial types like  briquettes , and  intelligent  deduction  of  impurities  were  expounded  on.   At  the  same time , it points  out  the  engineering  challenges  faced  in  aspects  such  as  the  identification  of  chemical compositions , the recognition of small material types , and the internal quality inspection of briquettes .Finally , it puts forward the development process of the application of the scrap steel intelligent grading system in enterprises and its future trends .

The semi-quantitative detection of metal coating thickness on scrap steel surfaces using laser- induced breakdown spectroscopy  (LIBs) was investigated , based on the characteristic regular changes in spectral intensity and ablation crater morphology during the laser ablation process .  Firstly , by stud- ying the morphology  of laser  ablation  craters ,  a  mathematical  model  was  established  to  correlate  the depth of the ablation crater with the number of laser pulses .   subsequently , by combining the judgment of the critical point of laser penetration through the coating , the thickness of the metal coating on the scrap steel surface is quantified.  Anovel method for the rapid assessment of coating thickness on scrap steel surfaces was provided.

Real-time identification of the  thickness  of the  scrap  steel  material  line  is  of  great  signifi- cance for its transportation and the electric arc furnace steelmaking process .  Currently , there is no al- gorithm research based on deep learning methods for this scenario.  Aiming at the above problems , the current mainstream segmentation model networks were  compared  and  PP-Liteseg for the  segmentation of the edge  contours  of  scrap  steel  was  used.    Furthermore ,  optimiZation  strate gies  such  as  Dice loss and LovasZsoftmax loss were introduced to improve the original cross-entropy loss function of PP-Lite- seg , achieving the best balance between global supervision and local optimiZation of local details .  The mean intersection over union  (mIoU)  of  segmentation  reaches 81 . 11% .   Finally , based  on  the  opti- miZed model , a calculation method for the height of the scrap steel material line was designed , which extracts key reference lines by using the segmentation results , enabling real-time monitoring and quan- titative evaluation of the  maximum  and  average  stacking  heights  of  scrap  steel.   The  experimental  re- sults show that this method has the ability of high-precision segmentation under complex working condi- tions and excellent generaliZation performance for different positions , providing reliable technical sup-port for the accurate  identification of the  thickness  of the  scrap  steel  material  line  in  the  electric  arc furnace steelmaking process .

The laser-induced  breakdown  spectroscopy  ( LIBS)  technology  to  analyZe the  evolution  of spectral characteristics of scrap steel with and without coatings  as  a  function  of laser pulse  count  was utiliZed.  The research reveals that for coated scrap steel , the spectral line intensity of coating elements initially increases and then decreases with the increase in pulse count , while the spectral line intensity of Fe element gradually increases .   For uncoated  scrap steel , the  spectral line  intensity  of Fe  element rapidly increases after the initial pulses and then stabiliZes .   Based on these patterns , using the  stand- ard deviation threshold of the normaliZed intensity of Fe element ( set at 0 . 02) to identify the presence of coatings on scrap steel surfaces was proposed.  Furthermore , by analyZing the changes in the normal- iZed intensity of coating elements , a method based on the cumulative value of normaliZed spectral line intensities to determine the type of coating element was introduced , where the element corresponding to the maximum cumulative value was identified as the coating element .  An effective technical means for the rapid identification and classification of coatings on scrap steel surfaces was provided.

Intelligent scrap steel grading can optimiZe the  scrap steel grading process of steel enterpri- ses and scrap steel bases in the process of purchasing scrap steel , improve the efficiency of scrap steel grading , and reduce the emotional quality of scrap steel grading .   For the  intelligent  scrap steel grad- ing , based on the original target detection algorithm , a method based on the target tracking algorithm was proposed to optimiZe the automatic termination of the grading function in the process of intelligent scrap steel grading. The target detection algorithm was used to first detect and identify the grading vehicles appea- ring in the gun-type camera  ( gun camera)  screen.   There may be incomplete vehicles , multiple grad- ing vehicles , etc .  , especially  in  these  special  cases ,  it  is  necessary  to  design  a  target  tracking  algo- rithm for vehicle identification to improve the accuracy of identifying scrap steel vehicles to be graded.The target tracking algorithm to realiZe the automatic termination function of the  scrap steel intelligent grading system was adopted.  compared with the original single target detection algorithm , the kalman filter and the Hungarian algorithm were used to solve the state prediction and trajectory matching asso- ciation of the targets detected in the image , so as to improve the robustness of the target tracking algo- rithm  in complex scenes .  The accuracy of the  automatic resolution function of the scrap steel intelligent grading system has been increased from 86%  to 93% .   The  multi-object tracking by associating every detection box  (ByteTrack) algorithm explored in this article has improved the multiple object tracking accuracy(MOTA) evaluation index by 17. s%  and 1s. 3%  respectively compared with the simple on- line and realtime tracking(SORT)  and deep learning for simple  online  and realtime tracking  (Deep- SORT) algorithms , and the IDF1  evaluation index has  improved by  16. 1%  and 9. 9%  respectively .  The automatic termination function  based  on  the  target  tracking  algorithm  can  improve  its  accuracy , especially in complex scenarios , and more accurately determine whether the grading vehicle is moving , so as to trigger whether the scrap steel grading of the current vehicle needs to be terminated.

With the rapid development of industry 4 . 0 and intelligent manufacturing , the  steelmaking industry is facing great opportunities  and challenges of transformation and upgrading .   The  application of steelmaking  intelligent  technology  not  only  improves  production  efficiency  and  optimiZes  product quality , but  also  significantly  reduces  energy  consumption  and  environmental  pollution ,  which  pro- motes the development of the steel industry in the direction of green , intelligent and sustainable.  The current status and future development trend ofsteelmaking intelligence was reviewed , and the integrat- ed application of artificial intelligence , big data , internet of things and other advanced technologies in the electric arc furnace , converter and refining process was focused on.  The key technologies of steel- making intelligence , including intelligent control system , data acquisition and monitoring technology , machine learning  algorithms , refining  process  optimiZation ,  etc.  ,  were  detailed.   It  also  summariZes the application cases of various types of intelligent technologies in improving the control of molten steel composition , smelting process precision , and energy utiliZation efficiency .  In addition , the challenges and bottlenecks ofsteelmaking intelligence in practical applications were discussed , such as data qual-ity , system integration , real-time and adaptability issues .  Finally , the future development direction of intelligent steelmaking technology was looked into , especially the potential in the fields of digital twin , automation control , intelligent prediction and optimiZation , aiming to  provide  reference  and guidance for the technological innovation and intelligent transformation of the steel industry .

Three persistent challenges in steelmaking-continuous casting process optimiZation were  ad- dressed : inefficient process modeling , imprecise parameter control , and inadequate  multi-process co- ordination.  Three core technological innovations were developed : intelligent ladle furnace  (LF) metal- lurgy , precision-controlled continuous casting solidification , and dynamic multi-process collaboration.Firstly , an intelligent LFcontrol system through the  integration of metallurgical principles with inter- pretable machine  learning  was  established.    This  system  enables  accurate  regulation  of molten  steel  temperature , alloy composition  ( si/Mn content) , and  argon  stirring parameters , substantially reduc - ing both material consumption and energy usage while eliminating traditional reliance on empirical ad- justments and repeated sampling.  secondly , for continuous casting optimiZation , a solidification cool- ing strategy based on steel phase-transformation characteristics was proposed.  Through noZZle arrange- ment optimiZation and secondary-phase  precipitation control , this  approach effectively mitigates  crack formation and segregation defects in microalloyed steels .  A predictive model combining principal com- ponent analysis with deep neural networks further enhances process control by guiding real-time param- eter adjustments .   To  resolve  production  coordination  challenges  in  complex  steelmaking  workshops , dynamic collaborative operation technologies rooted in metallurgical process engineering theory was de- veloped.  The implementation of quantitative  coordination metrics  ensures  efficient  material flow  man- agement , significantly improving operational synchroniZation across multiple processes .

The  process  of  steelmaking  and  continuous  casting   ( Scc)  was  accompanied  by  complex chemical reactions and physical changes , and has high uncertainty .  The production scheduling of the Scc  process  has always been one of the difficult problems in iron and steel enterprises .  In theory , this kind of problem can be reduced to the dynamic flexible flow shop scheduling problem , which is one of the hot issues in the academic research.   Genetic programming  (GP)  has  been successfully applied to the automatic design of heuristic rules for dynamic scheduling problems .  However , in traditional GP, the selection of the parent solution through the tournament mode will slow down the convergence speed of the  algorithm.   A two-stage  semantic  selection  mechanism  was  proposed  to  accelerate  GPconver-gence by providing more convergent parent solutions for GP.  In  addition ,  in  order  to  retain  the  valid information in  the  parent  individual more  efficiently , a  correlation based variation  strate gy was  de- signed , which replaces the negative correlation subtrees in the individual tree with randomly generated subtrees .   The  performance  of the  proposed  algorithm  was  compared  and  analyZed in  8 scenarios  with different configurations , and  the  results  show  that  the  proposed algorithm has a good competitiveness compared with the existing algorithms .

The mold is a critical component of continuous  casting , playing a vital role  in maintaining the safety and efficiency  of the  casting  process .   Due  to  the  complex  heat  transfer  mechanisms  within the mold , direct investigation of its thermal behavior is quite challenging .  Therefore ,the interior-point method to solve the inverse heat flux distribution problem along the broadside of the mold and performs longitudinal fitting of the results was employed.  The results indicate that the relative root mean square error between the  calculated  and  measured  temperatures  is  only  0. 94%  , demonstrating  the  accuracy and effectiveness of the proposed inverse problem model.  The solution to the inverse problem reveals a consistent relationship between the non-uniform heat flux distribution at the mold-casting interface and the uneven temperature distribution of the copper plate.  Further longitudinal fitting of the results  shows that the heat flux increases , and  its  fluctuations  along  the  broadside  become  more  pronounced  as  the distance to the meniscus decreases .   Conversely , both the heat flux and its fluctuations decrease as the distance from the meniscus increases .

In the blast furnace ironmaking process , the silicon content in hot metal is a crucial indica- tor for evaluating  production  stability  and  iron  quality .    However ,  the  prediction  accuracy  of  silicon content in hot metal  is  affected ,  due  to  the  complex  mechanism  of  blast  furnace  ironmaking ,  multi - field and multi-phase coupling , data loss and noise.  with the development of industrial intelligence ,industrial foundation models have shown application potential.  Based on the concept of industrial foun- dation models , a silicon content prediction method that integrates probabilistic distributions and graph convolutional networks was proposed.  The proposed method consists of a multi-channel univariate fea- ture extractor , a graph neural network , and a predictor.  The multi-channel univariate feature extractor employs a variational autoencoder to extract probabilistic representations of variables .  The graph neural network captures  variable  interactions  and  compensates  for  missing  data.    Real-world  data  collected from a blast furnace  digital  twin  system  was  utiliZed  for  validation.   Comparative  experiments  demon- strate that the  proposed  method  improves  the  prediction  accuracy  of  silicon  content  by  4. 31%   com- pared to existing approaches .  The novel practical reference for the application of industrial foundation models in blast furnace ironmaking quality prediction was provided.

In the hot rolling production process , the imbalance in industrial data distribution increases the difficulty of diagnosing  strip  crown  abnormalities , which  severely  affects  product  quality .   To  ad- dress this issue , a method combining the Extra Trees algorithm with the resampling technique SMOTE- Tomek link was proposed.  The approach effectively solves the  class  imbalance  problem  in  multi-class data.   On this basis , an improved chaotic genetic algorithm  (ICGA)  was  used to optimiZe the  model/s hyperparameters and determine the optimal hyperparameter combination.   Model interpretability analy- sis was  conducted  using  the  local  interpretable  model-agnostic  explanations   ( LIME)  method ,  which further reveals the key process parameters affecting the strip crown and their contributions .  Experimen-tal results  show  that  the   proposed  method  achieves  accuracy ,  recall ,  precision ,  and   F1   Score  of0.9907,0.9908,0.9910 , and 0. 9908 , respectively , significantly outperforming traditional mod- els , which verifies the effectiveness and advantages of the  method in strip crown diagnosis and provides support for fault diagnosis in the hot rolling production process .

In the process of hot strip rolling , the temperature rise  curve  of billet in reheating furnace has a significant impact on the quality of products .   Due to the bad conditions in the furnace , the actual temperature of the billet is difficult to be directly monitored , so it is necessary to establish a real-time temperature prediction model for the billet heating process .  The traditional mechanism model based on partial differential equation  (PDE) is usually difficult to meet the needs of real-time prediction due to its high computational  complexity ; The  neural  network  model  has  the  characteristics  of  low  accuracy and relying on a large number of training labels based on the actual temperature of billet , so it has not been well applied in practice.  Aprediction model of hot rolling billet temperature based on generaliZed physics-informed  (GPINN)  was  proposed.   Firstly , the  position  of the  billet  in  the  reheating  furnace was tracked according to the action signal of the walking beam , and the temperature of the whole fur- nace was  predicted  by  cubic  spline  interpolation ;  secondly ,  the  PDE describing  billet  heating  was solved by PINN, and  combined with the“ branch  net-trunk  net ”structure  of deep  operator  network (DeepoNe t ) , the different initial temperatures and specific time and space positions of billet were co- ded respectively , which effectively realiZes the real-time prediction of billet temperature in the heating process with different initial temperatures ; Finally , the effectiveness of the method was verified by the case  study of a real steel plant .  Compared with the traditional neural network method and mechanism - based method , GPINN integrates  the  advantages  of  physical  information  and  neural  network ,  better captures the heat conduction characteristics in the process of billet heating and improves the interpret- ability and prediction accuracy of the model.