To address the challenge of high carbon emissions in China's steel industry and promote green and low-carbon transformation, this study systematically reviews the carbon emission reduction pathways, technological advancements, and empirical achievements of long-process steelmaking under the framework of steel-chemical integration technology. It further reveals the potential and challenges in achieving near-zero carbon emissions. In terms of process optimization, steel-chemical integrated production utilizes the blast furnace gas (CO) in a directional manner, transforming the carbon resources traditionally emitted through combustion into chemical products such as formic acid and ethylene glycol, thus achieving a "use instead of discharge" carbon cycle mode. This can reduce carbon emissions by up to 79.68 kg/t in the steel industry and 259.16 kg/t in the chemical industry (calculated by molten steel). Simultaneously, the ironmaking process, through the collaborative application of hydrogen-based direct reduction iron (DRI), iron coke technology, and high-pellet ore smelting, can reduce the carbon emissions of molten iron from 1.7 t/t to 0.8 t/t. The BOF process, through low-carbon raw materials, energy substitution, and low-carbon smelting technologies, can reduce its carbon emissions from 159.6 kg/t to -165.95 kg/t (calculated by molten steel). Furthermore, dynamic models based on carbon flow analysis indicate that through the multi-path collaboration of steel-chemical integrated production, CCUS (carbon capture, utilization, and storage), and scrap ratio optimization, the carbon emissions of the BF-BOF long process can be reduced from the current 1 625.35 kg/t to 287.73 kg/t (calculated by molten steel). While the electric arc furnace (EAF) short process has an ultra-low-carbon potential of 64 kg/t, the long process will remain the primary decarbonization method until 2035.

Based on public data from 2014 to 2023, this paper analyzes 11 factors affecting CO₂ emissions in the iron and steel industry using the Generalized Divisia Index Method (GDIM). It identifies five main drivers of CO₂ emissions: industrial output, energy intensity, CO₂ intensity of energy consumption, crude steel production, and investment intensity. The research reveals that CO₂ emissions in the iron and steel industry have undergone three distinct stages: rapid growth, stabilization, and decline. Production expansion emerges as the primary driver of CO₂ emission growth, while improvements in energy efficiency and technological advancements play a pivotal role in emission reduction. Through the construction of integrated dynamic models and complex interaction models, the study quantifies the cumulative and synergistic effects of these drivers, highlighting the significant impact of technological progress and energy structure optimization on emission reduction. A multi-scenario forecasting model projects China's CO₂ emission trends from 2019 to 2035 under business as usual scenario, low-CO₂ scenario, and sustainable scenario. The findings indicate that business as usual scenario is unlikely to achieve the dual- CO₂ targets, whereas the low-CO₂scenario and sustainable scenario could reduce emissions to 1 239 million and 1 050 million tons, respectively, by 2035 through targeted policy interventions and technological innovation. To facilitate these outcomes, the paper proposes policy recommendations, including enhancing the CO₂ reduction management system, accelerating technological innovation, optimizing the energy structure, and promoting international cooperation. These measures aim to provide a scientific foundation and actionable pathway for the steel industry's transition toward a low-CO₂ future.

Under the background of low-carbon transformation of China's iron and steel industry, the development of electric arc furnace (EAF) short process faces multiple constraints such as resources and energy. From a provincial perspective, a comprehensive evaluation system is constructed, which encompasses 20 indicators, such as scrap supply, energy supply, and economic feasibility. By combining the analytic hierarchy process (AHP) and entropy weight method (EWM) for weight assignment and employing the TOPSIS model, the development potential of electric furnace process is evaluated. The results show that the development potential of short process steelmaking in Hebei, Guangdong, Shandong and other provinces is high, while the development potential of Xizang Autonomous Region, Beijing and Hainan is low. Guangdong and Hebei provinces lead in the resource dimension due to their mature scrap supply chains and large consumer markets. In terms of economic feasibility, the northwest, north, and southwest regions excel with their reliable power supply and cost advantages. Meanwhile, coastal and southwestern regions have established strong institutional support through policies such as capacity replacement subsidies. In the future, the development of EAF short processes in China requires the implementation of regionally differentiated strategies to promote the large-scale development of EAF steelmaking and achieve the industry's deep decarbonization goals.

In the journey to promote the national industrialization process and build a modern industrial system，the steel industry plays a pivotal role. However，this process is also accompanied by the generation of a large amount of solid waste，posing a severe challenge to China ecological environment.Comprehensive analyses were conducted on the physical and chemical properties， current utilization status， and potential value of typical refractory solid wastes in the steel industry，namely steel slag， zinc-bearing dust， and solid residues from flue gas treatment. The application prospects of steel slag in fields such as building materials are broad，but its resource utilization rate is low，at only 30%，urgently requiring the development of efficient utilization technologies and the expansion of application channels. The potential of steel slag for sensible heat recovery and carbon sequestration has not been fully utilized，and the paper proposes that future research should consider these factors comprehensively to promote green and efficient utilization. The treatment of zinc-containing dust currently relies mainly on pyrometallurgical processes，but the high energy consumption and costs are prominent issues that necessitate technological innovation to reduce treatment costs and improve recovery efficiency. For waste derived from flue gas treatment，the application of resource utilization technologies has achieved certain results in fields such as building materials and precious metal recycling，but still faces challenges in terms of economic value and industrialization process. In summary，through technological innovation and comprehensive resource utilization，it is not only possible to enhance the environmental friendliness of the steel industry but also to promote the coordinated development of economic benefits and environmental protection，providing strong support for the sustainable development of China steel industry.

China's steel industry is entering a critical period of structural adjustment, facing multiple constraints such as market, environmental protection, and energy carbon emissions. Under the background of global climate governance, how to comprehensively respond to domestic and foreign carbon constraints has become a topic that steel enterprises must face and deal with positively. This study systematically reviews the development process of China's carbon market, analyzes the current situation and trends of the mandatory carbon market and the voluntary emission reduction market, and proposes carbon trading strategy suggestions for the carbon emission constraints faced by steel enterprises. The research shows that the steel industry will face the problem of rising export costs under the framework of carbon trade barriers mainly constructed by Europe and the United States. After entering the national carbon market, steel enterprises with weak carbon control capabilities will have a quota gap in 2026. By constructing an optimal compliance model and comparing the economic performance of different carbon trading strategies, it is found that the strategy of purchasing quotas in advance and in excess can significantly reduce the long-term compliance costs of enterprises. Parameter sensitivity analysis shows that free quotas, crude steel output, and emission intensity are highly sensitive parameters, and small changes can lead to significant fluctuations in total compliance costs. The construction of China's carbon market needs to seek a dynamic balance between responding to international rule competition and ensuring industrial competitiveness. It is comprehensively suggested that steel enterprises should establish a carbon asset management system as soon as possible, set up a dedicated carbon asset management department, formulate dynamic trading strategies, explore carbon financial tools, accelerate the research and development of low-carbon technologies and green transformation, and establish an intelligent carbon asset management and control platform to cope with market changes.

As CO₂ emissions from pre-iron processes account for over 70% of the total emissions in the iron and steel industry, carbon reduction in pre-iron processes is crucial for achieving the industry's "carbon peak" and "carbon neutrality" goals. An overview of development status of existing hydrogen metallurgy technology is summarized, followed by a description of the pure hydrogen shaft furnace reduction process route proposed by CISRI (China Iron and Steel Research Institute Group). Through the development of pure hydrogen shaft furnace reduction technology and core equipment, CISRI established the first industrial demonstration line for pure hydrogen shaft furnace reduction in the world. Through optimization of process parameters and continuous hydrogen supply, a metallization rate of 97.0%-99.4% was achieved. Melting tests of carbon-free sponge iron from the pure hydrogen shaft furnace were conducted using well-chamber furnace, electric arc furnace, and vacuum induction furnace. Results show that the melting cycle for cold-pressed balls and blocks is shorter in the electric arc furnace, while gas content decreases significantly in the vacuum induction furnace. The high-purity iron obtained by vacuum induction furnace exhibits a purity (mass fraction) of 99.9%. The pure hydrogen metallurgy technology developed by CISRI offers a solution for achieving low or zero CO₂ emissions in green hydrogen metallurgy in China. Hydrogen metallurgy is anticipated to exhibit greater competitiveness than carbon metallurgy in the future.

As a superior reducing agent and clean energy source, hydrogen (H₂) presents an ideal alternative to carbon-based reducing agents. However, controversies persist concerning hydrogen utilization efficiency and carbon reduction effectiveness in hydrogen-enriched smelting processes. The reduction characteristics of H₂ and CO under varying temperatures were systematically investigated through comparative analysis. Hydrogen utilization rates during blast furnace hydrogen-enriched smelting were subsequently evaluated through both experimental studies and industrial trials. Results show that hydrogen utilization rates in industrial trials range from 43.15% to 44.40%, which are higher than those in experimental studies. Finally, limiting factors of blast furnace hydrogen-enriched smelting were analyzed, and carbon reduction effects under different operation strategies were calculated. Results indicate that maintaining a constant theoretical combustion temperature through oxygen enrichment limits hydrogen injection and results in insignificant carbon reduction. Conversely, preheating hydrogen effectively mitigates the problem of excessively low theoretical combustion temperature during hydrogen-enriched smelting, enabling large-scale hydrogen injection and significant carbon reduction.

Driven by the "dual carbon" strategic objectives，the iron and steel industry，as a major carbon emission sector，underwent technological innovations in low-carbon ironmaking that became crucial pathways for achieving carbon neutrality. This led to the development of several key technologies including fluxed pellet production with large pellet ratio blast furnace operation，composite iron coke，hydrogen-enriched carbon cycling blast furnace，and hydrogen-based shaft furnace short process. The application of fluxed pellets and high pellet ratio in blast furnace optimized the burden structure by partially replacing sinter. Composite iron coke served as highly reactive material that substituted for conventional coke. The core principle of hydrogen-enriched blast furnace technology involved injecting hydrogen or hydrogen-rich reducing gas through tuyeres to replace carbon-based fuels. Hydrogen-based shaft furnace short process emerged as an alternative ironmaking process that utilized hydrogen as reducing agent，demonstrating advantages in efficiency，environmental friendliness and energy conservation. All the aforementioned low-carbon ironmaking processes demonstrated carbon emission reduction potential. A systematic investigation of low-carbon ironmaking technology systems was required，involving both the advancement of traditional blast furnace processes and the development of alternative ironmaking technologies. The selection of appropriate low-carbon technological pathways was projected to contribute significantly to achieving carbon neutrality in China's steel industry.

Sintering optimization ore blending is a critical step in steelmaking production, aiming to achieve efficient resource utilization, cost control, and optimized smelting performance through multi-mineral blending. This process plays a significant role in improving resource utilization rates, enhancing sinter quality, and reducing production costs and energy consumption. However, with the increasing scarcity of high-quality iron ore resources and the growing complexity of raw material structures, traditional blending methods, which rely on static linear models and empirical decision-making, struggle to address challenges such as ore composition fluctuations, multi-objective optimization, dynamic coupling of multi-processes, and nonlinear constraints. This is particularly evident when attempting to simultaneously optimize chemical composition, cost, and metallurgical performance. Intelligent algorithms, such as genetic algorithms and particle swarm optimization, by integrating data-driven approaches with mechanistic models, can significantly enhance multi-objective optimization capabilities, providing new pathways to overcome the bottleneck of dynamic multi-objective optimization. The research progress in sintering optimization ore blending is systematically reviewed, the applicability of traditional methods and intelligent algorithms is compared, and the technical directions for constructing an intelligent low-carbon blending system are proposed, addressing core issues such as dynamic responses, cross-process coordination, and data-mechanism fusion. The goal is to provide theoretical support and practical reference for the steel industry in achieving resource-intensive utilization and intelligent transformation.

The sintering process is one of the important sources of CO emissions in the steel industry，accounting for 40% to 50% of the total emissions in the steel industry. The generation mechanism，emission law and emission reduction path of CO in sintering flue gas have been systematically summarized. CO is mainly derived from the incomplete combustion of solid fuels and high-temperature reduction reactions，and its emission shows periodic characteristics，that is，from low-temperature，low-oxygen and low-CO in the initial stage，gradually to low-temperature，low-oxygen and high-CO，and finally presents high-temperature，high-oxygen and low-CO. This paper summarizes the existing CO emission reduction strategies and research results from four perspectives: source emission reduction，process control，end-of-pipe treatment and collaborative treatment，and makes a prospect. On the basis of systematically combing the existing CO emission reduction strategies，data-driven methods have been introduced into the field of sintering flue gas treatment. The characteristic maps of the generation and change of CO in sintering flue gas are used，and corresponding big data models are constructed according to different steel mills，so as to realize the online optimization and predictive control of sintering process parameters and make process adjustments in advance. This provides a new research method for sintering flue gas CO emission reduction on the basis of sorting and summarizing.

In the hydrogen-rich direct reduction process, the adjustment of reduction process parameters directly affects the quality of the final product, direct reduced iron (DRI). It is imperative to elucidate the influence laws of reduction temperature, reduction gas flow rate and reduction gas ratio (H₂/CO) on the reduction behavior of pellets, as this is of great significance for the optimization of the hydrogen-rich reduction process and the enhancement of product quality. Hematite with a total iron mass fraction of 66.67% was used as the iron-containing raw material to prepare pellets with basicity of 0.3. Referring to the Midrex method's gas-based direct reduction standard, the pellets were directly reduced by gas under different reduction conditions. The metallization rate, pulverization rate, expansion rate, compressive strength after reduction and phase changes were systematically studied. It has been demonstrated that increasing the reduction temperature from 750 ℃ to 900 ℃ results in a substantial enhancement of the metallization rate from 86.63% to 96.66%. At 900 ℃, the generation of a substantial number of iron whiskers within the pellet leads to an anomalous expansion rate of over 25%. It was also determined that a low reduction gas flow rate (10 L/min) led to a reduction in the reduction rate of the pellet. Furthermore, the internal structure of the pellet was found to be non-homogeneous, resulting in increased pulverization and expansion. However, it was also demonstrated that by increasing the gas flow rate appropriately, all the indexes were significantly improved. An increase in the hydrogen ratio in the reduction gas has been demonstrated to enhance the reduction performance of the pellets significantly. In an environment devoid of carbon monoxide (CO), the metallization rate and compressive strength of the reduced pellets attained 96.83% and 615 N/P, respectively, while the pulverization rate and reduction expansion rate were recorded at 0.11% and 4.11%, respectively. In addition, XRD and SEM analyses demonstrated that the iron grains of the reduced pellets under a pure hydrogen atmosphere exhibited uniformity and good crystallinity, and the metallic iron precipitated out in a laminated form, thus indicating that the pellets possessed a more stable structure. The research results provide a theoretical foundation for the optimization of the hydrogen-rich direct reduction process.

The sintering-blast furnace ironmaking process is identified as the highest carbon-emitting process in the iron and steel industry. Its low-carbon development is of critical importance for achieving the "Carbon Peaking and Carbon Neutrality" goals of the industry. In recent years，the production of low-carbon blast furnace burden materials via cold-bonded pellet technology has garnered extensive attention. A comprehensive review is presented on research advancements in preparing low-carbon cold-bonded pellets using conventional iron ore fines and iron-containing solid wastes. It is revealed that current cold-bonded iron ore pellets are predominantly applied in ironmaking furnaces with low mechanical strength requirements，such as rotary kilns，rotary hearth furnaces，and converters. To meet the low-carbon production requirements of blast furnaces，the strengthening mechanism of solid wastes as auxiliary materials in enhancing the strength indices of cold-bonded pellets is thoroughly discussed. Additionally，a systematic analysis is conducted on the energy consumption and carbon emissions associated with the application of cold-bonded pellet technology across different ironmaking equipment. The results show that this technology significantly reduces energy consumption and carbon emissions in blast furnace ironmaking while effectively utilizing iron-containing solid waste resources，thus delivering notable environmental benefits. Based on these findings，prospects for the development of low-carbon multi-source solid waste-iron ore fines cold-bonded pellets are put forward，aiming to provide references for the green and low-carbon development of the sintering-blast furnace ironmaking process.

During continuous casting processes，cast blooms are prone to quality defects such as corner cracks，longitudinal and transverse surface cracks，internal segregation，and bulging deformation. The analysis and control of these defects are critical for enhancing product qualification rates and remain central research topics in both academia and industry. Among these defects，strain field distribution data serves as a fundamental basis for predicting cracks and damage，thereby underscoring the academic and engineering significance of monitoring cross-sectional strain in cast blooms. To address the challenges and precision demands of strain monitoring in continuous casting，an innovative solution based on the self-attention conditional generative adversarial network (SACGAN) is proposed. The model uses the temperature field generated by real-time operating parameters as input and the strain field of the slab section as output to explore its feasibility as an efficient alternative to traditional finite element analysis (FEA). Experimental results demonstrate that this approach effectively bridges the gap between physical mechanism analysis and model design，providing rapid and precise strain prediction capabilities for continuous casting processes. Furthermore，the deep learning framework exhibits notable advantages and broad application prospects in addressing complex material mechanics problems compared to conventional FEA.

Driven by China's "Dual Carbon" strategy and international carbon tariff pressures，the steel industry is accelerating its green transformation toward electric arc furnace(EAF) short-process steelmaking for high-quality sheet production. While steel enterprises are currently exploring demonstration projects involving hydrogen-based direct reduced iron(DRI)，green electricity coupling，and intelligent process upgrades，critical technical challenges remain，including carbon-phosphorus-oxygen reaction regulation，residual element purification，and full-process dynamic coordination. Compared with converter steelmaking，EAF steelmaking exhibits inferior dephosphorization efficiency due to constraints from charged raw materials，molten bath geometry，and suboptimal thermal/dynamic conditions. Specifically，carbon-oxygen reactions significantly impact dephosphorization kinetics in high-scrap-ratio EAF steelmaking. The low carbon content in raw materials leads to insufficient reaction intensity and weakened bath agitation，resulting in incomplete phosphorus removal. To solve these problems，optimization of temperature control，molten pool agitation，and strengthening of model applications are necessary，along with coordinated improvement of the furnace structure to balance efficient dephosphorization and high-quality steel production in EAF. Additionally，attention should be paid to the coordinated regulation of the composition and properties of dephosphorization slag，development of a slag system with low melting point and high phosphorus capacity，and improvement of the phosphorus distribution ratio between slag and metal by optimizing process parameters to enhance the stability of dephosphorization.Process parameter optimization is also critical to enhance slag-metal phosphorus partition ratios and stabilize dephosphorization performance. However，multi-variable，nonlinear，and stochastic disturbances during steelmaking-particularly raw material fluctuations，electrode heating disturbances，and varying dephosphorization kinetics in high-scrap-ratio EAF processes-pose challenges for establishing precise process control models. Future research should focus on multi-phase reaction mechanisms in EAF steelmaking，co-design of slag systems and processes，full-process dynamic optimization，and development of intelligent control technologies. These advancements will promote the industrial-scale application of high-scrap-ratio EAF short processes in premium steel manufacturing，ultimately supporting China's green transition in steelmaking and enhancing global competitiveness.

Carbon peaking and carbon neutrality are major national strategic decisions. Under the comprehensive implementation of the new development philosophy, traditional industries are accelerating their transition toward green and low-carbon practices. The heat treatment area of medium and thick plates plays a decisive role in the final performance of the product, but the production involves multiple processes, and the traditional production mode has scattered control parts, dense operator configuration, and insufficient coordination between processes, resulting in limited production rhythm and efficiency, which is difficult to meet the intelligent development of metallurgical industry and the demand for multi-variety, small-batch, and customized production modes. Meanwhile, heat treatment of medium and thick plates is also a process with high energy consumption and emissions, and its intelligent upgrade is an important way to achieve low-carbon production. To address these challenges, a furnace health assessment model, medium and thick plates seal recognition system and a physics-informed neural networks (PINN) based material temperature prediction model were developed, forming a high-precision intensive production technology based on collaborative perception, breaking the information island between various processes, achieving full process tracking and optimization control of the heat treatment area, and improving the automation and intelligence level of production. Practical production results show that after implementing the proposed technology, the energy consumption decreases by 18.38% year-on-year, output increases by 9.24%, heat treatment performance qualification rate improves by 1.95%, and workforce requirements reduces by 10%, achieving the goals of workforce optimization, energy savings, cost reduction, and product quality improvement.

The mechanical strength of pellet is the key index to evaluate its quality, and it is also the key control target in the pellet production process. Because of the long detection cycle and the lag of control response, it is very important to predict the pellet performance in real time and accurately for optimizing the pellet production process. An efficient gradient lifting decision tree model (GBDT) based on Pearson and Spearman algorithm was proposed to predict the pellet performance, and combined with machine learning (CatBoost) algorithm to build the pellet optimal blending model. The model was trained and tested by using the field production data and laboratory basic performance data. The prediction results show that the model has high accuracy in predicting the compressive strength and falling strength of green balls. The mean absolute error (MAE) for predicting the compressive strength of green balls is 0.163 7 N and R-square (R²) is 0.921 2, the MAE of green ball falling strength is 0.170 6 N and R² is 0.912 8, the MAE for predicting the compressive strength of oxidized pellets is 24.812 3 N and R² is 0.863 5. Model prediction combined with optimized ore blending model, 95% of the ore blending optimization cases are verified to have improved effect on the pellet production process.

In response to the bottleneck issues of poor real-time performance and weak anti-interference in current coke powder moisture detection methods，this study integrates SHAP interpretability analysis with multi-feature fusion technology to construct a coke powder moisture recognition model. The model mitigates the interference of brightness in texture feature extraction by employing a brightness correction algorithm. It extracts 42-dimensional multi-scale features，including brightness and texture of coke powder images，using wavelet transform and constructs a comprehensive moisture prediction model based on the extreme gradient boosting algorithm (XGBoost). SHAP analysis revealed that the moisture content of coke powder exhibited a significant negative correlation with image brightness and a significant positive correlation with texture roughness. The analysis identified the mean value of coke powder images，brightness contrast，and wavelet standard deviation as key moisture-sensitive parameters. A feature recursive elimination algorithm was applied to select the optimal 9-dimensional feature subset，which was then used to construct the moisture prediction model. The prediction accuracy metrics include mean absolute error (MAE) of 0.219，mean squared error (MSE) of 0.12，and mean absolute percentage error (MAPE) of 3.235×10^-2. These results significantly improved prediction speed and accuracy，facilitated precise control of coke powder moisture，and ensured sintering combustion efficiency and the quality of sintered ore.

The silicon content in hot metal is a critical indicator for gauging blast furnace operational efficiency and product quality. Precise prediction and control of this content are vital for ironmaking process optimization. Relying solely on empirical models and the professional knowledge of operators is insufficient to meet the requirements of high precision and adaptability in complex production environments. With the development of big data and artificial intelligence, data-driven machine learning methods have provided new ideas for predicting silicon content. An intelligent decision-making system that integrates deep learning and reinforcement learning has been developed for the precise and intelligent control of silicon content in hot metal. A CNN-Informer hybrid model was used for prediction, achieving an accuracy of 91%. The SHAP method was employed to analyze the interpret ability of the model, identifying the main influencing factors and providing a basis for control strategies. A self-learning, continuously optimized rule base was built by integrating expert experience and historical data. Rule effectiveness was quantified via confidence assessment and dynamically optimized using a Q-learning algorithm. The system automatically generates optimal control recommendations based on real-time data and model prediction results. Operators can precisely control the silicon content in hot metal by making minor adjustments to key parameters according to the prompts. After the implementation of the system, the fluctuation of silicon content in hot metal has been significantly reduced, with a stability rate exceeding 80% and a utilization coefficient reaching 2.73 t/(m³·d). Production stability has been enhanced, equipment failure rates have decreased, and economic benefits have been significantly improved.

As a recyclable green resource，scrap steel is an important alternative raw material for iron ore and plays a key role in promoting the sustainable development of the steel industry. The grade of scrap is related to the production cost and quality，and its accurate classification and rating are very important for scrap recycling. The existing research on scrap classification generally has problems such as poor detection effect of small target scrap and background interference，which affect the accuracy of scrap classification. Therefore，an improved scrap recognition algorithm YOLO-SNBP for dense small targets based on YOLOv5 is proposed. Firstly，a small target detection layer is added to improve the recognition effect of the model on dense small target scrap in the acceptance scene. Secondly，the BiFormer attention mechanism is introduced to enhance the model's ability to extract small target features in complex backgrounds. Finally，the Soft-NMS algorithm is used to replace the traditional NMS to reduce the problem of missed detection of scrap due to overlap. The YOLO-SNBP model is trained and verified on the self-built scrap data set，and different detection algorithms are compared and analyzed. The experimental results show that compared with the basic model，the P(Precision)，R(Recall) and P_mA (mean Average Precision) values of the YOLO-SNBP model are increased by 2.8%，7.2% and 7.4%，respectively. Compared with the previous algorithms，the P_mA values are increased by 29.0%，18.2% and 18.8%，respectively. It shows significant advantages in the accurate identification of scrap steel，and provides effective support for the efficient rating during acceptance.

The accuracy of the preset rolling force in the skin pass mill of a cold-rolled galvanizing line directly determines whether the product performance and surface quality meet customer requirements. Based on actual production data from a cold-rolled galvanizing line, this study employs machine learning algorithms to predict the preset rolling force. By comparing multiple machine learning models, the light gradient boosting machine (LightGBM) algorithm was identified as the optimal prediction model. To address challenges such as the high dimensionality of hyperparameters requiring optimization and the coupling between parameters in the LightGBM algorithm, the northern goshawk optimization (NGO) algorithm was applied to optimize its hyperparameters. This resulted in the improved NGO-LGBM algorithm, achieving a prediction R² value of 0.826, a mean absolute percentage error (E_MAP) of 9.47%, and a root mean square error (E_RMS) of 452 kN. A self-learning model was further introduced to refine the rolling force prediction, compensating for the limitations of data-driven models in handling small-sample predictions and aligning with real-time equipment and process variations. Experimental results demonstrate that after self-learning correction, the E_MAP decreases from 9.47% to 6.65%, and the R² value increases to 0.913. The optimization significantly improves the fitting of small-sample data with larger prediction errors and reduces the average prediction error.

The steel industry is one of the major sources of energy consumption and carbon emissions. In the context of promoting low-carbon intelligent manufacturing, achieving efficient control of multivariable nonlinear coupled systems through optimization algorithms has become a key challenge. Based on the theoretical model of such systems, a dynamic-weight-based multi-objective reinforcement learning optimization algorithm (DW-MORL) is proposed. Within the Markov decision process (MDP) framework, the algorithm defines states, actions, and reward functions, and employs a policy gradient method for multi-objective learning. A dynamic weight allocation mechanism is introduced to adjust the reward combination ratios according to the historical performance of each objective, followed by normalization at each update step. Experimental results show that, compared with traditional PID control and particle swarm optimization methods, the proposed DW-MORL algorithm demonstrates superior performance in typical multi-objective control tasks of steel reheating furnaces: reducing unit energy consumption by 8.7%, minimizing outlet temperature fluctuation by 12.3%, and improving throughput by 5.4%. These results validate the effectiveness and advancement of the proposed method in intelligent optimization of multi-objective industrial systems. This research provides important technical support and reference for the practical implementation of low-carbon intelligent manufacturing in the steel industry.

As humans face the depletion of non-renewable resources，the difficult disposal of solid waste caused by the consumption of these resources must also be confronted. Through a thorough investigation of the characteristics of carbon steel slag，its primary components are analyzed，and its potential nutrient elements are harnessed to develop a soilless cultivation technology for leisure horticulture. This approach is promising for achieving large-scale resource utilization of steel slag beyond the construction materials sector，offering significant economic value and practical significance. The soilless cultivation technology using carbon steel slag for leisure horticulture is characterized by a compact texture，excellent porosity，an appropriate pH level，and potential nutritional value，demonstrating broad application prospects. The characteristics of carbon steel slag are first briefly introduced，and the advantages and disadvantages of various treatment technologies are discussed. It is identified that roller slag and hot braised slag treatment technologies meet the requirements for application in soilless cultivation. The feasibility of using carbon steel slag in potted seedling cultivation and turf establishment was verified，and the current limitations of applying carbon steel slag in soilless cultivation were detailed. Based on this，it is highlighted that establishing classification standards for carbon steel slag，improving its management，and formulating unified substrate standards are critical challenges for future research and policy formulation.

In the context of carbon emissions in steel industry，the development of adsorbents in carbon capture technology was crucial. Blast furnace slag based zeolite molecular sieves were focused to investigate the kinetic of CO₂ adsorption and desorption in detail. It was shown that the Bangham model was suitable to describe the CO₂ adsorption process with an apparent activation energy of -4.085 8 kJ/mol，suggesting that the process was a physical process of binding between CO₂ molecules and the adsorbent surface through weak interaction forced such as van der Waals forces. Isoconversional methods (Starink and Tang) and the master curve method were used to determine that the desorption process conformed to the Avrami-Eroféev model and showed that the activation energy increased and then decreased with increasing conversion rate. The reaction rate was initially controlled by the diffusion of CO₂ on the surface of zeolite molecular sieves. As the reaction progressed，the apparent activation energy increased due to the decrease in resistance to CO₂ diffusion caused by the increase in temperature. Finally，the drive for desorption decreases with the increase in conversion rate，resulting in a decrease in the apparent activation energy. This study aimed to clarify the reaction mechanism of CO₂ adsorption and desorption of zeolite molecular sieves and to provide a theoretical basis for the practical application of this material.

To address the significant fluctuations in zinc mass fraction and low resource utilization efficiency of converter secondary sludge，a synergistic control strategy integrating scrap steel classification，diversion，and sludge gradient utilization was developed based on zinc element balance analysis. Analysis of zinc characteristics in scrap steel from Shagang's converter shops revealed mass fractions of 0.51% in standard scrap，1.14% in shredded scrap，and 2.20% in premium/clean scrap. A targeted scrap diversion plan was subsequently implemented. Shop 1 replaced high-zinc scrap with standard and other low-zinc scrap，reducing secondary sludge zinc mass fraction from 5.53% to 3.20%. High-zinc scrap was concentrated in Shop 2，elevating sludge zinc mass fraction to 12.20%. Medium-zinc shredded scrap was allocated to Shop 4，increasing sludge zinc content to 6.72%. Post-implementation results demonstrate that low-zinc sludge from Shop 1 reduced zinc loading in sinter by 42.09% when used in sintering，corresponding to a 33.67% reduction in blast furnace zinc loading. High-zinc sludge processed through a rotary hearth furnace achieved 5.70% zinc loading in mixed feed while boosting zinc oxide powder production by approximately 30%. This "Classification-Diversion-Gradient Utilization" approach established closed-loop zinc management，providing an efficient solution for resource utilization of iron and steel solid wastes.

To address the dual challenges of high carbon emissions in the steel industry and low utilization of steel slag solid waste，a synergistic "slag carbonization-pavement conversion" technology is proposed. Steel slag is carbonated by flue gas CO₂ and applied to open-graded friction course (OGFC) asphalt mixtures to achieve high-value utilization of steel slag. Through a high-temperature and low-pressure carbonation process under conditions of 600 ℃，0.5 MPa，2.5 h，and a flow rate of 9 L/h for water vapor and CO₂，the free calcium oxide content in steel slag is reduced from 3.20% to 1.15%，meeting the requirements of road material specifications. When 50% carbonated steel slag is incorporated into OGFC-10 asphalt mixtures，mix design optimization and systematic pavement performance tests show that the high-temperature stability，low-temperature crack resistance，water stability，drainage capacity，and skid resistance of the carbonated steel slag mixtures are improved by 11.34%，10.89%，9.30%，10.81%，and 3.90%，respectively，while the water immersion expansion rate is reduced by 66.93%. Carbon emission reduction mechanism analysis indicates that approximately 16.10 kg CO₂ can be directly fixed per ton of steel slag. Additionally，carbon emissions are reduced by 11.69，7.28，59.65 kg/t (in terms of CO₂) through substituting natural aggregates，extending road service life，and reducing resurfacing construction，respectively，achieving significant carbon emission reduction benefits. The research results provide technical support for the high-value utilization of metallurgical solid waste and low-carbon development of road engineering，facilitating the green transformation of the steel industry.