Method for producing ferrotitanium powder by reducing ilmenite

Abstract

The invention provides a method for producing ferrotitanium powder by reducing ilmenite. The method comprises the following steps: carrying out homogenization pretreatment on raw materials; dynamically proportioning a biomass-coal mixed reducing agent; multi-source online spectrum monitoring closed-loop feedback control is carried out; adding a nano CaO-Na2O composite catalyst in a buffer manner; and performing multi-stage magnetic separation reforming on the finished product. The method has the beneficial effects that through the synergistic effect of various innovative units, the fluctuation of raw materials and reducing atmosphere is reduced through pretreatment homogenization and dynamic proportioning, the real-time optimum reaction in the furnace is guaranteed through spectrum monitoring closed-loop control, the reaction rate is increased through the nano-catalyst, iron particle formation and magnetic separation reforming are fully separated, and products are utilized. According to the process, the direct reduction process of the ilmenite is precisely regulated and controlled, the content of TiO2 in the produced titanium-rich material is stabilized at 90-92%, the FeO residue is lower than 2%, the total recovery rate of metallic iron reaches 90% or above, and the energy consumption of unit output is reduced by about 15% compared with that of a traditional process. The process provides a new way for clean and efficient utilization of titanium resources.

Description

Technical Field

[0001] This invention belongs to the field of metallurgical technology, and in particular relates to a method for reducing ilmenite to produce ilmenite powder. Background Technology

[0002] In the direct reduction process of ilmenite (FeTiO3), a carbonaceous reducing agent is needed to reduce the iron oxides to metallic iron, while retaining as much TiO2 as possible in the titanium-rich slag. Traditional processes typically use coal or coke powder as the reducing agent. However, coal-based direct reduction suffers from high reaction temperatures, long reaction times, and poor product quality. Furthermore, full-scale biomass reduction also faces challenges: biomass is prone to pyrolysis at high temperatures, generating large amounts of volatiles and tar, which, if not properly handled, can lead to energy waste and environmental pollution. Existing technologies have attempted to use biomass pyrolysis tar to replace part of the coal, improving the quality of directly reduced iron and reducing dependence on fossil fuels, but incomplete biomass pyrolysis and high tar yields still persist. Therefore, how to fully leverage the low-temperature, high-efficiency advantages of biomass reduction while avoiding the adverse effects of excessive volatiles has become a key concern in the industry.

[0003] On the other hand, in the industrial practice of ilmenite reduction, a fixed coal-to-feed ratio is often used, lacking timely response to changes in raw materials and furnace conditions. The requirements for the ratio due to biomass participation and real-time changes in the furnace reducing atmosphere are not considered. Traditional processes rely heavily on operators' experience to judge the timing and ratio of feed, resulting in high labor intensity and significant fluctuations.

[0004] Furthermore, the direct reduction of ilmenite is a complex multiphase reaction involving the gradual reduction of the furnace charge and dynamic changes in the furnace atmosphere. In traditional processes, monitoring the furnace conditions relies primarily on intermittent manual sampling and experience-based judgment. For example, in titanium slag smelting, workers often periodically take samples of the furnace charge to analyze FeO content or observe flame color to adjust operations. This offline monitoring lags by several hours, making it difficult to reflect changes within the furnace in a timely manner. Especially in rotary kiln direct reduction, the complex temperature difference along the kiln length and the distribution of the atmosphere mean that the lack of effective online monitoring often leads to uneven reduction and significant fluctuations in product quality. Previously, ilmenite reduction production has suffered from insufficient FeO reduction or localized over-reduction: the former results in high FeO content in titanium-rich slag and reduced metallic iron recovery, while the latter may reduce TiO2 to low-valence titanium oxides or carbides, affecting TiO2 enrichment. A major reason for these problems is the lack of real-time monitoring and control of key reaction indicators within the furnace.

[0005] Furthermore, the reduction of ilmenite is commonly used to produce products such as high-titanium slag or ilmenite powder. In traditional processes, slag-forming agents such as lime (CaO) or alkali metal compounds are often added as auxiliaries to improve the degree of reduction and control impurities. However, the lime used in this method is in millimeter-sized particles, and its main function is desulfurization and adjusting slag basicity, failing to improve reaction kinetics at the nanoscale. Existing processes generally suffer from uneven reduction, localized overheating, or insufficient reduction, leading to unstable TiO2 encapsulation phases formed during reduction, high FeO residue, and affecting the quality of the final product.

[0006] Furthermore, the rotary kiln is a key piece of equipment commonly used in the reduction roasting of ilmenite. Parameters such as the calcination zone temperature, furnace atmosphere, and material residence time (determined by the rotational speed) directly affect the reduction reaction. In actual production, these parameters are interdependent and dynamically change. Traditional control methods relying on manual experience are difficult to coordinate in a timely and accurate manner, often leading to problems such as localized over- or under-reduction within the furnace, uneven temperature distribution, and decreased thermal efficiency.

[0007] Furthermore, in the process of reducing ilmenite to produce ilmenite powder (titanium-rich material), it is necessary to strictly control the stability of the reducing atmosphere (mainly composed of CO and CO2) in the furnace. However, in actual production, due to fluctuations in raw material composition, fuel ratio, air and oxygen supply, the CO / CO2 ratio and residual O2 content in the furnace often fluctuate. If the atmosphere is too reducing (too high CO, too low O2), it will lead to reduction overshoot, not only excessively reducing FeO to metallic iron, but also potentially generating low-valence oxides of Ti, affecting the stability of TiO2 enrichment. Conversely, if the atmosphere is insufficiently reducing (too high CO2, present O2), FeO residue will increase, and the quality of the titanium-rich material will decline. Therefore, achieving dynamic and stable control of the reducing atmosphere in the furnace is crucial to ensuring product quality. Summary of the Invention

[0008] In view of this, the present invention aims to provide a method for the reduction of ilmenite to produce ilmenite powder, in order to solve at least one of the problems existing in the prior art.

[0009] To achieve the above objectives, the technical solution of the present invention is implemented as follows: A method for producing ilmenite powder by reducing ilmenite includes the following steps: Raw material homogenization pretreatment: Ilmenite raw materials from different ore sources or batches are batched and mixed, and the contents of key components such as TiO2 and Fe are detected by online sampling and online XRF or LIBS analyzer. The ratio is adjusted to keep the composition of the raw materials entering the furnace constant within the set range. The dynamic proportioning of biomass-coal mixed reducing agent is carried out based on a biomass-coal mixed reducing agent dynamic proportioning system: a set proportion of biomass carbon is used to replace part of the coal powder as a reducing agent; the reducing atmosphere in the furnace is monitored by installing an online atmosphere analyzer. When the FeO content is detected to be too high, the proportion of biomass reducing agent is automatically increased; otherwise, it is reduced to about 20%, dynamically maintaining the carbon-oxygen balance in the furnace and reducing the fluctuation range of FeO. Multi-source online spectral monitoring closed-loop feedback control is implemented based on a multi-source online spectral monitoring closed-loop feedback system: A laser-induced breakdown spectroscopy (LAS) system is installed in the reaction section of the rotary kiln, and a laser-Raman spectroscopy combined monitoring system is installed in the kiln tail flue to analyze the reactants and atmosphere inside the furnace. LAS is used to quantitatively detect the degree of iron reduction in the furnace charge, while the Raman spectroscopy combined monitoring system is used to monitor the composition of the furnace gas. The control system automatically adjusts the kiln speed and the amount of combustion air / oxygen supplied based on the FeO content trend measured by LAS and the atmosphere composition measured by Raman spectroscopy: when reduction is insufficient, the speed is appropriately reduced to prolong the reaction time and the oxygen supply is reduced to maintain reducing properties; when reduction is excessive, the speed is increased or a small amount of air is introduced to suppress it, thus forming a closed-loop control. The buffering addition of nano-CaO-Na2O composite catalyst based on nano-scale catalyst buffering addition device: During the raw material mixing stage, a composite additive composed of nano-calcium oxide and sodium oxide is added at 0.5–1% of the weight of ilmenite. Multi-stage magnetic separation and reforming of finished products: After the reduction product is cooled by water quenching, large iron particles are first screened out, and the remaining material is ground in two stages to <0.074. Then, it is passed through a cascade system of weak magnetic separation and strong magnetic separation in sequence. Weak magnetic separation captures larger iron particles and metallized magnetic materials, while strong magnetic separation recovers fine metallic iron powder. Multi-stage magnetic separation ensures that the iron phase is separated from the titanium phase as much as possible, resulting in two products: titanium-rich material and iron-containing powder. Subsequently, the magnetically separated metallic iron particles and iron powder are mixed and pressed into shape to obtain titanium iron powder product.

[0010] Furthermore, in the multi-source online spectral monitoring closed-loop feedback control, the closed-loop control achieves basic monitoring and control through a multi-source online spectral monitoring closed-loop feedback system; Method 1: Based on the reducing atmosphere fluctuation steady-state control algorithm module, the precise regulation of the CO / CO2 / O2 atmosphere in the furnace is realized; Method 2: Based on the rotary kiln intelligent oxygen-temperature-speed three-parameter coordinated regulation system, the linkage control of oxygen supply, temperature and kiln speed is realized.

[0011] Furthermore, the biomass-coal mixed reducing agent dynamic proportioning system includes: at least two raw material silos, used to store and supply coal-based reducing agent and biomass-based reducing agent respectively; a feeding device with independently adjustable output flow rate, connected to the outlets of the coal-based silo and the biomass silo respectively, to deliver the two reducing agents to the reaction zone inside the furnace; a furnace atmosphere monitoring module, including a gas sampling component and a multi-component gas sampler installed inside the furnace or at the flue gas duct, used to measure the concentrations of CO, CO2, and O2 in the furnace gas online; a raw material characteristic monitoring module, used to detect the calorific value and moisture characteristics of biomass and coal; and a central control unit, which receives real-time data from the furnace atmosphere monitoring module and the raw material characteristic monitoring module. The central control unit has a pre-stored control program that calculates the target proportion of biomass to coal based on the deviation of the furnace gas composition and the changes in raw material characteristics, and outputs control signals to the feeding device.

[0012] Furthermore, the multi-source online spectral monitoring closed-loop feedback system includes: a solid charge spectral monitoring device for real-time acquisition of information on iron and titanium elements and their combined states in the furnace material; a furnace gas composition spectral monitoring device for real-time measurement of the content of CO, CO2, and O2 gas components in the furnace or furnace gas; a central control module, which is communicatively connected to the solid charge spectral monitoring device and the furnace gas composition spectral monitoring device, for fusing and analyzing the charge spectral data and the furnace gas spectral data, and generating control commands according to a preset control strategy; and an actuator, including a kiln speed control device and a combustion gas supply adjustment device, which receives commands from the central control module and automatically adjusts the kiln speed and oxygen supply.

[0013] Furthermore, the nano-catalyst buffer addition device includes a nano-catalyst storage and supply system, a pre-reduction section injection mechanism, a dynamic dispersion component, an alkalinity-pH monitoring probe, and a central control cabinet. The nano-catalyst storage and supply system includes a nano-catalyst storage and supply tank and a screw feeder for storing CaO–Na2O nano-dry powder catalyst and conveying it at a set flow rate. The pre-reduction section injection mechanism includes multiple high-temperature resistant pre-reduction section injection nozzles installed in the pre-reduction zone of the rotary kiln, connected to the storage and supply system to inject the nano-catalyst onto the surface of the material layer inside the rotary kiln via a carrier gas. The dynamic dispersion component is located inside the rotary kiln cylinder, corresponding to the position of the pre-reduction section injection nozzles, and is used to promote the uniform dispersion of the nano-catalyst in the material after it is injected. The alkalinity-pH monitoring probe includes an alkalinity sensor and a gas composition sensor installed at different positions inside the kiln to detect the alkalinity of the slag and the content of CO, CO2, and O2 in the reducing atmosphere in real time, and transmits the detection signals to the central control cabinet. The central control cabinet generates control commands to adjust the conveying rate of the feeder and / or the start and stop of the injection mechanism based on the deviation between the monitoring signal and the preset target.

[0014] Furthermore, the rotary kiln intelligent oxygen-temperature-speed three-parameter coordinated control system includes: at least one set of temperature sensors for detecting the temperature of the rotary kiln calcination zone; at least one flue gas composition analyzer for detecting the concentrations of CO, CO2, and residual O2 in the kiln exhaust; one oxygen-enriched air supply device with an adjustable oxygen-enriched valve for supplying oxygen-enriched combustion air into the rotary kiln; one variable frequency drive motor connected to the rotary kiln for adjusting the kiln rotation speed; and one industrial controller electrically connected to the temperature sensors, the oxygen-enriched air supply device, and the variable frequency drive motor. The industrial controller has a built-in fuzzy control module, a PID control module, and a neural network self-learning module for collaboratively calculating the oxygen supply valve opening, kiln temperature adjustment, and kiln rotation speed based on the kiln temperature and atmosphere deviation.

[0015] Furthermore, the reducing atmosphere fluctuation steady-state control algorithm module includes: a state perception unit, a reinforcement learning decision-making unit, an expert control unit, and an execution output unit; the state perception unit collects CO concentration, CO2 concentration, O2 concentration, and material temperature parameters in the reducing furnace to form a state vector; the reinforcement learning decision-making unit uses a deep neural network to construct a reinforcement learning agent and outputs control action commands based on the state vector; the expert control unit is used to verify and correct the actions output by the reinforcement learning decision-making unit; the execution output unit converts the verified control actions into adjustments to the on-site actuators.

[0016] Compared with existing technologies, the method for producing ilmenite powder by reducing ilmenite according to the present invention has the following advantages: (1) The present invention provides a method for producing ilmenite powder by reducing ilmenite. The biomass-coal mixed reducing agent dynamic ratio system can flexibly switch and adjust the ratio of biomass and coal during the ilmenite reduction process, making the release of carbon source more stable and controllable, and significantly improving the stability of the reducing atmosphere and the uniformity of the reaction.

[0017] (2) The method for producing ilmenite powder by reducing ilmenite according to the present invention, the multi-source online spectral monitoring closed-loop feedback system of the present invention innovatively applies LIBS and Raman spectroscopy to control the ilmenite reduction process, which significantly improves the level of process monitoring and adjustment, and achieves the production goals of high quality, high efficiency and green controllability while ensuring stable reduction of FeO and enrichment of TiO2.

[0018] (3) The method for producing ilmenite powder by reducing ilmenite as described in this invention utilizes a rotary kiln intelligent oxygen-temperature-velocity three-parameter coordinated control system. This system enables intelligent optimization control of the rotary kiln for ilmenite reduction, significantly reducing the frequency of manual intervention and improving control accuracy and reaction uniformity. Compared to the system without a rotary kiln intelligent oxygen-temperature-velocity three-parameter coordinated control system, the temperature distribution within the kiln is more stable, the average reduction degree of the material is more uniform, and the residual FeO is significantly reduced.

[0019] (4) The method for producing ilmenite powder by reducing ilmenite according to the present invention, the steady-state control algorithm module of the reducing atmosphere of the present invention can: adaptively maintain the CO / CO2 concentration ratio in the optimal range, stabilize the residual oxygen at the kiln tail at the ppm level and above the safety lower limit, greatly reduce the fluctuation range of the reducing atmosphere; at the same time, avoid the unstable attempts that may occur in pure RL and the rigidity of pure expert system, and achieve the effect of steady-state accuracy and dynamic optimization.

[0020] (5) The method for producing ilmenite powder by reducing ilmenite according to the present invention has significant innovations in the following aspects: the catalyst is added at the nanoscale, with high activity and large reaction contact area, achieving a micro-catalytic effect that cannot be achieved by macro-additives in the past. Adding alkaline substances such as Na2CO3 can lower the reduction initiation temperature and accelerate the growth of iron grains; increasing the basic oxide CaO to improve the slag basicity also helps to further reduce FeO. Attached Figure Description

[0021] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a schematic diagram of the dynamic proportioning system of biomass-coal mixed reducing agent according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the proportioning control process of the biomass-coal mixed reducing agent dynamic proportioning system according to an embodiment of the present invention; Figure 3 This is a schematic diagram showing the comparison curves of the change in the CO / CO2 ratio in the furnace before and after the biomass-coal mixed reducing agent dynamic proportioning system described in this embodiment of the invention over time. Figure 4 This is a schematic diagram of the overall layout of the multi-source online spectral monitoring closed-loop feedback system described in this embodiment of the invention on a rotary kiln; Figure 5 This is a schematic diagram of the control logic flow of the multi-source online spectral monitoring closed-loop feedback system according to an embodiment of the present invention; Figure 6 This is a schematic diagram comparing key indicators of the multi-source online spectral monitoring closed-loop feedback system before and after the restoration process according to an embodiment of the present invention; Figure 7 This is a schematic diagram of the structure of the nanocatalyst buffer addition device on a rotary kiln according to an embodiment of the present invention; Figure 8 This is a schematic diagram of the process flow of the nanocatalyst buffer addition device method according to an embodiment of the present invention; Figure 9 This is a schematic diagram of the composition and signal flow of the intelligent oxygen-temperature-velocity three-parameter coordinated control system for rotary kilns according to an embodiment of the present invention; Figure 10 This is a schematic diagram of the control strategy of the intelligent oxygen-temperature-speed three-parameter coordinated control system for rotary kilns according to an embodiment of the present invention; Figure 11 This is a schematic diagram of the fuzzy control decision rule variables according to an embodiment of the present invention; Figure 12 This is a schematic diagram illustrating the operation of the neural network PID parameter self-tuning described in an embodiment of the present invention; Figure 13 This is a schematic diagram of the integrated logic of the neural network and PID controller described in an embodiment of the present invention; Figure 14 This is a schematic diagram of the functional architecture of the reducing atmosphere fluctuation steady-state control algorithm module according to an embodiment of the present invention; Figure 15 This is a schematic diagram of the decision-making process of the deep reinforcement learning agent according to an embodiment of the present invention; Figure 16 This is a schematic diagram of the reward function curve described in an embodiment of the present invention; Figure 17 This is a schematic diagram of the control effect curve of the algorithm module described in the embodiment of the present invention under actual working conditions.

[0022] Explanation of reference numerals in the attached figures: 1. Biomass silo; 2. Coal silo; 3a. Adjustable speed feeder one; 3b. Adjustable speed feeder two; 4. Conveying pipeline; 6. Atmosphere analysis sampling port; 7. Gas sampler; 8. Control unit; 10. Rotary kiln body; 11. LIBS spectral probe; 12. Raman sampling probe; 13. LIBS spectrometer; 14. Raman spectrometer; 15. Central control module; 16. Rotary kiln drive motor frequency converter; 17. Blower / oxygen valve actuator. Detailed Implementation

[0023] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0024] like Figures 1 to 17 As shown, a method for producing ilmenite powder from ilmenite by reduction involves uniformly mixing ilmenite concentrate with a biomass-coal mixed reducing agent and nano-scale additives, followed by reduction in a rotary kiln under controlled temperature and oxygen atmosphere. The product is then water-quenched and cooled, and subsequently separated into metallic iron particles and titanium-rich material through multi-stage grinding and magnetic separation. Both are then processed into ilmenite powder. The method integrates innovative units such as dynamic proportioning of the biomass-coal mixed reducing agent, online monitoring and closed-loop control using LIBS+Raman spectroscopy, buffer addition of a nano-CaO-Na2O composite catalyst, and multi-stage magnetic separation and reforming of the finished product.

[0025] 1. Raw Material Homogenization Pretreatment: Ilmenite raw materials from different ore sources or batches are batched and mixed, and the content of key components such as TiO2 and Fe is detected by online sampling and online XRF or LIBS analyzers. The proportions are adjusted to keep the composition of the raw materials entering the furnace constant within a set range. For example, a raw material pre-homogenization silo is set up, and automatic batching control is achieved through PLC or industrial control computer combined with PID and expert rule algorithms, so that the TiO2 content of each batch of raw materials entering the furnace fluctuates by ≤±0.4%. This reduces product quality fluctuations caused by raw material fluctuations.

[0026] 2. Dynamic proportioning of biomass-coal mixed reducing agent based on a biomass-coal mixed reducing agent system: A certain proportion (20–40%) of biomass carbon (such as biomass coke powder or charcoal powder) is used to replace part of the coal powder as a reducing agent. The reducing atmosphere in the furnace is monitored by an online atmosphere analyzer (such as a CO / CO2 infrared sensor). When an upward trend in FeO content is detected, the proportion of biomass reducing agent is automatically increased (up to 30%); conversely, it is reduced to around 20%, dynamically maintaining the carbon-oxygen balance in the furnace and reducing FeO fluctuations. Biomass coke powder has the advantages of carbon neutrality and low sulfur content; its incorporation can reduce CO2 emissions and the sulfur content of the finished product. Closed-loop control of the mixed reducing agent proportion can reduce the FeO fluctuation of the product to within ±0.5%, while fully utilizing biomass energy.

[0027] 3. Multi-source online spectral monitoring closed-loop feedback control: Laser-induced breakdown spectroscopy (LIBS) is installed in the reaction section of the rotary kiln, and a laser-Raman spectroscopy combined monitoring system is installed in the kiln tail flue to rapidly analyze the reactants and atmosphere in the furnace. LIBS is used to quantitatively detect the degree of iron reduction in the furnace charge (by monitoring the spectral intensity of Fe and Ti elements in the slag / powder), and Raman spectroscopy is used to monitor the furnace gas components such as CO, CO2, and O2 concentrations, achieving key parameter acquisition every <2 minutes. The control system automatically adjusts the kiln speed and combustion air / oxygen feed rate based on the FeO content trend measured by LIBS and the atmosphere composition measured by Raman spectroscopy: when reduction is insufficient (FeO is too high), the speed is appropriately reduced to prolong the reaction time and the oxygen supply is reduced to maintain reducing properties; when reduction is excessive (metallization rate is too high and sintering may occur), the speed is increased or a small amount of air is introduced to suppress it. This forms a closed-loop control, achieving coordinated optimization of furnace temperature, oxygen potential, and material rate. This closed-loop control system achieves basic monitoring and control through a multi-source online spectral monitoring closed-loop feedback system, and can be further integrated with advanced control strategies for optimization. Method 1: Based on a steady-state control algorithm module for reducing atmosphere fluctuations, it focuses on precise regulation of the CO / CO2 / O2 atmosphere within the furnace; Method 2: Based on a rotary kiln intelligent oxygen-temperature-speed three-parameter coordinated regulation system, it achieves linked control of oxygen supply, temperature, and kiln speed. Through this intelligent control system, traditional manual experience-based adjustments can be replaced, improving the automation level of the reduction process and the stability of product quality.

[0028] 4. Nano-CaO-Na2O Composite Catalyst Buffer Addition Based on a Nanoscale Catalyst Buffer Addition Device: During the raw material mixing stage, a composite additive consisting of nano-calcium oxide and sodium oxide (or their precursors such as CaCO3+Na2CO3) with a particle size <50 nm is added at 0.5–1% of the weight of the ilmenite. The nano-additive provides an alkaline oxide buffer in the early stages of reduction, increasing the basicity of the material layer to approximately 0.5, breaking the FeO–TiO2 solid solution bond, and promoting the formation of a low-melting-point molten phase from non-ferrous oxides. Furthermore, its ultra-small particle size provides a highly reactive catalytic surface, accelerating the reaction kinetics of CO on FeO. In particular, the Na2O component can form a low-melting-point eutectic or sodium ferrate with FeO, lowering the temperature required for FeO reduction. Adding the nano-composite agent can reduce the reduction temperature by approximately 50°C, shorten the reduction time by more than 15%, and simultaneously facilitate the aggregation and growth of iron particles. Experimental comparisons show that the FeO residue in the titanium-rich material obtained without a catalyst is about 6%, while the FeO residue is reduced to <3% after adding nano-CaO-Na2O, and the metallization rate is increased by 5 to 8 percentage points.

[0029] 5. Multi-stage magnetic separation and reforming of the finished product: After water quenching and cooling, the reduction product is first screened to remove large iron particles (metallic iron particle size 1~5 mm). The remaining material is ground in two stages to <0.074 mm, and then passed through a cascade system of weak and strong magnetic separation. Weak magnetic separation captures larger iron particles and metallized magnetic materials, while strong magnetic separation further recovers fine metallic iron powder. Multi-stage magnetic separation ensures that the iron phase is separated from the titanium phase as much as possible, resulting in two products: titanium-rich material (TiO2 > 90%, FeO < 0.5%) and iron-containing powder (TFe > 88%, TiO2 < 6%). Subsequently, the magnetically separated metallic iron particles and iron powder are mixed and pressed into a molding compound to obtain titanium iron powder product—in which the iron phase mainly exists in a metallic state, and the titanium phase mainly exists in the form of high-titanium slag with a small amount of unreduced oxides. This titanium iron powder can be used as a Ti-containing alloy material for steelmaking and as a raw material for further extraction of sponge titanium. The introduction of a multi-stage magnetic separation process maximizes the enrichment rate of TiO2 and the recovery rate of iron: the recovery rate of TiO2 in titanium-rich materials reaches over 90%, and the recovery rate of iron in granular iron + iron powder is >92%.

[0030] In summary, the synergistic effect of each innovative unit: pretreatment homogenization and dynamic proportioning reduced fluctuations in raw materials and reducing atmosphere; closed-loop control via spectral monitoring ensured real-time optimization of the in-furnace reaction; nanocatalysts enhanced the reaction rate and iron particle formation; and magnetic separation and reforming fully separated and utilized the products. Under the process of this invention, the direct reduction process of ilmenite is precisely controlled, producing titanium-rich material with a stable TiO2 content of 90–92%, FeO residue below 2%, and a total iron recovery rate of over 90%, while energy consumption per unit output is reduced by approximately 15% compared to traditional processes. This process provides a new pathway for the clean and efficient utilization of titanium resources.

[0031] Beneficial effects: This biomass-coal mixed reducing agent dynamic proportioning system can flexibly switch and adjust the ratio of biomass and coal during the reduction of ilmenite, making carbon source release more stable and controllable, and significantly improving the stability of the reducing atmosphere and the uniformity of the reaction. Industrial trials show that compared with traditional fixed proportions, this biomass-coal mixed reducing agent dynamic proportioning system can reduce the fluctuation range of CO / CO2 in the kiln by more than 30%, reduce the residual FeO content in the titanium-rich feedstock by about 15%, and make the TiO2 content distribution more stable. Since biomass partially replaces coal, the fossil carbon consumption per unit product is reduced, and the net CO2 emissions are reduced, which is conducive to reducing the carbon footprint. In addition, intelligent feeding avoids the risk of drastic temperature fluctuations and slagging caused by the enrichment of volatile matter in biomass, keeping the furnace temperature at a lower but effective level, which helps to reduce the entry of impurities such as vanadium into the metallic phase and improve the purity of the titanium-rich feedstock. The biomass-coal mixed reducing agent dynamic proportioning system has a high degree of automation, reduces the burden on operators, and achieves stable and continuous production. In summary, this biomass-coal mixed reducing agent dynamic proportioning system not only achieves green metallurgy but also improves reduction efficiency and product quality, demonstrating significant economic and environmental benefits. It can serve as an important component of the process, working in synergy with other subsystems to optimize the reduction process of ferro-titanium powder.

[0032] This multi-source online spectral monitoring closed-loop feedback system deeply integrates advanced spectral analysis technology into the control of the ilmenite reduction process, bringing several beneficial effects. First, real-time monitoring of the furnace charge FeO and furnace gas CO / CO2 / O2 enables direct observation of the reduction reaction degree and atmospheric conditions, overcoming the shortcomings of traditional methods relying on indirect parameters and human experience. Compared to the several-hour lag of post-analysis, this multi-source online spectral monitoring closed-loop feedback system acquires data at a frequency of seconds to minutes, enabling timely capture of changes in the furnace state and a control response within seconds, significantly shortening control delays and reducing fluctuation accumulation. Second, the closed-loop feedback of multi-source information makes control more precise and coordinated. For example, when the furnace charge reduction is insufficient, the system not only reduces the kiln speed to prolong the reaction but also simultaneously adjusts the oxygen supply to enhance the atmosphere, with both aspects working together to ensure that FeO continues to decrease; similarly, when local over-reduction is detected, the kiln speed and ventilation are simultaneously increased to suppress it and avoid excessive TiO2 reduction. This coordinated control is superior to the independent operation of traditional single-loop control and can solve the problem of multi-variable coupling. Industrial test data shows that after the multi-source online spectral monitoring closed-loop feedback system was put into operation, the fluctuation range of FeO content in titanium-rich materials was reduced by about 30%, and the TiO2 grade of the product remained stable within the target range of ±0.5%. The CO / CO2 ratio in the kiln remained stable, and the ring formation and large fluctuations in furnace conditions caused by atmosphere imbalance in the past no longer occurred. Furthermore, this multi-source online spectral monitoring closed-loop feedback system improves resource utilization and energy efficiency: through spectral feedback, oxygen supply and temperature can be strictly controlled to meet the minimum requirements for reduction, avoiding heat loss and nitrogen oxide generation caused by excessive combustion air; sufficient reduction of FeO ensures that more iron enters the metal, improving metal yield, while more thorough enrichment of TiO2 reduces the burden of subsequent processing. Finally, the multi-source online spectral monitoring closed-loop feedback system realizes intelligent and labor-saving reduction of the reduction process: the work that originally required repeated manual sampling and experience-based adjustments is now automatically completed by the system, reducing labor costs and error risks, and providing a guarantee for continuous and stable production.

[0033] This intelligent oxygen-temperature-rate three-parameter coordinated control system for rotary kilns enables intelligent and optimized control of ilmenite reduction rotary kilns, significantly reducing the frequency of manual intervention and improving control accuracy and reaction uniformity. Compared to rotary kilns without this system, the temperature distribution within the kiln is more stable, the average reduction degree of the material is more uniform, and FeO residue is significantly reduced. Practical application shows that this system can stably control FeO residue below 0.5% and the TiO2 grade fluctuation of titanium-rich materials within ±0.4%, avoiding excessive metallization due to over-reduction or excessive FeO due to under-reduction. Simultaneously, due to optimized oxygen supply and combustion, fuel consumption is reduced, and thermal efficiency is increased by over 20%, achieving energy saving and consumption reduction. This intelligent oxygen-temperature-velocity three-parameter coordinated control system for rotary kilns, as an independent module, can be easily integrated into the existing reduction rotary kiln control system (multi-source online spectral monitoring closed-loop feedback system). It works in synergy with the multi-source online spectral monitoring closed-loop feedback system: it receives the production settings (such as output and temperature targets) from the multi-source online spectral monitoring closed-loop feedback system and outputs optimized control commands to the specific actuators, thereby ensuring that the main process is always in the best state.

[0034] The reducing atmosphere steady-state control algorithm module of this invention can: adaptively maintain the CO / CO2 concentration ratio within the optimal range (e.g., empirically estimated at around 90:10), stabilize the residual oxygen at the kiln tail at the ppm level above the safety lower limit, and greatly reduce the fluctuation range of the reducing atmosphere; at the same time, it avoids the unstable attempts that may occur with pure RL and the rigid limitations of pure expert systems, achieving a steady-state precision and dynamic optimization effect. After applying this module, the average residual FeO in the ilmenite reduction process can be reduced to ≤0.5%, and the batch-to-batch fluctuations are significantly reduced, with the TiO2 enrichment rate remaining stable within ±0.4% of the target value for a long period. More importantly, while improving quality, the reducing atmosphere steady-state control algorithm module reduces ineffective reactions and energy losses by optimizing the atmosphere, and works synergistically with the main control system (multi-source online spectral monitoring closed-loop feedback system) to achieve a thermal efficiency improvement of ≥20% and ensure safe and stable production.

[0035] This nanoscale catalyst buffer addition device features significant innovations in the following aspects: First, the catalyst is added at the nanoscale, resulting in high activity and a large reaction contact area, achieving microscopic catalytic effects that were previously unattainable with macroscopic additives. Adding alkaline substances such as Na₂CO₃ lowers the reduction initiation temperature and accelerates iron grain growth; increasing the basic oxide CaO improves the slag's basicity and also facilitates further FeO reduction. This nanoscale catalyst buffer addition device utilizes the synergistic catalysis of nano-CaO and Na₂O to form a stable alkaline film on the particle surface, slowing TiO₂ nucleus growth and maintaining the integrity of the cladding; on the other hand, the nano-additives promote the reaction between CO and FeO, allowing FeO to be more fully converted into metallic iron. Experimental verification shows that without catalyst addition, the FeO residue in titanium slag is often higher than 5%; however, after adding the catalyst from this nanoscale catalyst buffer addition device, FeO can be reduced to <2%, meeting the main process requirements. Second, this nanoscale catalyst buffer addition device achieves intelligent control of the catalyst addition process. By monitoring the kiln atmosphere composition and alkalinity index at multiple points, the catalyst feed rate is adjusted in a closed loop, offering greater flexibility than traditional fixed-ratio addition methods. For example, existing one-stage reduction roasting processes lack online control, making it difficult to handle fluctuations in CO concentration within the furnace in a timely manner, leading to unstable reduction rates. This nano-scale catalyst buffer addition device can instantly increase the Na2O-based catalyst dosage when the CO / CO2 ratio suddenly decreases (indicating insufficient reducing power), thereby improving the CO regeneration rate; conversely, it can reduce catalyst dosage when signs of over-reduction appear (excess CO, abnormal local temperature rise), preventing excessive accumulation and sintering of metallic iron. Furthermore, due to its dry powder injection method, it avoids the additional steps required for additive premixing and granulation, allowing direct integration into rotary kiln production lines and demonstrating strong engineering applicability. The modular design of the nano-scale catalyst buffer addition device allows for modification according to different kiln types. While ensuring the normal operation of the main kiln, the catalyst injection system can be installed externally without affecting raw material feeding and product discharge. In summary, this nano-scale catalyst buffer addition device is highly independent and can be used for both new production lines and the upgrading of existing rotary kilns. When used in conjunction with other devices, it can significantly improve product quality and energy efficiency.

[0036] In a preferred embodiment of the present invention, a biomass-coal mixed reducing agent dynamic proportioning system is used in a rotary kiln or furnace process for reducing ilmenite to produce ilmenite powder. The biomass-coal mixed reducing agent dynamic proportioning system includes: at least two raw material silos (biomass silo 1 and coal silo 2), one for storing and supplying coal-based reducing agent and the other for storing and supplying biomass reducing agent; a feeding device (adjustable speed feeder) with independently adjustable output flow rate, respectively connected to the outlets of the coal silo and the biomass silo to feed the two reducing agents to the reaction zone inside the furnace; and a furnace atmosphere monitoring module, including components installed inside the furnace or... The gas sampling component and multi-component gas sampler at the flue gas duct are used to measure the concentrations of CO, CO2, and O2 in the furnace gas online; the raw material characteristic monitoring module is used to detect characteristic parameters such as calorific value and moisture content of the biomass and coal; and the central control unit receives real-time data from the atmosphere monitoring module and the raw material characteristic monitoring module. The central control unit has a pre-stored control program that calculates the target ratio of biomass to coal based on the deviation of the gas composition in the furnace and the changes in raw material characteristics, and outputs control signals to the feeding device to dynamically adjust the instantaneous feeding rate of the two materials, thereby maintaining the reducing atmosphere in the furnace within a predetermined range.

[0037] The gas sampler uses Raman spectroscopy analysis technology, which can simultaneously obtain the CO, CO2, and O2 content of the gas in the furnace within 1-2 seconds, enabling rapid feedback control of the reducing atmosphere.

[0038] The raw material characteristic monitoring module includes: an online calorific value analyzer, used to measure the lower heating value of the current batch of biomass by near-infrared or other means; and a moisture sensor, used to detect the moisture content of biomass and coal, wherein the control unit corrects the influence model of biomass on furnace temperature and atmosphere based on the calorific value and moisture data to optimize the proportioning calculation.

[0039] The control program of the central control unit includes a closed-loop control algorithm with feedforward compensation: when the CO / CO2 ratio in the furnace is detected to be lower than the preset lower limit or the O2 concentration is higher than the preset upper limit, the feeding rate of biomass is increased and / or the feeding rate of coal is decreased; conversely, when the CO / CO2 ratio is higher than the upper limit or there are signs of incomplete combustion, the feeding rate of biomass is decreased and / or the feeding rate of coal and the oxygen supply in the furnace are increased, so that the reducing atmosphere parameters return to the target range.

[0040] The biomass reducing agent is uncarbonized plant biomass, including but not limited to straw, sawdust, coconut shells or rice husks, with a particle size of 10-50 mm after crushing and sieving; the coal reducing agent is bituminous coal or anthracite powder with a particle size of less than 5 mm; the two materials are metered by the feeding device and simultaneously or separately fed into the rotary kiln to achieve mixed combustion and reduction.

[0041] It includes a mixing and homogenization system, located downstream of the coal and biomass feeding device, for fully mixing the two materials before they are fed into the furnace; the mixing and homogenization system is a dynamic mixer with a propeller or stirring blades to prevent local imbalances caused by fluctuations in a single material flow.

[0042] A method for reducing ilmenite using a dynamic proportioning system includes the following steps: ilmenite concentrate and auxiliary additives are mixed according to conventional methods to form green pellets or directly added to a rotary kiln; the kiln is started and heated, and a biomass and coal mixed reducing agent with an initially set proportion is introduced; during the reduction process, the concentrations of CO, CO2, and O2 in the furnace are continuously monitored by an atmosphere monitoring module. Based on the deviation between the monitored values ​​and the target values, the central control unit calculates adjustment commands to dynamically change the instantaneous feeding rate of biomass and coal, achieving real-time matching of the carbon source release rate; when raw material batches are changed or the load changes, the control unit automatically adjusts the biomass / coal baseline proportion with reference to a self-learning model, smoothly transitioning to the new operating conditions; the above control continues until the reduction process ends. This method can maintain the furnace temperature and reducing atmosphere stable within the optimized range throughout the entire reduction cycle, avoiding incomplete or excessive reduction caused by fluctuations in raw materials or operating conditions, and improving the quality and consistency of the final product.

[0043] The target reducing atmosphere parameters are set as CO / CO2 volume ratio of 1.5 to 2.5 and O2 concentration of <1%. The central control unit calculates the ratio adjustment every 1 to 5 seconds to ensure that the instantaneous deviation of the furnace atmosphere parameters does not exceed ±10%, thereby controlling the FeO content in the titanium-rich slag within the expected range and improving the TiO2 enrichment effect.

[0044] Figure 1 This is a schematic diagram of the dynamic proportioning system for the biomass-coal mixed reducing agent of the present invention. Figure 1 The diagram shows a biomass silo 1, a coal silo 2, adjustable-speed feeders (adjustable-speed feeder 3a and adjustable-speed feeder 3b), a conveying pipeline 4, a mixing and homogenization system, an atmosphere analysis sampling port 6, a gas sampler 7, a control unit 8, and signal / control lines connecting the control unit to each actuator. Biomass silo 1 and coal silo 2 each send materials out via their respective adjustable-speed feeders (3a and 3b), which are then combined via the mixing and homogenization system before being fed into the rotary kiln. The atmosphere analysis sampling port 6 is located in the middle or rear of the kiln and is used to continuously collect gas from inside the rotary kiln, which is then fed into the gas sampler 7 for CO, CO2, and O2 content detection. The control unit 8 receives gas analysis data and raw material characteristic sensor data, and outputs control signals to the adjustable-speed feeders (3a and 3b) to dynamically adjust the instantaneous feeding rate and ratio of the two materials. Figure 2This is a schematic flowchart of the proportioning control process of the biomass-coal mixed reducing agent dynamic proportioning system. At the start of the process, the system reads the calorific value and composition parameters of the raw materials, and sets the target reducing atmosphere parameter range (e.g., CO / CO2 ratio) and kiln temperature range. In the main circulation control, the furnace gas composition is collected in real time, and the current oxygen potential level is determined. If insufficient reducing atmosphere is detected (e.g., CO concentration below the lower limit or O2 increase), the biomass feed rate is increased or the coal feed rate is decreased to improve CO yield; conversely, if the reducing atmosphere is too strong and there are signs of incomplete combustion, the biomass feed rate is decreased or the coal feed rate is increased to suppress excess volatiles. Subsequently, the total feed rate is recalculated based on the adjusted proportion to maintain thermal balance within the furnace. After the adjustment is executed, the process returns to continue the cycle monitoring. This process also includes a disturbance prediction module, which predicts trends caused by changes in raw material batches or kiln speed and ventilation. Once it is predicted that the future atmosphere will deviate from the target, the proportion is fine-tuned in advance, acting as feedforward control. Figure 2 This demonstrates how combining feedback control with predictive compensation enables adaptive dynamic optimization of the mix ratio. Figure 3 The diagram illustrates the change in the CO / CO2 ratio in the furnace over time before and after adopting this biomass-coal mixed reducing agent dynamic proportioning system. Curve A represents the situation where the CO / CO2 ratio fluctuates drastically under the traditional fixed proportioning method; curve B is the stable curve after applying dynamic proportioning control. It is evident that this biomass-coal mixed reducing agent dynamic proportioning system significantly reduces atmosphere fluctuations.

[0045] In a preferred embodiment of the present invention, a multi-source online spectral monitoring closed-loop feedback system is used for process control of the direct reduction of ilmenite to ilmenite powder. The system includes: a solid charge spectral monitoring device for real-time acquisition of information on iron and titanium elements and their combined states in the furnace material; a furnace gas composition spectral monitoring device for real-time measurement of the content of gas components such as CO, CO2, and O2 in the furnace or furnace gas; a central control module 15, communicatively connected to the above two monitoring devices, for fusing and analyzing the charge spectral data and furnace gas spectral data, and generating control commands according to a preset control strategy; and an actuator, including a kiln speed control device and a combustion gas supply adjustment device, which receive commands from the central control module 15 and automatically adjust the kiln speed and oxygen supply to achieve closed-loop control of the reduction reaction process within the furnace.

[0046] The solid furnace charge spectral monitoring device adopts laser-induced breakdown spectroscopy (LIBS) technology, including a LIBS spectral probe 11 installed on the furnace wall and a LIBS spectrometer 13 connected to it. The LIBS device emits high-energy laser pulses to the surface of the furnace charge at predetermined intervals to generate plasma spectra and analyzes in real time the content of iron and titanium elements and the proportion of oxides in the furnace charge, thereby assessing the degree of FeO reduction and TiO2 enrichment.

[0047] The furnace gas composition spectral monitoring device adopts Raman spectroscopy gas analysis technology. The device includes a Raman sampling probe 12 and a Raman spectrometer 14 located at the kiln tail or flue. It can simultaneously measure the concentration of multiple components such as CO, CO2, O2, N2 and H2 in the furnace gas. Its single sampling analysis time does not exceed 5 seconds, and the relative standard deviation of the detection accuracy of CO, CO2 and other concentrations is ≤0.1%. It is used to provide real-time indication parameters of oxygen potential in the furnace atmosphere.

[0048] The central control module 15 includes an industrial computer or a programmable logic controller (PLC), whose built-in control algorithm has fusion decision logic: when the LIBS signal of the furnace charge indicates that the rate of decrease of FeO content has slowed down or the residual FeO is too high, and the Raman signal of the furnace gas shows that the CO / CO2 ratio is lower than the predetermined value and the O2 concentration is increased, it is judged that the reduction is insufficient and the atmosphere is too oxidizing. The central control module 15 issues an instruction to reduce the rotary kiln speed and reduce the supply of combustion air; when the LIBS signal shows that FeO has dropped to a low level and the Ti element content has dropped abnormally (indicating over-reduction), and the CO / CO2 ratio of the furnace gas is higher than the upper limit and the O2 is close to zero, it is judged that there is a tendency for over-reduction. The central control module 15 issues an instruction to increase the kiln speed and increase the amount of combustion oxygen to suppress over-reduction; if both the LIBS and Raman signals are within the normal range, the current operating conditions are maintained.

[0049] The control algorithm employs fuzzy control or expert system rules to comprehensively evaluate multi-source spectral data. It includes a rule base, for example: "If FeO decreases more slowly and CO2 increases, then slow down the kiln speed," and "If FeO is extremely low and CO is significantly high, then increase the kiln speed." Each rule corresponds to one or a group of actions. The central control module matches the corresponding rule to trigger the action based on the spectral data and adjusts the action amplitude according to the degree of deviation.

[0050] The central control module 15 includes an adaptive correction unit that can automatically adjust control parameter thresholds based on historical spectral data to adapt to different raw materials and operating conditions. This adaptive unit continuously learns the FeO reduction curve and gas composition change patterns of the furnace charge, optimizing parameters such as the CO / CO2 target value and the FeO rate threshold. This allows the control strategy to automatically switch at different times: maintaining a moderate oxygen potential in the early stages of reduction to avoid excessive reaction, and eliminating FeO tails and preventing Ti reduction in the later stages, dynamically optimizing the control scheme based on spectral feedback.

[0051] The actuator includes: a rotary kiln drive motor frequency converter 16, whose speed is adjusted in real time by analog signals or communication commands output by the central control module 15; and an auxiliary combustion blower / oxygen valve actuator 17, whose opening degree is controlled by the central control module 15 to adjust the flow rate of air or oxygen entering the kiln. The adjustment accuracy of the actuator is preferably achieved as follows: minimum kiln speed adjustment step ≤ 0.01 rpm, minimum oxygen supply valve adjustment step ≤ 1%, to achieve fine control of process parameters.

[0052] It also includes a data visualization and alarm unit, which displays the estimated values ​​of FeO content and TiO2 content obtained from spectral monitoring, as well as key parameters such as CO, CO2, and O2 concentrations, to the operators in real time, and generates alarm prompts when they deviate from the normal range or the rate of change is abnormal, so that the operators can understand the process status and intervene as a reference.

[0053] A method for controlling a rotary kiln for ilmenite reduction includes the following steps: During kiln operation, a solid charge spectral monitoring device and a furnace gas spectral monitoring device are activated to acquire continuous data on charge and gas composition; the acquired data is sent to a central control module for fusion analysis and compared with predetermined targets or thresholds; when insufficient FeO reduction is detected in the charge, the kiln speed is automatically reduced and the oxygen supply is decreased to enhance the reducing atmosphere and prolong the reaction time; when an over-reduction trend is detected in the charge or the furnace gas is too reducing, the kiln speed is automatically increased and the oxygen supply is increased to prevent excessive reduction of titanium and burn off excess carbon monoxide; under normal conditions, the current operating settings are maintained; the above adjustments are performed continuously in a closed loop until the reduction process ends. This method achieves precise control of the entire process of FeO reduction and TiO2 enrichment through multi-source spectral feedback, ensuring that the oxygen potential and temperature inside the furnace are always within the optimal window.

[0054] The interval between laser excitations for each charge spectral monitoring session is 1-5 minutes. The ratio of Fe to Ti spectral line intensities obtained from each analysis is used to calculate the residual FeO content. The furnace gas spectral monitoring frequency is above 1 Hz, continuously obtaining CO and CO2 concentrations to calculate the real-time CO / CO2 ratio. When the reduction in residual FeO is lower than a preset value and the CO / CO2 ratio is lower than the target lower limit within two adjacent charge spectral acquisition cycles, an adjustment action is triggered. The preset value is set based on production experience; for example, FeO needs to decrease by no less than 0.5 percentage points every 5 minutes, and the target lower limit for the CO / CO2 ratio is, for example, 1.8. Through the above method, the FeO content in titanium-rich slag can be controlled below 2%, minimizing fluctuations in the composition of ferrotitanium powder products and achieving highly stable production control. Figure 4This diagram illustrates the overall layout of a multi-source online spectral monitoring closed-loop feedback system on a rotary kiln. The diagram shows the rotary kiln body 10, a LIBS spectral probe 11 installed on the side wall in the middle of the kiln, a Raman sampling probe 12 installed in the kiln tail flue, and the corresponding LIBS spectrometer 13 and Raman spectrometer 14. A laser beam from the LIBS spectral probe 11 (indicated by the arrow in the diagram) passes through a window in the kiln wall and illuminates the surface of the material inside the kiln, generating a plasma light radiation signal, which is transmitted via optical fiber to the LIBS spectrometer 13 for analysis. The Raman sampling probe 12 is inserted into the center of the flue to collect the high-speed flue gas and is connected to the Raman spectrometer 14. The LIBS spectrometer 13 and the Raman spectrometer 14 are connected to a central control module 15 via an industrial Ethernet network. The central control module 15 also connects to the rotary kiln drive motor frequency converter 16 and the blower / oxygen valve actuator 17. Figure 4 The dashed line in the middle represents the signal flow: the spectrometer transmits the analyzed component data to the central control module 15, and the central control module 15 calculates and adjusts the signal according to the algorithm and sends it to the actuators (rotary kiln drive motor frequency converter 16 and blower / oxygen valve actuator 17) to adjust the kiln speed and oxygen supply. Figure 5 This is a schematic diagram of the control logic flow of this multi-source online spectral monitoring closed-loop feedback system. The process begins with initialization. Step S1 reads the initial spectral monitoring values ​​and sets the target range (e.g., FeO target decrease rate, CO / CO2 ratio range). Step S2 continuously acquires LIBS solid-state data and Raman gas data. Step S3 compares the current FeO level with the previous value or the target value to determine if the reduction rate has slowed down or if FeO is too high. Step S4 compares the current CO / CO2 and O2 with the set range to determine if the atmosphere is more oxidizing or reducing. Step S5 is the decision node: if a slower decrease in FeO is detected and the atmosphere is trending towards oxidation (CO / CO2 is low), then proceed to step S6a to implement "enhanced reduction" measures, including issuing commands to reduce the kiln speed and decrease the blast volume; if FeO is already very low and the atmosphere is extremely reducing (CO too high, O2≈0), then proceed to step S6b to implement "suppressed over-reduction" measures, including slightly increasing the kiln speed and increasing the oxygen supply to burn excess CO; otherwise, if all are normal, then maintain the current operation (S6c). After the adjustment is executed, return to step S2 and loop. The dashed box in the process indicates the adaptive module, which can adjust the decision thresholds (such as the FeO change threshold and the CO / CO2 threshold) based on historical data to optimize control sensitivity. Figure 5 The demonstrated closed-loop logic ensures that when the status of the furnace charge and furnace gas indicators exceeds the normal range, the system promptly takes corrective measures with dual-parameter linkage to restore the process to steady state. Figure 6This is a comparison chart of key indicators of the reduction process before and after applying this multi-source online spectroscopic monitoring closed-loop feedback system. (a) shows the change curve of FeO content in the furnace over time without using this multi-source online spectroscopic monitoring closed-loop feedback system. It can be seen that the decrease in FeO is unstable, and it rebounds midway due to atmospheric fluctuations. (b) shows the change in FeO after using this multi-source online spectroscopic monitoring closed-loop feedback system. The curve smoothly and monotonically decreases until it approaches zero, showing that the reduction process is stable and controlled. (c) shows a comparison of the CO / CO2 ratio. Under traditional control, the ratio fluctuates greatly, while under the control of this multi-source online spectroscopic monitoring closed-loop feedback system, the CO / CO2 ratio remains almost constant. These comparisons quantitatively demonstrate the effectiveness of this multi-source online spectroscopic monitoring closed-loop feedback system in stabilizing FeO reduction and the atmosphere.

[0055] In a preferred embodiment of the present invention, a nano-catalyst buffer addition device is used in the rotary kiln ilmenite reduction process, which includes a pre-reduction section and a main reduction section. The device comprises a nano-catalyst storage and supply system, a pre-reduction section injection mechanism, a dynamic dispersion component, an alkalinity-pH monitoring probe, and a central control cabinet. The nano-catalyst storage and supply tank includes a nano-catalyst storage and supply tank a and a screw feeder b, used to store CaO–Na2O nano-dry powder catalyst and deliver it at a set flow rate. The pre-reduction section injection mechanism includes multiple high-temperature resistant pre-reduction section injection nozzles arranged in the pre-reduction zone of the rotary kiln, connected to the storage and supply system to inject the nano-catalyst through a carrier gas into the pre-reduction zone. The surface of the material layer inside the rotary kiln; the dynamic dispersion component is located inside the rotary kiln cylinder, corresponding to the position of the injection nozzle in the pre-reduction section, and is used to promote the uniform dispersion of the nano-catalyst in the material after it is injected; the alkalinity-pH monitoring probe includes an alkalinity sensor and a gas composition sensor installed at different positions inside the kiln, which detects the alkalinity of the molten slag and the content of CO, CO2 and O2 in the reducing atmosphere in real time, and transmits the detection signals to the central control cabinet; the central control cabinet generates control commands based on the deviation between the monitoring signal and the preset target to adjust the conveying rate of the quantitative feeder and / or the start and stop of the injection mechanism, thereby dynamically controlling the injection amount and timing of the nano-catalyst.

[0056] The mass ratio of CaO to Na2O in the CaO–Na2O nanopowder catalyst is (3–5):1, the volume average particle size of the catalyst particles is less than 50 nm, and the specific surface area is not less than 30 m² / g.

[0057] The dynamic dispersion component is a heat-resistant vibrator or dynamic dispersion blade installed on the inner wall of the rotary kiln. It can rotate with the rotary kiln and apply mechanical disturbance or vibration to the material layer to prevent the agglomeration and deposition of nanoparticles.

[0058] The alkalinity-pH monitoring probe includes: a sampling tube for collecting molten slag samples from inside the kiln, a rapid condensation device connected to the sampling tube, and an online alkalinity meter connected to the sampling tube for measuring the alkalinity index of the slag sample; the gas composition sensor includes a CO / CO2 infrared gas analyzer and an electrochemical oxygen sensor, which are respectively arranged at the kiln tail and the middle of the kiln.

[0059] The central control cabinet includes a programmable logic controller (PLC) and host computer monitoring software. The PLC has a pre-stored control model based on PID and fuzzy control algorithms. When the CO / CO2 ratio is lower than the preset lower limit, the catalyst feed rate is increased. When the CO / CO2 ratio is higher than the preset upper limit or the oxygen content is lower than the lower limit, the feed rate is reduced or the injection is suspended. The opening and closing of each nozzle is adjusted according to the alkalinity difference at different axial positions in the kiln to maintain the stability of the pre-reduction section reaction.

[0060] A method for adding nano-scale catalysts to improve reduction efficiency and product stability in the rotary kiln reduction process of ilmenite includes the following steps: (a) When the material temperature in the rotary kiln reaches the pre-reduction stage temperature, the injection of nano-CaO–Na2O catalyst is initiated, with the initial injection rate corresponding to a catalyst addition amount of 0.5–1% of the mass of the raw material entering the kiln; (b) The nano-catalyst is sprayed onto the surface of the material in the pre-reduction section of the rotary kiln via carrier gas, and the catalyst is uniformly diffused in the material using a dynamic dispersion component within the kiln; (c) The basicity of the slag and the CO and CO2 concentrations in the reducing atmosphere are continuously monitored in the rotary kiln. O2 concentration: When the detected value deviates from the preset range, the central control cabinet adjusts the catalyst injection amount: when the CO / CO2 ratio decreases or the slag basicity is lower than the target value, the catalyst feeding rate is increased or the number of nozzles opened is increased; when the CO / CO2 ratio is too high or the oxygen content in the kiln is too low, the feeding rate is reduced or the catalyst injection is temporarily stopped; (d) Maintain the above feedback control until the end of the pre-reduction stage, and then continue to supplement the catalyst in small amounts or stop the injection as needed in the main reduction stage; (e) Remove the material that has been reduced and cool it, and carry out subsequent screening and magnetic separation as usual to obtain metallic iron and titanium-rich slag products.

[0061] In step (a), the pre-reduction stage temperature is 700–950℃, and the preferred addition amount of the catalyst CaO–Na2O is 0.8% of the raw material mass, which is injected starting when the material precipitates some FeO upon heating. In step (c), the preset range for the CO / CO2 ratio is 4.0–6.0, the preset lower limit for the O2 concentration is 0.8%, and the preset target range for the slag basicity is 1.0–1.3. The central control cabinet collects and calculates the deviation every 5 minutes, and uses an incremental PID algorithm combined with fuzzy rules to adjust the catalyst feeding command.

[0062] The catalyst CaO–Na2O undergoes a partial chemical reaction at high temperatures to form catalytically active intermediate products, including Na2FeO2, Na2SiO3, or Ca–Fe–O composite oxides. These products accelerate the reduction of FeO to metallic Fe at the microscopic interface and enhance slag flowability. By adjusting the opening and closing of nozzles in different pre-reduction sections, the catalyst can be selectively injected along the rotary kiln axis to adapt to changes in material temperature gradient and composition. Specifically, the kiln tail section primarily injects CaO components for early desulfurization and dephosphorization, the middle section injects a CaO–Na2O combination to enhance iron reduction, and the kiln head section receives a small amount of supplementary injection as needed to prevent over-reduction.

[0063] Figure 7 This is a schematic diagram of the nanocatalyst buffer addition device of the present invention on a rotary kiln. The diagram shows the relative arrangement of the following components: nanocatalyst storage and supply tank (a), screw metering feeder (b), carrier gas conveying pipeline (c), rotary kiln body (d), pre-reduction section injection nozzle (e), dynamic dispersion blades inside the kiln (f), alkalinity-pH monitoring probe (g), and central control cabinet (h). Figure 8 This is a process flow diagram of the nanocatalyst buffer addition device method. It includes the following steps: raw material and reducing agent preparation (S1), pre-reduction heating (S2), dynamic injection of nanocatalyst (S3), online monitoring of alkalinity and atmosphere (S4), closed-loop regulation and control (S5), main reduction completion and cooling (S6), etc. Solid arrows indicate the process flow sequence, and dashed arrows indicate the control feedback signal path.

[0064] In a preferred embodiment of the present invention, a rotary kiln intelligent oxygen-temperature-speed three-parameter coordinated control system is used for the automatic control of the rotary kiln process of ilmenite reduction. The system includes: at least one set of temperature sensors for detecting the temperature of the calcination zone of the rotary kiln; at least one flue gas composition analyzer for detecting the concentrations of CO, CO2, and residual O2 in the kiln exhaust; an oxygen-enriched air supply device (which can be a blower / oxygen valve actuator 17) with an adjustable oxygen-enriched valve for supplying oxygen-enriched combustion air into the rotary kiln; a variable frequency drive motor (which can be a rotary kiln drive motor inverter 16) connected to the rotary kiln for adjusting the kiln rotation speed; and an industrial controller (which can be a central control module 15) electrically connected to the aforementioned sensors and actuators. The industrial controller has a built-in fuzzy control module, a PID control module, and a neural network self-learning module for collaboratively calculating the oxygen supply valve opening, kiln temperature adjustment, and kiln rotation speed based on the kiln temperature and atmosphere deviation, thereby coordinating the oxygen supply, calcination temperature, and rotation speed within the rotary kiln.

[0065] The fuzzy control module stores expert rules and fuzzy logic for the rotary kiln reduction process. It uses kiln temperature deviation, temperature rise rate, and flue gas residual oxygen deviation as fuzzy input variables and outputs fuzzy decision signals for the adjustment direction and magnitude of oxygen supply and kiln speed. When a local over-reduction or over-oxygenation trend occurs, the fuzzy module can promptly identify and output composite adjustment commands to prioritize the correction of abnormal operating conditions.

[0066] The PID control module includes at least three PID loops, used for oxygen-enriched valve opening control, fuel / combustion air volume control, and kiln speed control, respectively. The oxygen-enriched valve opening PID loop uses the residual oxygen content in the flue gas as feedback, takes the deviation between the residual oxygen and the target value as input, and adjusts the oxygen supply valve opening to maintain a constant residual oxygen level at the kiln tail. The furnace temperature PID loop uses the temperature sensor reading as feedback to control the kiln temperature within a preset window (1100–1200℃). The kiln speed PID loop adjusts the rotary kiln speed according to the temperature distribution along the kiln or other linkage requirements. The setpoints of each PID loop are dynamically given by the fuzzy control module, and their PID parameters are tuned online by the neural network module, thereby achieving decoupled and coordinated control of multiple parameters.

[0067] The neural network self-learning module employs an Adaptive Neural Fuzzy Inference System (ANFIS) or an artificial neural network to adjust the control parameters of the PID control module online. The neural network module takes kiln temperature error, oxygen content error, and their rate of change as inputs, and outputs correction values ​​for the PID controller parameters. By continuously learning from historical data and current control performance, it automatically optimizes the PID parameters, improving control response speed and steady-state accuracy.

[0068] It also includes a safety expert rule base to limit the safe range of control parameters and intervene in control when anomalies occur. The rule base includes at least: a residual oxygen content lower limit threshold rule, which forcibly opens the oxygen-enriching valve to maintain a minimum oxygen supply when the residual oxygen in the flue gas is below a preset lower limit; a kiln maximum temperature threshold rule, which automatically reduces fuel supply and increases kiln speed to cool down when the kiln temperature exceeds a preset upper limit; and a sensor fault detection rule, which issues an alarm and switches the control mode to safe operating condition when key sensor signals deviate from a reasonable range or fluctuate abnormally, preventing misoperation.

[0069] A method for controlling the reduction of ilmenite using a rotary kiln intelligent collaborative control system includes the following steps: (a) collecting temperature data from multiple locations within the rotary kiln and data on the concentrations of O2, CO, and CO2 in the kiln tail flue gas; (b) comparing the collected data with preset targets to obtain indicators characterizing the current operating conditions, such as temperature deviation, residual oxygen deviation, and their rate of change; (c) inputting the above indicators into a fuzzy control module, and calculating based on a preset fuzzy rule set to obtain fuzzy control outputs for the opening degree of the oxygen-enriched gas supply valve, fuel / air volume adjustment, and kiln speed adjustment; (d) the fuzzy control outputs serving as the basis for each PID control loop. (e) The PID control module calculates specific control action commands based on the error between the set value and the real-time value, and adjusts the oxygen-enriched valve, fuel supply and motor frequency; (d) The neural network module monitors the control error in step (d), adjusts the PID parameters online to optimize the control process, and the neural network continuously learns new data to update its model; (f) The above control process is continuously executed in a loop to coordinately regulate the oxygen supply, temperature and speed of the rotary kiln, and stabilize the weak reducing atmosphere and temperature field in the kiln within the safety constraints, so as to achieve low FeO residue and high TiO2 enrichment rate of the reduction product.

[0070] Figure 9 This is a schematic diagram of the composition and signal flow of the intelligent oxygen-temperature-speed three-parameter coordinated control system for rotary kilns of this invention. The diagram shows that sensors (thermocouples, gas analyzers) acquire kiln temperature and atmosphere information. After receiving the information, the controller processes it through a fuzzy control module, a PID control module, and a neural network module, and outputs control signals to actuators such as oxygen-enriched valves and variable frequency motors to achieve coordinated control of oxygen supply, furnace temperature, and rotation speed of the rotary kiln. Figure 10 This is a flowchart of the control strategy for the intelligent oxygen-temperature-speed three-parameter coordinated control system of this rotary kiln. The flowchart describes the working steps of the fuzzy-PID-neural network coupled control: First, key parameters such as kiln temperature and residual oxygen are collected and compared with the set target to generate deviations; then, the fuzzy decision module determines the current operating condition category based on the rule base and outputs a preliminary adjustment plan (such as oxygenation, cooling, or speed increase); next, the PID control module calculates the specific control quantity changes based on the new set value given by the fuzzy logic; the neural network module adjusts the PID parameters in real time to adapt to changes in operating conditions; finally, the corrected control commands are sent to the actuators (adjusting valve opening and motor frequency). Simultaneously, the system records the control effect for online training of the neural network, and the above process is executed cyclically. Figure 11 This is a schematic diagram of fuzzy control decision rule variables, showing an example variable relationship of the fuzzy control module inferring the oxygen supply adjustment amount using two input variables: temperature deviation and residual oxygen deviation. It is used to illustrate how fuzzy decision-making comprehensively considers multiple factors in output control decisions. Figure 12 This is a schematic diagram of the working process of neural network PID parameter self-tuning. Figure 13This is a schematic diagram of the integrated logic of a neural network and a PID controller. Figure 12 , Figure 13 This includes a simplified diagram of the neural network structure and its integration with the PID controller, illustrating how the network automatically corrects the PID parameters based on control deviations, thereby improving dynamic control performance.

[0071] In a preferred embodiment of the present invention, a reducing atmosphere fluctuation steady-state control algorithm module is used for automatic steady-state control of the furnace atmosphere during the reduction of ilmenite. The module includes: a state perception unit, a reinforcement learning decision-making unit, an expert control unit, and an execution output unit. The state perception unit collects parameters such as CO concentration, CO2 concentration, O2 concentration, and material temperature within the reducing furnace to form a state vector. The reinforcement learning decision-making unit constructs a reinforcement learning agent using a deep neural network and outputs control action commands based on the state vector. The expert control unit stores process expert rules for verifying and correcting the actions output by the reinforcement learning agent. The execution output unit converts the verified control actions into adjustments to on-site actuators, such as adjusting the fuel feed rate, the opening degree of the oxygen-enriched valve, or the rotary kiln speed, to regulate the reducing atmosphere within the furnace in real time.

[0072] The reinforcement learning decision-making unit is constructed using a deep reinforcement learning algorithm, including a deep neural network structure of policy network and value network. It utilizes reinforcement learning methods with continuous state and continuous action space to train and optimize the furnace atmosphere control strategy. Its reward function rewards lower FeO residual content and more stable TiO2 enrichment rate as positive rewards, and atmosphere deviation from the target composition and excessive control action amplitude as negative rewards. The agent interacts with the environment and adjusts the internal policy network parameters after receiving reward feedback, thereby gradually learning a control strategy that keeps the CO / CO2 ratio and residual oxygen level stable within the target range.

[0073] The expert control unit includes preset safety constraint rules and control experience rules. When the control action output by the reinforcement learning decision unit may violate the safety constraints, the expert control unit restricts or replaces the action. The safety constraint rules include at least: the residual oxygen content in the furnace must not be lower than a safety threshold, the CO / CO2 ratio must not exceed a predetermined upper limit, and the single control adjustment range must not exceed the limit, to ensure that the reducing atmosphere is controlled within a safe and controllable range. In addition, the expert control unit has compensation control logic based on process experience. For example, when the detected FeO residue is continuously high, it outputs an additional compensation action to enhance reduction; when the detected temperature is significantly low, it outputs a compensation action to extend the reaction time. The outputs of these expert rules can be superimposed with or preferentially selected from the actions of the reinforcement learning agent to improve the reliability of the control effect.

[0074] The state perception unit includes an anomaly monitoring submodule, which is used to verify the authenticity and consistency of the collected signals such as CO, CO2, O2, and temperature. When a sensor malfunction or data anomaly is detected (such as the oxygen content reading suddenly dropping to zero or not changing with control), the anomaly monitoring submodule sends a signal to the expert control unit, triggering the expert unit to temporarily take over control or enable a safety mode to avoid improper actions by the reinforcement learning agent due to erroneous information output.

[0075] The reinforcement learning decision-making unit and the expert control unit work collaboratively in parallel. The reinforcement learning decision-making unit is responsible for optimizing control within the normal range, while the expert control unit is responsible for ensuring safety under abnormal or extreme conditions. The control commands output by the algorithm module act on the key actuators of the rotary kiln reduction process, thereby maintaining a stable atmosphere and reaction conditions inside the furnace. Specifically, this includes: adjusting the fuel (or reducing agent) feed rate to control the CO production rate, adjusting the opening of the oxygen-enriched air valve to control the oxygen supply, and optionally adjusting the kiln speed or bed thickness to affect the reaction time. Through comprehensive adjustment, the CO / CO2 concentration ratio inside the furnace is maintained at a predetermined target (e.g., around 90:10), the residual oxygen content at the kiln tail is kept above the safe lower limit and as low as possible, so that the FeO residue is always ≤0.5%, and the TiO2 enrichment rate fluctuates ≤±0.4%.

[0076] After applying this module, the reducing atmosphere can be controlled in a steady state even under disturbances and load changes, resulting in highly stable quality of ilmenite reduction products and an improvement in process thermal efficiency of ≥20%. Specifically, the standard deviation of the CO / CO2 concentration ratio is reduced by more than 50% compared to conventional control; the residual O2 concentration remains stable within a reasonable trace range for an extended period (fluctuations reduced to within ±0.05%); and the average residual FeO value in titanium-rich products is reduced, with a significant decrease in batch standard deviation. Correspondingly, the waste of excess reducing agent and oxygen is reduced, and energy consumption per unit product decreases, achieving a balance between energy saving, consumption reduction, and quality improvement.

[0077] A steady-state control method for reducing atmosphere utilizes an algorithm module to control the atmosphere inside an ilmenite reduction furnace, comprising the following steps: (a) real-time acquisition of furnace atmosphere and temperature parameters to constitute the current system state; (b) inputting the state into a deep reinforcement learning agent model, which calculates and outputs control actions, including adjustments to fuel feeding and oxygen supply; (c) verifying whether the action meets preset expert rules, and modifying the action until safety constraints are met if a violation occurs; (d) issuing and executing the final determined control action command to adjust the reaction conditions inside the furnace; (e) collecting the new state and product quality feedback after execution, calculating the reward value based on the degree of deviation from the target, and using it to update the reinforcement learning model; (f) repeating the above process, continuously interactively training to gradually optimize the control strategy and maximize long-term rewards without rule violations.

[0078] By employing the above method and utilizing a control strategy that combines deep reinforcement learning with expert experience, dynamic and stable control of the CO / CO2 / O2 composition in the furnace can be achieved. This maintains a constant reducing atmosphere under various disturbances, ensuring low FeO residue and stable TiO2 enrichment in the titanium-rich material.

[0079] Figure 14 This is a functional architecture diagram of the atmospheric fluctuation steady-state control algorithm module. The diagram shows the main components of the module, including a signal input interface (acquiring sensor data such as CO, CO2, O2 concentrations, and temperature), a deep reinforcement learning agent, an expert control unit, and a decision output interface (connected to fuel feed control and actuators). The actuators can be a blower / oxygen valve actuator 17 or a rotary kiln drive motor frequency converter 16. Arrows indicate information flow: the agent obtains the environmental state from the input, calculates the output action through a neural network, verifies it using expert rules, and sends it to the execution unit. Simultaneously, it obtains rewards from feedback for policy updates. Figure 15 This is a flowchart of the decision-making process for a deep reinforcement learning agent. It includes a cyclical process of state preprocessing, policy network computation of actions, comparison and correction with expert rules, execution of actions, receiving environmental feedback and rewards, and updating the policy. This flowchart emphasizes the integration of reinforcement learning policies and expert intervention. Figure 16 This is a schematic reward function curve, using FeO residue and CO / CO2 ratio as examples, showing how the reward changes with FeO deviation and atmosphere deviation: the curve reaches its maximum value near FeO=0 and CO / CO2=target value, and the reward decreases when deviating from these values. This helps in understanding the agent's optimization direction. Figure 17 These are the control effect curves of the algorithm module under actual operating conditions. The curves showing the changes in CO / CO2 ratio and FeO residue over time are plotted for both cases with and without this algorithm (traditional control). The curve comparison shows that this algorithm module significantly reduces fluctuation amplitude and improves stability.

[0080] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for producing ilmenite powder by reducing ilmenite, characterized in that: Includes the following steps: Raw material homogenization pretreatment: Ilmenite raw materials from different ore sources or batches are batched and mixed, and the contents of key components such as TiO2 and Fe are detected by online sampling and online XRF or LIBS analyzer. The ratio is adjusted to keep the composition of the raw materials entering the furnace constant within the set range. The dynamic proportioning of biomass-coal mixed reducing agent is carried out based on a biomass-coal mixed reducing agent dynamic proportioning system: a set proportion of biomass carbon is used to replace part of the coal powder as a reducing agent; the reducing atmosphere in the furnace is monitored by installing an online atmosphere analyzer. When the FeO content is detected to be too high, the proportion of biomass reducing agent is automatically increased; otherwise, it is reduced to about 20%, dynamically maintaining the carbon-oxygen balance in the furnace and reducing the fluctuation range of FeO. Multi-source online spectral monitoring closed-loop feedback control is implemented based on a multi-source online spectral monitoring closed-loop feedback system: A laser-induced breakdown spectroscopy (LAS) system is installed in the reaction section of the rotary kiln, and a laser-Raman spectroscopy combined monitoring system is installed in the kiln tail flue to analyze the reactants and atmosphere inside the furnace. LAS is used to quantitatively detect the degree of iron reduction in the furnace charge, while the Raman spectroscopy combined monitoring system is used to monitor the composition of the furnace gas. The control system automatically adjusts the kiln speed and the amount of combustion air / oxygen supplied based on the FeO content trend measured by LAS and the atmosphere composition measured by Raman spectroscopy: when reduction is insufficient, the speed is appropriately reduced to prolong the reaction time and the oxygen supply is reduced to maintain reducing properties; when reduction is excessive, the speed is increased or a small amount of air is introduced to suppress it, thus forming a closed-loop control. The buffering addition of nano-CaO-Na2O composite catalyst based on nano-scale catalyst buffering addition device: During the raw material mixing stage, a composite additive composed of nano-calcium oxide and sodium oxide is added at 0.5–1% of the weight of ilmenite. Multi-stage magnetic separation and reforming of finished products: After the reduction product is cooled by water quenching, large iron particles are first screened out, and the remaining material is ground in two stages to <0.

074. Then, it is passed through a cascade system of weak magnetic separation and strong magnetic separation in sequence. Weak magnetic separation captures larger iron particles and metallized magnetic materials, while strong magnetic separation recovers fine metallic iron powder. Multi-stage magnetic separation ensures that the iron phase is separated from the titanium phase as much as possible, resulting in two products: titanium-rich material and iron-containing powder. Subsequently, the magnetically separated metallic iron particles and iron powder are mixed and pressed into shape to obtain titanium iron powder product.

2. The method for producing ilmenite powder by reducing ilmenite according to claim 1, characterized in that: The biomass-coal mixed reducing agent dynamic proportioning system includes: at least two raw material silos, used to store and supply coal-based reducing agent and biomass-based reducing agent respectively; a feeding device with independently adjustable output flow rate, connected to the outlets of the coal-based silo and the biomass silo respectively, to deliver the two reducing agents to the reaction zone inside the furnace; a furnace atmosphere monitoring module, including a gas sampling component and a multi-component gas sampler installed inside the furnace or at the flue gas duct, used to measure the concentrations of CO, CO2, and O2 in the furnace gas online; a raw material characteristic monitoring module, used to detect the calorific value and moisture characteristics of biomass and coal; and a central control unit, which receives real-time data from the furnace atmosphere monitoring module and the raw material characteristic monitoring module. The central control unit has a pre-stored control program that calculates the target proportion of biomass to coal based on the deviation of the furnace gas composition and the changes in raw material characteristics, and outputs control signals to the feeding device.

3. The method for producing ilmenite powder by reducing ilmenite according to claim 1, characterized in that: The multi-source online spectral monitoring closed-loop feedback system includes: a solid charge spectral monitoring device for real-time acquisition of information on iron and titanium elements and their combined states in the furnace material; a furnace gas composition spectral monitoring device for real-time measurement of the content of CO, CO2, and O2 gas components in the furnace or furnace gas; a central control module, which is communicatively connected to the solid charge spectral monitoring device and the furnace gas composition spectral monitoring device, for fusing and analyzing the charge spectral data and furnace gas spectral data, and generating control commands according to a preset control strategy; and an actuator, including a kiln speed control device and a combustion gas supply adjustment device, which receives commands from the central control module and automatically adjusts the kiln speed and oxygen supply to control the process of the reduction reaction in the furnace in a closed loop.

4. The method for producing ilmenite powder by reducing ilmenite according to claim 1, characterized in that: The nano-catalyst buffer addition device includes a nano-catalyst storage and supply system, a pre-reduction section injection mechanism, a dynamic dispersion component, an alkalinity-pH monitoring probe, and a central control cabinet. The nano-catalyst storage and supply system includes a nano-catalyst storage tank and a screw feeder for storing CaO–Na2O nano-dry powder catalyst and conveying it at a set flow rate. The pre-reduction section injection mechanism includes multiple high-temperature resistant pre-reduction section injection nozzles located in the pre-reduction zone of the rotary kiln, connected to the storage and supply system to inject the nano-catalyst onto the surface of the material layer inside the rotary kiln via a carrier gas. The dynamic dispersion component is located within the rotary kiln cylinder. Inside, corresponding to the position of the injection nozzle in the pre-reduction section, it is used to promote the uniform dispersion of the nano-catalyst in the material after it is injected. The alkalinity-pH monitoring probe includes an alkalinity sensor and a gas composition sensor installed at different positions in the kiln, which detect the alkalinity of the slag in the kiln and the content of CO, CO2 and O2 in the reducing atmosphere in real time, and transmits the detection signals to the central control cabinet. The central control cabinet generates control commands based on the deviation between the monitoring signal and the preset target to adjust the conveying rate of the quantitative feeder and / or the start and stop of the injection mechanism, thereby dynamically controlling the injection amount and timing of the nano-catalyst.

5. The method for producing ilmenite powder by reducing ilmenite according to claim 1, characterized in that: In multi-source online spectral monitoring closed-loop feedback control, the closed-loop control achieves basic monitoring and control through a multi-source online spectral monitoring closed-loop feedback system; Method 1: Based on the steady-state control algorithm module for reducing atmosphere fluctuations, precise control of the CO / CO2 / O2 atmosphere inside the furnace is achieved; Method 2: Based on the intelligent oxygen-temperature-speed three-parameter coordinated control system of rotary kiln, the linkage control of oxygen supply, temperature and kiln speed is realized.

6. The method for producing ilmenite powder by reducing ilmenite according to claim 5, characterized in that: The rotary kiln intelligent oxygen-temperature-speed three-parameter coordinated control system includes: at least one set of temperature sensors for detecting the temperature of the rotary kiln calcination zone; at least one flue gas composition analyzer for detecting the concentrations of CO, CO2, and residual O2 in the kiln exhaust; one set of oxygen-enriched air supply device with an adjustable oxygen-enriched valve for supplying oxygen-enriched combustion air into the rotary kiln; one set of variable frequency drive motor connected to the rotary kiln for adjusting the kiln rotation speed; and one set of industrial controller electrically connected to the temperature sensors, the oxygen-enriched air supply device, and the variable frequency drive motor. The industrial controller has built-in fuzzy control module, PID control module, and neural network self-learning module for coordinating the calculation of the oxygen supply valve opening, kiln temperature adjustment, and kiln rotation speed based on the kiln temperature and atmosphere deviation, so that the oxygen supply, calcination temperature, and rotation speed in the rotary kiln are coordinated and controlled.

7. A method for producing ilmenite powder by reducing ilmenite according to claim 5, characterized in that: The steady-state control algorithm module for reducing atmosphere fluctuations includes: a state perception unit, a reinforcement learning decision-making unit, an expert control unit, and an execution output unit. The state perception unit collects parameters such as CO concentration, CO2 concentration, O2 concentration, and material temperature in the reduction furnace to form a state vector. The reinforcement learning decision-making unit uses a deep neural network to construct a reinforcement learning agent and outputs control action commands based on the state vector. The expert control unit stores process expert rules to verify and correct the actions output by the reinforcement learning decision-making unit. The execution output unit converts the verified control actions into adjustments to the on-site actuators.

Images