A dynamic optimization method for double-side blown copper smelting slag type based on molten pool working condition real-time monitoring and model driving

Abstract

The present application belongs to copper pyrometallurgy technical field, specifically discloses a kind of dynamic optimization method of double-side blowing copper smelting slag type based on molten pool working condition real-time monitoring and model driving.The method includes: real-time acquisition molten pool temperature, flue gas oxygen content, slag Fe3O4 Content and viscosity and other multiple key working condition parameters;Parameter is input into the preset control model, the expert rule based on metallurgical thermodynamics and production data is embedded in the model, and the dynamic adjustment amount of SiO2, CaO, MgO three kinds of slag forming agent is calculated accordingly;According to the adjustment amount, the proportion of slag forming agent is adjusted in real time;Periodic acquisition of slag sample analysis, compare measured value with model expected value, and feedback correction model parameters, realize closed-loop optimization.The present application can dynamically adjust slag type according to molten pool real-time working condition, reduce the fluctuation of slag fluidity by more than 40%, stabilize the copper content of slag below 1.5%, and improve the direct recovery rate of copper to more than 97.5%, significantly improve the process stability and metal recovery rate.

Description

Technical Field

[0001] This invention belongs to the field of copper pyrometallurgical technology, and particularly relates to a dynamic optimization method for slag type in double-sided copper blowing based on real-time monitoring of molten pool conditions and model-driven approach. Background Technology

[0002] In modern pyrometallurgical processes such as double-sided blowing pool copper smelting, the selection and control of slag pattern are crucial factors determining smelting efficiency, metal recovery rate, furnace lining life, and production stability. An ideal slag should possess suitable melting point and viscosity to ensure good fluidity, while effectively capturing impurities such as gangue and iron oxide, and minimizing physical entrainment and chemical dissolution losses to the target metallic copper. Traditional slag pattern control methods typically rely on static thermodynamic phase diagram calculations and production experience, pre-setting a fixed slag-forming agent ratio. However, the actual production process is a complex and dynamically changing system. Fluctuations in raw material composition, adjustments in oxygen supply intensity, and changes in production load can all cause real-time changes in key operating parameters within the molten pool, such as temperature, oxygen potential, and matte grade. A fixed slag pattern often struggles to adapt to these dynamic changes, leading to a mismatch between the slag pattern and real-time operating conditions. This results in a series of problems, including excessively high slag viscosity, excessive Fe3O4 content, and increased copper content in the slag, severely impacting production indicators.

[0003] To address the limitations of static control, existing technologies have explored various approaches. For example, patent CN120032735A discloses a slag optimization method for a dual-furnace continuous copper smelting process. This method optimizes the slag shape for smelting and blowing by calculating slag viscosity and drawing isothermal phase diagrams. It combines theoretical calculations with production practice, providing an optimized slag composition range. Patent CN120272733A discloses a method for optimizing the blowing process. It analyzes the effects of different slag shapes, oxygen concentrations, and temperatures on the copper content of the blowing slag through equilibrium calculations to determine optimal blowing parameters. These methods have made progress in slag shape optimization under static or pre-set conditions, but still mainly rely on pre-calculation or fixed proportions, failing to achieve dynamic adjustments based on real-time molten pool conditions. Furthermore, other technological development directions include feedforward control based on raw material analysis and intelligent optimization based on historical data and machine learning models. However, feedforward control lacks the ability to adjust the real-time state of the molten pool; and existing intelligent optimization models are often too general, failing to specifically address slag optimization by constructing a clear and executable closed-loop technical solution that integrates multi-dimensional real-time operating condition monitoring and the synergistic dynamic control of multiple slag-forming agents. In particular, existing technologies have yet to provide an effective solution for combining multiple key parameters that directly reflect the slag state, such as molten pool temperature, oxygen potential, slag viscosity, and Fe3O4 content, and using this as a basis to dynamically adjust the proportions of various slag-forming agents such as SiO2, CaO, and MgO.

[0004] Therefore, the present invention needs to provide a method that can dynamically adjust the slag shape according to the real-time operating conditions of the molten pool to achieve adaptive optimization. Summary of the Invention

[0005] This invention aims to solve the technical problem of slag shape control in double-sided copper blowing smelting, and provides a dynamic optimization method for slag shape in double-sided copper blowing smelting based on real-time monitoring of molten pool conditions and model-driven approach, including the following steps:

[0006] Step S1: Real-time acquisition of molten pool operating parameters in the double-sided blowing smelting furnace. The molten pool operating parameters include at least the molten pool temperature, the O2 content in the flue gas, the Fe3O4 content in the slag, and the measured slag viscosity.

[0007] Step S2: Input the molten pool operating parameters collected in real time in step A into the preset control model. The control model is embedded with expert rules constructed based on historical production data and metallurgical thermodynamics principles. Based on the expert rules and the input real-time parameters, calculate the dynamic adjustment amount of the three slagging agents, SiO2, CaO and MgO.

[0008] Step S3: Based on the dynamic adjustment amount calculated in step B, adjust the proportions of SiO2, CaO, and MgO added to the smelting furnace in real time.

[0009] Step S4: Periodically collect slag samples for composition and performance analysis. Compare the measured slag composition or performance data obtained from the analysis with the expected target value or predicted value of the control model described in Step B. Based on the comparison deviation, make feedback corrections to the expert rule parameters in the control model to achieve closed-loop optimization.

[0010] Furthermore, the expert rules of the control model in step B include at least the following: when the real-time collected molten pool temperature is higher than a first threshold and the O2 content in the flue gas is higher than a second threshold, a first rule is triggered, and the output of the first rule is to increase the SiO2 addition ratio; when the real-time collected Fe3O4 content in the slag is higher than a third threshold, a second rule is triggered, and the output of the second rule is to increase the CaO addition amount; when the real-time collected measured slag viscosity is higher than a fourth threshold, a third rule is triggered, and the output of the third rule is to increase the MgO addition ratio.

[0011] Further, the first threshold is 1350℃, the second threshold is 80% volume concentration; in the first rule, the addition ratio of SiO2 is increased by 0.5% to 1.0%; the third threshold is 5% by mass; in the second rule, the addition ratio of CaO is increased by 0.3% to 0.5%; the fourth threshold is 0.3 Pa·s; in the third rule, the addition ratio of MgO is increased by 0.2% to 0.4%.

[0012] Furthermore, in step C, the iron-silicon ratio in the final slag is controlled within the range of 1.8 to 2.2 by dynamically adjusting the addition ratio of SiO2, CaO and MgO.

[0013] Furthermore, the time interval for periodically collecting slag samples in step D is 30 minutes.

[0014] Furthermore, after dynamic adjustment in step C, the mass percentage range of each component in the slag is: SiO2 20% to 25%, CaO 2.5% to 4%, and MgO 1.5% to 2.5%.

[0015] Furthermore, the molten pool operating parameters collected in real time in step A also include the grade of matte.

[0016] Furthermore, when the third rule is triggered, the control model also outputs instructions to increase the target set value of the molten pool temperature of the smelting furnace.

[0017] Compared with the prior art, the method of the present invention has the following advantages:

[0018] (1) Improved stability of the double-sided copper smelting process: By real-time monitoring of multiple key parameters of the molten pool and dynamic adjustment of the slag-forming agent, dynamic matching between the slag type and real-time operating conditions was achieved, effectively solving the problem of mismatch between static slag type and dynamic operating conditions. The results show that the fluctuation of slag fluidity can be reduced by 40%, and the process is more stable and controllable.

[0019] (2) Improved copper recovery rate: By precisely controlling the slag type, especially by keeping the Fe3O4 content in the slag stable within the ideal range of 3% to 5%, the copper content in the slag was significantly reduced, keeping it stable below 1.5%, thereby increasing the direct copper recovery rate to over 97.5%, resulting in significant economic benefits.

[0020] (3) Extended equipment life: Effective control of Fe3O4 content reduces its corrosive effect on furnace lining, which helps to extend the service life of smelting furnace lining and reduce maintenance costs.

[0021] (4) Enhanced production adaptability: The closed-loop feedback optimization mechanism established by this invention enables the system to adapt to fluctuations in raw material composition and production load, enhances the robustness of the entire production system, reduces human intervention, and provides an effective technical approach for intelligent production of copper smelting. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the dynamic optimization method for double-sided copper blowing slag type according to the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer and more explicit, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the scope of protection of the invention.

[0024] This invention provides a dynamic optimization method for slag patterns in dual-sided copper blowing smelting based on real-time monitoring of molten pool conditions and model-driven approach. (See attached diagram) Figure 1 As shown, this method constitutes a complete "perception-decision-execution-learning" closed-loop control system, aiming to solve the fundamental problem that traditional static slag shape control is unable to adapt to the dynamic changes in the molten pool conditions. Its core lies in real-time monitoring of multi-dimensional key parameters, using an intelligent model with embedded expert rules for decision-making, coordinating the adjustment of various slag-forming agents, and continuously optimizing the model itself based on feedback results, thereby achieving accurate and dynamic matching between the slag shape and real-time operating conditions.

[0025] First, step S1 is executed to collect molten pool operating parameters in real time. To ensure the accuracy of control, key indicators reflecting the overall metallurgical state of the molten pool need to be monitored online or analyzed via high-frequency sampling. In a specific implementation scenario, these parameters and their acquisition methods include: continuously measuring the molten pool temperature online using a sheathed thermocouple inserted into the molten pool; installing a zirconia oxygen analyzer in the smelting furnace flue to monitor the O2 content in the flue gas in real time, which characterizes the oxidizing atmosphere (oxygen potential) of the molten pool; periodically collecting slag samples (e.g., every 20-30 minutes) through water quenching, and after drying, analyzing their chemical composition using X-ray fluorescence spectrometry (XRF) to obtain the Fe3O4 content in the slag; simultaneously, using a high-temperature rotational viscometer to measure the actual slag viscosity of the remelted slag sample or an online dedicated viscosity probe, as a direct indicator for evaluating slag fluidity. In addition, as a preferred implementation, samples can also be periodically taken from the matte tap and the matte grade can be quickly obtained through XRF analysis. All these parameters are transmitted in real time to the central control unit via an industrial network, forming a dynamically updated operating condition dataset.

[0026] Secondly, step S2 is executed to calculate the adjustment amount of slagging agent based on the control model. The central control unit inputs the real-time parameters collected in step S1 into a preset control model. This model is not a simple calculation formula, but an expert rule system with a clear "IF-THEN" logic, which integrates historical production big data, metallurgical thermodynamic principles such as the Fe-O-Si-Ca-Mg multi-component phase diagram, and the experience of experts in the field. The core function of the model is to analyze the working condition type represented by the current combination of multiple parameters, and accordingly calculate the dynamic adjustment amount required for the three key slagging agents, SiO2, CaO, and MgO. A typical expert rule base supporting claims 2-3 is as follows: when the system determines that the molten pool temperature is >1350℃ and the flue gas O2 content is >80%, the first rule is triggered, and the model outputs the instruction "increase the proportion of SiO2 addition". The metallurgical basis for this model is that the high temperature and high oxygen potential conditions easily promote the over-oxidation of FeO in the slag to form high-melting-point Fe3O4. Increasing SiO2 can reduce the iron-silicon ratio (Fe / SiO2) of the slag, thereby inhibiting the formation of Fe3O4. The preferred range for increasing this ratio is 0.5% to 1.0%. When the Fe3O4 content in the slag is >5%, the second rule is triggered, and the output instruction is "increase the amount of CaO added". CaO can react with Fe3O4 to form low-melting-point phases such as calcium iron olivine, promoting the decomposition and melting of solid Fe3O4. The preferred increase is 0.3% to 0.5%. When the measured slag viscosity is >0.3 Pa·s, the third rule is triggered, and the output instruction is "increase the proportion of MgO added". The addition of MgO can effectively reduce the liquidus temperature of the slag and significantly improve its fluidity. The preferred increase is 0.2% to 0.4%. This model can simultaneously process multiple triggered rules and comprehensively calculate the synergistic adjustment scheme for SiO2, CaO, and MgO.

[0027] Next, step S3 is executed to dynamically adjust the slag-forming agent addition ratio. Based on the precise instructions output by the control model in step S2, the central control unit drives the automatic feeding system (such as a quantitative belt scale or pneumatic conveying device) to adjust the feeding rate of SiO2 (usually in the form of quartz sand), CaO (usually in the form of lime), and MgO (usually in the form of dolomite) added to the reaction zone of the smelting furnace in real time. The macroscopic process objective of this dynamic adjustment is to achieve a more optimal stable slag composition through the synergistic effect of multiple components. Specifically, this means stabilizing the iron-silicon ratio of the slag within the ideal range of 1.8 to 2.2 (corresponding to claim 4), and ensuring that the mass percentages of each component in the slag are: SiO2 20% to 25%, CaO 2.5% to 4%, and MgO 1.5% to 2.5% (corresponding to claim 6). Slag within this composition range has advantages such as a suitable melting point, low viscosity, and minimal copper dissolution loss.

[0028] Finally, step S4 is executed to achieve closed-loop feedback and optimization. The system establishes a fixed feedback cycle, for example, every 30 minutes (corresponding to claim 5). At the end of each cycle, the latest slag sample is automatically or manually collected and subjected to the same composition and performance analysis as in step S1. The obtained measured data (such as Fe3O4 content and viscosity) are compared with the target values ​​expected when the model was triggered for adjustment in step S2 (e.g., the desired reduction of Fe3O4 to 4%). If a systematic deviation is found, this deviation information is fed back to the control model. The model or its associated learning algorithm can then correct the threshold parameters in the expert rules (e.g., fine-tuning the viscosity trigger threshold from 0.3 Pa·s to 0.28 Pa·s) or the calculation coefficients for the adjustment amount. Through this continuous "evaluation-correction" cycle, the control model can continuously evolve, better adapting to the individual characteristics of specific furnaces and the long-term fluctuations of raw materials, thereby achieving closed-loop optimization.

[0029] Example 1

[0030] This embodiment demonstrates how the method of the present invention can perform multi-rule coordinated control when a typical high temperature and high oxygen potential abnormal condition occurs during copper smelting in a double-sided blown pool.

[0031] A double-sided blowing smelting furnace with a capacity of 2000 tons / day processes copper concentrate containing 25% Cu and 30% Fe. At a certain moment, the control system executes step S1 to collect real-time operating conditions: molten pool temperature T is 1360℃, flue gas oxygen content O2% is 82%, and matte grade is 72%. Simultaneous rapid analysis data from slag samples show that the Fe3O4 content in the slag is 5.2%, and the measured slag viscosity η is 0.32 Pa·s.

[0032] The control system then executes step S2, inputting the data into the control model. The model determines that: temperature (1360℃>1350℃) and oxygen content (82%>80%) trigger the first rule; Fe3O4 content (5.2%>5%) triggers the second rule; and viscosity (0.32Pa·s>0.3Pa·s) triggers the third rule. After comprehensive calculation, the model outputs a coordinated adjustment command: increasing the SiO2 addition ratio by 1.0 percentage point, the CaO addition ratio by 0.5 percentage point, and the MgO addition ratio by 0.3 percentage point.

[0033] The system then executes step S3: Based on the instructions, the automatic feeding system increases the SiO2 addition ratio from the baseline of 22% to 23.0%, CaO from 3% to 3.5%, and MgO from 1.8% to 2.1%. This adjustment aims to optimize the slag iron-silicon ratio from 2.1 to 1.9.

[0034] After 30 minutes of adjustment (i.e., one feedback cycle), step S4 was executed: re-sampling and analysis. Results showed that the Fe3O4 content in the slag was effectively reduced to 4.3%, and the slag viscosity significantly improved to 0.25 Pa·s. Continuous monitoring indicated that the copper content in the slag steadily decreased from 1.6% before the adjustment to approximately 1.3%.

[0035] This embodiment demonstrates that, in the face of adverse working conditions that easily lead to magnetite precipitation, the method of the present invention can automatically identify and initiate multi-rule synergistic regulation, quickly and accurately reverse the trend of furnace condition deterioration, and effectively control the Fe3O4 content and slag viscosity, thereby reducing the loss of copper in the slag.

[0036] Example 2

[0037] This embodiment demonstrates how the method of the present invention can proactively adjust the slag shape to maintain optimal production conditions when the composition of raw materials changes significantly (such as low-grade raw materials).

[0038] The same smelting furnace was switched to a low-grade copper concentrate containing 18% Cu, which has a higher content of gangue (SiO2, Al2O3). After the raw material switch, the system monitored the following molten pool conditions: T=1330℃, O2%=78%, and slag viscosity η=0.28Pa·s. These parameters themselves did not trigger the abnormal rules in Example 1.

[0039] At this point, the control model operates based on steady-state optimization logic (step S2). Its goal is to find the optimal slag type that can maintain the target iron-silicon ratio (2.0) and qualified viscosity (<0.3 Pa·s) under the new raw material conditions. After calculation, the model outputs the following instructions: to adapt to the high gangue content in the raw material and avoid excessive slag volume, the SiO2 addition ratio is reduced to 21%; at the same time, to improve the slag basicity to ensure fluidity, the CaO addition ratio is increased to 3.8% (step S3).

[0040] After the above adjustments, the slag fluidity remained good (η<0.3Pa·s) during the processing of low-grade ore, and the direct copper recovery rate remained stable at a high level of 97.5%.

[0041] This embodiment demonstrates that the method of the present invention can not only "correct deviations" in abnormal situations, but also actively "find the best" for different raw material conditions, showing strong production adaptability and robustness, and can effectively cope with raw material fluctuations.

[0042] Example 3

[0043] This embodiment demonstrates the method's precise response to atypical operating conditions, such as high viscosity due to raw material impurities.

[0044] Suppose that the Al2O3 content in a certain batch of copper concentrate is abnormally high. The system collected the following operating conditions: T=1340℃, O2%=75%, Fe3O4 in slag=4.0%, matte grade=70%, but the measured slag viscosity η is as high as 0.35 Pa·s.

[0045] The control model determines (step S2): only the viscosity parameter (0.35 Pa·s > 0.3 Pa·s) triggers the third rule. Based on this, the model outputs an adjustment instruction: increase the MgO addition ratio by 0.4 percentage points to 2.1% to utilize MgO to improve the fluidity of the high-alumina slag system. Simultaneously, as a preferred synergistic measure, the model also outputs an auxiliary instruction: increase the target setpoint of the molten pool temperature by 20°C to 1360°C to assist in reducing viscosity through physical heating (step S3).

[0046] Feedback after 30 minutes (step S4) showed that the slag viscosity had successfully dropped to 0.26 Pa·s and the slag fluidity had returned to normal.

[0047] This embodiment demonstrates the flexibility of the control model in handling complex problems. It can not only perform multi-factor coupled control, but also precisely target single parameter anomalies, and can combine "chemical slag adjustment" and "physical temperature control" synergistic strategies to quickly restore production stability.

[0048] Comparative Example 1

[0049] The same initial harsh operating conditions as in Example 1 were used (T=1360℃, O2%=82%, Fe3O4=5.2%, η=0.32Pa·s, slag copper content 1.6%). The method of this invention was applied for intervention, and after 30 minutes, the results were as described in Example 1, with significant improvement in all indicators.

[0050] Comparative Example 2

[0051] Under the same initial operating conditions, a traditional static slag control strategy was adopted, which involved strictly adhering to a fixed preset ratio (SiO2=22%, CaO=3%, MgO=1.8%) to add slag-forming agents without responding to real-time operating conditions. After 30 minutes, the furnace conditions continued to deteriorate: the Fe3O4 content in the slag rose to 6.5%, the slag viscosity worsened to 0.40 Pa·s, and the copper content in the slag increased to 2.0% due to poor slag fluidity and severe mechanical entrainment.

[0052] As can be seen from the comparison of the embodiments and comparative examples, the dynamic optimization method based on real-time monitoring of molten pool conditions and model-driven approach provided by the present invention can effectively overcome the drawbacks of static control. Through real-time perception, intelligent decision-making and closed-loop learning, it significantly improves the process stability, metal recovery rate and production adaptability of double-sided copper blowing smelting.

[0053] The above description is a preferred embodiment of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention shall fall within the protection scope of the present invention.

Claims

1. A dynamic optimization method for slag type in dual-sided copper blowing smelting based on real-time monitoring of molten pool conditions and model-driven approach, characterized in that, Includes the following steps: S1. Real-time acquisition of molten pool operating parameters in the double-sided blowing smelting furnace, the molten pool operating parameters including at least molten pool temperature, O2 content in flue gas, Fe3O4 content in slag, and measured slag viscosity; S2. Input the molten pool operating parameters collected in real time in step A into a preset control model. The control model is embedded with expert rules constructed based on historical production data and metallurgical thermodynamics principles. Based on the expert rules and the input real-time parameters, calculate the dynamic adjustment amount of the three slagging agents, SiO2, CaO and MgO. S3. Based on the dynamic adjustment amount calculated in step B, adjust the proportions of SiO2, CaO and MgO added to the smelting furnace in real time; S4. Periodically collect slag samples for composition and performance analysis. Compare the measured slag composition or performance data obtained from the analysis with the expected target value or predicted value of the control model described in step B. Based on the comparison deviation, make feedback corrections to the expert rule parameters in the control model to achieve closed-loop optimization.

2. The dynamic optimization method for double-sided copper blowing slag pattern according to claim 1, characterized in that, The expert rules for the regulation model described in step B include at least the following: When the real-time collected molten pool temperature is higher than the first threshold and the O2 content in the flue gas is higher than the second threshold, the first rule is triggered, and the output of the first rule is to increase the SiO2 addition ratio. When the Fe3O4 content in the slag collected in real time is higher than the third threshold, the second rule is triggered, and the output of the second rule is to increase the amount of CaO added. When the measured slag viscosity collected in real time is higher than the fourth threshold, the third rule is triggered, and the output of the third rule is to increase the proportion of MgO added.

3. The dynamic optimization method for double-sided copper blowing slag pattern according to claim 2, characterized in that: The first threshold is 1350℃, the second threshold is 80% volume concentration; in the first rule, the addition ratio of SiO2 is increased by 0.5% to 1.0%; the third threshold is 5% by mass; in the second rule, the addition ratio of CaO is increased by 0.3% to 0.5%; the fourth threshold is 0.3 Pa·s; in the third rule, the addition ratio of MgO is increased by 0.2% to 0.4%.

4. The dynamic optimization method for double-sided copper blowing slag pattern according to claim 1, characterized in that, In step C, the iron-silicon ratio in the final slag is controlled within the range of 1.8 to 2.2 by dynamically adjusting the addition ratio of SiO2, CaO and MgO.

5. The dynamic optimization method for double-sided copper blowing slag pattern according to claim 1, characterized in that, The time interval for periodically collecting slag samples in step D is 30 minutes.

6. The dynamic optimization method for double-sided copper blowing slag pattern according to claim 1, characterized in that, After dynamic adjustment in step C, the mass percentage range of each component in the slag is: SiO2 20% to 25%, CaO 2.5% to 4%, and MgO 1.5% to 2.5%.

7. The dynamic optimization method for double-sided copper blowing slag type according to any one of claims 1 to 6, characterized in that, The molten pool operating parameters collected in real time in step A also include the grade of matte.

8. The dynamic optimization method for double-sided copper blowing slag pattern according to claim 2 or 3, characterized in that, When the third rule is triggered, the control model also outputs instructions to increase the target set value of the molten pool temperature of the smelting furnace.

Images