Temperature and electromagnetic synergic control method for high-temperature electromagnetic extraction process of hazardous and precious metal

Abstract

The present application relates to the field of hazardous waste resource utilization and intelligent control technology, especially the temperature and electromagnetic synergistic control method of hazardous waste precious metal high-temperature electromagnetic extraction process, the technical scheme comprises: through the real-time acquisition of multi-source sensor array, the multi-dimensional data such as molten pool temperature distribution, electromagnetic parameters and molten state image, the electromagnetic-thermal-flow multi-physical field coupling mechanism model is constructed and online corrected, the digital twin model is formed and updated synchronously with the physical system; the intelligent prediction of process parameters is carried out by using deep learning method, and the double closed-loop synergistic self-adaptive control architecture of mechanism model correction closed loop and AI optimization closed loop is established, the temperature and electromagnetic frequency in high-temperature electromagnetic melting process are dynamically adjusted, the accurate separation of metal phase and slag phase and the stable vitrification of residual slag are realized; the extraction rate of precious metal and non-ferrous metal is effectively improved, the product quality uniformity of glass body is guaranteed, the energy consumption is reduced, and the intelligent control of hazardous waste high-temperature electromagnetic melting extraction process is suitable.

Description

Technical Field

[0001] This invention relates to the field of hazardous waste resource utilization and intelligent control technology, and in particular to a method for temperature and electromagnetic coordinated control in the high-temperature electromagnetic extraction process of precious metals from hazardous waste. Background Technology

[0002] Industries such as non-ferrous metal smelting, electroplating, and electronic waste dismantling generate large amounts of hazardous waste rich in precious and non-ferrous metals. Traditional landfill and solidification disposal methods not only pose a risk of heavy metal leakage and pollution but also waste valuable metal resources.

[0003] High-temperature melting technology can achieve the harmlessness and resource utilization of hazardous waste. Among existing technologies, plasma melting can convert inorganic materials into vitrified slag; microwave heating can utilize the difference in melting points of metals to achieve stepwise separation; electromagnetic induction heating has advantages such as fast heating speed, high thermal efficiency, and built-in stirring effect, and shows promising application prospects in the recycling of precious metals from electronic waste. The residue after high-temperature melting and vitrification can be exempted from general solid waste classification and used as building materials. GB / T 41015-2021 provides a standard basis for the quality judgment of vitrified products.

[0004] However, the existing technology still has shortcomings: (1) Electromagnetic melting lacks systematic special process design in the field of precious metal extraction from hazardous waste; (2) Traditional chemical methods for recovering precious metals result in serious emissions of waste and low recovery rates. Existing high-temperature melting processes mostly use plasma or fuel heating, which consumes a lot of energy and makes it difficult to achieve integrated synergy between the extraction of multiple valuable metals and the vitrification of residues; (3) There is a lack of systematic research on the compatibility of process parameters with different materials, and batch stability is difficult to guarantee.

[0005] In view of this, we propose a temperature and electromagnetic synergistic control method for the high-temperature electromagnetic extraction process of precious metals from hazardous waste to solve the existing problems. Summary of the Invention

[0006] The purpose of this invention is to provide a method for temperature and electromagnetic coordinated control in the high-temperature electromagnetic extraction process of precious metals from hazardous waste, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for coordinated temperature and electromagnetic control in a high-temperature electromagnetic extraction process of precious metals from hazardous waste, comprising: Step S1: Real-time acquisition and preprocessing of multi-source data: Through a multi-source sensor array deployed on a high-temperature electromagnetic melting equipment, process parameter data during the melting process are acquired in real time. The process parameter data includes molten pool temperature distribution data, electromagnetic frequency data, electromagnetic power data, material melting state image data, furnace atmosphere data, and crucible wall temperature data. The acquired multi-source data is cleaned, normalized, and time-aligned for preprocessing to construct a multi-source sensing dataset. Step S2, Construction of Electromagnetic-Thermal-Fluid Multiphysics Coupling Mechanism Model and Digital Twin Modeling: Based on the electromagnetic induction heating principle, heat conduction equation, and fluid dynamics equation in the high-temperature electromagnetic melting process, an electromagnetic-thermal-fluid multiphysics coupling mechanism model is constructed; the multi-source sensing dataset collected in real time in step S1 is used to correct the parameters and calibrate the boundary conditions of the coupling mechanism model, forming a digital twin model that is updated in real time with the physical system; the digital twin model is used to dynamically simulate and predict the temperature field distribution, electromagnetic field distribution, and melt flow state inside the molten pool; Step S3, Intelligent prediction of process parameters based on deep learning: The preprocessed multi-source sensing dataset in step S1 and the physical field simulation data output by the digital twin model in step S2 are used as training samples to construct and train a process parameter prediction model based on deep learning; the deep learning model takes the process parameter sensing data at the current moment and the control parameters at the previous moment as input, and takes the optimal temperature setpoint and optimal electromagnetic frequency setpoint at the future moment as output to generate the predicted value of the process parameters. Step S4, Dual-Loop Collaborative Adaptive Control: The dual loops include an activation mechanism model correction loop and a synchronous activation AI optimization loop. For the activation mechanism model correction loop, the real-time collected actual operating data of the physical system is compared with the output of the digital twin model to calculate the model deviation. Based on the model deviation feedback, the inherent parameters and boundary conditions in the electromagnetic-thermal-fluid coupling mechanism model are updated to ensure that the digital twin model continuously conforms to the actual working conditions. For the synchronous activation AI optimization loop, new production data and the corrected mechanism model output are used to incrementally train the deep learning-based process parameter prediction model and dynamically update the model parameters. Through the collaborative iteration of the dual-loop mechanism model correction loop and the AI ​​optimization loop, collaborative adaptive control of temperature and electromagnetic parameters is achieved during the high-temperature electromagnetic melting process. Step S5, Control Command Output and Execution: The temperature control command and electromagnetic frequency control command generated in step S4 are output to the actuator of the high-temperature electromagnetic melting equipment to adjust the electromagnetic heating power and electromagnetic frequency, so as to achieve coordinated control of the molten pool temperature and electromagnetic field until the electromagnetic melting extraction of hazardous waste precious metals and the vitrification treatment of residue are completed.

[0008] Furthermore, the multi-source sensor array includes: multiple thermocouples or infrared temperature sensors distributed along the axial and radial directions of the crucible for acquiring molten pool temperature distribution data; electromagnetic frequency sensors and electromagnetic power sensors for acquiring electromagnetic frequency data and electromagnetic power data; a high-speed industrial camera or infrared thermal imager for acquiring image data of the molten state of the material; a gas composition sensor for acquiring furnace atmosphere data; and a temperature sensor attached to the outer wall of the crucible for acquiring crucible wall temperature data.

[0009] Furthermore, the electromagnetic-thermal-fluid multiphysics coupling mechanism model in step S2 includes: an electromagnetic field model based on Maxwell's equations to describe the magnetic field distribution and eddy current effect during electromagnetic induction heating; a temperature field model based on Fourier's heat conduction equation to describe the heat transfer and temperature distribution inside the molten pool; and a flow field model based on Navier-Stokes' equations to describe the flow behavior of molten metal and slag. The electromagnetic field model, temperature field model, and flow field model achieve joint solution of the multiphysics field through coupling of physical property parameters and boundary conditions.

[0010] Furthermore, in step S3, the deep learning-based process parameter prediction model adopts a physical information neural network architecture. The mechanistic boundaries and physical constraints provided by the digital twin model are embedded into the neural network structure in the form of a loss function. The physical constraint loss function is introduced to suppress the prediction output that deviates from the physical laws, thereby forming a process parameter prediction value that conforms to the physical laws.

[0011] Furthermore, the deep learning-based process parameter prediction model adopts a hybrid neural network architecture that combines convolutional neural networks and bidirectional long short-term memory networks. The convolutional neural network is used to extract the spatial features of the molten state image data, while the bidirectional long short-term memory network is used to capture the temporal dynamic characteristics of the temperature sequence data and the electromagnetic parameter sequence data.

[0012] Furthermore, the dual-loop collaborative adaptive control in step S4 also includes: when the mechanism model correction closed loop detects that the deviation between the digital twin model output and the actual operating data exceeds a preset threshold, an alarm signal is triggered and the incremental training of the AI ​​optimization closed loop is suspended. After the mechanism model completes the correction and the deviation falls back to within the threshold, the operation of the AI ​​optimization closed loop is resumed.

[0013] Furthermore, the temperature control commands include the target temperature value of the molten pool, the set value of the heating rate, and the set value of the holding time; the electromagnetic frequency control commands include the set value of the electromagnetic frequency and the set value of the electromagnetic power; the actuator includes the frequency adjustment module and the power adjustment module of the high-frequency electromagnetic induction power supply.

[0014] Furthermore, it also includes a visualization and interaction step: the industrial-grade visualization and interaction platform displays in real time the temperature field distribution map of the molten pool, the electromagnetic field distribution map, the simulation status of the digital twin model, the predicted trend of process parameters, and the operating status of the dual closed-loop collaborative control, supporting operators to manually intervene and optimize the control parameters.

[0015] Furthermore, it is applicable to one or more of the following stages in the high-temperature electromagnetic melting extraction process of hazardous waste precious metals: material heating and melting stage, metal sedimentation and separation stage, and residue vitrification stage; in different process stages, the dual closed-loop collaborative adaptive control adopts temperature setpoints and electromagnetic frequency setpoints that match the material characteristics of that stage.

[0016] Furthermore, in step S1, the acquisition frequency of multi-source data is not less than 1Hz, the calculation update cycle of the digital twin model does not exceed 1 second, and the control cycle of the dual closed-loop collaborative adaptive control does not exceed 5 seconds, so as to achieve real-time dynamic control of the high-temperature electromagnetic melting process.

[0017] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention employs a high-temperature electromagnetic induction heating method, utilizing an electromagnetic field to directly generate heat within the molten pool, accompanied by electromagnetic stirring. This effectively breaks down the slag-metal interface barrier, allowing micron-sized precious metal particles dispersed in the slag phase to fully contact the molten metal. Actual measurements show that the extraction rate of non-ferrous metals such as copper and nickel can reach over 95%, while the enrichment and recovery rate of precious metals such as gold, silver, palladium, and platinum can reach over 98%.

[0018] 2. This invention fully vitrifies the slag phase after metal extraction at high temperature, forming a dense tetrahedral network structure that firmly solidifies heavy metal ions within it. The vitrified product has a vitrified mass fraction of over 99%, an acid dissolution rate of less than 1%, and a heavy metal leaching concentration far below the limit of GB / T 41015-2021 standard. It can be managed as general solid waste and sold as building materials.

[0019] 3. The electromagnetic induction heating of the present invention generates heat directly inside the material, with a thermal efficiency of 60% to 75%. It does not require preheating of the furnace and can be turned on and off immediately. Compared with traditional resistance heating or fuel heating methods, energy consumption is significantly reduced, and fossil fuels are not used, thus reducing flue gas emissions.

[0020] 4. This invention achieves precise coordinated regulation of temperature and electromagnetic frequency through a dual closed-loop collaborative adaptive control architecture that combines real-time data acquisition from multiple sources of sensors, digital twin modeling, and deep learning prediction. This effectively overcomes the product quality fluctuation problem caused by traditional processes relying on experience, and ensures the uniformity of the processing effect of different batches of materials. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the working process of the temperature and electromagnetic synergistic control method for the high-temperature electromagnetic extraction process of hazardous waste precious metals according to the present invention. Detailed Implementation

[0022] The technical solution of the present invention will be further described below with reference to the accompanying drawings and specific embodiments. Example 1

[0023] This embodiment provides a temperature and electromagnetic synergistic control method for the high-temperature electromagnetic extraction process of precious metals from hazardous waste, which is used to perform melt extraction and vitrification treatment of hazardous waste rich in precious metals (mainly a mixture of waste circuit board fragments and electroplating sludge) generated by an electronic waste dismantling company.

[0024] The raw materials used in this embodiment are a mixture of waste circuit board fragments and electroplating sludge generated by an electronic waste dismantling company. The mixture contains approximately 18% to 25% copper, 3% to 6% nickel, 80 to 150 g / t gold, 200 to 500 g / t silver, 20 to 50 g / t palladium, and 5 to 15 g / t platinum. The moisture content of the raw materials is approximately 35% to 50%.

[0025] The above raw materials were subjected to the following pretreatment in sequence: (1) Crushing: Jaw crusher is used to crush large materials to a particle size ≤5mm; (2) Screening: Vibrating screen is used to screen the crushed material to control the particle size within the range of 0.5 to 3 mm. The material under the screen enters the next process, and the material over the screen is returned to the crusher for re-crushing. (3) Drying: The screened material is sent to a rotary dryer and dried at 105 to 150°C until the moisture content is ≤3%.

[0026] Mix the pretreated material with flux and reducing agent in the following proportions until homogeneous: The flux addition amount is 8% to 15% of the total material mass. The flux used in this embodiment is a composite flux, comprising 20 parts SiO2, 4.5 parts H3BO3, and 7.0 parts CaF2 by mass. The flux's function is to lower the melting temperature of the material system and promote the glass formation of the slag phase. The reducing agent addition amount is 3% to 8% of the total material mass. The reducing agent used in this embodiment is coke powder (fixed carbon content ≥85%, particle size ≤1mm), used to reduce the metal oxides in the material to elemental metals, thereby improving the metal recovery rate.

[0027] The pretreated materials, flux, and reducing agent are put into a forced mixer and mixed at a speed of 30 to 60 r / min for 15 to 30 minutes at room temperature until the components are evenly distributed.

[0028] The uniformly mixed materials are fed into a high-temperature electromagnetic melting furnace. The high-temperature electromagnetic melting furnace used in this embodiment is a medium-frequency induction melting furnace with an adjustable operating frequency of 110kHz and a maximum operating temperature of 1800℃. The crucible is made of clay-graphite, which has excellent thermal conductivity and thermal shock resistance, making it suitable for melting precious metals and non-ferrous metals such as copper and nickel. The furnace atmosphere is controlled to be a reducing atmosphere (nitrogen is introduced as a protective gas at a flow rate of 25L / min, while CO is generated by the combustion of coke powder as a reducing agent to form a reducing atmosphere) to prevent metal oxidation.

[0029] After the materials are added, the high-temperature electromagnetic melting furnace is turned on for heating and melting. Throughout the melting process, adaptive regulation is performed using a temperature and electromagnetic coordinated control method.

[0030] like Figure 1 As shown, the temperature and electromagnetic coordinated control method for the high-temperature electromagnetic extraction process of precious metals from hazardous waste includes the following steps: S1: Real-time acquisition and preprocessing of multi-source data; S2: Construction of electromagnetic-thermal-fluid multiphysics coupling mechanism model and digital twin modeling; S3: Intelligent prediction of process parameters based on deep learning; S4: Dual closed-loop cooperative adaptive control; S5: Control command output and execution.

[0031] In step S1, process parameter data are acquired in real time using a multi-source sensor array deployed on a high-temperature electromagnetic melting furnace. The sensor array specifically includes: Nine S-type thermocouples (PtRh10-Pt) are distributed along the crucible axis (three layers from top to bottom) and radially (three rings in total: center, half radius, and wall) to collect molten pool temperature distribution data. The electromagnetic frequency sensor and the electromagnetic power sensor are respectively installed at the output end of the intermediate frequency power supply to collect electromagnetic frequency data and electromagnetic power data. A high-speed industrial camera (1920×1080 resolution, 30fps) is installed at the observation window on the top of the furnace to collect image data of the material in the molten state. Gas composition sensors (O2, CO, CO2 sensors) are installed at the furnace exhaust port to collect furnace atmosphere data; Three K-type thermocouples (located at the top, middle, and bottom of the crucible) are attached to the outer wall of the crucible to collect crucible wall temperature data.

[0032] The acquisition frequency of the above multi-source data is 2Hz. The acquired data is transmitted to the industrial control computer through the data acquisition card, and preprocessed by cleaning (removing outliers that are significantly beyond the range), normalization (mapping each sensor data to the [0,1] interval), and time alignment (interpolating and synchronizing based on the system clock) to construct a multi-source sensing dataset.

[0033] In step S2, based on the physical principles of the high-temperature electromagnetic melting process, a multi-physics coupling mechanism model of electromagnetic-thermal-fluid fields is constructed, specifically including: Electromagnetic field model: Based on Maxwell's equations, it describes the distribution of the alternating magnetic field generated by the induction coil in the crucible and material, as well as the resulting eddy current effect. The input parameters include electromagnetic frequency, electromagnetic power, coil geometric parameters, and electromagnetic properties of the material (conductivity, permeability). Temperature field model: Based on the Fourier heat conduction equation, it is used to describe the heat transfer and temperature distribution inside the molten pool. The input parameters include the thermal conductivity, specific heat capacity, density and latent heat of molten material. Flow field model: Based on the Navier-Stokes equations, it is used to describe the flow behavior of molten metal and slag under the action of gravity and electromagnetic stirring. The input parameters include the density, viscosity and interfacial tension of the metal phase and the slag phase.

[0034] The three models mentioned above achieve joint solution of multiple physics fields by coupling physical property parameters (temperature affects physical property parameters such as conductivity and viscosity) and coupling boundary conditions (joule heat generated by electromagnetic field serves as the heat source term of temperature field, and flow field affects convective heat transfer of temperature field).

[0035] In step S3, the multi-source sensing dataset acquired in real time in step S1 is used to correct parameters and calibrate boundary conditions for the above-mentioned coupling mechanism model: the model simulation output is compared with the actual sensor measurements, and the Kalman filter algorithm is used to correct the inherent parameters (such as physical property parameters) and boundary conditions (such as the heat dissipation coefficient of the crucible wall) in the model online, forming a digital twin model that is updated in real time with the physical system. This digital twin model can dynamically simulate and predict the temperature field distribution, electromagnetic field distribution, and melt flow state inside the molten pool, with a simulation calculation update cycle of 0.5 seconds.

[0036] Using the preprocessed multi-source sensing dataset from step S1 and the physical field simulation data (including temperature field distribution matrix, electromagnetic field distribution matrix, and melt flow velocity field matrix) output by the digital twin model from step S2 as training samples, a process parameter prediction model based on physical information neural network (PINN) is constructed.

[0037] This model employs a hybrid neural network architecture combining convolutional neural networks (CNNs) and bidirectional long short-term memory (LSTM) networks. The CNN portion (3 convolutional layers + 2 pooling layers + 1 fully connected layer) extracts spatial features from molten state image data (such as the degree of material melting and the location of the slag-gold interface). The LSTM portion (2 BiLSTM layers, each with 128 hidden units) captures the temporal dynamics of temperature and electromagnetic parameter sequence data. Simultaneously, the mechanistic boundaries and physical constraints (such as energy conservation and momentum conservation) provided by the digital twin model are embedded into the neural network structure as a loss function. This physical constraint loss function suppresses predictions that deviate from physical laws.

[0038] The deep learning model takes the current process parameter sensing data and the control parameters (temperature setpoint and electromagnetic frequency setpoint) from the previous moment as input, and outputs the optimal temperature setpoint and optimal electromagnetic frequency setpoint for the next 30 seconds to generate predicted process parameter values.

[0039] In step S4, for opening the closed loop of the mechanism model correction, the real-time collected actual operating data of the physical system is compared with the output of the digital twin model, and the model deviation is calculated (temperature deviation ≤ ±10℃ and electromagnetic field strength deviation ≤ ±5% are considered acceptable). Based on the feedback of the model deviation, the inherent parameters and boundary conditions in the electromagnetic-thermal-fluid coupling mechanism model are updated so that the digital twin model continuously conforms to the actual working conditions.

[0040] To simultaneously initiate the AI ​​optimization closed loop, new production data and the corrected mechanism model output are used to incrementally train the deep learning-based process parameter prediction model (incremental training is performed after each batch is processed, with the learning rate set to 0.001 and the batch size to 64), and the model parameters are dynamically updated.

[0041] By employing a dual-loop collaborative iteration of mechanistic model correction and AI optimization, adaptive control of temperature and electromagnetic parameters is achieved during high-temperature electromagnetic melting. When the mechanistic model correction loop detects a deviation between the digital twin model output and the actual operating data exceeding a preset threshold (temperature deviation > ±10℃ for more than 10 seconds, or electromagnetic field strength deviation > ±5% for more than 10 seconds), an alarm signal is triggered, and incremental training of the AI ​​optimization loop is paused. Once the mechanistic model has completed its correction and the deviation has fallen back below the threshold, the AI ​​optimization loop resumes operation.

[0042] In step S5, the temperature control command (including the target temperature value of the molten pool, the heating rate setting value, and the holding time setting value) and the electromagnetic frequency control command (including the electromagnetic frequency setting value and the electromagnetic power setting value) generated by the dual closed-loop collaborative control are output to the actuator of the high-temperature electromagnetic melting equipment (including the frequency adjustment module and the power adjustment module of the medium frequency induction power supply) to adjust the electromagnetic heating power and electromagnetic frequency, thereby realizing the coordinated control of the molten pool temperature and the electromagnetic field.

[0043] In this embodiment, the specific process parameters for the melt treatment are dynamically adjusted and stabilized within the following range by the aforementioned intelligent control system: Melting temperature: Gradually increase to 1450 to 1550℃; Electromagnetic frequency: 1 to 5 kHz (dynamically adjustable according to the state of the molten pool); Heating rate: 10 to 20 °C / min; Holding time: After reaching the target temperature, hold for 60 to 120 minutes to allow the valuable metal to fully melt and separate by sedimentation due to density differences (metal phase density is approximately 7 to 9 g / cm³). 3 It settles at the bottom, with a slag phase density of approximately 2.5 to 3.0 g / cm³. 3 (Floating on top).

[0044] Throughout the melting process, the temperature control accuracy of the molten pool reaches ±5℃.

[0045] After the melting and holding period is completed, the metallic phase and slag phase are collected separately using the following methods: (1) First, the bottom molten metal phase (copper, nickel and other non-ferrous metals and precious metals enriched) is discharged through the bottom outlet of the furnace and flows into the ingot casting machine to cast ingots, and copper-nickel alloy ingots are obtained (copper content is about 70% to 80%, nickel content is about 10% to 20%, and precious metals are enriched in it); (2) The upper slag phase is then discharged from the slag outlet on the furnace side and enters the vitrification process.

[0046] The separated slag phase is then held in a high-temperature electromagnetic melting furnace for 30 to 60 minutes at a temperature of 1450 to 1550°C to allow for complete vitrification. During this process, components such as SiO2, CaO, and Al2O3 in the slag phase form a dense [SiO4] tetrahedral network structure, which firmly fixes the residual heavy metal ions within the glass-forming mesh.

[0047] The high-temperature molten slag (approximately 1450 to 1550°C) after vitrification is discharged from the furnace and rapidly cooled by water quenching (water quenching temperature 20 to 30°C, the product temperature drops to <100°C after water quenching), yielding black glass particles. The water-quenched glass particles are then crushed and sieved, with the particle size controlled within the range of 0.5 to 5 mm, to obtain glass aggregate suitable for use as building materials.

[0048] For metal products, the extraction rates of copper and nickel in copper-nickel alloy ingots were 96.3% and 94.8%, respectively. For precious metals, the enrichment and recovery rates were 98.6% for gold, 97.2% for silver, 96.5% for palladium, and 95.8% for platinum.

[0049] For the vitreous products, the tests were conducted according to GB / T 41015-2021 "Technical Requirements for Vitrification Products of Solid Waste", and the results are as follows: the vitreous mass fraction was 98.6% (standard requirement ≥85%), determined by polarized light microscopy; the acid loss rate was 0.8% (standard requirement ≤3%), determined by hydrochloric acid solution immersion method; the leaching toxicity of heavy metals was tested according to "Identification Standard for Hazardous Waste - Leaching Toxicity Identification" (GB 5085.3—2007), and the leaching concentrations of lead <0.1mg / L, chromium <0.05mg / L, cadmium <0.01mg / L, arsenic <0.01mg / L, and zinc <0.5mg / L, which are far below the standard limits.

[0050] The above test results show that the vitreous product obtained in this embodiment meets the requirements of GB / T 41015-2021 standard and can be sold as building materials (such as concrete aggregate, road filler, etc.). Example 2

[0051] The difference between this embodiment and Embodiment 1 lies in the type of raw materials and some process parameters.

[0052] Regarding the raw materials: The raw materials used in this embodiment are a mixed dried material of nickel-containing electroplating sludge and copper-containing etching waste liquid generated from an electroplating industrial park. The copper content is approximately 10% to 15%, the nickel content is approximately 8% to 12%, the gold content is approximately 5 to 15 g / t, and the silver content is approximately 50 to 100 g / t. The moisture content of the raw materials is approximately 5% to 10% (pre-dried).

[0053] For pretreatment: Since the moisture content of the raw materials is already low, only crushing and screening are required to control the particle size of the material within the range of 0.5 to 2 mm. No drying process is required.

[0054] For ingredient mixing: the amount of flux added is 10% to 18% of the total mass of materials (because the SiO2 content in electroplating sludge is low, the amount of flux needs to be increased appropriately to adjust the slag system composition and make it easier to vitrify); the amount of reducing agent added is 5% to 10% of the total mass of materials (because the oxide content of nickel and copper is high, the amount of reducing agent needs to be increased).

[0055] Regarding the melting temperature: The melting temperature is dynamically adjusted by the intelligent control system and stabilized at 1500 to 1600℃ (because the nickel content in the electroplating sludge is high, and the melting point of nickel is higher than that of copper, a higher melting temperature is required).

[0056] The electromagnetic frequency is 2 to 8 kHz.

[0057] The heat preservation time is 90 to 150 minutes (because the composition of electroplating sludge is more complex, a longer heat preservation time is required to ensure that the metal settles and separates fully).

[0058] The vitrification temperature and time are 1500 to 1600℃, and the holding time is 40 to 60 minutes.

[0059] The product testing results are as follows: the extraction rate of copper is 95.1%, and the extraction rate of nickel is 96.7%. Regarding precious metals, the enrichment and recovery rate of gold is 97.8%, and the enrichment and recovery rate of silver is 96.5%. The vitreous product has a vitreous mass fraction of 97.2%, an acid loss rate of 1.1%, and the heavy metal leaching toxicity is also far below the limits of GB 5085.3—2007, meeting the requirements of GB / T 41015-2021. Example 3

[0060] The difference between this embodiment and Embodiment 1 is that the material being processed is a single type of hazardous waste precious metal material, and different temperature and electromagnetic synergistic control strategies are adopted at different process stages.

[0061] For the raw materials: The raw materials used in this embodiment are platinum group metal-containing waste catalysts produced by a precious metal refining enterprise, wherein the platinum content is approximately 500 to 1000 g / t, the palladium content is approximately 300 to 600 g / t, the rhodium content is approximately 50 to 150 g / t, and the support is mainly Al2O3 (content approximately 80% to 90%). The moisture content of the raw materials is <2%.

[0062] For pretreatment: the waste catalyst is crushed and ball-milled to control the particle size within the range of 0.1 to 0.5 mm, without the need for drying.

[0063] For ingredient mixing: the amount of flux added is 5% to 10% of the total mass of the materials (because Al2O3 carrier itself has good glass formation ability, the amount of flux can be appropriately reduced); the amount of reducing agent added is 2% to 5% of the total mass of the materials.

[0064] For phased intelligent collaborative control, this embodiment adopts temperature setpoints and electromagnetic frequency setpoints that match the material characteristics of each of the following three process stages, and the stages are automatically switched by the intelligent control system: (1) Material heating and melting stage (heating from room temperature to 1450℃): The target heating rate is 15 to 25℃ / min, and the electromagnetic frequency is set to 5 to 10kHz (higher frequency helps to heat up quickly). The key control point is to heat up quickly and shorten the melting time. (2) Metal sedimentation and separation stage (heat preservation at 1450 to 1550℃): The electromagnetic frequency is set to 1 to 3 kHz (lower frequency is beneficial to electromagnetic stirring and promotes slag-metal separation). The key control is to maintain stable temperature and promote full sedimentation of the metal phase. (3) Residue vitrification stage (1550 to 1600℃ heat preservation): The electromagnetic frequency is set to 3 to 5 kHz. The key control is to fully vitrify the slag phase and ensure that the glass mass fraction meets the standard.

[0065] The product testing results are as follows: the enrichment recovery rate of platinum is 98.2%, the enrichment recovery rate of palladium is 97.5%, and the enrichment recovery rate of rhodium is 96.1%. The vitreous product has a vitreous mass fraction of 99.1% and an acid dissolution rate of 0.6%. All indicators are better than the requirements of GB / T 41015-2021 standard.

[0066] During the process operation of the above embodiments, the following information is displayed in real time through an industrial-grade visualization interaction platform (an HMI human-machine interface developed based on LabVIEW): molten pool temperature field distribution map (displaying the axial and radial temperature distribution of the crucible in the form of a pseudo-color cloud map); electromagnetic field distribution map (displaying the magnetic field strength distribution generated by the induction coil); digital twin model simulation status (displaying the comparison curve between the model simulation output and the actual sensor data); process parameter prediction trends (displaying the future predicted values ​​and confidence intervals of temperature and electromagnetic frequency by the deep learning model); and dual-loop collaborative control operation status (displaying the current operation status, deviation values, and correction records of the mechanism model correction loop and the AI ​​optimization loop).

[0067] Operators can manually intervene and optimize control parameters via the touch screen (such as manually modifying temperature setpoints, switching control modes, etc.), and the system records all operation logs for traceability.

[0068] The above specific embodiments are merely several preferred embodiments of the present invention. Based on the technical solutions of the present invention and the relevant teachings of the above embodiments, those skilled in the art can make various alternative improvements and combinations to the above specific embodiments.

Claims

1. A method for coordinated temperature and electromagnetic control in a high-temperature electromagnetic extraction process of precious metals from hazardous waste, characterized in that, include: Step S1: Real-time acquisition and preprocessing of multi-source data: Through a multi-source sensor array deployed on a high-temperature electromagnetic melting equipment, process parameter data during the melting process are acquired in real time. The process parameter data includes molten pool temperature distribution data, electromagnetic frequency data, electromagnetic power data, material melting state image data, furnace atmosphere data, and crucible wall temperature data. The acquired multi-source data is cleaned, normalized, and time-aligned for preprocessing to construct a multi-source sensing dataset. Step S2, Construction of Electromagnetic-Thermal-Fluid Multiphysics Coupling Mechanism Model and Digital Twin Modeling: Based on the electromagnetic induction heating principle, heat conduction equation, and fluid dynamics equation in the high-temperature electromagnetic melting process, an electromagnetic-thermal-fluid multiphysics coupling mechanism model is constructed; the multi-source sensing dataset collected in real time in step S1 is used to correct the parameters and calibrate the boundary conditions of the coupling mechanism model, forming a digital twin model that is updated in real time with the physical system; the digital twin model is used to dynamically simulate and predict the temperature field distribution, electromagnetic field distribution, and melt flow state inside the molten pool; Step S3, Intelligent prediction of process parameters based on deep learning: The preprocessed multi-source sensing dataset in step S1 and the physical field simulation data output by the digital twin model in step S2 are used as training samples to construct and train a process parameter prediction model based on deep learning; the deep learning model takes the process parameter sensing data at the current moment and the control parameters at the previous moment as input, and takes the optimal temperature setpoint and optimal electromagnetic frequency setpoint at the future moment as output to generate the predicted value of the process parameters. Step S4, Dual-Loop Collaborative Adaptive Control: The dual loops include an activation mechanism model correction loop and a synchronous activation AI optimization loop. For the activation mechanism model correction loop, the real-time collected actual operating data of the physical system is compared with the output of the digital twin model to calculate the model deviation. Based on the model deviation feedback, the inherent parameters and boundary conditions in the electromagnetic-thermal-fluid coupling mechanism model are updated to ensure that the digital twin model continuously conforms to the actual working conditions. For the synchronous activation AI optimization loop, new production data and the corrected mechanism model output are used to incrementally train the deep learning-based process parameter prediction model and dynamically update the model parameters. Through the collaborative iteration of the dual-loop mechanism model correction loop and the AI ​​optimization loop, collaborative adaptive control of temperature and electromagnetic parameters is achieved during the high-temperature electromagnetic melting process. Step S5, Control Command Output and Execution: The temperature control command and electromagnetic frequency control command generated in step S4 are output to the actuator of the high-temperature electromagnetic melting equipment to adjust the electromagnetic heating power and electromagnetic frequency, so as to achieve coordinated control of the molten pool temperature and electromagnetic field until the electromagnetic melting extraction of hazardous waste precious metals and the vitrification treatment of residue are completed.

2. The temperature and electromagnetic coordinated control method for the high-temperature electromagnetic extraction process of precious metals from hazardous waste according to claim 1, characterized in that, The multi-source sensor array includes: multiple thermocouples or infrared temperature sensors distributed along the axial and radial directions of the crucible for acquiring molten pool temperature distribution data; electromagnetic frequency sensors and electromagnetic power sensors for acquiring electromagnetic frequency data and electromagnetic power data; a high-speed industrial camera or infrared thermal imager for acquiring image data of the molten state of the material; a gas composition sensor for acquiring furnace atmosphere data; and a temperature sensor attached to the outer wall of the crucible for acquiring crucible wall temperature data.

3. The temperature and electromagnetic coordinated control method for the high-temperature electromagnetic extraction process of precious metals from hazardous waste according to claim 1, characterized in that, The electromagnetic-thermal-fluid multiphysics coupling mechanism model in step S2 includes: an electromagnetic field model based on Maxwell's equations to describe the magnetic field distribution and eddy current effect during electromagnetic induction heating; a temperature field model based on Fourier's heat conduction equation to describe the heat transfer and temperature distribution inside the molten pool; and a flow field model based on Navier-Stokes' equations to describe the flow behavior of molten metal and slag. The electromagnetic field model, temperature field model, and flow field model achieve joint solution of the multiphysics field through coupling of physical property parameters and boundary conditions.

4. The temperature and electromagnetic coordinated control method for the high-temperature electromagnetic extraction process of precious metals from hazardous waste according to claim 1, characterized in that: In step S3, the deep learning-based process parameter prediction model adopts a physical information neural network architecture. The mechanism boundary and physical constraints provided by the digital twin model are embedded into the neural network structure in the form of a loss function. The physical constraint loss function is introduced to suppress the prediction output that deviates from the physical law, thus forming the process parameter prediction value that conforms to the physical law.

5. The temperature and electromagnetic coordinated control method for the high-temperature electromagnetic extraction process of precious metals from hazardous waste according to claim 1, characterized in that: The deep learning-based process parameter prediction model adopts a hybrid neural network architecture that combines convolutional neural networks and bidirectional long short-term memory networks. The convolutional neural network is used to extract the spatial features of the molten state image data, while the bidirectional long short-term memory network is used to capture the temporal dynamic characteristics of the temperature sequence data and the electromagnetic parameter sequence data.

6. The temperature and electromagnetic coordinated control method for the high-temperature electromagnetic extraction process of precious metals from hazardous waste according to claim 1, characterized in that, The dual-loop collaborative adaptive control in step S4 also includes: when the mechanism model correction closed loop detects that the deviation between the digital twin model output and the actual operating data exceeds a preset threshold, an alarm signal is triggered and the incremental training of the AI ​​optimization closed loop is suspended. After the mechanism model completes the correction and the deviation falls back to within the threshold, the operation of the AI ​​optimization closed loop is resumed.

7. The temperature and electromagnetic coordinated control method for the high-temperature electromagnetic extraction process of precious metals from hazardous waste according to claim 1, characterized in that: Temperature control commands include target temperature of the molten pool, heating rate setting, and holding time setting; electromagnetic frequency control commands include electromagnetic frequency setting and electromagnetic power setting; the actuator includes a frequency adjustment module and a power adjustment module for the high-frequency electromagnetic induction power supply.

8. The method for coordinated temperature and electromagnetic control in the high-temperature electromagnetic extraction process of precious metals from hazardous waste according to claim 1, characterized in that, It also includes a visualization and interactive step: the industrial-grade visualization and interactive platform displays in real time the temperature field distribution map of the molten pool, the electromagnetic field distribution map, the simulation status of the digital twin model, the predicted trend of process parameters, and the operating status of the dual closed-loop collaborative control, supporting operators to manually intervene and optimize the control parameters.

9. The temperature and electromagnetic coordinated control method for the high-temperature electromagnetic extraction process of precious metals from hazardous waste according to claim 1, characterized in that, It is applicable to one or more of the following stages in the high-temperature electromagnetic melting extraction process of precious metals from hazardous waste: material heating and melting stage, metal sedimentation and separation stage, and residue vitrification stage; in different process stages, the dual closed-loop collaborative adaptive control adopts temperature setpoints and electromagnetic frequency setpoints that match the material characteristics of that stage.

10. The temperature and electromagnetic coordinated control method for the high-temperature electromagnetic extraction process of precious metals from hazardous waste according to claim 1, characterized in that: In step S1, the acquisition frequency of multi-source data is no less than 1Hz, the calculation and update cycle of the digital twin model is no more than 1 second, and the control cycle of the dual closed-loop collaborative adaptive control is no more than 5 seconds, so as to achieve real-time dynamic control of the high-temperature electromagnetic melting process.

Images