Coal gangue magnetization fluidized roasting device

Abstract

The invention relates to the technical field of coal gangue magnetic separation, in particular to a coal gangue magnetization fluidized roasting device which comprises a raw material preprocessing module, a fluidized magnetization roasting module, a magnetic separation module, a data center module and a mass feed module. The raw material preprocessing module realizes gradient crushing and dynamic screening of coal gangue through a multi-stage crushing unit and a closed cycle pneumatic classification device, and the online particle size analysis device monitors particle size data in real time and transmits the particle size data to the data center module; the fluidization and magnetization roasting module adopts a double-cavity reaction tower structure, and is combined with an infrared thermal imager array and a PI (Proportional Integral) controller to regulate and control the mixing proportion of fluidization gas and inhibit the oxidation of ferric oxide; through multi-module data driving optimization, the technical problems of insufficient multi-variable coupling regulation, low separation precision and high energy consumption are solved, and low-carbon efficient resourceful treatment of the coal gangue is realized.

Description

Technical Field

[0001] The invention relates to the technical field of magnetic separation of coal gangue, in particular to a magnetic fluidized roasting device for coal gangue. Background Art

[0002] Gangue, as a solid waste generated during coal processing, faces technical challenges in resource processing, such as high energy consumption, unstable product quality, and low iron oxide removal efficiency. Traditional roasting processes mostly use a single parameter for control, which makes it difficult to balance the multivariable coupling relationship between temperature, gas flow rate, and magnetization intensity, resulting in limited reaction efficiency, high energy consumption, and the active structure of the iron, aluminum, and silicon components in the product is easily destroyed by high-temperature calcination. In addition, the existing magnetic separation technology has insufficient separation accuracy for submicron iron impurities, which restricts the preparation of high-purity aluminum silicon materials. Therefore, there is an urgent need for a gangue processing device that can achieve coordinated optimization of multiple process parameters, improve magnetic separation accuracy, and reduce energy consumption to meet the needs of low-carbon and efficient resource utilization. Summary of the Invention

[0003] In response to the shortcomings of the existing technology, the present invention provides a magnetized fluidized roasting device for coal gangue, which solves the technical problems of unstable product quality and low iron oxide removal efficiency caused by insufficient control of multivariable coupling relationships, low separation accuracy of submicron iron impurities and high energy consumption in the process of coal gangue resource processing.

[0004] In order to solve the above technical problems, the specific technical solutions of the present invention are as follows:

[0005] The present invention provides a gangue magnetized fluidized roasting device, comprising: a multi-stage crushing unit, an online particle size analyzer, a component analysis probe, and a control device, wherein the control device comprises:

[0006] The raw material pre-processing module crushes and grinds the gangue through a multi-stage crushing unit, collects material particle size data in real time using an online particle size analyzer, and sends the particle size data to the data center module;

[0007] The fluidized magnetization roasting module receives the material output by the raw material pretreatment module, adjusts the fluidization gas flow rate and the power of the electric heating element according to the roasting parameter instructions generated by the data hub module, and outputs the product to the magnetic separation module after completing the magnetization roasting;

[0008] A magnetic separation module receives the magnetized roasted product, dynamically adjusts the magnetic pole spacing and the electromagnetic coil current according to the magnetic field strength control instruction generated by the data hub module, separates the iron phase impurities, and outputs the iron-poor material to the mass feed module;

[0009] The data center module integrates the digital twin platform and the optimization engine, receives the real-time particle size data of the raw material pretreatment module, the temperature field monitoring data of the fluidized magnetization roasting module, and the hysteresis loop data of the magnetic separation module, and generates particle size control instructions, roasting temperature correction instructions, and magnetic field strength threshold instructions through a multi-objective nonlinear programming algorithm, and distributes them to the corresponding modules;

[0010] A mass feed module detects the phase composition data of the iron-depleted material by combining a component analysis probe, extracts characteristic vectors using a principal component analysis algorithm, matches process parameters in a historical database to generate gas ratio correction instructions, and feeds the correction instructions back to the data hub module;

[0011] The data hub module updates the parameter weights of the multi-objective nonlinear programming algorithm according to the gas ratio correction instruction of the mass feed module, triggering the fluidized magnetization roasting module to adjust the mixing ratio of hydrogen and carbon monoxide to form a closed-loop control.

[0012] Furthermore, in the gangue magnetized fluidized roasting device of the present invention, the raw material pretreatment module includes:

[0013] The multi-stage crushing unit module controls the multi-stage crushing unit to perform impact crushing and high-pressure roller grinding to process the coal gangue in stages, and outputs the crushed materials to the closed-circuit pneumatic classification device;

[0014] A closed-circuit pneumatic classifier dynamically screens the crushed material through a cyclone separator and an airflow screener, and the substandard particles are returned to the multi-stage crushing unit;

[0015] The particle size analysis device collects the output material particle size data of the closed-circuit pneumatic classification device in real time, inputs the particle size data into the particle size prediction model of the data center module, generates a crusher roller gap adjustment instruction and sends it to the multi-stage crushing group.

[0016] Furthermore, the gangue magnetized fluidized roasting device of the present invention, the fluidized magnetized roasting module comprises:

[0017] The double-chamber reaction tower has a conical fluidized bed preheating zone in the front section and a U-shaped cooling chamber in the rear section. The preheating zone receives the qualified materials from the raw material pretreatment module and heats them to the set temperature.

[0018] An infrared thermal imager array is embedded in the side wall of the reaction tower to collect three-dimensional temperature field distribution data in real time and send it to the data center module;

[0019] The gas distribution plate receives the roasting parameter instructions from the data center module, triggers the PID controller to adjust the mixing ratio of hydrogen and carbon monoxide according to the temperature field distribution data, and forms a gradient cooling zone in the U-shaped cooling cavity to inhibit the oxidation of ferric oxide.

[0020] Furthermore, in the gangue magnetized fluidized roasting device of the present invention, the magnetic separation module comprises:

[0021] The permanent magnetic drum roughing section receives the roasting product output by the fluidized magnetization roasting module and performs primary magnetic separation to separate large iron-phase impurities;

[0022] The electromagnetic high-gradient selection section is equipped with an adjustable-pitch magnetic pole array, which receives the magnetic field strength threshold instruction from the data hub module and dynamically adjusts the magnetic pole spacing and electromagnetic coil current;

[0023] The hysteresis detection device is integrated at the outlet of the electromagnetic high-gradient selection section, and monitors the saturation of magnetic particles in real time to generate hysteresis loop data, which is sent to the data center module as a basis for correcting the magnetic field strength threshold instruction.

[0024] Furthermore, in the gangue magnetized fluidized roasting device of the present invention, the data hub module includes:

[0025] The digital twin platform integrates a discrete element material motion model and a thermal field distribution simulation model, receives real-time particle size data from the raw material pretreatment module, temperature field monitoring data from the fluidized magnetization roasting module, and hysteresis loop data from the magnetic separation module, and generates material motion trajectory and thermal field distribution prediction results;

[0026] An optimization engine, based on the prediction results of the digital twin platform, calculates particle size control instructions, roasting temperature correction instructions, and magnetic field strength threshold instructions through a multi-objective nonlinear programming algorithm;

[0027] The risk prediction unit dynamically corrects the fluidizing gas flow rate instruction output by the optimization engine according to the material residence time distribution data of the fluidized magnetization roasting module.

[0028] Furthermore, in the gangue magnetized fluidized roasting device of the present invention, the mass feed module includes:

[0029] Combined with a component analysis probe, continuous X-ray diffraction and fluorescence spectroscopy detection is performed on the iron-depleted material output by the magnetic separation module to obtain phase composition data;

[0030] A multispectral imaging system, which quantifies the whiteness value of the iron-depleted material through diffuse reflectance spectroscopy analysis to generate whiteness detection data;

[0031] The process adjustment engine inputs the phase composition data and the whiteness detection data into the principal component analysis algorithm, extracts the feature vector and matches it with the historical database of the data center module, generates a gas ratio correction instruction and feeds it back to the data center module.

[0032] Furthermore, the gangue magnetized fluidized roasting device of the present invention further comprises:

[0033] The phase composition data and whiteness detection data of the mass feed module trigger the parameter weight update of the data hub module;

[0034] The optimization engine reallocates the particle size control priority of the raw material pretreatment module, the preheating zone temperature gradient threshold of the fluidized magnetization roasting module, and the magnetic field strength threshold of the magnetic separation module through a multi-objective nonlinear programming algorithm according to the updated weights;

[0035] The fluidized magnetization roasting module receives the redistributed preheating zone temperature gradient threshold, adjusts the power of the electric heating element, and simultaneously adjusts the mixing ratio of hydrogen and carbon monoxide in the cooling zone according to the magnetic field strength threshold.

[0036] Beneficial effects of the present invention:

[0037] The present invention effectively solves the technical problems of insufficient multi-variable coupling control, low separation accuracy of submicron iron impurities and high energy consumption in the resource processing of coal gangue through modular collaborative control and data-driven optimization mechanism. The data center module integrates the digital twin platform and the multi-objective nonlinear programming algorithm, and integrates the particle size data of the raw material pretreatment module, the temperature field distribution data of the fluidized magnetization roasting module and the hysteresis loop data of the magnetic separation module in real time. Through multi-physical field coupling simulation and Pareto optimal solution calculation, it dynamically generates particle size control instructions, roasting temperature correction instructions and magnetic field strength threshold instructions to achieve multi-dimensional collaborative optimization of temperature, gas flow rate and magnetic field strength. The magnetic separation module adopts permanent magnet-electromagnetic composite sorting technology, combined with a hysteresis detection device and an adjustable spacing magnetic pole array, and significantly improves the capture efficiency of submicron iron impurities through magnetic domain state analysis and dynamic adjustment of magnetic field gradient. The mass feed module generates gas ratio correction instructions through feature extraction of phase composition and whiteness data and historical parameter matching, drives the optimization engine to update weight parameters, and closes the loop to adjust the electric heating power and gas mixing ratio, thereby maintaining the active structural stability of aluminum silicon materials while reducing energy consumption, and realizing low-carbon and efficient resource processing of coal gangue. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solution of the present invention, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, for ordinary technicians in this field, other drawings can be obtained based on the drawings without paying any creative labor.

[0039] Figure 1 A system architecture diagram of a gangue magnetized fluidized roasting device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the specific embodiments of the present invention and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. The technical solutions provided by each embodiment of the present invention are described in detail below in conjunction with the drawings. In order to better understand the purpose of the present invention, the present invention is further described in detail below.

[0041] See also Figure 1 The present invention provides a coal gangue magnetized fluidized roasting device, comprising: a multi-stage crushing unit, an online particle size analyzer, a component analysis probe, and a control device, wherein the control device establishes a communication connection with the multi-stage crushing unit, the online particle size analyzer, and the component analysis probe, and the control device comprises:

[0042] The raw material pre-processing module crushes and grinds the gangue through a multi-stage crushing unit, collects material particle size data in real time using an online particle size analyzer, and sends the particle size data to the data center module;

[0043] The raw material pretreatment module crushes and grinds the gangue through a multi-stage crushing unit. An impact crusher performs primary crushing, using high-speed impact force to break down large lumps into coarse particles. The coarsely crushed material enters a high-pressure roller mill, where roller pressure applied by a hydraulic system provides fine crushing, producing a crushed product with a uniform particle size distribution. Wear-resistant liners are installed within the crushing unit to prevent contamination from contact between metal components and the material, maintaining material purity for subsequent magnetic separation. The crushed material is conveyed through a closed-loop pneumatic classifier. Substandard particles are returned to the multi-stage crushing unit through a return pipe for reprocessing, forming an iteratively optimized crushing and classification process.

[0044] The closed-circuit pneumatic classifier dynamically screens the crushed material using a cyclone separator and an airflow screener. The cyclone separator uses centrifugal force to separate particles of varying sizes based on their inertia, initially selecting material within the preset size range. The airflow screener then performs a secondary screening of the separated material to further remove particles exceeding the specified size. Particles that fail the screener are pneumatically conveyed back to the multi-stage crushing unit, forming a closed-circuit system until the particle size meets the fluidized bed roasting process requirements. The screened, qualified material then enters the subsequent processing module through a sealed pipeline, preventing contamination from the external environment.

[0045] The particle size analyzer integrates a laser diffraction sensor and an optical imaging unit to collect real-time particle size distribution data from the closed-loop pneumatic classifier. The laser diffraction sensor calculates particle size by analyzing scattered light intensity from multiple angles, while the optical imaging unit captures the material's morphological characteristics. This collected particle size data is transmitted to the data hub module via an industrial communication protocol and fed into a particle size prediction model for real-time analysis. The data hub module generates roller gap adjustment instructions based on the particle size deviation output by the prediction model, dynamically adjusting the roller gap and crushing pressure of the high-pressure roller grinding roll.

[0046] After receiving the adjustment instructions from the data center module, the multi-stage crushing unit drives the hydraulic system through the servo motor to change the roller gap and synchronously adjust the rotor speed and feed rate of the impact crusher. The adaptive adjustment of the crushing parameters and the dynamic screening of the closed-circuit pneumatic classification device form a coordinated control to stabilize the material particle size within the threshold range of the fluidized bed roasting process. The particle size analysis device continuously monitors the optimized crushed material data, forming a closed-loop control chain of crushing-classification-feedback, providing raw materials with uniform particle size for the subsequent magnetization roasting module. The pre-treated material enters the fluidized bed magnetization roasting module through the pneumatic conveying system to complete the subsequent heat treatment and magnetization separation process.

[0047] The fluidized magnetization roasting module receives the material output by the raw material pretreatment module, adjusts the fluidization gas flow rate and the power of the electric heating element according to the roasting parameter instructions generated by the data hub module, and outputs the product to the magnetic separation module after completing the magnetization roasting;

[0048] After receiving the qualified materials output by the raw material pretreatment module, the fluidized bed magnetization roasting module implements heat treatment and atmosphere control through a dual-chamber reaction tower structure. The front section of the dual-chamber reaction tower is a conical fluidized bed preheating zone, which is equipped with a porous gas distribution structure. Fluidizing gas is evenly injected from the bottom to form a material suspension state. The rear section is a U-shaped cooling chamber, which suppresses the ferric oxide oxidation reaction through an annular jet hole array and a gradient cooling zone. The electric heating elements distributed circumferentially in the preheating zone adjust the power output according to the roasting temperature correction instructions issued by the data center module, raising the material temperature to the set temperature threshold.

[0049] An array of infrared thermal imagers, evenly spaced, is embedded in the reactor's sidewall at the interface between the preheating zone and the cooling chamber. This data is collected in real time, capturing three-dimensional surface temperature data. This data is transmitted via optical fiber to the data hub module, which drives the thermal field simulation model on the digital twin platform to identify areas of localized overheating or abnormal temperature gradients. The simulation results are compared with a preset temperature curve to generate temperature compensation instructions, which are then sent to the controller of the electric heating element in the preheating zone, establishing a dynamic temperature control mechanism.

[0050] The gas distribution plate is integrated at the bottom of the U-shaped cooling chamber. After receiving the roasting parameter instructions from the data center module, it triggers the PID controller to adjust the mixing ratio of hydrogen and carbon monoxide. Hydrogen is injected at high speed through the central nozzle to form a reducing atmosphere, and carbon monoxide diffuses through the outer annular channel to form a concentration gradient distribution from the inside to the outside in the cooling chamber. The PID controller dynamically corrects the gas mixing ratio based on the real-time temperature field data of the infrared thermal imager array to match the cooling rate with the fluidizing gas flow rate, thereby avoiding microstructural defects in the material due to rapid cooling. A gas composition sensor is configured at the outlet of the cooling chamber to monitor the residual oxygen concentration and feed it back to the data center module to form a closed-loop adjustment link for the gas ratio.

[0051] The fluidizing gas flow rate and the power of the electric heating element are linked and controlled through the optimization engine of the data hub module. The temperature gradient threshold of the preheating zone and the gas mixing ratio instruction of the cooling chamber are updated synchronously to maintain the thermodynamic equilibrium of the magnetic roasting process. The processed materials enter the magnetic separation module through the pneumatic conveying system to complete the separation process of iron phase impurities. The data hub module receives the hysteresis loop data of the magnetic separation module in real time, verifies the effect of the magnetic field through the magnetic circuit simulation model of the digital twin platform, and dynamically corrects the temperature gradient threshold and gas mixing ratio coefficient in the roasting parameter instruction to form a cross-module collaborative optimization mechanism.

[0052] A magnetic separation module receives the magnetized roasted product, dynamically adjusts the magnetic pole spacing and the electromagnetic coil current according to the magnetic field strength control instruction generated by the data hub module, separates the iron phase impurities, and outputs the iron-poor material to the mass feed module;

[0053] The magnetic separation module receives the calcined product from the fluidized magnetization calcination module and then performs primary magnetic separation in the permanent magnetic drum roughing section. The drum, covered with an array of high-remanence permanent magnets, separates large iron-phase impurities through magnetic field adsorption during rotation. Non-magnetic material is evenly distributed via a vibrating feeder onto a conveyor belt before entering the electromagnetic high-gradient separation section. The separated iron-phase impurities are collected by a scraper device and sent to a waste bin. The remaining material is pneumatically conveyed to subsequent separation processes.

[0054] The electromagnetic high-gradient separation section is equipped with an adjustable-pitch magnetic pole array and electromagnetic coil assembly. Upon receiving magnetic field strength threshold commands from the data hub module, a servo motor drives a lead screw mechanism to adjust the magnetic pole spacing. Reducing the magnetic pole array spacing enhances the local magnetic field gradient, improving the capture efficiency of submicron iron impurities. The electromagnetic coil current is dynamically adjusted using PWM modulation technology to match the magnetic field strength to the current iron content of the material. The data hub module generates a current fluctuation compensation coefficient based on a model predicting the residual iron content in the magnetic separation product to optimize the magnetic field gradient distribution.

[0055] The hysteresis detection device, integrated at the exit of the electromagnetic high-gradient concentrator section, consists of a fluxgate sensor and a data acquisition unit. The sensor monitors the remanent magnetization and coercivity of the magnetic particles in real time, generating a characteristic hysteresis loop spectrum. This spectrum is Fourier transformed to extract harmonic components, which are then uploaded to the data hub module. This is then input into a magnetic domain analysis model to calculate the magnetic field intensity threshold deviation. Based on this deviation, the data hub module generates pole spacing adjustment instructions and current correction values, which are then sent to the electromagnetic coil controller and servo drive mechanism, forming a dynamic closed-loop control loop for magnetic field parameters.

[0056] After the selection, the iron-depleted material enters the mass feed module via a pneumatic conveying pipeline. Simultaneously, hysteresis loop data and magnetic field adjustment parameters are transmitted back to the digital twin platform of the data hub module. The digital twin platform uses a magnetic circuit simulation model to verify the correlation between the magnetic field gradient distribution and the material's magnetization effect. The platform then optimizes the magnetic pole array layout strategy based on historical data on the harmonic components of the hysteresis loop. The optimized magnetic field intensity threshold command is then simultaneously updated to the electromagnetic high-gradient selection section, improving the separation accuracy of iron-phase impurities.

[0057] The coordinated control of the permanent magnetic drum roughing section and the electromagnetic high-gradient cleaning section achieves process parameter linkage through a data hub module. Separation efficiency data from the roughing section is fed back to the optimization engine via a current sensor, dynamically adjusting the magnetic field action time and pole-gap range in the cleaning section. The magnetic separation module's magnetic field gradient, pole-gap, and current parameters form a multi-dimensional collaborative optimization mechanism, ensuring the separation accuracy of submicron iron impurities meets the requirements for the preparation of high-purity aluminum-silicon materials. After composition testing, the iron-depleted material enters the resource utilization stage, completing the closed-loop control of the entire gangue processing process.

[0058] The data center module integrates the digital twin platform and the optimization engine, receives the real-time particle size data of the raw material pretreatment module, the temperature field monitoring data of the fluidized magnetization roasting module, and the hysteresis loop data of the magnetic separation module, and generates particle size control instructions, roasting temperature correction instructions, and magnetic field strength threshold instructions through a multi-objective nonlinear programming algorithm, and distributes them to the corresponding modules;

[0059] The data center module receives in real time the material particle size data transmitted by the raw material pretreatment module, the temperature field monitoring data uploaded by the fluidized magnetization roasting module, and the hysteresis loop data fed back by the magnetic separation module through the industrial communication protocol. After the data is normalized and time-stamp aligned by the preprocessing unit, it is input into the digital twin platform to build a multi-physics field coupling simulation environment. The discrete element material motion model integrated in the digital twin platform analyzes the particle size data and simulates the flow trajectory and classification efficiency of the crushed material; the thermal field distribution simulation model predicts the heat conduction state and temperature gradient change trend of the preheating zone and the cooling chamber based on the three-dimensional temperature field data of the infrared thermal imager array; the hysteresis loop data is mapped to the magnetic domain evolution analysis model after Fourier transformation to analyze the correlation between the magnetization characteristics of iron phase impurities and the separation efficiency.

[0060] The optimization engine uses a multi-objective nonlinear programming algorithm, using the simulation prediction results of the digital twin platform as constraints. The algorithm optimizes the crusher roller gap adjustment, the fluidized bed electric heating power compensation value, and the electromagnetic coil current threshold, with minimizing overall energy consumption and maximizing iron phase separation efficiency as optimization goals. Particle size control instructions are implemented by dynamically adjusting the roller pressure of the high-pressure roller mill and the airflow velocity of the closed-loop pneumatic classifier. The roasting temperature correction instruction drives the PID controller of the preheating zone electric heating element to adjust the power output and optimizes the thermodynamic equilibrium in combination with the hydrogen and carbon monoxide mixing ratio of the cooling chamber gas distribution plate. The magnetic field strength threshold instruction triggers the electromagnetic high-gradient selection section's magnetic pole spacing servo mechanism to work in conjunction with the electromagnetic coil PWM controller, improving the capture accuracy of submicron iron impurities.

[0061] The risk prediction unit collects material residence time distribution data from the fluidized bed magnetization roasting module in real time and constructs a residence time probability density function using a kernel density estimation algorithm. When the residence time of a material of a specific particle size deviates from the preset distribution range, a fluidizing gas flow rate correction mechanism is triggered. The correction coefficient is dynamically generated based on a correlation analysis between the residence time deviation and the fluidized bed pressure differential data. This correction coefficient is then transmitted to the proportional control valve on the gas distribution board via an edge computing node, simultaneously updating the thermal field simulation boundary conditions of the digital twin platform.

[0062] Generated particle size control instructions, roasting temperature correction instructions, and magnetic field strength threshold instructions are distributed via Industrial Ethernet to the actuators of the corresponding modules. After receiving these instructions, the raw material preprocessing module uses servo motors to drive the roller gap and rotor speed of the multi-stage crushing unit. The fluidized bed magnetization roasting module adjusts the power of the electric heating element and the gas mixing ratio. The magnetic separation module responds to the magnetic field strength threshold instructions and dynamically optimizes the magnetic pole spacing and electromagnetic coil current. The execution status data of each module is transmitted back to the data hub module via the OPC-UA protocol. This calibrates the prediction accuracy of the digital twin model and iteratively optimizes the algorithm parameter weights, forming a closed-loop control chain from data acquisition, simulation prediction, instruction generation, and execution feedback.

[0063] The digital twin platform uses a three-dimensional visualization interface to display material motion trajectories, temperature field distribution, and magnetic domain state evolution in real time, providing visual decision support for process parameter optimization. The optimization engine combines process parameter-product quality correlation data from the historical database, screens the optimal solution set through a sliding time window mechanism, and dynamically adjusts the weight allocation strategy of the multi-objective nonlinear programming algorithm. The hysteresis loop data of the magnetic separation module is cross-validated with the phase composition detection results of the mass feed module, driving the optimization engine to reallocate the particle size control priority and magnetic field gradient threshold, achieving cross-module parameter collaborative optimization and steady-state control of the coal gangue resource processing process.

[0064] A mass feed module detects the phase composition data of the iron-depleted material by combining a component analysis probe, extracts characteristic vectors using a principal component analysis algorithm, matches process parameters in a historical database to generate gas ratio correction instructions, and feeds the correction instructions back to the data hub module;

[0065] The mass flow module continuously monitors the iron-depleted material output from the magnetic separation module using a component analysis probe. This probe integrates an X-ray diffractometer and a fluorescence spectrometer, mounted above the material conveyor belt for contactless scanning. The X-ray diffractometer analyzes the material's crystalline structure, while the fluorescence spectrometer measures the ratios of iron, aluminum, and silicon, generating phase composition data including residual ferroferric oxide and the aluminum oxide / silicon dioxide ratio. This data is transmitted via an industrial bus to the process adjustment engine, where it serves as input for the principal component analysis algorithm.

[0066] The multispectral imaging system, equipped with a high-resolution CCD array and multi-band filters, collects diffuse reflectance spectral data from the material surface in the visible to near-infrared range. This spectral data is processed using a CIE Lab color space conversion algorithm, extracting the luminance component L to quantify the material's whiteness. This data is then combined with the chromaticity coordinates a and b* to generate a whiteness measurement dataset. The imaging system and component analysis probe are triggered synchronously to establish a spatiotemporal correlation mapping between phase composition and whiteness indicators, providing multi-dimensional data support for feature vector extraction.

[0067] After receiving the phase composition data and whiteness test data, the process adjustment engine performs dimensionality reduction using a principal component analysis algorithm to extract feature vectors reflecting the residual iron phase and aluminum-silicon activity. These feature vectors are then input into a K-nearest neighbor classifier and matched against process parameter-product quality correlation data stored in the data hub module's historical database. Successfully matched parameter combinations are optimized using a Gaussian process regression model to generate correction instructions for the hydrogen-carbon monoxide mixture ratio. These instructions are then fed back to the data hub module's optimization engine via Industrial Ethernet.

[0068] After receiving the gas ratio correction command, the data center module triggers the weight coefficient update mechanism of the multi-objective nonlinear programming algorithm. The updated algorithm reallocates the particle size control priority of the raw material pretreatment module, the preheating zone temperature gradient threshold of the fluidized bed magnetization roasting module, and the magnetic field strength threshold of the magnetic separation module. The optimization engine combines the thermal field simulation results and magnetic circuit distribution model of the digital twin platform to calculate the coordinated adjustment of the fluidizing gas mixing ratio and the electric heating power, generating new roasting parameter commands and sending them to the actuator.

[0069] After the correction instructions are executed, the iron-depleted material enters a new cycle of phase composition and whiteness testing. This testing data updates the matching rules of the historical database through a sliding time window mechanism. The digital twin platform of the data hub module cross-validates the current phase data with thermal field prediction results, optimizing the feature extraction dimensions of the principal component analysis algorithm. The effectiveness of process parameter adjustments is dynamically evaluated using hysteresis loop data from the magnetic separation module and whiteness indicators from the mass spectrometry module. This forms a closed-loop control chain from quality testing, parameter correction, and effect verification, maintaining the stability of the coal gangue resource processing process and the consistency of product quality.

[0070] The data hub module updates the parameter weights of the multi-objective nonlinear programming algorithm according to the gas ratio correction instruction of the mass feed module, triggering the fluidized magnetization roasting module to adjust the mixing ratio of hydrogen and carbon monoxide to form a closed-loop control.

[0071] After receiving the gas ratio correction instructions from the mass feedback module, the data center module dynamically adjusts the constraints of the multi-objective nonlinear programming algorithm through the parameter weight update unit. The correction instructions include the optimization coefficients for the hydrogen and carbon monoxide mixing ratio. The weight update unit normalizes the residual ferroferric oxide index in the phase composition data and the L* value component of the whiteness test data, and inputs them into the fuzzy logic controller to generate the weight adjustment parameters. The process parameter-product quality correlation data stored in the historical database is filtered through a sliding window mechanism to select the data set with the highest match to the current feature vector, driving the generation of the parameter weight update instructions.

[0072] The optimization engine loads the updated parameter weights and reconstructs the objective function and constraints of the multi-objective nonlinear programming algorithm. The algorithm uses the particle size distribution variance of the raw material pretreatment module, the standard deviation of the temperature gradient in the preheating zone of the fluidized bed magnetization roasting module, and the coefficient of variation of the magnetic field intensity of the magnetic separation module as optimization variables to calculate the Pareto optimal solution set for each module's control parameters. The particle size control priority is reordered based on the weighted ratio of the crusher unit's energy consumption to its magnetization efficiency. Roller gap adjustment instructions and airflow screening frequency correction values are generated and synchronously updated to the raw material pretreatment module's actuators.

[0073] After the fluidized bed magnetized roasting module receives the preheating zone temperature gradient threshold adjustment command from the optimization engine, the PID controller of the electric heating element adjusts the power output based on the real-time temperature field data from the infrared thermal imaging camera array. The axial temperature compensation coefficient of the preheating zone is mapped to the thermal field simulation model of the digital twin platform to generate a dynamic temperature control curve. The gas distribution plate in the cooling chamber uses a stepper motor to drive the proportional control valve opening. The injection of hydrogen and carbon monoxide is synchronously adjusted according to the updated mixing ratio command, forming a reducing atmosphere gradient distribution from the center outward.

[0074] The fluidizing gas velocity and cooling rate after the gas mixture ratio is adjusted are collaboratively verified using the thermal field distribution prediction model on the digital twin platform. A gas composition sensor at the cooling chamber outlet monitors the residual oxygen concentration in real time, providing feedback to the optimization engine in the data hub module, triggering the secondary correction mechanism of the PID controller. The effectiveness of the adjusted process parameters is cross-evaluated using hysteresis loop data from the magnetic separation module and phase detection results from the mass flow module, forming a closed-loop control chain from weight update and parameter optimization to effect verification.

[0075] The data hub module continuously receives execution status data from each module via the OPC-UA protocol, calibrating the prediction accuracy of the digital twin model and iteratively optimizing algorithm parameter weights. The historical database uses a time series analysis model to record product quality trends before and after parameter adjustments, optimizing the rule base of the fuzzy logic controller. The gas mixing ratio in the fluidized bed magnetization roasting module and the temperature gradient in the preheating zone form a multi-dimensional parameter linkage, inhibiting the oxidation of ferric oxide while maintaining the active structural stability of the aluminum-silicon material, achieving low-carbon and efficient operation of the coal gangue treatment process.

[0076] The gangue magnetized fluidized bed roasting device provided by the present invention realizes the coordinated optimization of multiple process parameters through modular design. The raw material pretreatment module uses a multi-stage crushing unit to crush the gangue, in which an impact crusher completes the coarse crushing and a high-pressure roller mill implements the fine crushing. The crushed material enters a closed-circuit pneumatic classification device for dynamic screening, and particles that do not meet the standards are returned to the crushing unit for reprocessing. The online particle size analysis device collects the particle size distribution data of the crushed product in real time, transmits it to the data center module via the industrial bus, and provides input parameters for subsequent particle size control.

[0077] The dual-chamber reaction tower of the fluidized magnetized roasting module is divided into a conical fluidized bed preheating zone and a U-shaped cooling chamber. After the preheating zone receives qualified material, the electric heating element adjusts its power based on the roasting temperature correction command issued by the data center module, raising the material to the set temperature. An infrared thermal imager array monitors the three-dimensional temperature distribution within the reaction tower in real time, and a PID controller dynamically adjusts the fluidizing gas flow rate. The cooling chamber uses a gas distribution plate to create a gradient cooling zone based on the hydrogen and carbon monoxide mixing ratio, suppressing the oxidation reaction of ferric oxide.

[0078] The magnetic separation module consists of a permanent magnet drum roughing section and an electromagnetic high-gradient cleaning section. The roughing section performs preliminary magnetic separation on the roasted product, removing large iron-phase impurities. The cleaning section is equipped with an adjustable-pitch magnetic pole array. Upon receiving magnetic field intensity threshold commands from the data hub module, it optimizes the magnetic field gradient by adjusting the electromagnetic coil current. A hysteresis detection device collects real-time saturation data on magnetic particles, generating a hysteresis loop that is fed back to the data hub module as a basis for adjusting magnetic field intensity.

[0079] The data hub module integrates a digital twin platform and an optimization engine. The digital twin platform simultaneously receives particle size data from the raw material pretreatment module, temperature field data from the fluidization module, and hysteresis loop data from the magnetic separation module to construct a discrete element material motion model and a thermal field distribution simulation model. The optimization engine, based on a multi-objective nonlinear programming algorithm, calculates particle size control instructions, temperature correction instructions, and magnetic field threshold instructions. It also dynamically adjusts the fluidizing gas flow rate using a risk prediction unit combined with the material residence time distribution function.

[0080] The mass spectrometry module uses a component analysis probe to perform X-ray diffraction and fluorescence spectroscopy on iron-depleted materials, acquiring phase composition data. A multispectral imaging system analyzes the material's diffuse reflectance spectrum and quantifies whiteness values using a CIE Lab color space conversion algorithm. A principal component analysis algorithm extracts eigenvectors from the phase and whiteness data, matches them to the data center module's historical process parameter database, generates gas ratio correction instructions, and feeds them back to the optimization engine.

[0081] The data hub module updates the weight parameters of the multi-objective nonlinear programming algorithm based on the gas ratio correction instructions, reallocating the particle size control priority and magnetic field strength thresholds. The fluidized bed magnetization roasting module receives the updated temperature gradient thresholds and adjusts the electric heating power in the preheating zone and the gas mixing ratio in the cooling zone, forming a closed-loop control process from raw material crushing, magnetization roasting, magnetic separation, to quality feedback. Through collaborative optimization of the data hub, each module achieves active structural protection of the iron, aluminum, and silicon components in the gangue and efficient separation of submicron iron impurities.

[0082] Specifically, the gangue magnetized fluidized roasting device of the present invention, the raw material pretreatment module includes:

[0083] The multi-stage crushing unit module controls the multi-stage crushing unit to perform impact crushing and high-pressure roller grinding to process the coal gangue in stages, and outputs the crushed materials to the closed-circuit pneumatic classification device;

[0084] A closed-circuit pneumatic classifier dynamically screens the crushed material through a cyclone separator and an airflow screener, and the substandard particles are returned to the multi-stage crushing unit;

[0085] The particle size analysis device collects the output material particle size data of the closed-circuit pneumatic classification device in real time, inputs the particle size data into the particle size prediction model of the data center module, generates a crusher roller gap adjustment instruction and sends it to the multi-stage crushing group.

[0086] In the magnetized fluidized bed roasting device for coal gangue described in this invention, the raw material pretreatment module achieves gradient crushing of the coal gangue through a multi-stage crushing unit. An impact crusher performs coarse crushing of the raw coal gangue, using high-speed impact force to break up bulky material. The coarsely crushed particles enter a high-pressure roller mill, where roller pressure applied by a hydraulic system completes fine crushing, producing crushed material with a uniform particle size. Wear-resistant liners are installed within the crushing unit to prevent metal contact with the material and ensure purity during subsequent magnetic separation.

[0087] Crushed material is conveyed through a conveyor into a closed-circuit pneumatic classifier for dynamic screening. A cyclone separator uses centrifugal force to separate particles of varying sizes. An airflow screener performs a secondary screening of particles that meet the standards. Substandard particles are returned to the multi-stage crushing unit through a return pipe for further processing. This closed-circuit system iteratively optimizes crushing and classification until the material particle size meets the requirements for fluidized bed roasting.

[0088] The particle size analyzer integrates a laser diffraction sensor and an optical imaging unit to monitor the particle size distribution of the output material from the closed-loop pneumatic classifier in real time. The collected particle size data is transmitted to the data hub module via an industrial communication protocol. This data is then fed into a particle size prediction model to calculate the deviation between the current crushing efficiency and the target particle size. Based on this deviation, the data hub module generates roller gap adjustment instructions, dynamically adjusting the roller gap and crushing pressure of the high-pressure grinding rollers (HPGR), achieving adaptive optimization of crushing parameters.

[0089] Upon receiving adjustment commands, the multi-stage crushing unit uses a servo motor to drive the hydraulic system to change the roller gap and simultaneously adjust the rotor speed of the impact crusher. A particle size analyzer continuously collects optimized crushed material data, forming a closed-loop control chain of crushing, grading, and detection, ensuring that the material particle size remains within the threshold required by the fluidized bed roasting process. Pre-treated, qualified materials are conveyed through a sealed conveying pipeline to the fluidized bed magnetizing roasting module, providing uniformly sized raw materials for the subsequent magnetizing reaction.

[0090] Specifically, the gangue magnetized fluidized roasting device of the present invention, the fluidized magnetized roasting module includes:

[0091] The double-chamber reaction tower has a conical fluidized bed preheating zone in the front section and a U-shaped cooling chamber in the rear section. The preheating zone receives the qualified materials from the raw material pretreatment module and heats them to the set temperature.

[0092] An infrared thermal imager array is embedded in the side wall of the reaction tower to collect three-dimensional temperature field distribution data in real time and send it to the data center module;

[0093] The gas distribution plate receives the roasting parameter instructions from the data center module, triggers the PID controller to adjust the mixing ratio of hydrogen and carbon monoxide according to the temperature field distribution data, and forms a gradient cooling zone in the U-shaped cooling cavity to inhibit the oxidation of ferric oxide.

[0094] The fluidized bed magnetization roasting module achieves thermal treatment and atmosphere control for coal gangue through a dual-chamber reaction tower. The conical fluidized bed preheating zone utilizes a porous gas distribution structure. Fluidizing gas is evenly injected from the bottom to suspend the material. Electric heating elements distributed circumferentially along the tower wall raise the preheating zone temperature to a set threshold. Driven by the fluidizing gas, the qualified material is continuously transported to the U-shaped cooling chamber, forming a continuous processing channel.

[0095] An array of infrared thermal imagers is embedded in the reactor's sidewalls at equal intervals, covering the interface between the preheating zone and the cooling chamber. The cameras collect real-time surface temperature distribution data via optical fiber transmission, generating a three-dimensional thermal map of the temperature field and uploading it to the data center module. This temperature data is then mapped to the thermal field simulation model on the digital twin platform to identify areas of localized overheating or abnormal temperature gradients.

[0096] The gas distribution plate is integrated into the bottom of the U-shaped cooling chamber and features an annular array of jet holes and a proportional control valve. The roasting parameter commands from the data center module trigger the PID controller, which dynamically calculates the hydrogen and carbon monoxide mixing ratio based on temperature field distribution data. Hydrogen is injected at high speed through the central jet hole to create a reducing atmosphere, while carbon monoxide diffuses through the peripheral annular channel, forming a concentration gradient from the inside out of the cooling chamber.

[0097] The temperature change rate in the gradient cooling zone is linked to the gas mixing ratio to suppress the oxidation reaction of ferric oxide in the high-temperature section. A gas composition sensor is installed at the cooling chamber outlet to monitor the residual oxygen concentration in real time and feed it back to the PID controller, forming a closed-loop adjustment of the gas ratio. The fluidizing gas flow rate is optimized to match the cooling rate to avoid microstructural defects in the material caused by rapid cooling.

[0098] The data center module dynamically updates the temperature gradient thresholds in the roasting parameter instructions based on magnetic detection data from the cooling chamber outlet. The preheating zone's electrical heating power is linked to the air pressure of the cooling chamber's gas distribution plate to maintain thermodynamic equilibrium during the magnetized roasting process. The processed material is then pneumatically conveyed to the magnetic separation module to separate iron impurities.

[0099] Specifically, the magnetic separation module of the gangue magnetized fluidized roasting device of the present invention includes:

[0100] The permanent magnetic drum roughing section receives the roasting product output by the fluidized magnetization roasting module and performs primary magnetic separation to separate large iron-phase impurities;

[0101] The electromagnetic high-gradient selection section is equipped with an adjustable-pitch magnetic pole array, which receives the magnetic field strength threshold instruction from the data hub module and dynamically adjusts the magnetic pole spacing and electromagnetic coil current;

[0102] The hysteresis detection device is integrated at the outlet of the electromagnetic high-gradient selection section, and monitors the saturation of magnetic particles in real time to generate hysteresis loop data, which is sent to the data center module as a basis for correcting the magnetic field strength threshold instruction.

[0103] The magnetic separation module achieves efficient separation of iron-phase impurities through multi-stage magnetic field control. The permanent magnetic drum roughing section utilizes an inclined drum design, covered with an array of high-remanence permanent magnets. The roasted product is evenly distributed onto the drum surface via a vibrating feeder. Large iron-phase impurities are pulled out of the material flow by the permanent magnets and collected by a scraper device into a waste bin. The remaining material then flows via a conveyor belt into the electromagnetic high-gradient separation section.

[0104] The electromagnetic high-gradient concentrator section is equipped with a Halbach-type magnetic pole array. The pole spacing is dynamically adjusted by a servo motor-driven lead screw mechanism. The magnetic field intensity threshold command issued by the data center module triggers the electromagnetic coil current adjustment to match the magnetic field gradient to the current iron content of the material. The adjustable-pitch magnetic pole array expands or contracts the pole spacing based on particle size distribution data, optimizing the capture efficiency of submicron iron impurities.

[0105] The hysteresis detection device integrates a fluxgate sensor and a data acquisition unit and is installed at the material outlet of the concentrating section. The sensor monitors the remanent magnetization and coercivity of the magnetic particles in real time, generating a characteristic hysteresis loop spectrum. This spectrum is Fourier transformed to extract the harmonic components, which are then uploaded to the magnetic domain analysis model in the data hub module to calculate the deviation between the current magnetic field strength threshold and the target value.

[0106] The data hub module adjusts the current fluctuation compensation coefficient in the magnetic field intensity threshold command based on the harmonic component data of the hysteresis loop characteristic spectrum. The updated threshold command is simultaneously transmitted to the electromagnetic coil controller and the magnetic pole spacing adjustment mechanism, forming a dynamic optimization mechanism that links magnetic field gradient and pole spacing. The selected iron-depleted material enters the mass feed module through a pneumatic conveying pipeline for composition detection and process parameter feedback.

[0107] During the separation process, the rotation speed of the permanent magnet drum is coordinated with the magnetic field gradient of the electromagnetic section. Separation efficiency data from the coarse section is fed back to the data hub module via a current sensor, dynamically optimizing the magnetic field application time in the fine section. The real-time data stream from the hysteresis detection device is interactively verified with the magnetic circuit simulation model on the digital twin platform, improving the precision of iron phase separation from submicron to nanometer levels. The processed iron-depleted material meets the requirements for the preparation of highly active aluminum-silicon components, achieving the goal of resource utilization of coal gangue.

[0108] Specifically, the gangue magnetized fluidized roasting device of the present invention, the data hub module includes:

[0109] The digital twin platform integrates a discrete element material motion model and a thermal field distribution simulation model, receives real-time particle size data from the raw material pretreatment module, temperature field monitoring data from the fluidized magnetization roasting module, and hysteresis loop data from the magnetic separation module, and generates material motion trajectory and thermal field distribution prediction results;

[0110] An optimization engine, based on the prediction results of the digital twin platform, calculates a particle size control instruction, a roasting temperature correction instruction, and a magnetic field strength threshold instruction through a multi-objective nonlinear programming algorithm;

[0111] The risk prediction unit dynamically corrects the fluidizing gas flow rate instruction output by the optimization engine according to the material residence time distribution data of the fluidized magnetization roasting module.

[0112] The data hub module achieves full process control through multi-source data fusion and simulation optimization. The digital twin platform receives real-time particle size data from the raw material pretreatment module and inputs it into a discrete element material motion model to simulate the flow trajectory of crushed particles. It also simultaneously integrates infrared temperature field data from the fluidized bed magnetization roasting module to drive a thermal field distribution simulation model to predict the material's thermal conductivity. Hysteresis loop data from the magnetic separation module undergoes Fourier transformation and is mapped to a magnetic properties analysis model to analyze the evolution of iron-phase magnetic domains. Multi-physics coupled simulation results are updated in real time to a 3D visualization interface, providing dynamic data support for optimization decisions.

[0113] The optimization engine incorporates a multi-objective nonlinear programming algorithm, using the digital twin platform's predicted material trajectory data, thermal field distribution data, and magnetic domain evolution data as constraints. With the optimization objectives of minimizing energy consumption and maximizing magnetization efficiency, the algorithm calculates the crusher roller gap adjustment, the fluidized bed electric heating power compensation value, and the electromagnetic coil current threshold, generating instructions for particle size control, roasting temperature correction, and magnetic field intensity threshold. The optimized instruction set is distributed to the corresponding module's actuators via an industrial communication protocol, establishing cross-module parameter linkage.

[0114] The risk prediction unit collects real-time material residence time distribution data from the fluidized bed magnetization roasting module and constructs a residence time probability density function using kernel density estimation. When the residence time of a material of a specific particle size deviates from the normal distribution, a fluidizing gas flow rate correction mechanism is triggered. The correction coefficient is dynamically adjusted based on the correlation analysis between the residence time deviation and the fluidized bed pressure differential data. The updated flow rate command is then sent to the proportional control valve on the gas distribution board via the edge computing node.

[0115] The digital twin platform simultaneously receives corrected gas flow rate data from the risk prediction unit and recalculates the boundary conditions of the thermal field distribution simulation model. The optimization engine iteratively calculates multi-objective parameter combinations based on the updated simulation results, forming a closed-loop feedback loop from data acquisition, simulation prediction, optimization decision-making, and risk control. Status data from each module's actuators is transmitted back to the data hub via the OPC-UA protocol, continuously calibrating the digital twin model's prediction accuracy and enabling dynamic optimization of coal gangue processing process parameters.

[0116] Specifically, the gangue magnetized fluidized roasting device of the present invention, the mass feed module includes:

[0117] Combined with a component analysis probe, continuous X-ray diffraction and fluorescence spectroscopy detection is performed on the iron-depleted material output by the magnetic separation module to obtain phase composition data;

[0118] A multispectral imaging system, which quantifies the whiteness value of the iron-depleted material through diffuse reflectance spectroscopy analysis to generate whiteness detection data;

[0119] The process adjustment engine inputs the phase composition data and the whiteness detection data into the principal component analysis algorithm, extracts the feature vector and matches it with the historical database of the data center module, generates a gas ratio correction instruction and feeds it back to the data center module.

[0120] The mass flow module dynamically optimizes process parameters through multi-dimensional testing and data analysis. A combined component analysis probe, integrating an X-ray diffractometer and a fluorescence spectrometer, is installed above the iron-depleted material outlet conveyor belt of the magnetic separation module. The probe performs non-contact scanning of the continuously passing material, while the X-ray diffractometer analyzes the crystal structure and the fluorescence spectrometer determines the elemental composition, generating phase composition data including the ferroferric oxide content and the aluminum oxide / silicon dioxide ratio.

[0121] The multispectral imaging system, equipped with a high-resolution CCD array and multi-band filters, captures the diffuse reflectance spectrum of the material surface from the visible to near-infrared wavelength range. The spectral data is processed using a CIE Lab color space conversion algorithm, extracting the luminance component (L) as a quantitative indicator of whiteness. This is then combined with the chromaticity coordinates a and b* to generate whiteness measurement data. The imaging system is synchronously triggered with the component analysis probe to establish a spatiotemporal correlation mapping between phase composition and whiteness values.

[0122] After receiving phase composition data and whiteness test data, the process adjustment engine uses principal component analysis (PCA) to reduce dimensionality and extract feature vectors reflecting the residual iron phase and aluminum-silicon activity. These feature vectors are then input into a K-nearest neighbor classifier and matched against process parameter-product quality correlation data stored in the data hub module's historical database. Successfully matched parameter combinations are then optimized using a Gaussian process regression model to generate instructions for adjusting the hydrogen and carbon monoxide mixture ratio.

[0123] The gas ratio correction command is fed back to the optimization engine of the data center module via industrial Ethernet, triggering an update of the weight coefficients of the multi-objective nonlinear programming algorithm. The updated algorithm recalculates the gas distribution plate opening parameters and the preheating zone temperature gradient threshold of the fluidized bed magnetization roasting module, forming a closed-loop control chain from product quality detection to process parameter adjustment.

[0124] The data center module's historical database continuously records product quality changes after correction instructions are executed, and updates feature vector matching rules through a sliding time window mechanism. The multispectral imaging system's whiteness detection data is cross-validated with the thermal field distribution predictions from the digital twin platform to optimize the feature extraction dimensions of the principal component analysis algorithm. The processed iron-depleted material is then pneumatically conveyed to a storage unit, completing the complete process for gangue resource processing.

[0125] Specifically, the gangue magnetized fluidized roasting device of the present invention further includes:

[0126] The phase composition data and whiteness detection data of the mass feed module trigger the parameter weight update of the data hub module;

[0127] The optimization engine reallocates the particle size control priority of the raw material pretreatment module, the preheating zone temperature gradient threshold of the fluidized magnetization roasting module, and the magnetic field strength threshold of the magnetic separation module through a multi-objective nonlinear programming algorithm according to the updated weights;

[0128] The fluidized magnetization roasting module receives the redistributed preheating zone temperature gradient threshold, adjusts the power of the electric heating element, and simultaneously adjusts the mixing ratio of hydrogen and carbon monoxide in the cooling zone according to the magnetic field strength threshold.

[0129] Phase composition data and whiteness test data from the mass feed module are transmitted via the industrial bus to the parameter weight update unit in the data hub module. The residual ferroferric oxide content in the phase composition data and the L* value component of the whiteness test data are normalized and input into the fuzzy logic controller to calculate the weight adjustment coefficient. A sliding window mechanism is used to select the data set with the highest similarity to the current feature vector from the matching process parameter combinations in the historical database, generating weight update instructions.

[0130] The optimization engine loads the updated parameter weights and reconstructs the objective function constraints within a multi-objective nonlinear programming algorithm. The algorithm uses the particle size distribution variance of the raw material pretreatment module, the standard deviation of the temperature gradient in the preheating zone of the fluidization module, and the coefficient of variation of the magnetic field strength of the magnetic separation module as optimization variables to calculate the Pareto optimal solution for each module's control parameters. The particle size control priority is reordered based on the weighted ratio of the crusher unit's energy consumption to its magnetization efficiency, generating roller gap adjustment instructions and airflow screening frequency correction values.

[0131] After the fluidized bed magnetized roasting module receives the reassigned preheating zone temperature gradient threshold, the electric heating element's PID controller adjusts its power output based on real-time temperature field data from the infrared thermal imaging camera array. The temperature gradient threshold is mapped to the thermal field simulation model on the digital twin platform to generate the preheating zone axial temperature compensation coefficient. The gas distribution plate in the cooling zone uses a stepper motor to drive the proportional valve opening, synchronously adjusting the hydrogen and carbon monoxide mixing ratio to ensure that the gas concentration gradient spatially matches the updated magnetic field strength threshold.

[0132] The electromagnetic coil controller in the magnetic separation module receives magnetic field strength threshold commands from the optimization engine and adjusts the current pulse width using PWM modulation technology. The servo mechanism of the adjustable-pitch magnetic pole array dynamically reduces the pole pitch based on magnetic field gradient distribution data, improving the capture efficiency of submicron iron phases. The data hub module monitors the synergistic effect of magnetic field strength and gas mixture ratio in real time, verifying the effectiveness of parameter adjustments using material magnetic detection data.

[0133] After process parameter adjustments, the iron-depleted material enters the mass-feed module for a new round of composition testing. Phase composition data and whiteness test data trigger another weight update cycle. The data center module's historical database uses a time series analysis model to record product quality trends before and after parameter adjustments, optimizing the fuzzy logic controller's rule base. Closed-loop coordinated control across multiple modules continuously reduces residual ferroferric oxide, improves the activity index of the aluminum-silicon material, and achieves steady-state operation of the gangue resource processing process.

[0134] The specific embodiment of the present invention relates to a practical application process of a coal gangue magnetized fluidized roasting device. The coal gangue raw materials are subjected to gradient crushing by a multi-stage crushing unit, and the impact crusher performs coarse crushing on the raw materials to generate coarse particles with a particle size of 20-50 mm; the coarsely crushed materials enter the high-pressure roller mill for fine crushing, and the roller pressure applied by the hydraulic system grinds the materials to the target particle size range. Wear-resistant liners are installed inside the crushing unit to isolate the direct contact between metal parts and materials and avoid the mixing of iron impurities. The crushed products are dynamically screened by a closed-circuit pneumatic grading device, and the unqualified particles are returned to the crushing unit through the return pipe, forming a closed-loop optimization process of crushing-grading. The online particle size analysis device uses laser diffraction technology to monitor the material particle size distribution in real time. The particle size data is transmitted to the data center module via the industrial bus, driving the particle size prediction model to generate roller gap adjustment instructions to optimize the crushing efficiency.

[0135] The fluidized magnetization roasting module utilizes a dual-chamber reaction tower structure. The preheating zone features a conical fluidized bed. Fluidizing gas is uniformly injected through a porous distribution plate at the bottom, ensuring material suspension and uniform heating. Electric heating elements are distributed circumferentially along the preheating zone. Power output is adjusted based on roasting temperature correction commands issued by the data center module, heating the material to a set temperature range of 650-800°C. An infrared thermal imager array embedded in the sidewall of the reaction tower captures three-dimensional temperature field distribution data in real time. This data is transmitted via optical fiber to the thermal field simulation model on the digital twin platform, identifying localized overheating areas and generating temperature compensation commands. The cooling chamber utilizes a U-shaped structure. The gas distribution plate adjusts the hydrogen to carbon monoxide mixture ratio from 3:1 to 5:1, creating an internal-to-external concentration gradient that suppresses the oxidation of ferric oxide to ferroferric oxide. The cooling rate is controlled at 10-15°C / min to prevent microcracks in the material caused by rapid cooling.

[0136] The magnetic separation module consists of a permanent magnet drum coarse selection section and an electromagnetic high-gradient selection section. The coarse selection section uses a permanent magnet array to absorb and separate iron-phase impurities larger than 100 microns, with the remaining material entering the selection section. The electromagnetic high-gradient selection section is equipped with a Halbach-type magnetic pole array. The pole spacing is dynamically adjusted to 2-5 mm by a servo motor-driven lead screw mechanism. The electromagnetic coil current is regulated to 50-200 A using PWM modulation technology, and the magnetic field strength ranges from 0.8 to 1.5 Tesla. The hysteresis detection device integrates a fluxgate sensor to monitor the remanent magnetization and coercivity of magnetic particles in real time, generating hysteresis loop data and uploading it to the data hub module. The data hub uses Fourier transform to analyze the harmonic components of the hysteresis loop. Combined with the magnetic circuit simulation model of the digital twin platform, it dynamically adjusts the magnetic field gradient threshold and pole spacing parameters, increasing the capture efficiency of submicron iron impurities to over 95%.

[0137] The data hub module integrates a digital twin platform and an optimization engine, and receives real-time particle size distribution data from the raw material pretreatment module, temperature field data from the fluidization module, and hysteresis loop data from the magnetic separation module. The digital twin platform constructs a discrete element material motion model to simulate the flow trajectory of crushed particles, a thermal field distribution model to predict the temperature gradient changes in the preheating zone and the cooling chamber, and a magnetic domain analysis model to analyze the magnetization characteristics of the iron phase. The optimization engine is based on a multi-objective nonlinear programming algorithm, with the goals of minimizing energy consumption and maximizing magnetization efficiency. It generates particle size control instructions, temperature correction instructions, and magnetic field strength threshold instructions, and distributes them to the actuators of each module via industrial Ethernet. The risk prediction unit dynamically adjusts the gas flow rate instructions based on the fluidized bed material residence time distribution data to avoid local material accumulation or temperature anomalies.

[0138] The mass-feedback module detects the residual ferroferric oxide and the alumina / silicon dioxide ratio of the iron-depleted material by combining an X-ray diffractometer with a fluorescence spectrometer sensor, and the multispectral imaging system quantifies the material's whiteness value. The principal component analysis algorithm extracts the characteristic vectors of the phase and whiteness data, matches the process parameter combinations in the historical database, and generates gas mixing ratio correction instructions. The data center module updates the weight parameters of the optimization algorithm based on the feedback instructions, reallocating the temperature gradient threshold of the fluidization module and the magnetic field strength threshold of the magnetic separation module. The electric heating power of the preheating zone and the gas mixing ratio of the cooling chamber are adjusted in conjunction to form a closed-loop control of the process parameters, ultimately achieving efficient separation of iron-phase impurities in coal gangue and stable protection of the aluminum-silicon active structure.

[0139] The present invention solves the technical problem of insufficient multi-variable coupling control in gangue processing through modular collaborative control and data-driven optimization mechanism. The data center module integrates the digital twin platform and the multi-objective nonlinear programming algorithm, and receives the particle size data of the raw material pretreatment module, the temperature field data of the fluidized magnetization roasting module, and the hysteresis loop data of the magnetic separation module in real time. Through multi-physical field coupling simulation and Pareto optimal solution calculation, the particle size control instructions, roasting temperature correction instructions and magnetic field strength threshold instructions are dynamically generated to achieve multi-dimensional parameter collaborative optimization of temperature, gas flow rate and magnetic field strength, and balance the coupling relationship between thermodynamic reaction and magnetization efficiency.

[0140] To address the challenge of low separation accuracy for submicron iron impurities, the magnetic separation module utilizes a permanent magnet-electromagnetic hybrid separation technology. The permanent magnet drum coarse separation section separates large iron particles, while the electromagnetic high-gradient separation section utilizes an adjustable-pitch magnetic pole array linked to a hysteresis detection device to adjust the magnetic field intensity threshold in real time. Hysteresis loop data is Fourier transformed to analyze the magnetic domain state of the iron phase. Combined with the magnetic circuit simulation model of the digital twin platform, the magnetic pole spacing and coil current are dynamically adjusted to improve the gradient capture accuracy of submicron iron impurities.

[0141] To reduce energy consumption and stabilize product quality, the mass feed module and the data hub module form a closed-loop feedback loop. A component analysis probe is combined with a multispectral imaging system to detect the phase composition and whiteness value of the iron-depleted material. A principal component analysis algorithm extracts eigenvectors, matches historical process parameters, and generates gas ratio correction instructions. The optimization engine updates parameter weights based on feedback instructions, reallocating particle size priority, temperature gradient thresholds, and magnetic field strength thresholds. A PID controller and proportional valve coordinately adjust the fluidizing gas mixing ratio and electric heating power, achieving dynamic energy optimization and protecting the active aluminum-silicon structure.

Claims

1. A coal gangue magnetized fluidized roasting device, characterized in that: include: A multi-stage crushing unit, an online particle size analyzer, a component analysis probe, and a control device, wherein the control device includes: The raw material pre-processing module crushes and grinds the gangue through a multi-stage crushing unit, collects material particle size data in real time using an online particle size analyzer, and sends the particle size data to the data center module; The fluidized magnetization roasting module receives the material output by the raw material pretreatment module, adjusts the fluidization gas flow rate and the power of the electric heating element according to the roasting parameter instructions generated by the data hub module, and outputs the product to the magnetic separation module after completing the magnetization roasting; A magnetic separation module receives the magnetized roasted product, dynamically adjusts the magnetic pole spacing and the electromagnetic coil current according to the magnetic field strength control instruction generated by the data hub module, separates the iron phase impurities, and outputs the iron-poor material to the mass feed module; The data center module integrates the digital twin platform and the optimization engine, receives the real-time particle size data of the raw material pretreatment module, the temperature field monitoring data of the fluidized magnetization roasting module, and the hysteresis loop data of the magnetic separation module, and generates the particle size control instructions, roasting temperature correction instructions, and magnetic field strength threshold instructions through a multi-objective nonlinear programming algorithm; A mass feed module detects the phase composition data of the iron-depleted material by combining a component analysis probe, extracts characteristic vectors using a principal component analysis algorithm, matches process parameters in a historical database to generate gas ratio correction instructions, and feeds the correction instructions back to the data hub module; The data hub module updates the parameter weights of the multi-objective nonlinear programming algorithm according to the gas ratio correction instruction of the mass feed module, triggering the fluidized magnetization roasting module to adjust the mixing ratio of hydrogen and carbon monoxide.

2. The gangue magnetized fluidized roasting device according to claim 1, characterized in that: The raw material pretreatment module includes: The multi-stage crushing unit module controls the multi-stage crushing unit to perform impact crushing and high-pressure roller grinding to process the coal gangue in stages, and outputs the crushed materials to the closed-circuit pneumatic classification device; A closed-circuit pneumatic classifier dynamically screens the crushed material through a cyclone separator and an airflow screener, and the substandard particles are returned to the multi-stage crushing unit; The particle size analysis device collects the output material particle size data of the closed-circuit pneumatic classification device in real time, inputs the particle size data into the particle size prediction model of the data center module, generates a crusher roller gap adjustment instruction and sends it to the multi-stage crushing group.

3. The gangue magnetized fluidized roasting device according to claim 1, characterized in that: The fluidized magnetization roasting module includes: The double-chamber reaction tower has a conical fluidized bed preheating zone in the front section and a U-shaped cooling chamber in the rear section. The preheating zone receives the qualified materials from the raw material pretreatment module and heats them to the set temperature. An infrared thermal imager array is embedded in the side wall of the reaction tower to collect three-dimensional temperature field distribution data in real time and send it to the data center module; The gas distribution plate receives the roasting parameter instruction from the data center module and triggers the PID controller to adjust the mixing ratio of hydrogen and carbon monoxide according to the temperature field distribution data.

4. The gangue magnetized fluidized roasting device according to claim 1, characterized in that: The magnetic separation module comprises: The permanent magnetic drum roughing section receives the roasting product output by the fluidized magnetization roasting module and performs primary magnetic separation to separate large iron-phase impurities; The electromagnetic high-gradient selection section is equipped with an adjustable-pitch magnetic pole array, which receives the magnetic field strength threshold instruction from the data hub module and dynamically adjusts the magnetic pole spacing and electromagnetic coil current; The hysteresis detection device is integrated at the outlet of the electromagnetic high-gradient selection section, and monitors the saturation of magnetic particles in real time to generate hysteresis loop data, which is sent to the data center module as a basis for correcting the magnetic field strength threshold instruction.

5. The gangue magnetized fluidized roasting device according to claim 1, characterized in that: The data hub module includes: The digital twin platform integrates a discrete element material motion model and a thermal field distribution simulation model, receives real-time particle size data from the raw material pretreatment module, temperature field monitoring data from the fluidized magnetization roasting module, and hysteresis loop data from the magnetic separation module, and generates material motion trajectory and thermal field distribution prediction results; An optimization engine, based on the prediction results of the digital twin platform, calculates particle size control instructions, roasting temperature correction instructions, and magnetic field strength threshold instructions through a multi-objective nonlinear programming algorithm; The risk prediction unit dynamically corrects the fluidizing gas flow rate instruction output by the optimization engine according to the material residence time distribution data of the fluidized magnetization roasting module.

6. The gangue magnetized fluidized roasting device according to claim 1, characterized in that: The mass feed module includes: Combined with a component analysis probe, continuous X-ray diffraction and fluorescence spectroscopy detection is performed on the iron-depleted material output by the magnetic separation module to obtain phase composition data; A multispectral imaging system, which quantifies the whiteness value of the iron-depleted material through diffuse reflectance spectroscopy analysis to generate whiteness detection data; The process adjustment engine inputs the phase composition data and the whiteness detection data into the principal component analysis algorithm, extracts the feature vector and matches it with the historical database of the data center module, generates a gas ratio correction instruction and feeds it back to the data center module.

7. The gangue magnetized fluidized roasting device according to claim 1, characterized in that: Also includes: The phase composition data and whiteness detection data of the mass feed module trigger the parameter weight update of the data hub module; The optimization engine reallocates the particle size control priority of the raw material pretreatment module, the preheating zone temperature gradient threshold of the fluidized magnetization roasting module, and the magnetic field strength threshold of the magnetic separation module through a multi-objective nonlinear programming algorithm according to the updated weights; The fluidized magnetization roasting module receives the redistributed preheating zone temperature gradient threshold, adjusts the power of the electric heating element, and simultaneously adjusts the mixing ratio of hydrogen and carbon monoxide in the cooling zone according to the magnetic field strength threshold.

Images