A centrifugal atomization method and system for producing ultrafine soft magnetic powders

Abstract

The application relates to the field of soft magnetic material preparation, and particularly discloses a centrifugal atomization method and system for preparing superfine soft magnetic powder. The method comprises the following steps: controlling the superheat degree of a Fe-Si-Al alloy melt, injecting the melt into a high-speed centrifugal atomizer through a laser texturing centrifugal disc to generate refined liquid drops; controlling the atmosphere and performing multi-stage cooling under a specific hydrogen proportion and a low-oxygen environment; preparing the powder through ultrasonic-assisted rapid cooling and solidification; adopting an intelligent algorithm and multi-stage separation to accurately classify the particle size; detecting the morphology, oxygen content and magnetic performance of the powder to screen qualified products; and finally realizing process optimization through equipment deviation analysis and system calibration. The method can accurately control the particle size distribution of the powder, effectively reduce the oxygen content, and significantly improve the sphericity and soft magnetic performance of the powder.

Description

Technical Field

[0001] This invention relates to the field of soft magnetic material preparation, and in particular to a centrifugal atomization method and system for preparing ultrafine soft magnetic powder. Background Technology

[0002] In the field of soft magnetic material preparation, existing solutions for centrifugal atomization methods and systems used to prepare ultrafine soft magnetic powders typically employ traditional centrifugal atomization processes combined with inert atmosphere protection. These solutions suffer from limitations such as insufficient control of melt superheat, low droplet refinement efficiency, and difficulty in maintaining atmosphere stability. Existing methods largely rely on conventional centrifugal disc structures and single atmosphere control methods. In scenarios involving vacuum induction furnace melting or argon-hydrogen mixed atmospheres, they are prone to uneven droplet size distribution and increased oxide inclusions, making it difficult to achieve stable sphericity and low oxygen content in ultrafine soft magnetic powders. Regarding the combined processing of Fe-Si-Al alloy melts and argon-hydrogen mixed atmospheres during high-speed centrifugal atomization, existing technologies generally suffer from common shortcomings in dynamic matching of melt injection, real-time adjustment of atmosphere ratios, and coordinated control of cooling pressure. This makes it difficult to establish a consistent process of melt treatment—atomization parameters—atmosphere control—solidification into powder in continuous ultrafine powder preparation scenarios, resulting in unsatisfactory powder morphology control and insufficient batch stability. Summary of the Invention

[0003] This invention provides a centrifugal atomization method and system for preparing ultrafine soft magnetic powder, which solves the problem of how to prepare ultrafine soft magnetic powder based on Fe-Si-Al alloy melt in a vacuum induction furnace with an argon-hydrogen mixed atmosphere, using a magnetically levitated centrifugal atomizer and an atmosphere control unit under conditions of laser texturing treatment on the surface of the centrifugal disk.

[0004] To address the aforementioned technical problems, this invention provides a centrifugal atomization method for preparing ultrafine soft magnetic powder, comprising:

[0005] Fine droplets are generated from Fe-Si-Al alloy melt by controlling the melt superheat at 150℃ based on PID algorithm, using laser texturing centrifugal injection with surface roughness Ra≤0.1μm and rotation speed of 25000rpm, and centrifugal atomization treatment with real-time monitoring at 25000rpm.

[0006] Based on the refinement of droplets, a hydrogen ratio of 5% by volume is achieved, oxygen content is maintained at ≤200ppm by feedback regulation, and pressure balance is achieved by multi-stage cooling towers connected in series to generate an atomized environment.

[0007] From the atomization environment, the cooling rate of the ultrasonic transducer frequency range of 20kHz to 2MHz is configured, the ultrasonic field is optimized using a genetic algorithm, and ultrasonic-assisted droplet solidification is performed to generate rapidly cooled powder.

[0008] Based on rapidly cooled powder, a multi-stage cyclone separation process is performed, including setting a 5-20μm particle size threshold, optimizing intelligent algorithms such as support vector machines, artificial neural networks and fuzzy logic control, and setting cyclone separators in series, to generate powder with the target particle size.

[0009] From the target particle size powder, morphology detection with a sphericity threshold of not less than 95% is performed, oxygen content threshold comparison is maintained at ≤200ppm, and magnetization intensity test is performed using a vibrating sample magnetometer with a saturation magnetization intensity of not less than 1.8 Tesla to generate powder with qualified performance.

[0010] Based on the performance-compliant powder, we performed equipment deviation analysis using principal component analysis, laser power compensation to maintain a surface roughness Ra≤0.1μm, and magnetic levitation gap calibration to maintain a gap size of 50μm±2μm, thereby generating an optimized system configuration.

[0011] Furthermore, the Fe-Si-Al alloy melt specifically includes:

[0012] A homogeneous liquid metal fluid in a specific thermodynamic state, formed by complete melting under the protection of high-purity argon gas in a vacuum induction furnace:

[0013] Chemical composition: Its elemental composition strictly conforms to the preset Fe-Si-Al alloy composition design, and the error of the mass percentage of each element is controlled within ±0.5%;

[0014] Physical state: It is a high-temperature liquid phase with uniform composition, no metallurgical inclusions, no precipitated phases and low dissolved gas content, and has superheat that can be precisely controlled by PID algorithm and maintained at 150℃±10℃.

[0015] Environmental conditions: The entire preparation and maintenance process is carried out in an inert atmosphere with low oxygen partial pressure, which is guaranteed by a high-purity protective gas circulation system, and no oxide film is formed on the free surface of the melt.

[0016] Furthermore, the process of generating refined droplets also includes:

[0017] Fe-Si-Al alloy melt was obtained, and melt superheat control was performed based on PID algorithm to maintain superheat at 150℃, thus obtaining superheated melt.

[0018] Melt flow is extracted from superheated melt and laser-textured centrifugal disk is injected, wherein the surface roughness Ra of the centrifugal disk is ≤0.1μm and the rotation speed is 25000rpm, to generate initial droplets;

[0019] The initial droplets were centrifuged and atomized at 25,000 rpm, and real-time monitoring was performed using a high-speed imaging system and a laser particle size analyzer to generate refined droplets.

[0020] Furthermore, the process of generating a fogging environment also includes:

[0021] Based on the refinement of droplets, hydrogen is proportioned and prepared, with a hydrogen volume ratio of 5%, to obtain the initial atmosphere.

[0022] Oxygen content signals are extracted from the initial atmosphere and real-time feedback adjustment is performed to maintain oxygen content ≤200ppm and generate a stable atmosphere.

[0023] A multi-stage cooling tower pressure balancing process is used to stabilize the atmosphere, with the cooling towers connected in series to generate an atomized environment.

[0024] Furthermore, the process of generating quenched powder also includes:

[0025] The atomization environment is obtained, and the cooling rate is configured. The frequency range of the ultrasonic transducer array is 20kHz to 2MHz, and the ultrasonic field parameters are obtained.

[0026] Frequency features are extracted from ultrasonic field parameters, standing wave field optimization is performed, and genetic algorithm is used to optimize parameters and generate a directional cooling field.

[0027] The directional cooling field is subjected to droplet solidification treatment, and ultrasonic-assisted droplet solidification treatment is used to generate rapidly cooled powder.

[0028] Furthermore, the process of generating powder with the target particle size also includes:

[0029] Based on rapidly cooled powder, a 5-20μm threshold setting process was performed to obtain classification parameters;

[0030] Particle size distribution data is extracted from grading parameters and optimized using intelligent algorithms, including support vector machines, artificial neural networks, and fuzzy logic control, to generate dynamic adjustment instructions.

[0031] The dynamic adjustment command is subjected to multi-stage cyclone separation processing, in which the cyclone separators are set in series to generate powder of the target particle size.

[0032] Furthermore, the process of generating powder that meets performance standards also includes:

[0033] Based on the target particle size powder, a threshold setting process is performed, wherein the sphericity threshold is not less than 95%, and morphology detection parameters are obtained.

[0034] Oxygen content data is extracted from morphology detection parameters, and a 200ppm threshold comparison is performed to maintain oxygen content ≤200ppm, generating a quality assessment report.

[0035] The quality assessment report is processed by magnetization intensity testing. A vibrating sample magnetometer is used and the saturation magnetization intensity is not less than 1.8 Tesla to generate powder that meets the performance standards.

[0036] Furthermore, the process of generating an optimized system configuration also includes:

[0037] Based on the performance of the powder, we performed equipment deviation analysis and used principal component analysis algorithm to obtain a list of calibration requirements.

[0038] Laser power data is extracted from the calibration requirements list, texture accuracy compensation is performed to maintain surface roughness Ra≤0.1μm, and updated laser parameters are generated.

[0039] The updated laser parameters are calibrated using magnetic levitation gap calibration to maintain the gap size at 50μm±2μm, and an optimized system configuration is generated.

[0040] Furthermore, the expression for generating refined droplets includes:

[0041] An improved PID control algorithm is used to establish a superheat regulation model:

[0042]

[0043] in, This represents the superheat adjustment amount in the k-th iteration; This is the superheat error term; These are control coefficients optimized based on historical process data; For summation index; This represents the total number of iterations. To control the cycle time step;

[0044] Define melt flow properties:

[0045] in, This refers to the melt flow properties index; The dynamic viscosity of the melt; It is a two-dimensional temperature gradient; The density of the melt; For laser-textured surface roughness parameters;

[0046] Power compensation is triggered when Φ > 1.2;

[0047] Constructing an improved Navier-Stokes equation to describe melt flow:

[0048]

[0049] Among them, is the radial velocity component; is the time; is the local acceleration term of the melt radial velocity; ∇ is the Nabla operator; is the convective acceleration term of the melt radial velocity; is the material derivative or total acceleration of the fluid element; is the pressure gradient force acting on the fluid element per unit volume, and the negative sign indicates that the direction of the force is opposite to the direction of increasing pressure; is the dynamic viscosity of the melt; is the Laplace operator; is the viscous force or diffusion force acting on the fluid element per unit volume; is the centrifugal force term; ω is the angular velocity of the centrifugal disk; r is the radial distance of the melt element from the center of the disk;

[0050] Define the droplet size prediction model:

[0051]

[0052] Among them, is the droplet diameter; σ is the surface tension of the melt; Ra is the laser texturing surface roughness parameter;

[0053] Droplet diameter is judged to be qualified when the deviation from the high-speed camera data is ≤ 5%;

[0054] Use the improved Reynolds number to judge the secondary atomization effect:

[0055]

[0056] Among them, is the Reynolds number; is the droplet diameter; is the tangential velocity of the droplet; is the dynamic viscosity of the melt;

[0057] When > 200, droplet secondary breakup is triggered, and define the atomization efficiency optimization function:

[0058] Among them, is the atomization efficiency; is the residence time of the droplet in the atomization zone; is the critical breakup time threshold.

[0059] Furthermore, a centrifugal atomization system for preparing ultrafine soft magnetic powder, which is applied to the method described in any one of the above, includes:

[0060] The melt superheat control module is used to acquire Fe-Si-Al alloy melt and perform melt superheat control processing;

[0061] The laser texturing injection module, connected to the melt superheat control module, is used to process the initial droplet generation of the melt flow;

[0062] The centrifugal atomization module, connected to the laser texturing injection module, is used to perform centrifugal atomization processing at 25,000 rpm;

[0063] The gas mixing module is used to generate an initial argon-hydrogen mixture containing 5% H2;

[0064] The oxygen content feedback module, connected to the gas ratio module, is used to adjust the oxygen content to generate a stable atmosphere.

[0065] The multi-stage cooling tower actuator is connected to an oxygen content feedback module to implement pressure balance and generate an atomized environment.

[0066] The parameter alignment module is used to synchronize centrifugal atomization parameters with atomization environment parameters;

[0067] The dynamic arbitration module, connected to the oxygen content feedback module, is used to output pressure regulation signals;

[0068] The data recording module, connected to the dynamic arbitration module, is used to update the dynamic oxygen content threshold table.

[0069] The key innovations of this invention include:

[0070] (1) In a magnetically levitated centrifugal atomizer, the surface of the centrifugal disk is treated with laser texturing to improve the droplet refinement efficiency, which involves the injection of Fe-Si-Al alloy melt and the generation of initial droplets.

[0071] (2) Real-time feedback regulation of argon-hydrogen mixed gas is adopted to ensure atmosphere stability, which involves the generation of the initial atmosphere and the maintenance of the stable atmosphere.

[0072] (3) The atomization environment is optimized by pressure balancing treatment through multi-stage cooling towers, which involves the conversion of stable atmosphere and the formation of atomization environment.

[0073] The following are its main beneficial effects:

[0074] (1) The surface of the centrifuge disk treated by laser texturing plays a role in the droplet refinement process, enabling the initial droplets to be uniformly refined under high-speed rotation, improving the uniformity of droplet size distribution, and is suitable for the continuous preparation of ultrafine powders.

[0075] (2) Real-time feedback regulation of argon-hydrogen mixed gas plays a role in maintaining atmosphere stability, ensuring accurate extraction and regulation of oxygen content signal, suppressing the increase of oxidative inclusions, and is suitable for achieving stable powder morphology.

[0076] (3) The pressure balancing treatment of multi-stage cooling towers plays a role in optimizing the atomization environment, ensuring the selection of cooling medium and the coordinated control of cooling pressure, promoting the stable realization of powder sphericity and low oxygen content, and is suitable for the preparation of 3D printed soft magnetic composite materials. Attached Figure Description

[0077] Figure 1 A schematic flowchart of a centrifugal atomization method for preparing ultrafine soft magnetic powder provided in an embodiment of this application;

[0078] Figure 2 This is a structural block diagram of a centrifugal atomization system for preparing ultrafine soft magnetic powder, provided in an embodiment of this application. Detailed Implementation

[0079] Example 1: Refer to Figure 1 This is a schematic flowchart of a centrifugal atomization method for preparing ultrafine soft magnetic powder provided by an embodiment of the present invention. The process may include at least steps S100-S600:

[0080] S100, from Fe-Si-Al alloy melt, melt superheat control based on PID algorithm to maintain superheat at 150℃, laser texturing centrifugal injection with surface roughness Ra≤0.1μm and rotation speed of 25000rpm, and centrifugal atomization treatment with real-time monitoring at 25000rpm to generate fine droplets;

[0081] S200, based on refined droplets, performs hydrogen proportioning with a hydrogen volume ratio of 5%, maintains oxygen content feedback regulation of ≤200ppm, and performs multi-stage cooling tower pressure balance treatment with cooling towers connected in series to generate an atomized environment.

[0082] S300: From the atomization environment, the cooling rate of the ultrasonic transducer is configured in the frequency range of 20kHz to 2MHz, ultrasonic field optimization is performed using a genetic algorithm, and ultrasonic-assisted droplet solidification is used to generate rapidly cooled powder.

[0083] S400, based on rapidly cooled powder, performs intelligent algorithm optimization including support vector machine, artificial neural network and fuzzy logic control, and multi-stage cyclone separation processing with cyclone separators in series, setting a 5-20μm particle size threshold, to generate powder of the target particle size;

[0084] S500: From the target particle size powder, perform morphology detection with a sphericity threshold of not less than 95%, compare the oxygen content threshold of maintaining an oxygen content of ≤200ppm, and perform magnetization intensity testing using a vibrating sample magnetometer with a saturation magnetization intensity of not less than 1.8 Tesla to generate powder with qualified performance.

[0085] S600, based on performance-compliant powder, performs equipment deviation analysis using principal component analysis, laser power compensation to maintain surface roughness Ra≤0.1μm, and magnetic levitation gap calibration to maintain gap size within 50μm±2μm, generating an optimized system configuration.

[0086] Step S100 includes at least steps S110-S130:

[0087] S110. Obtain Fe-Si-Al alloy melt and perform melt superheat control processing based on PID algorithm to maintain superheat at 150℃ to obtain superheated melt.

[0088] The Fe-Si-Al alloy melt is the core initial material of this preparation method, specifically defined as a homogeneous liquid-phase metal fluid in a specific thermodynamic state, formed by complete melting in a vacuum induction furnace under the protection of high-purity argon gas. Its physicochemical state includes the following specific aspects:

[0089] Chemical composition: Its elemental composition strictly conforms to the preset Fe-Si-Al alloy composition design, and the mass percentage error of each element is controlled within ±0.5% to ensure the performance of the final soft magnetic powder.

[0090] Physical state: It is a high-temperature liquid phase with uniform composition, no metallurgical inclusions, no precipitated phases and low dissolved gas content. It has a superheat that is precisely controlled by a PID (Proportional-Integral-Derivative) algorithm and maintained at 150℃±10℃ to ensure that it has excellent flowability and atomization characteristics.

[0091] Environmental conditions: The entire preparation and maintenance process is carried out in an inert atmosphere with low oxygen partial pressure, protected by a high-purity protective gas (argon purity ≥99.999%) circulation system, and no oxide film is formed on the free surface of the melt.

[0092] Obtaining the Fe-Si-Al alloy melt is a prerequisite and direct input for performing step S110, "Melt Superheat Control Process".

[0093] Before performing step S110, the Fe-Si-Al alloy melt is first prepared, specifically including: providing Fe-Si-Al alloy raw materials with a specific composition ratio, the chemical composition of which meets the performance design requirements of ultrafine soft magnetic powder. The alloy raw materials are then loaded into the crucible of a vacuum induction furnace (VIF), the furnace body is sealed, and the vacuum system is activated to evacuate the furnace cavity to a vacuum level ≤10⁻²Pa to effectively remove air and moisture from the furnace. High-purity argon gas (purity ≥99.999%) is then introduced as a protective atmosphere to restore the furnace pressure and maintain it at a slightly positive pressure state, preventing oxidation of the melt during the smelting process. The electromagnetic induction heating system is started, and the alloy raw material is heated in a stepwise manner according to the preset heating curve. First, the temperature is raised to above the solidus temperature of the alloy for preheating, and then the heating continues until it is completely melted and reaches the target superheat control starting temperature. Finally, a Fe-Si-Al alloy melt with uniform composition, no pollution and in a completely liquid state is obtained, which serves as the input source for the "melt superheat control treatment" in step S110.

[0094] Specifically, Fe-Si-Al alloy melt is obtained as the input source for step S110. The Fe-Si-Al alloy is first placed in a vacuum induction furnace (VIF) for melting. The melting process includes introducing high-purity argon gas to remove oxygen from the furnace chamber and using electromagnetic induction heating to achieve uniform melting of the alloy. After melting, a high-precision thermocouple sensor is used to monitor the melt temperature in real time, and combined with numerical simulation of the melt temperature field, temperature distribution data of the melt is obtained. Further, the melt superheat control refers to the precise adjustment of the excess melt temperature relative to the liquidus temperature of the Fe-Si-Al alloy. Specifically, a PID (proportional-integral-derivative) control algorithm is used. By adjusting the power output of the induction furnace, the melt superheat is maintained at approximately 150°C to ensure good fluidity and a suitable thermodynamic state. The temperature signal collected by the thermocouple sensor is transmitted to the control unit via a data acquisition system. The control unit compares the preset superheat target value in real time and triggers a heating power adjustment command to complete closed-loop control. The superheated melt exhibits a uniform liquid phase structure, free of significant inclusions and bubbles, and possesses a stable chemical composition that meets the design requirements for Fe-Si-Al alloys. To prevent melt oxidation, the smelting environment is maintained under low oxygen partial pressure conditions, with a high-purity protective gas circulation system continuously removing oxygen. The superheated melt is used as the output field name for this step and is passed to the subsequent "Mel Injection" step in S120 for processing into a high-speed rotating centrifugal disc.

[0095] S120. Extract the melt flow from the superheated melt and inject it into a laser-textured centrifugal disk, wherein the surface roughness Ra of the centrifugal disk is ≤0.1μm and the rotation speed is 25000rpm, to generate initial droplets.

[0096] The superheated melt specifically refers to the intermediate product output after the "melt superheat control treatment" in step S110. It is specifically defined as: a Fe-Si-Al alloy liquid phase material in a specific thermodynamic state, whose temperature is precisely controlled at a superheat state of about 150°C above the alloy liquidus temperature; its physical characteristics are a pure melt with uniform composition, no obvious metallurgical inclusions and bubbles, and good fluidity; its chemical composition is stable and conforms to the preset alloy design; and its preparation and maintenance are carried out under the protection of an inert atmosphere with low oxygen partial pressure.

[0097] The melt stream specifically refers to a continuous or semi-continuous stream of liquid metal extracted from the superheated melt via a high-temperature corrosion-resistant metal pipe and guided to the laser-textured centrifugal disc injection port. This melt stream inherits all the physicochemical properties (such as composition, superheat, and purity) of the superheated melt and serves as the direct input material for performing the "laser-textured centrifugal disc injection" operation in step S120.

[0098] In step S120, the input source is the superheated melt output from S110. Specifically, the superheated melt is introduced into the inner cavity of the laser-textured centrifugal disk through a high-temperature corrosion-resistant metal pipe. The laser-textured centrifugal disk is a high-speed rotating disc-shaped device whose surface has been laser-textured, with the textured surface roughness (Ra) controlled within 0.1 μm. The texturing process uses a high-power pulsed laser beam to etch the disk surface with micron-level precision, forming a uniformly distributed micro-texture structure. The texture parameters include the width, depth, and spacing of the microgrooves, and the specific values ​​are set according to the powder particle size refinement requirements. The rotation speed of the laser-textured centrifugal disk is set to 25,000 rpm, and the rotation drive system uses magnetic levitation bearings to reduce mechanical wear and vibration, achieving high-stability rotation. Under the action of gravity and centrifugal force, the superheated melt enters the central region of the rotating centrifugal disk through the injection port. Driven by the high-speed rotational centrifugal force, the melt moves radially outward. The laser-textured surface texture promotes the shear separation effect of melt flow by increasing surface energy and microscopic disturbances, causing the melt to form initial droplets under centrifugal force. The droplet formation process includes liquid film stretching, fracture, and droplet detachment. A high-speed camera system is used for real-time morphological monitoring to capture droplet size, velocity, and distribution characteristics. The image data is analyzed using image recognition algorithms to ensure that the droplet size is uniform and within a preset range. To prevent droplet aggregation and adhesion, the surface of the centrifuge disc is coated with a high-temperature resistant, non-stick coating, and the coating thickness and material are optimized. The initial droplet is the output field name of this step, which is passed to the subsequent "droplet refinement" step in S130 for high-speed centrifugal atomization processing.

[0099] S130. The initial droplets are centrifuged and atomized at 25,000 rpm, and real-time monitoring is performed using a high-speed imaging system and a laser particle size analyzer to generate refined droplets.

[0100] In step S130, the input source is the initial droplet output from S120. Specifically, the initial droplet is subjected to extremely high centrifugal acceleration under the high-speed rotation of the laser texturing centrifuge disc at 25000 rpm, resulting in further droplet refinement. The droplet refinement process includes droplet stretching, breakage, and secondary atomization. Centrifugal force accelerates the droplet radially, and the droplet breaks into smaller, refined droplets under the combined action of high-speed shear force and surface tension. The droplet refinement is monitored in real time using a high-speed imaging system and a laser particle size analyzer to measure the droplet size distribution, morphology, and velocity, ensuring that the refined droplet size is uniform and meets the requirements for ultrafine powder preparation. During the droplet refinement process, the system adjusts the centrifuge disc rotation speed and melt flow rate through closed-loop control, combined with laser texturing surface parameters, to achieve precise control of the droplet size. Abnormal situations during the droplet refinement process, such as droplet aggregation, splashing, and irregular breakage, are captured by sensor signals and fed back to the control unit, which automatically adjusts the rotation speed and flow rate parameters. The refined droplet is the output field name of this step, which serves as the input for the subsequent S210 "Atmosphere Environment Configuration" step, allowing for the control of the atomization environment under an argon-hydrogen mixed atmosphere. Steps S110 to S130 form a complete laser-textured centrifugal atomization chain, ensuring the coordinated operation of the melt state, injection process, and droplet refinement, supporting the implementation of subsequent atmosphere control and powder preparation processes.

[0101] In another embodiment, in step S110, the input source is the Fe-Si-Al alloy melt inside a vacuum induction furnace, and the melt temperature distribution data is acquired through a high-precision thermocouple sensor. Formula ① uses an improved PID control algorithm to establish a superheat control model:

[0102]

[0103] in: The superheat adjustment amount for the kth iteration is derived from the weighted average of the numerical simulation data of the temperature field and the measured values ​​of the thermocouple. The superheat error term is generated from a thermocouple sensor. These are control coefficients optimized based on historical process data; The summation index indicates the number of iterations;

[0104] This represents the total number of iterations. To control the cycle time step;

[0105] The adjustment value calculated by formula ① is output through the magnetic levitation bearing power adjustment module, forming a closed-loop control link. Formula ② defines the melt flow performance index:

[0106] in: It is a melt flow performance indicator used to verify the thermodynamic state of the melt before injection; The dynamic viscosity of the melt is determined by an alloy composition database. The temperature gradient is two-dimensional and is derived from numerical simulation of the temperature field. The density of the melt; These are the surface roughness parameters for laser texturing.

[0107] This indicator is used to verify the thermodynamic state before melt injection, and power compensation is triggered when Φ>1.2. The output field name "Superheated Melt" in this step is consumed by "Mel Injection" in S120.

[0108] Further, in step S120, the "superheated melt" is fed into a laser-textured centrifuge disc via a high-temperature pipeline. Equation ③ constructs an improved Navier-Stokes equation to describe the melt flow:

[0109]

[0110] in: The radial velocity component is the data source, which is the melt motion trajectory captured by a high-speed camera system. :time; ∇ represents the local acceleration term for the radial velocity of the melt; ∇ is the Nabla operator; The convective acceleration term for the radial velocity of the melt represents the acceleration generated when a fluid element moves from one point to another (a region with different velocities); For a fluid infinitesimal element, the mass derivative or total acceleration is given. This is the pressure gradient force on a unit volume fluid element. The negative sign indicates that the direction of the force is opposite to the direction of pressure increase (i.e., from high pressure to low pressure). The dynamic viscosity of the melt; For the Laplace operator; It represents the viscous force or diffusion force experienced by a unit volume fluid element, indicating the internal friction force generated due to the non-uniformity of fluid viscosity and velocity (shear action); For centrifugal force; ω = 25000 rpm is the angular velocity of the centrifugal disk; r is the radial distance of the melt element from the center of the disk;

[0111] Equation ③, compared to the standard Navier-Stokes equations, is "improved" by adding a volume force term in a rotating coordinate system. To accurately describe the melt flow during centrifugal atomization.

[0112] The ΔT value from formula ① is used to calculate the μ parameter update. Formula ④ defines the droplet size prediction model:

[0113]

[0114] Wherein: is the droplet diameter; σ is the surface tension of the melt, and the data source is the melt physical property database; Ra is the surface roughness parameter of the laser texture;

[0115] When the deviation between the predicted droplet size output by this model and the high-speed camera data is ≤ 5%, it is determined to be qualified. The output field name "initial droplet" in this step is consumed by "droplet refinement" in S130.

[0116] Furthermore, in step S130, the input "initial droplet" undergoes centrifugal atomization at 25,000 rpm. Formula ⑤ uses an improved Reynolds number to determine the secondary atomization effect:

[0117]

[0118] Wherein: is the Reynolds number, used to determine the secondary atomization effect; is the droplet diameter, and the data source is the laser particle size analyzer; is the tangential velocity of the droplet; is the dynamic viscosity of the melt;

[0119] When > 200, droplet secondary breakup is triggered. Formula ⑥ defines the atomization efficiency optimization function:

[0120] Wherein: is the atomization efficiency; is the residence time of the droplet in the atomization zone, inversely calculated from the gas flow velocity; is the critical breakup time threshold;

[0121] The output field name "refined droplet" in this step is consumed by "atmosphere environment configuration" in S210.

[0122] In S100, the formula chain forms a complete closed-loop control and forward prediction process:

[0123] (1) Real-time control: Formula ① (PID control) affects the physical property parameter (μ) by adjusting the temperature. Specifically, the change in superheat (temperature) will significantly affect the viscosity μ of the melt. Therefore, the output by the PID controller is an adjustment instruction, which changes the melt state through an actuator (magnetic levitation bearing), and its final effect is reflected in the change of the melt physical property parameter - viscosity μ. When calculating the melt flow, this updated μ value must be used in Formula ③ to more accurately simulate the actual flow situation. This is a key temperature-physical property-flow field coupling relationship.

[0124] (2) Quality Verification: Formula ② (Φ index) acts as a gating mechanism to ensure melt quality. Specifically, the Φ value is a verification index, not a variable directly input into the next formula. When Φ > 1.2, the system determines that the melt has insufficient fluidity and needs to trigger the power compensation mechanism. This compensation action will be fed back to the execution end of Formula ① (magnetic levitation bearing power adjustment module), optimizing the melt state by changing the energy input, thereby indirectly affecting all subsequent calculations. This is a quality gating node.

[0125] (3) Flow simulation: Formula ③ uses updated physical property parameters to simulate the flow field. Specifically, the key parameters in Formula ④ The tangential velocity (v) comes directly from the flow field described by Equation ③. Equation ③ simulates the acceleration process of the melt on the centrifugal disk, thus determining the velocity of the melt when it breaks at the edge of the disk. This velocity is used to calculate the droplet size. One of the core inputs.

[0126] (4) Size Prediction: Formula ④ uses the flow field results to predict the initial droplet size. Specifically, Formula ④ is a theoretical prediction model, and its output is... This represents the droplet size under ideal conditions. This value is fed into equation ⑤ as the calculation of the Reynolds number. When required Parameters. Simultaneously, the system compares the actual values ​​measured by a high-speed camera or laser particle size analyzer with the predicted values ​​from formula ④ (a deviation ≤ 5% is considered acceptable), thus achieving real-time verification and calibration of the model.

[0127] (5) Process Judgment: Formula ⑤ is used to calculate the Reynolds number and determine whether secondary crushing has occurred. Specifically, It is a key dimensionless number for determining the dynamic behavior of droplets (whether secondary breakup will occur). It is primarily used for process logic decisions. Meanwhile, the atomization efficiency η is highly dependent on the degree of fluid turbulence, and... It is the core parameter characterizing the flow state (laminar or turbulent), and therefore it naturally becomes the main variable in the efficiency optimization function.

[0128] (6) Efficiency evaluation: Formula ⑥ evaluates the efficiency of the entire atomization process based on the Reynolds number.

[0129] The link clearly demonstrates how "overheat control", "flow simulation", "atomization prediction" and "efficiency evaluation" are linked together to form a highly coupled and intelligent preparation process.

[0130] The technical effect of this section is that by using closed-loop temperature control, melt flow modeling, and atomization parameter optimization, the droplet size can be precisely controlled, providing input materials that meet the particle size requirements for argon-hydrogen atmosphere control.

[0131] Step S200 includes at least steps S210-S230:

[0132] S210. Based on the refined droplets, hydrogen proportioning treatment is performed, wherein the hydrogen volume ratio is 5%, to obtain the initial atmosphere.

[0133] The refined droplets output from step S130 are used as input. Specifically, a pre-proportioned argon-hydrogen mixture is supplied from a high-purity argon-hydrogen gas supply system. The hydrogen volume ratio of the argon-hydrogen mixture is set to 5%, which is achieved through joint adjustment by a high-precision gas flow meter and a mass flow controller (MFC). The mixture first passes through a multi-stage purification and filtration device to remove moisture, oxygen, and other impurities, ensuring that the gas purity meets or exceeds industrial-grade standards. The gas flow meter collects the flow signals of argon and hydrogen in real time, and the data is transmitted to the central control unit via an industrial fieldbus. The central control unit uses a digital proportional-integral-derivative (PID) control algorithm to perform closed-loop regulation of the argon and hydrogen flow rates, maintaining the hydrogen volume ratio stable within the range of 5% ± 0.1%. To prevent fluctuations in atmosphere composition, the system is equipped with dual redundant flow meters and an automatic switching mechanism; abnormal data automatically triggers an alarm and safety interlock. The mixture is then transported to the atomization reaction chamber via a corrosion-resistant pipeline, entering the atomization environment. The initial atmosphere is used as the output field name of this step and is passed to the subsequent "pressure regulation" step of S220 for pressure balance control and oxygen content monitoring.

[0134] S220: Extract the oxygen content signal from the initial atmosphere, perform real-time feedback adjustment, maintain the oxygen content ≤200ppm, and generate a stable atmosphere;

[0135] Further, the input source is the initial atmosphere output from step S210. Specifically, the initial atmosphere is used to acquire oxygen content signals through a multi-point oxygen sensor array. The oxygen sensors employ high-sensitivity electrochemical sensing elements, installed at different locations within the atomization reaction chamber to achieve real-time monitoring of local and overall oxygen content. The sensor signals are acquired by an analog-to-digital converter (ADC) and then transmitted to the atmosphere control unit. Based on real-time oxygen content data and a preset oxygen content threshold (≤200ppm), the atmosphere control unit uses a fast-response PID feedback control algorithm to adjust the argon and hydrogen flow ratio, dynamically adjusting the mixed gas composition. The system has an oxygen content abnormality alarm mechanism; when the oxygen content exceeds the set range, it automatically triggers a gas flow adjustment command and records the abnormal event log for subsequent analysis. To prevent powder oxidation caused by oxygen content fluctuations, the atmosphere control unit also coordinates with the exhaust system to adjust the chamber pressure, maintaining a negative pressure state to prevent outside air infiltration. The stable atmosphere is the output field name of this step, which is passed to the subsequent "environment maintenance" step S230 for multi-stage cooling tower pressure balancing and stable control of the atomization environment.

[0136] S230. Multi-stage cooling tower pressure balancing treatment is performed on a stable atmosphere, wherein the cooling towers are set in series to generate an atomized environment;

[0137] Specifically, the stable atmosphere output from step S220 is introduced into a multi-stage cooling tower integrated system for pressure balancing. The multi-stage cooling tower comprises multiple cooling units connected in series, each equipped with independent temperature and pressure sensors to achieve precise control of the atmosphere temperature gradient and pressure distribution. The system regulates airflow speed through airflow regulating valves and fan drives to ensure uniform atmosphere flow within each cooling unit. The pressure control unit, based on multi-point pressure sensor data, employs a closed-loop control algorithm to adjust the inlet and outlet pressures, maintaining stable pressure within the atomization reaction chamber and preventing atmosphere turbulence that could lead to droplet oxidation or decreased atomization efficiency. The cooling tower's inner wall is made of corrosion-resistant material and equipped with a high-efficiency heat exchanger to ensure the atmosphere temperature gradient meets the droplet condensation requirements. The system includes an atmosphere flow and temperature anomaly detection module; in case of anomalies, it automatically adjusts airflow parameters and records the operating status. The atomization environment, as the output field name of this step, is passed to the subsequent "Cooling Medium Selection" step in S310 for configuring the ultrasonic-assisted rapid cooling rate. Steps S210 to S230 form a complete argon-hydrogen atmosphere control link, enabling coordinated adjustment of gas ratio, oxygen content, and pressure and temperature environment, supporting the implementation of subsequent ultrasonic-assisted rapid cooling and powder preparation processes.

[0138] In another embodiment, in step S210, the input source is an argon-hydrogen mixed gas supply system, and formula ① establishes the gas proportioning control equation:

[0139]

[0140] in: The hydrogen volumetric flow rate at time t is the data source, which is a mass flow controller. To set the pressure value; This refers to the total volume of the gas container; It is the ideal gas constant; The gas temperature; The pressure deviation is derived from the pressure sensor array;

[0141] Formula ② defines the purity index of the mixed gas:

[0142] in: This is an indicator of the purity of the mixed gas; The argon purity is measured from a gas analyzer. This refers to the hydrogen concentration. The oxygen content threshold; Nitrogen concentration;

[0143] The output field name "Initial Atmosphere" of this step is consumed by "Pressure Regulation" of S220.

[0144] Furthermore, in step S220, the "initial atmosphere" is input and monitored by the oxygen sensor array. Formula ③ constructs the oxygen content feedback equation:

[0145]

[0146] in: For the first The control output value of the next iteration; The proportionality coefficient for controlling oxygen content; For the first The oxygen content error in the next iteration; The integral time constant; The sampling period of the controller;

[0147] Formula ④ defines the pressure-oxygen content coupling function:

[0148] in: The pressure-oxygen content coupling coefficient; To monitor pressure values ​​in real time; Current oxygen content; To set the pressure value; Set the oxygen content value;

[0149] When Ψ>1.1, the exhaust system linkage is triggered. The output field name "Stable Atmosphere" in this step is consumed by "Environmental Maintenance" of S230.

[0150] Furthermore, in step S230, a "stable atmosphere" is input into the multi-stage cooling tower. Formula ⑤ establishes the pressure balance equation:

[0151]

[0152] in: The summation index indicates the number of cooling tower stages; The pressure drop of the nth stage cooling tower; 2 represents the gas density; The flow rate of the nth stage cooling tower; The flow coefficient of the nth stage cooling tower; This represents the total gas flow rate; Let n be the cross-sectional area of ​​the nth stage cooling tower;

[0153] Formula 6 defines the temperature-pressure optimization function:

[0154] in: For temperature-pressure optimization function; This refers to the inlet temperature of the cooling tower. This refers to the inlet pressure of the cooling tower. This refers to the outlet temperature of the cooling tower. This refers to the outlet pressure of the cooling tower. The gas Reynolds number is calculated from gas flow parameters (such as velocity, viscosity, and density).

[0155] The output field name "Atomization Environment" of this step is consumed by the "Cooling Medium Selection" of S310.

[0156] Detailed explanation of each link point in step S200:

[0157] 1. Formula ① → Overall Environment: Provides basic input

[0158] Linkage: Hydrogen volumetric flow rate calculated using formula ① It is the most direct operational quantity for creating the "initial atmosphere".

[0159] Detailed explanation: The output of this formula directly drives the actuator (mass flow controller, MFC) to inject a specific ratio of argon-hydrogen mixed gas into the system, creating the initial, formula-configured gas environment for all subsequent steps. It is the fundamental input action for the entire S200 stage.

[0160] 2. Formula ② → Process Logic: Quality Gating

[0161] Linkage: Gas purity index calculated by formula ② It is a verification indicator used to determine whether the "initial atmosphere" is qualified.

[0162] Detailed explanation: The Γ value directly reflects the purity of the mixed gas. This is a quality gate node. If the Γ value does not meet the standard (e.g., due to impure gas source), an alarm or feedback is required to adjust the gas supply formula until it meets the standard before proceeding to the "pressure regulation" step of S220. This ensures that subsequent processes are carried out in a high-purity base atmosphere.

[0163] 3. Formula ③ → Formula ④: Core Feedback and Control

[0164] Linkage: The real-time control result of formula ③ directly affects the real-time oxygen content in formula ④. .

[0165] Detailed explanation: Formula ③ (PI controller) runs continuously, aiming to stabilize the oxygen content at a set value (200ppm) by adjusting gas flow rate and other means. Therefore, the CO2(k) value at any given time reflects the control effect of Formula ③. This real-time oxygen content value is directly used as the core input parameter in the coupling function Ψ of Formula ④ for calculation.

[0166] 4. Formula ④ → Process Logic: Safety Interlock

[0167] Linkage: Pressure-oxygen content coupling coefficient calculated by formula ④ It is a safety monitoring indicator used to trigger interlocking protection actions.

[0168] Detailed explanation: The Ψ value comprehensively reflects the abnormal state of pressure and oxygen content. When Ψ > 1.1, it indicates that the system may have a leak (pressure change) or contamination (increased oxygen content). This will trigger the exhaust system linkage, a safety interlock action to prevent unqualified atmosphere from entering the subsequent cooling and atomization stages, ensuring the safety and stability of the entire system.

[0169] 5. Formula ⑤ → Formula ⑥: System state transfer

[0170] Linkage: The pressure balance state of the multi-stage cooling tower described by formula ⑤ is the foundation of the entire system, and the resulting outlet pressure outlet temperature And the implicit parameters such as gas flow rate are the gas Reynolds number in formula ⑥. The foundation.

[0171] Detailed Explanation: Formula ⑤ ensures stable gas flow within the cooling tower system. The physical parameters (pressure, temperature, flow) exhibited by this stable flow state are the prerequisite for optimization calculations in Formula ⑥. In Formula ⑥... It needs to be calculated based on factors such as airflow velocity and cooling tower structural dimensions, all of which depend on the equilibrium state described by formula ⑤.

[0172] 6. Formula ⑥ → Process Optimization: Guiding Operation

[0173] Linkage: Temperature-pressure optimization function calculated by formula ⑥ It is a guiding indicator used to optimize the operating parameters of cooling towers.

[0174] Detailed explanation: The Ω value itself does not directly participate in control, but provides an evaluation value for operators or advanced optimization algorithms to determine the current operating efficiency of the cooling tower (such as heat exchange efficiency, pressure drop loss, etc.), and guides how to adjust parameters such as airflow distribution and cooling intensity to achieve optimal system energy consumption and performance.

[0175] The formula chain in the S200 section forms a complete closed loop of "formulation-control-monitoring-optimization":

[0176] Formula execution: Inject gas according to the set ratio using formula ① to establish the initial environment.

[0177] Quality inspection: Formula ② verifies the purity of the initial environment as a gate control.

[0178] Fine-tuning: Formula ③ uses closed-loop feedback to dynamically and precisely control oxygen content.

[0179] Safety monitoring: Formula ④ couples the monitoring of pressure and oxygen content to trigger a safety interlock.

[0180] System balance: Formula ⑤ ensures the pressure stability of the gas delivery system.

[0181] Operational optimization: Formula 6 evaluates system efficiency and guides the optimization of operation points.

[0182] This process ensures a clean, stable, low-oxygen, and highly efficient gaseous environment for the atomization process, which is crucial for guaranteeing that the final powder chemical composition meets the standards.

[0183] This section summarizes the technical effects: By precisely controlling the gas ratio, dynamically compensating for oxygen content, and balancing multiple pressure levels, a stable low-oxygen atomization environment is constructed, supporting the stable implementation of subsequent rapid cooling processes.

[0184] Step S300 includes at least steps S310-S330:

[0185] S310. Obtain the atomization environment and perform cooling rate configuration processing, wherein the frequency range of the ultrasonic transducer array is 20kHz to 2MHz, and obtain the ultrasonic field parameters.

[0186] The atomization environment output from step S230 is used as input. Specifically, the atomization environment is introduced into the cooling medium configuration unit within the ultrasonic-assisted rapid cooling device through a corrosion-resistant gas pipeline. The cooling medium consists of a circulating fluid of a mixture of high-purity argon and hydrogen. The fluid temperature is monitored in real time by multi-point thermocouple sensors, and the sensor data is transmitted to the central control unit via a data acquisition system. The central control unit adjusts the cooling medium flow rate through a frequency converter-driven pump based on a preset cooling rate target, achieving temperature gradient control of the cooling medium. The ultrasonic transducer array is installed on both sides of the cooling medium flow path. The transducer array consists of multiple piezoelectric ceramic elements, and the element spacing and arrangement are optimized through finite element analysis to ensure uniform distribution of ultrasonic energy in the cooling medium. The drive signal for the ultrasonic transducer array is output from a high-frequency power amplifier. The frequency and power parameters are set by a digital signal generator, with a frequency range covering 20kHz to 2MHz, and the power range dynamically adjusted according to cooling requirements. The cooling rate configuration involves adjusting the excitation frequency and power of the ultrasonic transducer array to generate a high-intensity ultrasonic field within the cooling medium, promoting micro-disturbances and acoustic cavitation on the droplet surface, thereby enhancing the droplet's heat transfer efficiency. The system is equipped with a multi-channel ultrasonic power monitoring module to collect the transducer input power and reflected power in real time, transmitting the data to the control unit for feedback adjustment. The ultrasonic field parameters are used as the output field name in this step and passed to the subsequent "Power Adjustment" step in S320 for dynamic adjustment of the ultrasonic power.

[0187] S320. Extract frequency features from ultrasonic field parameters, perform standing wave field optimization processing, use genetic algorithm to optimize parameters, and generate directional cooling field;

[0188] Further, frequency characteristic signals are extracted from the ultrasonic field parameters output in step S310. Specifically, the frequency characteristics include the main harmonic frequency, the second harmonic frequency, and the harmonic amplitude distribution. The data is acquired by an ultrasonic field spectrum analyzer with a sampling rate of over 10MHz. The frequency characteristic signals undergo Fast Fourier Transform (FFT) by a digital signal processor to generate a spectrum diagram. The spectrum diagram data is input into the standing wave field optimization module. The standing wave field optimization module, based on a finite element acoustic simulation model and combined with the spectrum data, adjusts the excitation parameters of the ultrasonic transducer array to optimize the formation of standing waves in the cooling medium and enhance the uniformity of sound pressure distribution within the cooling medium. The optimization process includes adjusting the transducer phase difference, amplitude ratio, and fine-tuning the excitation frequency. A genetic algorithm is used for multi-parameter iterative optimization to ensure the formation of a stable directional cooling field. The directional cooling field is defined as an ultrasonic sound field with a specific spatial distribution and directionality formed in the cooling medium, whose sound pressure peak value and node position are precisely controlled to promote rapid heat conduction and uniform cooling during droplet solidification. The system monitors the sound pressure distribution in real time using an ultrasonic field sensor array. The monitoring data is fed back to the control unit to form a closed-loop control, automatically adjusting the transducer drive signal to maintain the stability of the directional cooling field. The directional cooling field, as the output field name of this step, is passed to the subsequent "rapid cooling execution" step S330 for droplet solidification.

[0189] S330. Perform droplet solidification treatment on the directional cooling field, and perform ultrasonic-assisted droplet solidification treatment to generate rapidly cooled powder.

[0190] In conjunction with the directional cooling field output from step S320, specifically, the refined droplets output from step S130 are injected into the cooling medium flow region. Under the influence of the directional cooling field, the droplets undergo an ultrasonic-assisted rapid cooling process. The droplet solidification treatment includes acoustic cavitation on the droplet surface, enhanced micro-turbulence, and accelerated heat conduction. Under the action of ultrasound, the droplet surface forms fine oscillations, promoting rapid heat dissipation from the droplet interior. The droplet solidification process is monitored in real-time by a high-speed camera system and a laser scattering particle size analyzer, capturing droplet morphology changes and solidification time, and the data is transmitted to the process control unit. The process control unit adjusts the ultrasonic power and frequency based on the monitoring data, dynamically regulating the directional cooling field parameters to ensure the continuity and uniformity of the droplet solidification process. After droplet solidification, rapidly cooled powder is formed, with particle size and morphology meeting the preset requirements for ultrafine soft magnetic powder. This rapidly cooled powder is used as the output field name for this step and is passed to the subsequent "particle size screening" step in S410 for cyclone classification. Steps S310 to S330 form a complete ultrasonic-assisted rapid cooling chain, achieving efficient solidification of droplets and optimization of powder morphology.

[0191] Step S400 includes at least steps S410-S430:

[0192] S410. Based on rapidly cooled powder, a 5-20μm threshold setting process is performed to obtain classification parameters;

[0193] The quenched powder output from step S330 is used as input. Specifically, the quenched powder is introduced into a cyclone classifier for classification. The cyclone classifier is an integrated cyclone separator comprising multiple concentrically arranged cyclone separation units. Each unit is equipped with an independent inlet, outlet, and powder collection port. The cyclone classifier achieves particle size classification by adjusting the airflow path and rotation speed. Specifically, the classification parameters include a classification threshold, defined as the upper and lower limits of the powder particle size, set between 5 and 20 micrometers (μm). The classification threshold is set through the parameter input interface of the central control unit. A particle size sensor array is installed inside the cyclone classifier, employing laser scattering particle size analysis technology to collect real-time particle size distribution data of the powder entering the classifier. The sensor signals are transmitted to the cyclone classification control module after analog-to-digital conversion. The control module executes a dynamic airflow adjustment strategy based on the preset classification threshold, controlling the separation efficiency inside the cyclone by adjusting the airflow speed at the inlet and the opening of the exhaust port of the cyclone separation unit. The process of setting the grading parameters includes inputting, verifying, and storing the grading thresholds. The system provides a historical data recording function for the grading thresholds for traceability and optimization. In abnormal situations, such as when the particle size distribution deviates from the preset range, the system automatically triggers an alarm and records an abnormal event log for subsequent analysis. The grading parameters, as the output field names of this step, are passed to the subsequent "Airflow Adjustment" step in S420 for dynamic airflow adjustment based on the grading parameters.

[0194] S420. Extract particle size distribution data from the grading parameters and perform intelligent algorithm optimization processing. The intelligent algorithm includes support vector machine, artificial neural network and fuzzy logic control to generate dynamic adjustment instructions.

[0195] Based on the classification parameters output in step S410, specifically, powder particle size distribution data is extracted from these parameters and input into the intelligent algorithm optimization module. The intelligent algorithm is based on a machine learning model, incorporating a fusion of multiple algorithms including Support Vector Machine (SVM), Artificial Neural Network (ANN), and Fuzzy Logic Control. Algorithm inputs include real-time particle size distribution data, historical classification efficiency data, and environmental parameters. Data preprocessing and normalization are performed using multi-dimensional features to eliminate noise. The intelligent algorithm calculates the deviation index by comparing the current particle size distribution with the target classification threshold and adjusts the airflow parameters of the cyclone classifier using an adaptive optimization strategy. Specific adjustments include controlling the inlet velocity, the exhaust port opening of the cyclone unit, and the inner wall coating temperature, adjusting the rotation speed and flow pattern of the airflow to influence the inertial separation effect of the powder. The intelligent algorithm module continuously collects the operating status and particle size distribution changes of the cyclone classifier through a closed-loop feedback mechanism, updates the optimization model parameters in real time, and generates dynamic adjustment commands. The system has a safety threshold for adjustment commands to prevent over-adjustment from causing equipment malfunctions. All adjustment behaviors are recorded in the system log for subsequent tracking and analysis. The dynamic adjustment command is passed as the output field name of this step to the subsequent "tiered execution" step of S430 for the specific implementation of multi-stage cyclone separation.

[0196] S430: Perform multi-stage cyclone separation processing on the dynamic adjustment command, wherein the cyclone separators are set in series to generate powder with the target particle size.

[0197] In conjunction with the dynamic adjustment command output in step S420, specifically, the dynamic adjustment command is processed using multi-stage cyclone separation. The multi-stage cyclone separation system includes multiple cyclone separator units connected in series, each unit equipped with an independent airflow inlet, airflow outlet, and powder collection device. According to the dynamic adjustment command, the system adjusts the airflow velocity and separation threshold of each cyclone separator to precisely control the distribution and settling of powder in each stage of the cyclone separator. Within the cyclone separator, the powder is subjected to centrifugal force; larger particles are thrown towards the separator wall due to inertia and fall along the wall to the collection port, while smaller particles are discharged with the airflow to the outlet. The system uses a high-sensitivity particle mass sensor to monitor the powder mass flow rate at each collection port, and the data is transmitted to the central processing unit in real time. The central processing unit compares the sensor data with the dynamic adjustment command and automatically adjusts the airflow parameters to achieve closed-loop control of the separation efficiency. To prevent powder agglomeration and blockage within the cyclone separator, the system is equipped with an automatic backflush valve to periodically release accumulated powder and prevent abnormal equipment operation. The target particle size powder is defined as a powder aggregate with a particle size in the range of 5 to 20 micrometers and a morphology that meets the requirements of ultrafine soft magnetic powder. The powder is transported to the subsequent performance testing unit through a sealed conveying pipeline. The target particle size powder is used as the output field name of this step and is passed to the subsequent "Magnetic Performance Test" step of S510 for performance testing such as sphericity and oxygen content.

[0198] Step S500 includes at least steps S510-S530:

[0199] S510. Based on the target particle size powder, a threshold setting process is performed, wherein the sphericity threshold is not less than 95%, and the morphology detection parameters are obtained.

[0200] The target particle size powder output from step S430 is used as input. Specifically, the target particle size powder is introduced into a sphericity detector for morphology detection. The sphericity detector employs a high-resolution scanning electron microscope (SEM) combined with an image processing system, enabling precise measurement of the three-dimensional morphology of powder particles. The detector is equipped with an automatic sample delivery device, allowing for continuous acquisition of powder samples, ensuring high throughput and high repeatability in the detection process. Sphericity is defined as the ratio of the actual surface area of ​​a powder particle to the surface area of ​​an ideal sphere corresponding to its volume. The threshold is set to be no less than 95%, and this threshold is input through the control software parameter interface and stored in the morphology detection parameter database. Specifically, the powder sample is uniformly distributed on the sample stage through a sealed delivery pipe. The sample stage uses a combination of electric rotation and linear movement to achieve multi-angle, multi-region imaging acquisition. During imaging, the SEM system automatically adjusts the accelerating voltage and working distance to ensure that the image contrast and resolution meet the measurement requirements. The acquired image data is used for particle edge extraction and three-dimensional reconstruction using an image recognition algorithm to calculate the surface area and volume of each particle, and then calculate the sphericity. The system employs a batch processing mode to perform statistical analysis on tens of thousands of particles, generating sphericity distribution curves. During the detection process, the system monitors the sample delivery rate and image quality in real time, automatically discarding abnormal data and triggering alarms, with relevant information recorded in the detection log. The morphology detection parameters are used as output field names for this step and passed to the subsequent "Sample Acquisition" step in S520 for oxygen content data extraction and quality assessment.

[0201] S520. Extract oxygen content data from morphology detection parameters, perform 200ppm threshold comparison processing, maintain oxygen content ≤200ppm, and generate a quality assessment report.

[0202] Specifically, based on the morphology detection parameters output in step S510, oxygen content data of powder particles is extracted from these parameters for quality assessment. The oxygen content data is acquired by an energy dispersive spectrometer (EDS) integrated into a sphericity analyzer system, enabling qualitative and quantitative elemental analysis of the powder particle surface and near-surface layers. During morphology detection, the sample is automatically transferred to the EDS detection area, where an electron beam excites the powder sample, and the detector captures characteristic X-ray signals. The oxygen content is defined as the mass fraction of oxygen in the powder, with a threshold set to no more than 200 ppm (parts per million). This threshold is set and stored by the quality assessment software. The system automatically performs background subtraction, peak identification, and correction on the acquired spectral data to obtain accurate oxygen content values. The quality assessment module, combined with the morphology detection parameters, performs threshold comparison on the oxygen content data to determine the degree of oxidation of the powder. This module supports multi-point sampling and statistical analysis, and can identify the distribution characteristics and anomalies of oxygen content. Abnormal oxygen content data triggers an automatic alarm mechanism and records relevant detection data and environmental parameters for traceability. The quality assessment report is automatically generated by the system and includes statistical results of oxygen content, particle size distribution, sphericity distribution, and related charts. The report format conforms to industry standards and is transmitted to the central control unit via a data interface. This quality assessment report, as an output field name for this step, is passed to the subsequent "parameter feedback" step in S530 for magnetization strength testing and performance evaluation.

[0203] S530. The quality assessment report is processed by magnetization intensity test, using a vibrating sample magnetometer with a saturation magnetization intensity of not less than 1.8 Tesla, to generate powder with qualified performance.

[0204] Based on the quality assessment report output in step S520, specifically, the quality assessment report is processed by magnetization testing to generate performance-compliant powder. The magnetization test uses a vibrating sample magnetometer (VSM) to measure the saturation magnetization of the powder by measuring the change in magnetic flux caused by the vibration of the sample in an alternating magnetic field. Performance-compliant powder is defined as powder with a saturation magnetization of 1.8 Tesla or higher; this threshold is set and stored by the performance testing software parameters. Samples are extracted from the qualified batches shown in the quality assessment report using an automatic sampling device and placed in the VSM sample tank. During the test, the system automatically adjusts the magnetic field strength, records the magnetization curve, and calculates key performance indicators including coercivity, remanence, and saturation magnetization. The data acquisition module monitors the test status in real time; abnormal waveforms trigger warnings and automatic retesting; all test data is stored in the performance database. The performance analysis module, combined with morphology and oxygen content data, comprehensively evaluates the magnetic properties of the powder and generates a performance compliance confirmation document. This document contains the powder's magnetic properties, testing conditions, and batch information for subsequent process adjustments and quality control. Powders meeting performance standards are listed as the output field name for this step and passed to the subsequent "Calibration Baseline" step in S610, supporting system parameter calibration and process optimization. Steps S510 to S530 form a complete powder performance testing chain, enabling comprehensive evaluation of powder morphology, composition, and magnetic properties, supporting closed-loop quality management of the preparation system.

[0205] Step S600 includes at least steps S610-S630:

[0206] S610. Based on the performance-compliant powder, perform equipment deviation analysis and use principal component analysis algorithm to obtain a calibration requirement list.

[0207] Specifically, the performance-compliant powder is used as input, and the output of the performance-compliant powder is from step S530, including multi-dimensional quality indicators such as the powder's magnetic properties, particle size distribution, sphericity, and oxygen content. The historical production data includes, but is not limited to, records of process parameters for each batch of powder preparation, equipment operation logs, environmental monitoring data, and quality inspection results. The data acquisition system synchronously acquires the historical production data from the production database and equipment control system via an industrial Ethernet interface. The historical production data undergoes data cleaning through a preprocessing module, including outlier detection, missing value completion, and time series alignment, ensuring data quality meets the requirements of subsequent analysis. The data analysis unit uses statistical analysis and machine learning algorithms to perform equipment deviation analysis on the historical production data. Specifically, Principal Component Analysis (PCA) is used to reduce the dimensionality of multi-dimensional process parameters and identify key influencing factors; combined with a time series anomaly detection algorithm, deviation patterns and trend changes during equipment operation are identified. The equipment deviation analysis results are presented through a visualization interface, including deviation amplitude, deviation time distribution, and possible fault warnings. The system further automatically generates a calibration requirement list based on preset deviation thresholds, specifying the equipment parameters to be adjusted, the adjustment range, and the priority. The calibration requirement list is stored in a structured data format, supporting subsequent retrieval and version management. Abnormal events and calibration requirements are recorded in the production log database for quality traceability and auditing. The calibration requirement list, as the output field name of this step, is passed to the subsequent "Parameter Adjustment" step in S620 for updating laser power and system parameters.

[0208] S620. Extract laser power data from the calibration requirement list, perform texturing accuracy compensation processing to maintain surface roughness Ra≤0.1μm, and generate updated laser parameters.

[0209] Specifically, the laser power-related data in the calibration requirement list is the primary processing target. The calibration requirement list includes the laser power deviation range, textured surface quality feedback, and historical laser parameter configuration records. The laser power data is extracted through a data parsing module and input into the textured surface accuracy compensation processing unit. The textured surface accuracy compensation processing employs a feedback control-based algorithm, combining a laser power adjustment model with a textured surface roughness (Ra) control target to dynamically calculate the compensation value of the laser output power. Specifically, the laser power adjustment model includes parameters such as laser current, voltage, pulse frequency, and pulse width, and establishes a mapping relationship between power and textured effect through multivariate regression analysis. The target value for textured surface roughness is set to Ra≤0.1μm. The compensation algorithm adjusts the laser drive signal according to the deviation amplitude, regulating the energy density and scanning path of the pulsed laser. The system uses a high-precision power sensor to collect the laser output power in real time. The feedback signal is filtered by a digital signal processor and then input into the compensation algorithm to form a closed-loop control. The textured surface accuracy compensation processing unit also incorporates auxiliary parameters such as ambient temperature, laser cooling status, and optical component cleanliness to comprehensively adjust the laser power output. The compensated laser parameters are formatted and converted by the parameter update module to generate an update command conforming to the laser controller interface protocol. The update command includes the laser power setting, laser beam scanning speed, and path parameters, and is stored in the updated laser parameter field. The system employs a multi-level safety verification mechanism to prevent laser parameters from exceeding the device's capacity; in case of an anomaly, it automatically reverts to historical stable parameters and triggers an alarm. The updated laser parameters are used as the output field name for this step and are passed to the subsequent "System Reconfiguration" step in S630 for overall system configuration optimization.

[0210] S630. Perform magnetic levitation gap calibration on the updated laser parameters to maintain the gap size at 50μm±2μm and generate an optimized system configuration.

[0211] Specifically, the updated laser parameters are used as the core input for the magnetic levitation gap calibration process. This calibration process controls the gap size of the magnetic levitation centrifugal atomizer, setting it to 50 μm, which is a key process parameter affecting droplet formation and powder particle size distribution. The system uses a high-precision laser ranging sensor array to measure the actual gap between the magnetic levitation centrifugal disk and the atomizer structure in real time. The ranging data is transmitted to the calibration control unit after analog-to-digital conversion. Based on the updated laser parameters and the ranging data, the calibration control unit uses a Model Predictive Control (MPC) algorithm to adjust the gap. Specifically, the MPC algorithm establishes a magnetic levitation force field model and a mechanical structure response model to predict the impact of different laser parameters on the magnetic levitation gap and calculate the optimal adjustment scheme. The system drives the magnetic levitation bearing controller to adjust the electromagnetic force, achieving micron-level adjustment of the gap. The gap adjustment process includes four stages: start-up preheating, dynamic measurement, real-time adjustment, and stability verification. The preheating phase ensures stable thermal expansion of the equipment. The dynamic measurement phase collects gap data from multiple points. The real-time adjustment phase continuously optimizes the gap based on feedback signals. The stability verification phase detects gap fluctuation amplitude and frequency to ensure the gap remains within the range of 50μm±2μm. During gap calibration, the system automatically records adjustment parameters, timestamps, and environmental conditions, storing them in the equipment calibration log. The anomaly detection module monitors abnormal gap fluctuations and electromagnetic drive anomalies, triggering alarms and executing a safety shutdown procedure. After calibration, the system generates an optimized system configuration file, containing updated laser parameters, magnetic levitation gap settings, and related equipment status information. This optimized system configuration serves as the output field name for this step, passed to the subsequent "melt superheat control" step in S110, forming a closed-loop feedback loop to support parameter updates and optimizations for the next round of melt preparation and centrifugal atomization processes.

[0212] Example 2: Figure 2 A structural block diagram of a centrifugal atomization system for preparing ultrafine soft magnetic powder according to an embodiment of the present invention is shown. Figure 2 As shown, the structure may include:

[0213] The melt superheat control module 01 is used to acquire Fe-Si-Al alloy melt and control its superheat. Specifically, it receives Fe-Si-Al alloy melt from a vacuum induction furnace, monitors the melt temperature in real time using a high-precision thermocouple sensor, and obtains temperature distribution data by combining numerical simulation of the melt temperature field. A PID control algorithm is used to precisely adjust the excess melt temperature relative to the Fe-Si-Al alloy liquidus temperature, adjusting the induction furnace power output to maintain superheat at approximately 150°C, thus forming a superheated melt. This superheated melt is recorded as an output field name and maintains a consistent association with the preset superheat target value. This output field name is passed to the laser texturing injection module as input, while temperature monitoring logs are retained for subsequent traceability.

[0214] The laser texturing injection module 02, connected to the melt superheat control module, is used to process the melt flow and generate initial droplets. Specifically, it receives superheated melt from the melt superheat control module and introduces it into the inner cavity of the laser texturing centrifuge disk through a high-temperature corrosion-resistant metal pipe. The surface of the laser texturing centrifuge disk is laser texturing-treated, with a surface roughness controlled within 0.1 μm. A high-power pulsed laser beam is used to etch the disk surface, forming a fine texture structure. Under the action of gravity and centrifugal force, the superheated melt enters the central region of the rotating centrifuge disk and, driven by high-speed centrifugal force, moves radially outward, forming initial droplets. The initial droplet is the output field name, transmitted to the centrifugal atomization module as input, and the droplet size and velocity information are registered in the high-speed camera system for subsequent modules to read.

[0215] Centrifugal atomization module 03, connected to the laser texturing injection module, is used to perform 25000rpm centrifugal atomization processing. Specifically, it receives initial droplets from the laser texturing injection module. Under the high-speed rotation of the laser texturing centrifuge disk, the droplets are subjected to extremely high centrifugal acceleration, further refining them. The droplet refining process includes stretching, breakage, and secondary atomization. Centrifugal force causes the droplets to accelerate radially, splitting into smaller, refined droplets. The system adjusts the centrifuge disk speed and melt flow rate through closed-loop control, combined with laser texturing surface parameters, to achieve precise control of droplet size. The refined droplet is the output object name, which is used as input by the gas proportioning module and recorded in the particle size analyzer as the correspondence with the droplet refining strategy.

[0216] Gas proportioning module 04 is used to generate an initial argon-hydrogen mixed atmosphere containing 5% H2. Specifically, based on the refined droplets from the centrifugal atomization module, a preset ratio of argon-hydrogen mixed gas is introduced, with the hydrogen volume ratio set to 5%. The mixed gas passes through a multi-stage purification and filtration device to remove impurities and ensure gas purity. The gas flow meter collects the flow signals of argon and hydrogen in real time, and the central control unit uses a PID control algorithm to maintain a stable hydrogen volume ratio. The initial atmosphere is the output field name, which is transmitted to the oxygen content feedback module for oxygen content adjustment and its effective status is recorded in the flow control log.

[0217] The oxygen content feedback module 05, connected to the gas proportioning module, is used to adjust the oxygen content to generate a stable atmosphere. Specifically, it receives the initial atmosphere from the gas proportioning module and acquires oxygen content signals through a multi-point oxygen sensor array. Based on real-time oxygen content data and a preset oxygen content threshold, the atmosphere control unit uses a PID feedback control algorithm to adjust the flow ratio of argon and hydrogen, dynamically adjusting the composition of the mixed gas. The stable atmosphere is the output field name, transmitted to the multi-stage cooling tower actuator for pressure balancing, and the adjustment status is recorded in the oxygen content monitoring log.

[0218] The multi-stage cooling tower actuator 06, connected to the oxygen content feedback module, is used to implement pressure balance to generate an atomization environment. Specifically, a stable atmosphere is introduced into the multi-stage cooling tower integrated system for pressure balancing. The cooling tower includes multiple cooling units arranged in series, each equipped with an independent temperature and pressure sensor to adjust the airflow speed and achieve uniform atmosphere flow. The pressure control unit, based on multi-point pressure sensor data, uses a closed-loop control algorithm to adjust the inlet and outlet pressures to maintain stable pressure within the atomization reaction chamber. The atomization environment is the output field name, transmitted to the parameter alignment module for synchronizing atomization parameters and registering status information in the cooling tower operation log.

[0219] The parameter alignment module 07 is used to synchronize centrifugal atomization parameters with atomization environment parameters. Specifically, it receives the atomization environment from the multi-stage cooling tower actuator, combines it with the droplet refinement parameters from the centrifugal atomization module, and performs parameter synchronization processing. During parameter alignment, the system compares the current atomization parameters with the environmental parameters, adjusts the centrifugal disc speed and atmosphere composition to ensure parameter consistency. The synchronized parameters are output field names, passed to the dynamic arbitration module for invocation, and the synchronization status is recorded in the parameter alignment log.

[0220] The dynamic arbitration module 08, connected to the oxygen content feedback module, is used to output a pressure regulation signal. Specifically, based on the real-time oxygen content data from the oxygen content feedback module and the synchronization parameters from the parameter alignment module, dynamic arbitration processing is performed. The system generates a pressure regulation signal by comparing the current oxygen content with a preset threshold, thereby adjusting the atmosphere composition and pressure distribution. The regulation signal is the output field name, which is transmitted to the data recording module for registration, and the arbitration result is recorded in the arbitration log.

[0221] Data logging module 09, connected to the dynamic arbitration module, is used to update the dynamic oxygen content threshold table. Specifically, it receives pressure regulation signals from the dynamic arbitration module and updates the dynamic oxygen content threshold table based on real-time oxygen content data and historical records. The threshold table update process includes calculating, verifying, and storing the new thresholds to ensure the accuracy of oxygen content control. The updated threshold table is the output field name, returned to the melt superheat control module for parameter reinjection, and the update information is recorded in the data logging log.

Claims

1. A centrifugal atomization method for producing an ultrafine soft magnetic powder, characterized by, The process of generating the refined droplets further comprises: obtaining the Fe-Si-Al alloy melt, performing melt superheat control based on a PID algorithm, maintaining the melt superheat at 150°C, and obtaining a superheat melt; extracting a melt flow from the superheat melt, performing laser textured centrifugal disc injection, wherein the centrifugal disc surface roughness Ra is less than or equal to 0.1 μm and the rotation speed is 25000 rpm, and generating initial droplets; performing 25000 rpm centrifugal atomization processing on the initial droplets, and using a high-speed imaging system and a laser particle size analyzer for real-time monitoring, and generating refined droplets. The process of generating the atomized environment further comprises: based on the refined droplets, performing hydrogen proportioning, wherein the hydrogen volume ratio is 5%, and obtaining an initial atmosphere; ​ 2. The method of claim 1, wherein, ​ ​ ​ ​ ​ 3. The method of claim 1, wherein, ​ ​ ​ ​ 4. The method of claim 1, wherein, ​ ​ Extracting oxygen content signals from the initial atmosphere, real-time feedback regulation is performed to maintain the oxygen content ≤200ppm, and a stable atmosphere is generated; The stable atmosphere is subjected to multi-stage cooling tower pressure balance treatment, wherein the cooling towers are arranged in series to generate an atomization environment.

5. The method of claim 1, wherein, The process of generating the rapid-cooled powder further includes: Obtaining the atomization environment, and performing cooling rate configuration processing, wherein the frequency range of the ultrasonic transducer array is 20 kHz to 2 MHz, and an ultrasonic field parameter is obtained; Extracting frequency characteristics from the ultrasonic field parameter, and performing standing wave field optimization processing, wherein a genetic algorithm is used for parameter optimization to generate a directional cooling field; Performing droplet solidification processing on the directional cooling field, and performing ultrasonic-assisted droplet solidification processing to generate the rapid-cooled powder.

6. The method of claim 1, wherein, The process of generating the target particle size powder further includes: Based on the rapid-cooled powder, 5-20 μm threshold setting processing is performed to obtain grading parameters; Extracting particle size distribution data from the grading parameters, and performing intelligent algorithm optimization processing, wherein the intelligent algorithm includes support vector machines, artificial neural networks, and fuzzy logic control to generate dynamic adjustment instructions; The dynamic adjustment instructions are subjected to multi-stage cyclone separation processing, wherein the cyclone separators are arranged in series to generate the target particle size powder.

7. The method of claim 1, wherein, The process of generating the performance-compliant powder further includes: Based on the target particle size powder, threshold setting processing is performed, wherein the sphericity threshold is not less than 95%, and morphology detection parameters are obtained; Extracting oxygen content data from the morphology detection parameters, and performing 200ppm threshold comparison processing to maintain the oxygen content ≤200ppm, and generating a quality evaluation report; Performing magnetization intensity test processing on the quality evaluation report, using a vibrating sample magnetometer with a saturation magnetization intensity not less than 1.8 Tesla to generate the performance-compliant powder.

8. The method of claim 1, wherein, The process of generating the optimized system configuration further includes: Based on the performance-compliant powder, device bias analysis processing is performed using principal component analysis algorithm to obtain a calibration requirement list; Extracting laser power data from the calibration requirement list, and performing texture precision compensation processing to maintain a surface roughness Ra ≤0.1 μm, and generating updated laser parameters; Performing magnetic suspension gap calibration processing on the updated laser parameters to maintain a gap size of 50 μm ± 2 μm, and generating the optimized system configuration.

9. The method of claim 1, wherein, The expression for generating the refined droplets includes: An improved PID control algorithm is used to establish an overheating degree regulation model: wherein, is the superheat adjustment for the kth iteration; is the superheat error term; is the control coefficient optimized from historical process data; is the summation index; is the total number of iterations; is the control period time step; The melt flow properties are defined as follows: wherein, is a melt flow property index; is a melt dynamic viscosity; is a two-dimensional temperature gradient; is a melt density; is a laser textured surface roughness parameter; Power compensation is triggered when Φ>1.2; An improved Navier-Stokes equation is constructed to describe the melt flow: wherein, is the radial velocity component; is time; is the local acceleration term of the melt radial velocity; ∇ is the Nabla operator; is the convective acceleration term of the melt radial velocity; is the material derivative or total acceleration of the fluid element; is the pressure gradient force experienced by the unit volume fluid element, the negative sign indicates that the force is in the opposite direction of the pressure increase; is the dynamic viscosity of the melt; is the Laplace operator; is the viscous or diffusive force experienced by the unit volume fluid element; is the centrifugal force term; ω is the angular velocity of the centrifuge disk; r is the radial distance of the melt element from the center of the disk. A droplet size prediction model is defined: wherein, is the droplet diameter; σ is the melt surface tension; Ra is the laser textured surface roughness parameter; Droplet diameter Determined to be acceptable when the deviation from the high-speed camera data is ≤ 5%. An improved Reynolds number is used to determine the secondary atomization effect: wherein, Re is the Reynolds number; D is the droplet diameter; Vt is the droplet tangential velocity; μ is the melt dynamic viscosity; When >200 trigger droplet secondary break-up, define atomization efficiency optimization function: wherein, is the atomization efficiency; is the residence time of the droplets in the atomization zone; is the critical break-up time threshold.

10. A centrifugal atomization system for producing ultrafine soft magnetic powders for use in the method according to any one of claims 1 to 9, characterized in that It includes: A melt overheating degree control module is used to obtain Fe-Si-Al alloy melt and perform melt overheating degree control processing; A laser texturing injection module is connected to the melt overheating degree control module to process melt flow to generate initial droplets; A centrifugal atomization module is connected to the laser texturing injection module to perform 25000 rpm centrifugal atomization processing; A gas proportioning module is used to generate an argon-hydrogen mixed initial atmosphere containing 5% H2; An oxygen content feedback module is connected to the gas proportioning module to adjust the oxygen content to generate a stable atmosphere; A multi-stage cooling tower actuator is connected to the oxygen content feedback module to implement pressure balance to generate an atomization environment; A parameter alignment module is configured to synchronize the centrifugal atomization parameters with the atomization environment parameters. A dynamic arbitration module is connected to the oxygen content feedback module and configured to output a pressure adjustment signal. A data recording module is connected to the dynamic arbitration module and configured to update the dynamic oxygen content threshold table.

Images