Feeding speed adjusting method for magnetic material processing and control system thereof

Abstract

The invention relates to the technical field of magnetic material processing, discloses a feeding speed adjusting method for magnetic material processing and a control system, and mainly solves the problems of abnormal growth of crystal grains and inconsistent magnetic properties caused by non-uniform temperature distribution in traditional processing. The method comprises the following steps: arranging 6-8 multispectral cameras and a 16-point magnetic sensor array in a sintering furnace, and constructing a three-dimensional thermal field monitoring network with 0.05 mm / pixel resolution; a double-channel model is adopted to predict the deformation trend, a machining area is divided into independent control grids of 5 cm * 5 cm, partition speed regulation is achieved through a multi-channel servo system, and smooth transition is conducted on adjacent grids through a linear interpolation algorithm. The control system comprises a data acquisition module, an edge calculation unit and an execution mechanism, has the functions of raw material pre-inspection, real-time monitoring and exception handling, and can realize 50ms-level quick response and + / -1.5 DEG C temperature control precision. Compared with a traditional method, the machining efficiency is improved by 25% or above, the product percent of pass is improved, and reliable technical support is provided for production of high-performance magnetic materials.

Description

Technical Field

[0001] The present invention relates to the technical field of magnetic material processing, in particular to a feed speed adjustment method for magnetic material processing. Background Art

[0002] Feed speed adjustment refers to the control technology that dynamically adjusts the movement speed of the tool or processing equipment according to the real-time status of the material during the magnetic material processing to balance processing efficiency and quality. The real-time status includes temperature, deformation, and magnetic field changes.

[0003] During the processing of magnetic materials, especially for materials such as ferrite and NdFeB, during the sintering stage, the uneven size and composition fluctuations of the raw material particles cause uneven temperature distribution in the furnace. The fixed feed speed cannot adapt to local temperature changes, which can easily cause abnormal grain growth and affect the consistency of magnetic properties.

[0004] In this situation, traditional overall speed control methods cannot effectively address localized overheating or overcooling, resulting in an overall decrease in tool speed, reduced overall efficiency, and an inability to achieve precise control. Furthermore, traditional sensor monitoring relies primarily on a small number of sensors, which cannot capture subtle changes in temperature and deformation, resulting in delayed response. Actions are often not taken until deformation has already occurred, impacting the quality of the finished product.

[0005] Traditional methods install a small number of temperature or pressure sensors (usually no more than four) to monitor temperature and deformation changes in the processing area, then set thresholds to trigger global speed reduction or speed increase. Once a sensor detects that the temperature exceeds the set value or the deformation exceeds 0.1mm, the entire processing speed is immediately reduced. While this global adjustment method can prevent local overload in some areas, it reduces processing efficiency in other unaffected areas.

[0006] When the problem in the faulty area is alleviated, the speed will recover as a whole. However, the response in this process is untimely and inefficient, and every change point in the processing cannot be accurately controlled, thus limiting the further improvement of processing quality and yield rate.

[0007] Therefore, we propose a feed speed adjustment method for magnetic material processing to solve the problems in the above background. Summary of the Invention

[0008] The present invention provides a feed speed adjustment method for magnetic material processing, which can solve the problem that the existing magnetic material processing adopts an overall speed adjustment method, which cannot effectively deal with the problem of local temperature overheating or overcooling, resulting in reduced overall efficiency and inability to achieve precise control.

[0009] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0010] 1. A method for adjusting feed speed for magnetic material processing, comprising the following steps:

[0011] Step (1) multiple multispectral cameras are installed at equal intervals around the inner wall of the sintering furnace, each multispectral camera includes visible light, near infrared and thermal infrared bands, synchronously collects material surface temperature distribution data, and generates a real-time three-dimensional thermal field map with a resolution of 0.05 mm / pixel;

[0012] Step (2) 16 magnetic sensors are evenly arranged along the edge of the processing table in the sintering furnace to detect the change of the magnetic field intensity on the surface of the material in real time, and the magnetic field data is spatially aligned with the thermal field map of step (1);

[0013] Step (3) The material surface is divided into multiple independent control grids of 5 cm × 5 cm, and the safe processing speed is calculated based on the temperature difference ΔT and the predicted deformation variable ε_pred of each grid, where ΔT is the maximum temperature difference between the grid and the four adjacent grids;

[0014] Step (4) independently controlling the processing speed of each grid through a servo driver, wherein the speed difference between adjacent grids does not exceed 20 mm / s, and the single adjustment amplitude does not exceed 10% of the current speed;

[0015] Step (5) When it is detected that the local temperature gradient exceeds 30°C / cm, the corresponding grid speed is reduced to a safety threshold within 50ms, and the cooling of the area is accelerated.

[0016] 2. The feed speed adjustment method for magnetic material processing according to claim 1 is characterized in that each multispectral camera in step (1) comprises three independent imaging modules, covering the 500-700nm visible light, 800-1000nm near infrared and 8-14μm thermal infrared bands respectively; the overlap rate of the fields of view of adjacent cameras is made ≥30% by laser calibration, and a unified three-dimensional coordinate system is established in the furnace; at the same time, a fusion map containing temperature and deformation data is generated once every 5 seconds, and the deformation data is calculated by comparing the changes in the three-dimensional coordinates at adjacent moments.

[0017] 3. The feed speed adjustment method for magnetic material processing according to claim 2 is characterized in that the arrangement of the magnetic sensors in step (2) is as follows: a magnetic sensor is installed at every 10 cm on the edge of the processing table, forming a total of 16 detection points; each sensor detects a circular area with a radius of 5 cm, and the overlapping width of adjacent detection areas is ≥2 cm; the magnetic field intensity data is sampled at a frequency of 100 Hz.

[0018] 4. The feed rate adjustment method for magnetic material processing according to claim 1, wherein the predicted deformation variable ε_pred in step (3) is obtained by:

[0019] (a) A 3D convolutional neural network is used to extract temperature gradient features from the thermal field map and identify heat accumulation areas with a diameter ≥ 2 cm.

[0020] (b) Using a long short-term memory network to correlate 20 sets of historical processing parameters, the deformation trend in the next 30 seconds is predicted;

[0021] (c) Dynamically adjust the weight coefficients of temperature and deformation

[0022] 5. The feed speed adjustment method for magnetic material processing according to claim 1 is characterized in that the servo drive control method of step (4) includes: dividing the processing tool into control units corresponding to the material grid, each unit is equipped with an independent servo motor; and adjusting the feed speed by linear interpolation according to the distance within 2 cm of the adjacent grid boundaries.

[0023] 6. The feed rate adjustment method for magnetic material processing according to claim 1, wherein the cooling method of step (5) is: (a) locating abnormal grid coordinates within 20 ms by a high-speed processor;

[0024] (b) The atomizing cooling nozzle is activated within 1 cm of the abnormal grid, and the coolant flow rate is increased to 1.5 times the baseline value;

[0025] (c) When the temperature gradient returns to below 15°C / cm, the processing speed is gradually restored at a rate of 5% per second.

[0026] 7. The feed rate adjustment method for magnetic material processing according to claim 6, further comprising a crack detection step:

[0027] Install an acoustic emission sensor on the spindle of the processing equipment with a sampling frequency of ≥200kHz. When an acoustic emission signal with a characteristic frequency in the range of 150-250kHz is detected for 10ms, it is determined that a crack has occurred. Immediately stop the current grid processing and start the acoustic emission sensor within a 3cm range to conduct a flaw detection review.

[0028] 8. The feed rate adjustment method for magnetic material processing according to claim 1 is characterized in that a laser spectrometer is used to scan the raw material surface before sintering, and the iron and boron content is detected once every 2 cm×2 cm area; when the iron content deviation exceeds 0.3%, the corresponding grid base speed is adjusted and the pH value of the coolant is synchronously adjusted, and the compensation amount is ΔpH=0.1×ΔFe(%).

[0029] 9. A control system for implementing the feed speed adjustment method for magnetic material processing according to any one of claims 1 to 8, characterized in that it comprises: a data acquisition module, an edge computing unit, an actuator, and a human-machine interface;

[0030] The data acquisition module includes 6-8 multispectral cameras arranged in a ring, a 16-point magnetic sensor array and 3 laser displacement sensors; the edge computing unit has a built-in temperature-deformation coupling field reconstruction algorithm and deformation prediction model; the actuator includes a multi-channel servo drive system corresponding to the material grid; the human-machine interface is used to display the temperature, deformation prediction value and speed adjustment curve of each grid in real time, and abnormal areas are highlighted in red.

[0031] 10. The control system according to claim 9, characterized in that the laser displacement sensors are arranged as follows: one laser interferometer is installed in each of the three axes X / Y / Z of the processing table; and the measurement spot diameter of the laser interferometer is ≤0.5 mm.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] The feed rate adjustment method for magnetic material processing provided by this invention effectively addresses the problems of abnormal grain growth and inconsistent magnetic properties caused by uneven temperature distribution during traditional processing through an innovative closed-loop control system. This method first constructs a high-precision, multi-dimensional monitoring network, employing 6-8 multispectral cameras to achieve three-dimensional thermal field monitoring with a resolution of 0.05mm / pixel. Combined with a 16-point magnetic sensor array, this method increases temperature detection coverage and sensitivity to 100% and ±0.5°C, respectively, achieving a sixfold improvement in accuracy compared to traditional methods. In terms of intelligent control, the system uses a dual-channel model to predict deformation trends 30 seconds in advance. Combined with independent speed regulation on a 5cm×5cm grid, this system improves processing efficiency by over 25% while ensuring smooth transitions between adjacent grid speeds. A specially designed raw material pre-inspection system uses a 2cm×2cm grid to scan components and pre-compensate for areas with iron content deviations greater than 0.3%, reducing the frequency of subsequent process adjustments. During real-time processing, a 50ms-level rapid response mechanism, combined with an enhanced cooling system, increases product qualification rates from 82% to 97%. Furthermore, the acceleration limiter and partitioned drive design significantly reduced equipment wear and tear, extending tool life by 75%. This approach, through high-precision sensing, intelligent prediction, and partitioned execution, addresses the core issues of traditional overall speed regulation, including response lag, low efficiency, and insufficient control precision. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 The present invention is a flow chart of the feed speed adjustment method for magnetic material processing. DETAILED DESCRIPTION

[0035] The specific embodiments of the present invention are described in detail below, but it should be understood that the protection scope of the present invention is not limited by the specific embodiments.

[0036] Example 1:

[0037] See also Figure 1 , the present invention provides a technical solution:

[0038] A method for adjusting feed speed for magnetic material processing comprises the following steps:

[0039] Step (1) 6-8 multispectral cameras are installed equidistantly around the inner wall of the sintering furnace, each multispectral camera includes visible light, near infrared and thermal infrared bands, synchronously collects material surface temperature distribution data, and generates a real-time three-dimensional thermal field map with a resolution of 0.05 mm / pixel;

[0040] Step (2) 16 magnetic sensors are evenly arranged along the edge of the processing table in the sintering furnace to detect the change of the magnetic field intensity on the surface of the material in real time, and the magnetic field data is spatially aligned with the thermal field map of step (1);

[0041] Step (3) The material surface is divided into multiple independent control grids of 5 cm × 5 cm, and the safe processing speed is calculated based on the temperature difference ΔT and the predicted deformation variable ε_pred of each grid, where ΔT is the maximum temperature difference between the grid and the four adjacent grids;

[0042] Step (4) independently controlling the processing speed of each grid through a servo driver, wherein the speed difference between adjacent grids does not exceed 20 mm / s, and the single adjustment amplitude does not exceed 10% of the current speed;

[0043] Step (5) When it is detected that the local temperature gradient exceeds 30°C / cm, the corresponding grid speed is reduced to a safety threshold within 50ms, and the cooling of the area is accelerated.

[0044] The above scheme is an overall method for adjusting the feed speed for magnetic material processing, and the specific implementation method is as follows:

[0045] First, during the data acquisition phase, a three-dimensional thermal field map is constructed using 6-8 multispectral cameras arranged in a ring. The resolution of the three-dimensional thermal field map is 0.05mm / pixel. At the same time, 16 edge magnetic sensors form a detection network to synchronously monitor the full-surface temperature and magnetic field strength.

[0046] Through high-density sensing layout, compared with the traditional 4-point monitoring method, the temperature detection coverage rate is increased from less than 15% to 100%, and the thermal infrared band can identify micro-temperature differences of 0.5°C, thus solving the response lag problem caused by the traditional method of few sensor layout points and low sensitivity.

[0047] Furthermore, during the intelligent analysis phase, the system divides the material surface into independent control grids of 5cm x 5cm. A dual-channel neural network calculates the safe processing speed for each grid in real time. The spatial analysis channel uses 3D convolution to identify heat accumulation zones with a diameter ≥ 2cm, and the temporal prediction channel uses an LSTM model to correlate 20 sets of historical process parameters to predict deformation trends over the next 30 seconds. This predictive control approach shortens the intervention window by 8-10 seconds compared to traditional threshold response methods, enabling the system to implement control measures before abnormal grain growth occurs.

[0048] During the dynamic execution phase, the multi-channel servo system independently adjusts the speed of each grid based on the prediction results, using a linear interpolation algorithm to ensure that the speed difference between adjacent grids is ≤20mm / s. When a local temperature gradient exceeding 30°C / cm is detected, the FPGA controller completes target positioning and speed adjustment within 50ms, a 24-fold increase compared to the 1.2-second response speed of a traditional PLC system. This precise control of each area allows unaffected areas to maintain the highest processing speed, improving overall efficiency by over 25% compared to traditional global speed reduction methods.

[0049] Finally, during the abnormality handling phase, the system uses atomizing cooling nozzles to provide targeted, enhanced cooling of overheated areas, increasing the coolant flow rate to 1.5 times the baseline value. This, combined with the raw material pre-analysis module, allows for early compensation of iron content fluctuations, forming a closed-loop control loop for the entire process, from raw materials to finished product.

[0050] This solution changes the traditional overall speed regulation mode, realizes the technical improvement of magnetic material processing from passive remediation to active prevention, and provides technical support for the mass production of high-performance magnetic devices.

[0051] The formula for calculating the safe processing speed is:

[0052]

[0053] Where: v_safe = calculated safe processing speed;

[0054] v_base = base processing speed;

[0055] ΔT = the maximum temperature difference between the current grid and the four adjacent grids;

[0056] ΔT_max=maximum temperature difference allowed by the material;

[0057] ε_pred = predicted shape;

[0058] ε_max=the maximum deformation allowed by the material;

[0059] The calculation principle is that when the temperature difference (ΔT) increases, the speed will decrease, and when the predicted deformation (ε_pred) increases, the speed will also decrease. After the two effects are superimposed, the safe speed v_safe is obtained from the basic speed v_base.

[0060] In one embodiment of the above scheme, each multispectral camera in step (1) includes three independent imaging modules, covering the 500-700nm visible light, 800-1000nm near infrared and 8-14μm thermal infrared bands respectively; laser calibration is used to make the overlap rate of the fields of view of adjacent cameras ≥30%, and a unified three-dimensional coordinate system inside the furnace is established; at the same time, a fusion map containing temperature and deformation data is generated every 5 seconds, and the deformation data is calculated by comparing the changes in the three-dimensional coordinates at adjacent moments.

[0061] The arrangement of the magnetic sensors in step (2) is as follows: a magnetic sensor is installed at intervals of 10 cm on the edge of the processing table, forming a total of 16 detection points; each sensor detects a circular area with a radius of 5 cm, and the overlapping width of adjacent detection areas is ≥ 2 cm; the magnetic field intensity data is sampled at a frequency of 100 Hz, and the equipment interference noise is eliminated by Kalman filtering.

[0062] The predicted shape variable ε_pred in step (3) is obtained as follows:

[0063] (a) A 3D convolutional neural network is used to extract temperature gradient features from the thermal field map and identify heat accumulation areas with a diameter ≥ 2 cm.

[0064] (b) Using a long short-term memory network to correlate 20 sets of historical processing parameters, the deformation trend in the next 30 seconds is predicted;

[0065] (c) Dynamically adjust the weight coefficients of temperature and deformation;

[0066] The calculation formula of weight coefficient is α=0.7-0.1×(T curent / T Curie ) 2

[0067] Where T Curie is the material's Curie temperature. This weight coefficient formula dynamically adjusts the weights of the effects of temperature and deformation on processing speed. The temperature weight α gradually decreases as the current temperature (T_current) approaches the material's Curie temperature (T_Curie), while the deformation weight β increases accordingly (β=1-α). This allows for more focus on deformation control in the high-temperature stage and priority on temperature regulation in the low-temperature stage, ensuring that the material obtains the optimal speed adjustment strategy throughout the entire processing process.

[0068] The working principle of the above embodiment is: through the collaborative work of a multispectral camera array and a magnetic sensor network, the magnetic material processing process is accurately monitored, and the multispectral camera is equipped with three independent imaging modules for visible light, near infrared, and thermal infrared, which can fully capture the surface state of the material. Laser calibration is used to ensure that the field of view overlap rate is not less than 30%, and a unified three-dimensional coordinate system is established. The system generates a three-dimensional map that integrates temperature and deformation data every 5 seconds, where the deformation data is calculated by comparing the coordinate changes of adjacent time points. At the same time, a magnetic sensor is arranged every 10 cm on the edge of the processing table, with a total of 16 detection points. Each sensor covers an area with a radius of 5 cm, and adjacent areas maintain an overlap of at least 2 cm. The magnetic field data is collected at a frequency of 100 Hz, and Kalman filtering technology is used to eliminate interference noise to ensure data accuracy.

[0069] During data processing, the system uses a 3D convolutional neural network to analyze thermal field maps, identify heat accumulation areas with a diameter of more than 2 cm, and combines the analysis of 20 historical processing parameter sets with a long and short-term memory network to predict the deformation trend within the next 30 seconds. Through the dynamic weight coefficient formula, the system can automatically adjust the weight of the influence of temperature and deformation on the processing speed according to the relationship between the current temperature and the Curie temperature of the material; when the temperature is close to the Curie temperature, the temperature weight α is reduced and the deformation weight β (β=1-α) is increased accordingly, thereby giving priority to controlling deformation in the high temperature stage and focusing on temperature regulation in the low temperature stage. This intelligent dynamic adjustment strategy ensures that the material is always in the best process state throughout the entire processing process, and effectively solves the problems of abnormal grain growth and inconsistent magnetic properties caused by uneven temperature distribution in traditional methods.

[0070] In another embodiment, the servo drive control method of step (4) includes: dividing the machining tool into control units corresponding to the material grid, each unit is equipped with an independent servo motor; and adjusting the feed speed by linear interpolation according to the distance within 2 cm of the adjacent grid boundaries.

[0071] In conjunction with the first embodiment, the cooling method of step (5) is as follows: (a) locating abnormal grid coordinates within 20 ms by a high-speed processor;

[0072] (b) The atomizing cooling nozzle is activated within 1 cm of the abnormal grid, and the coolant flow rate is increased to 1.5 times the baseline value;

[0073] (c) When the temperature gradient returns to below 15°C / cm, the processing speed is gradually restored at a rate of 5% per second.

[0074] Furthermore, the above solution also includes a crack detection step:

[0075] Install an acoustic emission sensor on the spindle of the processing equipment with a sampling frequency of ≥200kHz. When an acoustic emission signal with a characteristic frequency in the range of 150-250kHz is detected for 10ms, it is determined that a crack has occurred. Immediately stop the current grid processing and start the acoustic emission sensor within a 3cm range to conduct a flaw detection review.

[0076] In the above embodiment, the speed is adjusted by linear interpolation within 2 cm of the adjacent grid boundaries. The interpolation formula is: 过渡 =(v A ×d B +v B ×d A ) / (d A +d B )

[0077] The principle of the above interpolation formula is: in the boundary area between adjacent grids A (speed v_A) and grid B (speed v_B), the transition speed is calculated proportionally based on the distance from the current position to the centers of the two grids (d_A and d_B); the speed transitions smoothly from v_A to v_B to avoid machining vibration or surface defects caused by sudden changes in speed. The role of the acceleration limiter is to limit the speed change rate to no more than 50mm / s 2 This prevents the servo motor from generating mechanical shock due to excessive speed adjustment, protects the equipment and ensures processing stability.

[0078] This embodiment achieves high-precision processing of magnetic materials through a multi-level collaborative control method. In terms of speed control, the system divides the processing tool into control units that match the material grid, and each unit is equipped with an independent servo motor. When the tool moves to within 2cm of the adjacent grid boundary, a linear interpolation algorithm is used to dynamically adjust the feed speed; based on the distance from the current position to the center of the adjacent grid (d_A and d_B), the transition speed is calculated according to the formula (v_A×d_B+v_B×d_A) / (d_A+d_B), so that the speed transitions smoothly from v_A to v_B, and the acceleration limiter is used to control the speed change rate to 50mm / s. 2 Within, effectively avoiding mechanical vibration and processing defects.

[0079] In terms of temperature anomaly processing, the system uses a high-speed processor to accurately locate the overheating grid coordinates within 20ms, and then activates the atomizing cooling nozzle within 1cm around the abnormal area, increasing the coolant flow to 1.5 times the baseline value for rapid cooling; after the temperature gradient returns to below 15℃ / cm, the processing speed is restored at a gradual rate of 5% per second to ensure process stability.

[0080] The system also integrates real-time crack detection, using an acoustic emission sensor mounted on the machining spindle for continuous monitoring at a frequency greater than 200kHz. When a characteristic frequency signal of 150-250kHz is detected for 10ms, it is immediately identified as a crack. The system immediately halts the current mesh processing and initiates ultrasonic flaw detection within a 3cm radius to further ensure quality. Through dynamic speed regulation, precise cooling, and defect detection, machining accuracy and product qualification rates are significantly improved.

[0081] In another embodiment, combined with the above solution, a laser spectrometer is used to scan the raw material surface before sintering, and the iron and boron content is detected once every 2cm×2cm area; when the iron content exceeds 0.3%, the corresponding grid base speed is adjusted according to the formula, v base-new =v base ×[1-0.2×(ΔFe / 0.3%)]; synchronously adjust the pH value of the coolant, and the compensation amount is ΔpH=0.1×ΔFe(%).

[0082] In the above scheme, during the raw material pre-analysis stage before sintering, a laser spectrometer is used to scan the material surface with a high-resolution 2cm×2cm grid to accurately detect the distribution of iron and boron content in each tiny area. When the iron content deviation in a certain area is detected to exceed the set threshold, the system automatically adjusts the baseline feed speed of the corresponding processing grid in that area according to the formula. For example, when the iron content is high, the speed is reduced, and the pH value of the coolant is simultaneously adjusted to compensate. The compensation amount ΔpH = 0.1×ΔFe%. Through the synergistic mechanism of composition detection and parameter pre-adjustment, the problem of raw material heterogeneity is compensated at the source, making the temperature field distribution in the subsequent sintering process more uniform, thereby reducing the risk of local overheating or deformation and improving the consistency of the magnetic properties of the final product. This step forms a closed loop with real-time processing control to ensure process stability.

[0083] A control system for implementing a feed speed adjustment method for magnetic material processing, comprising: a data acquisition module, an edge computing unit, an actuator, and a human-machine interface;

[0084] The data acquisition module includes 6-8 multispectral cameras arranged in a ring, a 16-point magnetic sensor array, and three laser displacement sensors. The edge computing unit incorporates a temperature-deformation coupling field reconstruction algorithm and a deformation prediction model. The actuator includes a multi-channel servo drive system corresponding to the material grid. The human-machine interface displays the temperature, deformation prediction values, and speed adjustment curves of each grid in real time, with abnormal areas highlighted in red. The laser displacement sensors are arranged with one laser interferometer installed in each of the three axes (X, Y, and Z) of the processing table. The laser interferometer's measurement spot diameter is ≤0.5mm.

[0085] In the above scheme, the deformation variable of the deformation prediction value is calculated as follows:

[0086]

[0087] Where Δx, Δy, and Δz are the coordinate changes between adjacent sampling periods.

[0088] In the above scheme, the control system realizes precise control of magnetic material processing through the collaborative work of multiple sensors. The multispectral cameras (6-8 units) and 16-point magnetic sensor array arranged in a ring in the data acquisition module are used to collect the temperature field and magnetic field data of the material surface in real time, and cooperate with the laser interferometer in the X / Y / Z three-axis direction of the processing table to monitor the deformation variable; the edge computing unit reconstructs the three-dimensional field through the temperature-deformation coupling algorithm and predicts the deformation trend in the next 30 seconds based on the historical 20 sets of process parameters (deformation variable ε=√(Δx 2 +Δy 2 +Δz 2 ), where Δx, Δy, and Δz are the coordinate changes between adjacent sampling periods); the actuator independently adjusts the speed of a 5cm×5cm grid through a multi-channel servo system based on the prediction results, and adjacent grids use linear interpolation transitions. Simultaneously, the human-machine interface visualizes temperature, deformation, and speed curves in real time, and issues red alerts in abnormal areas, forming a closed-loop control from monitoring and calculation to execution, ultimately achieving high-precision machining with a temperature difference control accuracy of ±1.5°C and a response delay of less than 50ms.

[0089] This control system adopts an intelligent closed-loop control scheme. Its working principle is as follows: the system collects temperature field and magnetic field data of the processing area in real time through 6-8 multispectral cameras and 16 magnetic sensors arranged in a ring, uses three high-precision laser measuring instruments (accuracy 0.5mm) to synchronously monitor the deformation state of the material, and then fuses multi-source data through the edge computing unit to establish a three-dimensional temperature-deformation coupling field model.

[0090] Then, the learning accumulation of 20 sets of historical process parameters is combined with advanced algorithms to predict the deformation trend in the next 30 seconds, where the deformation variable is calculated by the change of spatial coordinates: ε=√(Δx 2 +Δy 2 +Δz 2 );

[0091] Furthermore, the processing area is divided into 5cm×5cm independently controlled grids, and the speed of each grid is independently adjusted through a multi-channel servo system. A smooth transition algorithm is used between adjacent grids to ensure processing continuity.

[0092] Finally, the temperature field, deformation prediction and speed adjustment curve are intuitively displayed through the human-machine interface, with abnormal areas automatically highlighted in red as an early warning. The system's response delay is controlled within 50ms, and the temperature difference accuracy reaches ±1.5°C. This solution achieves high-precision intelligent control of the magnetic material processing process through closed-loop control of monitoring, prediction, execution and feedback.

[0093] Feed speed adjustment method for magnetic material processing:

[0094] First, a multi-dimensional monitoring network was constructed within the sintering furnace. Six to eight multispectral cameras were installed equidistantly around the furnace wall. Each camera integrated imaging modules for visible light, near-infrared, and thermal infrared. Laser calibration ensured over 30% field-of-view overlap, generating a three-dimensional thermal field map with a resolution of 0.05 mm / pixel every five seconds. Furthermore, magnetic sensors were placed every 10 cm around the edge of the processing table, forming a 16-point detection array. Each sensor covered a 5 cm radius area with a 2 cm overlap. Magnetic field data was collected at a 100 Hz frequency, denoised using a Kalman filter, and spatially aligned with the thermal field map.

[0095] During the data processing stage, the system divides the material surface into independent control grids of 5cm×5cm and performs real-time analysis through a dual-channel neural network. The spatial analysis channel uses 3D convolution to identify heat accumulation areas ≥2cm. The time prediction channel associates 20 sets of historical process parameters through the LSTM model to predict the deformation trend ε_pred in the next 30 seconds. The dynamic weight module automatically adjusts the parameter weight according to the ratio of the current temperature to the Curie temperature, focusing on deformation control in the high-temperature stage and prioritizing temperature regulation in the low-temperature stage.

[0096] The execution link adopts a multi-channel servo system, and each grid is equipped with an independent drive unit. The safe processing speed v_safe is calculated by a composite formula, and the influence of temperature difference ΔT and predicted deformation ε_pred is comprehensively considered. A 2cm transition zone is set between adjacent grids, and linear interpolation is performed according to the distance to achieve smooth speed transition, and the speed is ≤50mm / s 2 Acceleration limitations ensure mechanical stability.

[0097] The exception handling mechanism includes three levels of response measures. When the temperature gradient is greater than 30°C / cm, the FPGA controller locates the abnormal grid within 20ms, starts mist cooling within a range of 1cm, and gradually resumes work at a rate of 5% / s after the temperature returns to below 15°C / cm. The crack detection system captures the 150-250kHz characteristic frequency through an acoustic emission sensor with a sampling rate of ≥200kHz. When it persists for 10ms, it triggers a shutdown and starts an ultrasonic re-inspection within a range of 3cm.

[0098] Furthermore, during the raw material pre-processing stage, a laser spectrometer is used to perform a 2cm×2cm grid composition scan. When the iron content deviation is greater than 0.3%, the corresponding processing grid base speed is automatically adjusted, and the pH value of the coolant is compensated synchronously. Finally, the temperature field, deformation prediction curve and speed adjustment parameters are visualized in real time through the human-machine interface, and a red warning is issued in abnormal areas, forming a full-process closed-loop control from raw material detection, real-time monitoring to dynamic adjustment, achieving precise temperature control and rapid response, which improves processing efficiency by more than 25% compared with traditional methods.

[0099] The above disclosures are only a few specific embodiments of the present invention. However, the embodiments of the present invention are not limited thereto. Any changes that can be conceived by those skilled in the art should fall within the scope of protection of the present invention.

Claims

1. A method for adjusting feed speed for magnetic material processing, characterized in that: The following steps are involved: Step (1) multiple multispectral cameras are installed at equal intervals around the inner wall of the sintering furnace, each multispectral camera includes visible light, near infrared and thermal infrared bands, synchronously collects material surface temperature distribution data, and generates a real-time three-dimensional thermal field map with a resolution of 0.05 mm / pixel; Step (2) 16 magnetic sensors are evenly arranged along the edge of the processing table in the sintering furnace to detect the change of the magnetic field intensity on the surface of the material in real time, and the magnetic field data is spatially aligned with the thermal field map of step (1); Step (3) The material surface is divided into multiple independent control grids of 5 cm × 5 cm, and the safe processing speed is calculated based on the temperature difference ΔT and the predicted deformation variable ε_pred of each grid, where ΔT is the maximum temperature difference between the grid and the four adjacent grids; Step (4) independently controlling the processing speed of each grid through a servo driver, wherein the speed difference between adjacent grids does not exceed 20 mm / s, and the single adjustment amplitude does not exceed 10% of the current speed; Step (5) When it is detected that the local temperature gradient exceeds 30°C / cm, the corresponding grid speed is reduced to a safety threshold within 50ms, and the cooling of the area is accelerated.

2. The feed speed adjustment method for magnetic material processing according to claim 1, characterized in that: In step (1), each multispectral camera includes three independent imaging modules, covering the 500-700nm visible light, 800-1000nm near infrared and 8-14μm thermal infrared bands respectively; laser calibration is used to make the overlap rate of the fields of view of adjacent cameras ≥30%, and a unified three-dimensional coordinate system is established in the furnace; at the same time, a fusion map containing temperature and deformation data is generated every 5 seconds, and the deformation data is calculated by comparing the changes in the three-dimensional coordinates at adjacent moments.

3. The feed speed adjustment method for magnetic material processing according to claim 2, characterized in that: The arrangement of the magnetic sensors in step (2) is as follows: a magnetic sensor is installed at intervals of 10 cm on the edge of the processing table, forming a total of 16 detection points; each sensor detects a circular area with a radius of 5 cm, and the overlapping width of adjacent detection areas is ≥ 2 cm; the magnetic field intensity data is sampled at a frequency of 100 Hz.

4. The feed speed adjustment method for magnetic material processing according to claim 1, wherein: The predicted shape variable ε_pred in step (3) is obtained by the following method: (a) A 3D convolutional neural network is used to extract temperature gradient features from the thermal field map and identify heat accumulation areas with a diameter ≥ 2 cm. (b) Using a long short-term memory network to correlate 20 sets of historical processing parameters, the deformation trend in the next 30 seconds is predicted; (c) Dynamically adjust the weight coefficients of temperature and deformation.

5. The method for adjusting feed speed for magnetic material processing according to claim 1, wherein: The servo drive control method of step (4) includes: dividing the machining tool into control units corresponding to the material grid, each unit is equipped with an independent servo motor; and adjusting the feed speed by linear interpolation according to the distance within 2 cm of the adjacent grid boundaries.

6. The method for adjusting feed speed for magnetic material processing according to claim 1, wherein: The cooling method of step (5) is as follows: (a) locating abnormal grid coordinates within 20 ms by a high-speed processor; (b) The atomizing cooling nozzle is activated within 1 cm of the abnormal grid, and the coolant flow rate is increased to 1.5 times the baseline value; (c) When the temperature gradient returns to below 15°C / cm, the processing speed is gradually restored at a rate of 5% per second.

7. The method for adjusting feed speed for magnetic material processing according to claim 6, wherein: Also includes crack detection steps: Install an acoustic emission sensor on the spindle of the processing equipment with a sampling frequency of ≥200kHz. When an acoustic emission signal with a characteristic frequency in the range of 150-250kHz is detected for 10ms, it is determined that a crack has occurred. Immediately stop the current grid processing and start the acoustic emission sensor within a 3cm range to conduct a flaw detection review.

8. The method for adjusting feed speed for magnetic material processing according to claim 1, wherein: Before sintering, a laser spectrometer was used to scan the raw material surface, and the iron and boron content was detected once every 2cm×2cm area. When the iron content deviation exceeded 0.3%, the corresponding grid reference speed was adjusted: the pH value of the coolant was adjusted synchronously, and the compensation amount was ΔpH = 0.1×ΔFe (%).

9. A control system for implementing the feed speed adjustment method for magnetic material processing according to any one of claims 1 to 8, characterized in that: include: Data acquisition module, edge computing unit, actuator and human-machine interface; The data acquisition module includes 6-8 multispectral cameras arranged in a ring, a 16-point magnetic sensor array and 3 laser displacement sensors; the edge computing unit has a built-in temperature-deformation coupling field reconstruction algorithm and deformation prediction model; the actuator includes a multi-channel servo drive system corresponding to the material grid; the human-machine interface is used to display the temperature, deformation prediction value and speed adjustment curve of each grid in real time, and abnormal areas are highlighted in red.

10. The control system according to claim 9, characterized in that: The laser displacement sensor is arranged as follows: a laser interferometer is installed in each of the three axes X / Y / Z of the processing table; the measurement spot diameter of the laser interferometer is ≤0.5mm.

Images