A device for gripping a stack of magnetic yoke laminations

Abstract

The application relates to the technical field of magnetic yoke punching sheet stacking, and particularly discloses a grabbing and stacking device for magnetic yoke punching sheets, which comprises a base frame, a storage module, a stacking module and a grabbing module driven by a gantry robot, the grabbing module comprises a visual positioning assembly for collecting visible light and near-infrared data, a solving unit for solving adsorption point coordinates based on a mask matrix, multi-modal fusion and a thermal deformation coefficient, a plurality of vacuum suction cups with a moving module, and an adsorption state detection assembly for detecting flatness and real-time pressure, and an adsorption force control unit for performing graded initial adsorption based on the flatness data and dynamic compensation based on the real-time pressure and a double safety threshold, the stacking module comprises a center positioning column and a stacking alignment unit with a radial synchronous telescopic mechanism, and the alignment rod of the stacking alignment unit is provided with auxiliary wheels, a flexible polishing belt and a pressure sensor; the application realizes high-precision positioning, self-adaptive adsorption and high-coaxiality stacking.

Description

Technical Field

[0001] This invention relates to the field of magnetic yoke lamination stacking technology, and specifically discloses a gripping and stacking device for magnetic yoke laminations. Background Technology

[0002] Magnetic yoke laminations are core stacked components of electromagnetic equipment such as motors and generators. They are usually made of silicon steel sheets through precision stamping. Their surfaces have a specific density of slotted holes. The magnetic yoke assembly formed after stacking can realize magnetic flux transmission and mechanical structure reinforcement, directly affecting the electromagnetic performance and operational stability of the equipment.

[0003] In the magnetic yoke production process, the batch of stamped magnetic yoke blanks need to be precisely grasped, transferred and stacked according to preset precision. Therefore, automated grasping and stacking devices have become key equipment for improving production efficiency and ensuring stacking quality.

[0004] Existing magnetic yoke lamination gripping and stacking devices mostly use fixed array vacuum chucks combined with simple visual positioning, relying on a single visible light image for adsorption point positioning. This makes it difficult to cope with differences in the distribution of slots and holes on the lamination surface and slight thermal deformation after stamping, resulting in large deviations in adsorption point positioning and affecting stacking coaxiality. At the same time, the adsorption pressure of existing devices is mostly a fixed value, without considering the possible flatness deviations on the lamination surface, which can easily lead to problems such as weak adsorption causing lamination to fall off or excessive pressure causing damage to the lamination, thereby affecting the stacking accuracy and performance of the stacking device. Summary of the Invention

[0005] The purpose of this invention is to provide a gripping and stacking device for magnetic yoke laminations, so as to at least solve one of the above-mentioned problems existing in the prior art.

[0006] Specifically, the present invention is achieved through the following technical solution:

[0007] A gripping and stacking device for magnetic yoke laminations includes a base frame, with a storage module and a stacking module at each end of the base frame, and a gripping module on top of the storage and stacking modules. The gripping module is connected to a gantry robot mounted on the base frame.

[0008] The grasping module includes an execution unit, which includes a rotatable adsorption plate and an array of vacuum suction cups arranged around the bottom of the adsorption plate. A sensing and detection unit is also provided on the adsorption plate, which includes a visual positioning component.

[0009] The visual positioning component includes an image acquisition unit and a calculation unit; the image acquisition unit is used to acquire image data of the magnetic yoke blank when the execution unit grasps and adsorbs it and transmits it to the calculation unit, wherein the image data includes visible light channel data and near-infrared penetrating channel data.

[0010] The calculation unit calculates and generates the positioning coordinates of the suction point of the vacuum suction cup based on the image data;

[0011] Each of the vacuum suction cups is connected to the adsorption plate via a moving module, and the moving module is signal-connected to the calculation unit to drive the vacuum suction cup to the adsorption point positioning coordinates generated by the calculation unit.

[0012] Specifically, the process by which the calculation unit calculates and generates the positioning coordinates of the suction point of the vacuum suction cup based on image data is as follows:

[0013] Construct a mask matrix by obtaining the surface slot density of a standard image of a magnetic yoke lamination;

[0014] The visible light channel data and near-infrared penetration channel data of the image data are extracted, and the dual-channel data are fused using a dynamic weight allocation algorithm to generate fused data.

[0015] The mask matrix and fused data are input into the contour enhancement-noise separation network to calculate and output high-fidelity contour image data.

[0016] Extract the zeroth and first order geometric moments representing the centroid of the high-fidelity contour image data, as well as the second order central geometric moment representing the principal axis direction of the contour image, and correct them in real time in conjunction with the thermal deformation coefficient.

[0017] Based on the corrected geometric moments, the center coordinates and rotation angle of the magnetic yoke lamination are calculated.

[0018] Based on the corrected center coordinates and rotation angle, coordinate mapping is performed in combination with the preset array layout of the vacuum chuck on the adsorption plate. The priority weight of the mapped coordinates is adjusted according to the surface slot density of the magnetic yoke, and finally the adsorption point positioning coordinates of the vacuum chuck are generated.

[0019] Furthermore, the sensing and detection unit also includes an adsorption state detection component, which includes a flatness detection module and an adsorption force monitoring module.

[0020] The flatness detection module is used to detect the surface flatness data of the magnetic yoke punch near the positioning coordinates of the adsorption point before the vacuum suction cup performs the adsorption action.

[0021] The adsorption force monitoring module is used to monitor the real-time adsorption pressure value parameter during the grasping and stacking process after the vacuum suction cup adsorbs and grasps the magnetic yoke punch.

[0022] Furthermore, an adsorption force control unit is also provided on the adsorption plate. The adsorption force control unit is signal-connected to the flatness detection module and is used to generate an initial adsorption command for the vacuum suction cup by comparing the flatness data detected and calculated by the flatness detection module with a preset flatness deviation allowable threshold when receiving the flatness data. The initial adsorption command is:

[0023] When the surface flatness data meets the allowable flatness deviation threshold, the vacuum suction cup is controlled to adsorb the magnetic yoke punch at a first adsorption pressure, where the first adsorption pressure is the standard adsorption pressure threshold.

[0024] When the surface flatness data does not meet the allowable threshold for flatness deviation, the adsorption force control unit calculates a compensation pressure value based on the surface flatness data, and controls the vacuum suction cup to adsorb the magnetic yoke blank with a second adsorption pressure based on the compensation pressure value.

[0025] Furthermore, the adsorption force control unit is also signal-connected to the adsorption force monitoring module, and is used to generate a dynamic adsorption compensation command for the vacuum suction cup based on a safe adsorption pressure threshold when receiving the real-time adsorption pressure value parameter monitored by the adsorption force monitoring module; the dynamic adsorption compensation command is:

[0026] If the real-time adsorption pressure value is lower than the safe adsorption pressure threshold, the vacuum suction cup is controlled to increase the adsorption pressure until the real-time adsorption pressure value is not lower than the safe adsorption pressure threshold.

[0027] The safe adsorption pressure threshold includes a first safe adsorption pressure threshold corresponding to a first adsorption pressure and a second safe adsorption pressure threshold corresponding to a second adsorption pressure, wherein the second safe adsorption pressure threshold is greater than the first safe adsorption pressure threshold.

[0028] Specifically, the flatness detection module includes multiple laser distance sensors arrayed around each of the vacuum suction cups, and the adsorption force monitoring module includes vacuum pressure sensors correspondingly located inside each of the vacuum suction cups.

[0029] Furthermore, the stacking module includes a stacking platform, a support plate is provided in the middle of the upper part of the stacking platform, and a central positioning column is vertically provided in the middle of the support plate.

[0030] Furthermore, stacking alignment units are provided around the support plate. The stacking alignment units include stacking alignment rods arranged in a circular array, and the bottom end of the stacking alignment rods is provided with a radial synchronous telescopic mechanism.

[0031] Furthermore, the radial synchronous telescopic mechanism includes a rotating plate rotatably sleeved around the outside of the support plate. On the upper part of the rotating plate, there are multiple slide rail modules arranged in the radial direction and correspondingly stacked and aligned with the rods. Each slide rail module is connected to the bottom end of the corresponding stacked and aligned rod through a slide table that slides with it.

[0032] Preferably, an auxiliary wheel that contacts the outer edge contour of the magnetic yoke punch is rotatably sleeved outside the stacking alignment rod. A flexible grinding belt is attached to the outer wheel surface of the auxiliary wheel, and a pressure sensor is attached to the inside of the flexible grinding belt.

[0033] This invention employs a multimodal visual positioning component, including visible light and near-infrared fusion and a calculation unit, combined with a mask matrix and thermal deformation coefficient, to achieve high-precision calculation of the magnetic yoke lamination pose; and combines an adsorption state detection component and an adsorption force control unit to achieve active adsorption control from initial graded pressure to dynamic closed-loop compensation, cleverly solving the problems of inaccurate grasping and positioning and weak adsorption caused by reflection, slots, warping and vibration leakage;

[0034] Meanwhile, by employing a movable vacuum suction cup linked to the solution coordinates, which is adjusted and moved accordingly through a moving module, the adsorption point is ensured to accurately avoid the slots and land in the solid area, achieving stable gripping and adsorption. Furthermore, by using the central positioning column to ensure the rigid coaxiality benchmark of the stack, a stacking alignment unit with flexible force control and online deburring functions is used for secondary fine adjustment, which significantly improves the stacking accuracy of the magnetic yoke blanks and greatly improves the effect of the device. This results in improved coaxial accuracy, structural consistency and surface quality of the final stacked magnetic yoke blank assembly. Attached Figure Description

[0035] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings:

[0036] Figure 1 This is a schematic diagram of the overall structure of the gripping and stacking device of the present invention;

[0037] Figure 2 This is a signal block diagram of the visual positioning component of the present invention;

[0038] Figure 3 This is a bottom view of the adsorption plate structure of the present invention;

[0039] Figure 4 This is a signal block diagram of the adsorption state detection component of the present invention;

[0040] Figure 5 This is a top view of the rotating plate of the present invention.

[0041] Figure 6 For the present invention Figure 1 A partially enlarged schematic diagram of the auxiliary wheel structure.

[0042] In the above figures, the reference numerals represent: 1. Base frame; 2. Storage module; 3. Stacking module; 31. Stacking platform; 32. Support plate; 321. Central positioning post; 33. Stacking alignment rod; 341. Rotating plate; 342. Slide rail module; 343. Slide table; 35. Auxiliary wheel; 36. Flexible grinding belt; 37. Pressure sensor; 4. Gripping module; 41. Adsorption plate; 42. Vacuum suction cup; 431. Image acquisition device; 432. Laser distance sensor; 433. Vacuum pressure sensor; 44. Moving module. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. The illustrative embodiments and descriptions of this invention are for illustrative purposes only and are not intended to limit the invention. The embodiments described below are some, but not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0044] In the following description, numerous specific details are set forth to provide a thorough understanding of the invention. However, it will be apparent to those skilled in the art that these specific details are not necessary to practice the invention. In other embodiments, well-known structures, materials, or methods are not specifically described to avoid obscuring the invention. Unless otherwise specified, the materials, instruments, and reagents used in the following embodiments are commercially available. Unless otherwise specified, the techniques used in the embodiments are conventional methods well known to those skilled in the art.

[0045] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0046] Example 1:

[0047] Please refer to the following: Figure 1 and Figure 2As shown, this embodiment discloses a gripping and stacking device for magnetic yoke laminations, and provides an overall description of it. Specifically, the device includes a base frame 1, with a storage module 2 and a stacking module 3 respectively located at both ends of the base frame 1. A gripping module 4 is located on top of the storage module 2 and the stacking module 3. The gripping module 4 is connected to a gantry robot mounted on the base frame 1.

[0048] The grasping module 4 includes an execution unit, which includes a rotatable adsorption plate 41 and a vacuum suction cup 42 arrayed around the bottom of the adsorption plate 41. The adsorption plate 41 is also provided with a sensing and detection unit, which includes a visual positioning component.

[0049] The visual positioning component includes an image acquisition unit 431 and a calculation unit; the image acquisition unit 431 is used to acquire image data of the magnetic yoke blank when the execution unit grasps and adsorbs the magnetic yoke blank and transmit it to the calculation unit, wherein the image data includes visible light channel data and near-infrared transmission channel data.

[0050] The calculation unit calculates and generates the positioning coordinates of the adsorption point of the vacuum suction cup 42 based on the image data;

[0051] Each of the vacuum suction cups 42 is connected to the adsorption plate 41 via a moving module 44, and the moving module 44 is signal-connected to the calculation unit to drive the vacuum suction cup 42 to move to the adsorption point positioning coordinate position of the vacuum suction cup 42 generated by the calculation unit.

[0052] It should be noted that in existing magnetic yoke lamination gripping and stacking devices, the visual positioning process generally relies on a single visible light image for adsorption point location. However, this positioning method, when faced with magnetic yoke laminations in actual working conditions, has a limited sensing dimension and struggles to address the differences in slot distribution on the lamination surface, as well as slight thermal deformation and edge warping after stamping. Furthermore, the insulating coating on the lamination surface is prone to high-intensity reflection under visible light, leading to the loss of image features. Simultaneously, since the execution units of existing devices are mostly fixed array suction cups, they struggle to accurately locate the complex slot distribution on the surface of the magnetic yoke laminations during gripping and adsorption. This causes the suction cup array to easily misplace over the slots, failing to form an effective seal and ultimately resulting in the magnetic yoke laminations becoming loosely adhered and detaching, leading to poor device performance.

[0053] Therefore, based on the above problems, this embodiment explores the feasibility of combining multimodal fusion perception with flexible execution. Specifically, this solution introduces a multimodal visual positioning component, combining visible light channel data and near-infrared penetration channel data, to solve the problem of adsorption point positioning deviation caused by relying on a single visible light image in the prior art. Furthermore, by dynamically adjusting the position of the vacuum chuck 42, the adsorption point can adapt in real time to the differences in slot distribution and thermal deformation of the magnetic yoke blanks, thereby improving the accuracy of adsorption point positioning. In addition, the rotatable setting of the adsorption plate 41 and the coordinated operation of the arrayed vacuum chuck 42 enhance the adaptability of the gripping module 4 to magnetic yoke blanks with different postures, ensuring the stability of the stacking process.

[0054] Its working principle is as follows: This device achieves automated gripping and stacking of magnetic yoke laminations through the coordinated operation of the base frame 1, storage module 2, stacking module 3, and gripping module 4. The gripping module 4, as the core actuator, is driven by a gantry robot to move between the storage module 2 and the stacking module 3, ensuring spatial coverage of the operation. The execution unit in the gripping module 4 includes a rotatable adsorption plate 41 and vacuum suction cups 42 arrayed around the bottom of the adsorption plate 41. The rotational characteristics of the adsorption plate 41 allow it to adjust its angle according to the posture of the magnetic yoke laminations, thereby improving the adaptability of the adsorption. Furthermore, the visual positioning component in the sensing and detection unit plays a crucial role, and the image acquisition unit 4... During the grasping process, the visible light channel data and near-infrared penetrating channel data of the magnetic yoke blank are collected simultaneously. These two types of data capture surface details and internal structural information, respectively, effectively addressing the problem of differences in slot distribution. The collected image data is transmitted to the calculation unit, which calculates and generates the adsorption point positioning coordinates of the vacuum suction cup 42. This process compensates for the positioning deviation caused by thermal deformation through multimodal data fusion. Each vacuum suction cup 42 is connected to the adsorption plate 41 through the moving module 44, and the moving module 44 is signal-connected to the calculation unit. Based on the adsorption point positioning coordinates generated by the calculation unit, the moving module 44 drives the vacuum suction cup 42 to dynamically adjust to the target position, thereby achieving precise adsorption of the magnetic yoke blank.

[0055] Thus, by combining multi-source data acquisition, precise coordinate calculation, and dynamic position adjustment, the problem of adsorption point positioning deviation caused by relying on a single visible light image is solved, significantly improving the accuracy and reliability of grasping and stacking, and enhancing the performance of the stacking device.

[0056] It is also understandable that, in this embodiment, the positioning coordinates of the adsorption point of the vacuum chuck 42 generated by the calculation unit refer to a target physical position of the vacuum chuck 42 on the adsorption plate 41, which is the best adsorption position that can accurately align with the magnetic yoke blank and achieve stable adsorption when adsorbing the magnetic yoke blank.

[0057] The rotatable setting of the adsorption plate 41 is preferably achieved by a rotating mechanism, such as using a servo motor and a reducer to drive the adsorption plate 41 to adjust the angle, or using a pneumatic rotary cylinder to achieve its rotation function, mainly to adapt to the posture changes of the magnetic yoke blank.

[0058] The array arrangement of the vacuum suction cups 42 can be a regular arrangement, such as a rectangular array or a circular array. Alternatively, an irregular array layout can be designed based on the slot distribution characteristics of the magnetic yoke laminations to enhance the flexibility of adsorption. In this embodiment, a circular array arrangement is preferred. Figure 3 As shown.

[0059] The visible light channel data and near-infrared transmission channel data of the image acquisition device 431 can be realized through a dual-camera system, where one camera is responsible for capturing visible light images and the other camera is responsible for acquiring near-infrared images, or a single camera can be used in conjunction with a filter switching device to achieve dual-channel data acquisition.

[0060] The mobile module 44 preferably adopts an XY axis micro linear motor module. Each vacuum suction cup 42 corresponds to an independent XY axis micro linear motor module and is connected to the calculation unit via an Ethernet bus (or wireless gateway). The suction point positioning coordinates (X / Y axis absolute coordinates) output by the calculation unit are sent to each module in real time via the bus, thereby driving the vacuum suction cup 42 to move to the corresponding target position.

[0061] Obviously, in a further embodiment, the calculation unit accurately calculates the image data of the magnetic yoke lamination under the non-ideal working condition to ensure the accuracy and reliability of the subsequent device's positioning coordinates of the magnetic yoke lamination's adsorption point. To make the technical solution of this invention clearer, this embodiment specifically details the process by which the calculation unit calculates and generates the positioning coordinates of the vacuum suction cup 42 based on the image data. More specifically, it includes the following six sub-steps:

[0062] The first sub-step involves obtaining the surface slot density of the standard image of the magnetic yoke lamination and constructing a mask matrix.

[0063] The second sub-step involves extracting the visible light channel data and near-infrared penetration channel data from the image data, and then using a dynamic weight allocation algorithm to perform multimodal fusion of the dual-channel data to generate fused data.

[0064] The third sub-step involves inputting the mask matrix and fused data into the contour enhancement-noise separation network to calculate and output high-fidelity contour image data.

[0065] The fourth sub-step involves extracting the zeroth and first order geometric moments representing the centroid of the high-fidelity contour image data, as well as the second order central geometric moment representing the principal axis direction of the contour image, and then correcting them in real time using the thermal deformation coefficient.

[0066] The fifth sub-step involves calculating the center coordinates and rotation angle of the magnetic yoke lamination based on the corrected geometric moments.

[0067] The sixth sub-step involves mapping the coordinates of the vacuum suction cup 42 on the adsorption plate 41 based on the corrected center coordinates and rotation angle, combined with the preset array layout of the vacuum suction cup 42. The priority weight of the mapped coordinates is then adjusted according to the surface slot density of the magnetic yoke, ultimately generating the adsorption point positioning coordinates of the vacuum suction cup 42.

[0068] Understandably, in this embodiment, the mask matrix refers to an identifier matrix constructed based on the distribution characteristics of slots on the surface of the magnetic yoke lamination. It can be implemented using a pixel-level annotation method based on slot density, with the aim of providing a structured basis for subsequent weight adjustment of adsorption points.

[0069] The dynamic weight allocation algorithm can be understood as a method for adaptively adjusting the contribution ratio of the visible light channel and the near-infrared penetration channel. It can be achieved through a feedback mechanism based on ambient light intensity and slot density, aiming to optimize the dual-channel data fusion effect to cope with complex working conditions.

[0070] Contour Enhancement-Noise Separation Network is a deep learning model that guides the network to focus on effective contour regions and suppress slot interference through a mask matrix, while separating contour edges from background noise, thereby improving the accuracy of geometric feature extraction.

[0071] The thermal deformation coefficient is a correction parameter used to compensate for deformation errors caused by temperature changes during the stamping process. It can be obtained through experimental calibration or simulation analysis, with the aim of ensuring that the geometric moment reflects the true physical state.

[0072] In detail, this solution systematically solves the adsorption point positioning deviation problem through the multimodal image fusion and thermal deformation compensation mechanism described above, enabling the calculation unit to perform high-precision and robust positioning calculations on the image data of the magnetic yoke lamination. Specifically,

[0073] In this embodiment, the processing unit fuses the visible light and near-infrared dual-channel data in the image data using a dynamic weight allocation algorithm. This allows for real-time adjustment of weights based on reflective areas (e.g., enhancing the near-infrared channel weight in reflective areas to penetrate highlights), mitigating reflective interference from the insulating coating. This provides fused data with both detail integrity and anti-interference capabilities for subsequent processing. Then, a mask matrix is ​​constructed from the standard image of the magnetic yoke lamination. Combined with a contour enhancement-noise separation network, this network performs targeted separation and enhancement of effective contours (slot edges, lamination boundaries) and noise (burrs, reflective artifacts) in the fused data based on the standard slot density, resulting in the output. High-fidelity contour image data lays the foundation for geometric feature extraction; then, through real-time coupling correction of geometric moments and thermal deformation coefficients, the contour offset caused by thermal deformation of the stamping can be dynamically compensated, ensuring that the calculation benchmark of the center coordinates and rotation angles conforms to the actual shape; finally, based on the priority weight adjustment of slot density, the mapped suction cup coordinates avoid the high-density slot area on the surface of the magnetic yoke stamping and prioritize the solid area, thereby providing accurate and reliable position for the subsequent moving module 44 and adsorption force control unit, thus solving the problem that the fixed array suction cup is prone to erroneously adsorbing the slots and deformed warped areas on the surface of the magnetic yoke stamping, resulting in unstable adsorption.

[0074] It should be noted that, based on the above embodiments, the device ensures that the positioning coordinates of the adsorption point of the vacuum suction cup 42 can accurately fall on the solid area of ​​the magnetic yoke blank through the visual positioning component and the calculation unit, so as to ensure that the vacuum suction cup 42 can accurately adsorb and grasp the magnetic yoke blank. However, as is well known, the magnetic yoke blank, as a thin-walled stamped part, inevitably undergoes thermal deformation, warping or surface unevenness during stamping and transportation. At the same time, there is also a risk of adsorption pressure leakage due to vibration during high-speed grasping and movement.

[0075] Therefore, to further solve this technical problem, this embodiment proposes a preferred implementation of the sensing and detection unit, namely, the sensing and detection unit further includes an adsorption state detection component, and the adsorption state detection component includes a flatness detection module and an adsorption force monitoring module.

[0076] The flatness detection module is used to detect the surface flatness data of the magnetic yoke punch near the positioning coordinates of the adsorption point before the vacuum suction cup 42 performs the adsorption action.

[0077] The adsorption force monitoring module is used to monitor the real-time adsorption pressure value parameter during the grasping and stacking process after the vacuum suction cup 42 adsorbs and grasps the magnetic yoke punch.

[0078] As should be understood, in this embodiment, the purpose of the flatness detection module is to obtain the microscopic morphology information of the magnetic yoke sheet surface, thereby predicting the risk of adsorption.

[0079] The adsorption force monitoring module can collect the changes in internal adsorption pressure of the vacuum suction cup 42 in real time after adsorption, so as to capture the adsorption pressure fluctuations caused by vibration or displacement.

[0080] Specifically, the solution described in the above embodiments further achieves comprehensive monitoring and dynamic adjustment of the adsorption process by introducing an adsorption state detection component. That is, before the adsorption action is executed (while the vacuum suction cup 42 is still moving towards the attachment point positioning coordinate), the flatness detection module scans and detects the surface near the attachment point positioning coordinate of the magnetic yoke blank and outputs surface flatness data of the magnetic yoke blank. Based on the surface flatness data detected by the output, it can be determined whether there are uneven areas nearby, thereby triggering a pressure compensation mechanism to avoid adsorption failure due to poor local contact. The adsorption force monitoring module can further monitor the adsorption pressure value parameter after adsorption is completed. When pressure fluctuation is detected, the device can adjust the output in time according to the real-time feedback information to prevent the yoke detachment caused by sudden pressure drop or the magnetic yoke detachment caused by pressure accumulation. Thus, through the coordinated work of the above two modules, this embodiment enables the device's sensing capability to achieve a complete closed loop from the prediction of the physical state before grasping and adsorption to the dynamic process monitoring during grasping and adsorption. It also cleverly combines surface state perception with dynamic pressure monitoring, making the adsorption process change from passive execution to active adaptation, which significantly improves the stability and safety of grasping and stacking.

[0081] Furthermore, it is understandable that the adsorption state detection component and the execution unit in the gripping module 4 are linked. Its flatness detection module starts scanning when the adsorption plate 41 rotates to the target position to ensure that the positioning coordinates of the adsorption point are accurate. The adsorption force monitoring module enters the working state immediately after the vacuum suction cup 42 completes the adsorption action. Through signal interaction with the moving module 44, it ensures that the adsorption pressure is always maintained within a safe range. This design effectively solves the adsorption reliability problem caused by changes in surface conditions, while improving the stacking accuracy and usage effect of the device.

[0082] In a further preferred embodiment, such as Figure 4 As shown, an adsorption force control unit is also provided on the adsorption plate 41. The adsorption force control unit is signal-connected to the flatness detection module and is used to generate an initial adsorption command for the vacuum suction cup 42 by comparing the flatness data detected and calculated by the flatness detection module with a preset flatness deviation allowable threshold when receiving the flatness data. The initial adsorption command is:

[0083] When the surface flatness data meets the allowable threshold for flatness deviation, the vacuum suction cup 42 is controlled to adsorb the magnetic yoke punch at a first adsorption pressure, where the first adsorption pressure is the standard adsorption pressure threshold.

[0084] When the surface flatness data does not meet the allowable threshold for flatness deviation, the adsorption force control unit calculates a compensation pressure value based on the surface flatness data, and controls the vacuum suction cup 42 to adsorb the magnetic yoke blank with a second adsorption pressure based on the compensation pressure value.

[0085] Understandably, in this embodiment, the adsorption force control unit refers to a control module that can dynamically adjust the output command according to the input data, which can be implemented using an embedded controller or an industrial computer.

[0086] Flatness data refers to the quantitative value used to specifically characterize the deviation of the actual flatness from the ideal flatness (i.e., the zero-deviation benchmark is completely flat).

[0087] The allowable threshold for flatness deviation refers to a pre-set numerical standard used to judge whether the surface condition of the magnetic yoke lamination (i.e., its flatness difference) meets the adsorption requirements (this threshold can be calibrated to represent the maximum allowable unevenness, such as 1 mm), which can be determined through pre-experimental calibration.

[0088] In detail, the adsorption force control unit acquires the flatness data of the magnetic yoke sheet in real time through a signal connection with the flatness detection module, and compares it with a preset allowable flatness deviation threshold to generate a comparison result. Based on the comparison result, the vacuum chuck 42 is controlled to generate a graded initial adsorption command. Furthermore, the graded initial adsorption command includes the following two cases:

[0089] In one scenario, when the flatness data meets the preset flatness deviation allowable threshold (for example, the calculated flatness data is 0.5mm, which is less than the allowable threshold of 1mm), it indicates that the flatness data (deviation) near the adsorption point of the magnetic yoke is within the allowable range. Therefore, the suction cup can easily form a good seal. In this case, the adsorption force control unit can control the vacuum suction cup 42 to perform adsorption operation with the first adsorption pressure (that is, the standard adsorption pressure threshold).

[0090] In the second scenario, when the flatness data does not meet the preset allowable flatness deviation threshold (for example, the calculated flatness data is 1.2mm, which is greater than the allowable threshold of 1mm), it indicates that the flatness data (deviation) near the adsorption point of the magnetic yoke sheet exceeds the allowable range of unevenness. In this case, the vacuum suction cup 42 needs a larger initial adsorption pressure to overcome the gap and force a fit to form a stable adsorption seal. Thus, the adsorption force control unit will calculate a compensation pressure value based on the current flatness data and output a second adsorption pressure based on the compensation pressure value to perform the adsorption operation. It can be understood that the second adsorption pressure is the sum of the first adsorption pressure (standard adsorption pressure) and the compensation pressure value.

[0091] Therefore, the technical solution described in this embodiment solves the problem of adsorption reliability caused by changes in the surface flatness of the magnetic yoke blanks, realizes dynamic adaptive adjustment of adsorption pressure, and significantly improves the working efficiency and stacking accuracy of the gripping and stacking device.

[0092] In a further preferred embodiment, Figure 4 As shown, the adsorption force control unit is also signal-connected to the adsorption force monitoring module and is used to generate a dynamic adsorption compensation command for the vacuum suction cup 42 based on a safe adsorption pressure threshold when receiving the real-time adsorption pressure value parameter monitored by the adsorption force monitoring module; the dynamic adsorption compensation command is:

[0093] If the real-time adsorption pressure value is lower than the safe adsorption pressure threshold, the vacuum suction cup 42 is controlled to increase the adsorption pressure until the real-time adsorption pressure value is not lower than the safe adsorption pressure threshold.

[0094] The safe adsorption pressure threshold includes a first safe adsorption pressure threshold corresponding to a first adsorption pressure and a second safe adsorption pressure threshold corresponding to a second adsorption pressure, wherein the second safe adsorption pressure threshold is greater than the first safe adsorption pressure threshold.

[0095] Understandably, in this embodiment, the safe adsorption pressure threshold refers to the minimum pressure standard set in advance to determine whether the adsorption state of the magnetic yoke sheet is safe during dynamic transport (this threshold represents the minimum suction amplitude that must be maintained). Therefore, the second safe adsorption pressure threshold recorded in this solution is greater than the first safe adsorption pressure threshold, which means that the suction amplitude (or absolute value) corresponding to the second safe adsorption pressure threshold is greater than the suction amplitude (or absolute value) corresponding to the first safe adsorption pressure threshold.

[0096] In detail, the adsorption force control unit, through a signal connection with the adsorption force monitoring module, acquires the real-time adsorption pressure value parameter of the magnetic yoke after the initial adsorption is completed, and continuously compares it with a preset safe adsorption pressure threshold. At this time, the control unit will automatically retrieve the corresponding first or second safe adsorption pressure threshold as the currently effective safe adsorption pressure threshold based on the initial adsorption record for comparison, so as to generate a comparison result. Based on the comparison result, the control unit controls the vacuum suction cup 42 to generate a dynamic compensation command. Furthermore, the logic of this dynamic compensation command includes the following two cases:

[0097] In the first scenario, when the real-time adsorption pressure value is not lower than the preset safe adsorption pressure threshold, it indicates that the magnetic yoke is in a safe and stable adsorption state at the current moment, meaning the adsorption force is sufficient. For example, if the real-time adsorption pressure value is greater than or equal to the safe adsorption pressure threshold, the control unit determines that no intervention is needed. In this case, the adsorption force control unit does not trigger any compensation action, but only maintains continuous monitoring of the real-time adsorption pressure value. This ensures adsorption safety while avoiding unnecessary energy consumption.

[0098] In the second scenario, if the real-time adsorption pressure value is lower than the preset safe adsorption pressure threshold, it means that the magnetic yoke's adsorption state at the current moment is dangerous and unstable. In other words, if the amplitude of the real-time adsorption pressure value is lower than the safe adsorption pressure threshold, it indicates that pressure leakage has occurred and there is a risk of detachment. In this case, the vacuum suction cup 42 needs to immediately increase the adsorption pressure to re-ensure the seal and adsorption force. As a result, the adsorption force control unit will immediately trigger a dynamic adsorption compensation command, controlling the corresponding vacuum pump or proportional valve to continuously increase the adsorption pressure and continuously monitor the real-time adsorption pressure value until its pressure value amplitude returns to a level not lower than the safe adsorption pressure threshold.

[0099] Therefore, based on the technical solution described in the above embodiments, it solves the problem of dynamic detachment of magnetic yoke blanks caused by vibration or leakage during high-speed transport, realizes the fully closed-loop dynamic adaptive adjustment of the adsorption process, and significantly improves the working reliability and safety of the gripping and stacking device.

[0100] Furthermore, it can be understood that the safe adsorption pressure threshold in this embodiment is not fixed, but needs to be set according to the initial adsorption command (i.e., the first adsorption pressure and the second adsorption pressure), which includes the following two types:

[0101] The first safe adsorption pressure threshold is the first adsorption pressure applied to the magnetic yoke blanks that are "flat" for adsorption.

[0102] The second safe adsorption pressure threshold is the second adsorption pressure applied to the adsorption of "uneven" magnetic yoke blanks.

[0103] This is because the magnetic yoke piece adsorbed under the first adsorption pressure is relatively flat, so its adsorption state is relatively stable. However, the magnetic yoke piece adsorbed under the second adsorption pressure is not flat, so its adsorption state is naturally unstable and riskier. Therefore, a higher safety lower limit (i.e., the second safe adsorption pressure threshold) must be adopted to ensure adsorption reliability. Thus, this solution sets a dedicated safety threshold (i.e., the first safe adsorption pressure threshold and the second safe adsorption pressure threshold) for these two different initial adsorption commands, thereby achieving differentiated and highly reliable safety monitoring.

[0104] In some preferred embodiments, such as Figure 3 As shown, the flatness detection module includes multiple laser distance sensors 432 arrayed around each of the vacuum suction cups 42, and the adsorption force monitoring module includes vacuum pressure sensors 433 correspondingly disposed inside each of the vacuum suction cups 42.

[0105] Understandably, in this embodiment, the laser distance sensor 432 refers to a sensor that non-contactly measures the distance to a target by emitting and receiving laser beams. The array of the laser distance sensors 432 is arranged in a circular manner around the edge of the vacuum chuck 42, so that it can comprehensively and in real time measure the distance parameters between the sensor and the magnetic yoke piece located near the chuck, and convert the measured distance data into corresponding flatness data.

[0106] Vacuum pressure sensor 433 refers to a sensor used to accurately measure the gas pressure (i.e., vacuum degree) below atmospheric pressure in a sealed cavity. It has the characteristics of high sensitivity and fast response. The corresponding placement inside each vacuum suction cup 42 means that a single such sensor is directly integrated into the internal channel of the vacuum suction cup 42 for forming a sealed cavity, so that it can monitor the real pressure value in the sealed cavity formed by the suction cup in real time and independently.

[0107] In a further embodiment, such as Figure 1 and Figure 5 As shown, the stacking module 3 includes a stacking platform 31, a support plate 32 is provided in the middle of the upper part of the stacking platform 31, and a central positioning column 321 is vertically provided in the middle of the support plate 32.

[0108] Understandably, in this embodiment, the stacking platform 31 refers to the base platform used to carry and receive magnetic yoke laminations. It can be made of metal or high-strength composite materials, with the aim of providing a stable stacking support environment.

[0109] The support plate 32 is a component located in the middle above the stacking platform 31. It can be fixed to the stacking platform 31 by bolts or welding. Its purpose is to provide stacking support for the magnetic yoke laminations and to provide a high-precision reference point for subsequent positioning.

[0110] The center positioning post 321 is a positioning component that is vertically set in the middle of the support plate 32. It can adopt a cylindrical or prismatic structure. It is the core reference component for achieving high-precision stacking. Its physical center coordinates are accurately measured during equipment calibration and stored as a reference point to achieve tight fit with the center hole of the magnetic yoke during stacking and achieve precise alignment.

[0111] In practice, after the gripping module 4 has successfully gripped and adsorbed the magnetic yoke lamination, the gantry robot moves the gripping module 4 directly above the stacking module 3, ensuring that the calculated center coordinates of the lamination are perfectly aligned with the reference point coordinates of the central positioning post 321 on the horizontal plane. Once aligned, the gripping module 4 descends vertically, allowing the center hole of the magnetic yoke lamination to fit into the central positioning post 321, until the lamination is stably placed on the support plate 32 (or the already stacked laminations). Subsequently, the vacuum suction cup 42 releases pressure, and the gripping module 4 rises to its original position, ready for the next cycle of gripping and stacking. This significantly improves the coaxiality of the device during the stacking process of the magnetic yoke laminations. Thus, through the arrangement of the stacking platform 31, the support plate 32, and the central positioning post 321, precise stacking of the gripped and adsorbed magnetic yoke laminations is achieved.

[0112] In some preferred embodiments, Figure 1 and Figure 5 As shown in the figure, stacking alignment units are also provided around the support plate 32. The stacking auxiliary alignment units include stacking alignment rods 33 arranged in a ring array. The bottom end of the stacking alignment rods 33 is also provided with a radial synchronous telescopic mechanism.

[0113] Understandably, the stacked alignment rods 33 refer to the rod-shaped components distributed around the support plate 32 to provide circumferential constraints, and their position can be adjusted by a mechanical transmission structure or an electric push rod.

[0114] The radial synchronous telescopic mechanism is an actuator that can independently telescopically extend and retract along the radial direction of the support plate 32. It can be driven by hydraulic, pneumatic or electric motor. Its purpose is to adjust the position of the alignment rod in real time according to the actual contour change of the magnetic yoke lamination, thereby compensating for the small displacement caused by thermal deformation or position deviation.

[0115] Specifically, in this scheme, the stacking alignment unit forms a surrounding constraint frame through the stacking alignment rods 33 arranged in a ring array, thereby covering the outer edge area of ​​the magnetic yoke lamination. This allows it to disperse local stress based on the real-time position feedback during lamination stacking, using a multi-point uniformly distributed support force to avoid uneven stress caused by differences in slot density. At the same time, the radial synchronous telescopic mechanism at the bottom of the stacking alignment rods 33 serves as a core component, allowing the alignment rods to extend and retract along the radial direction of the support plate 32, thereby actively conforming to the outer edge contour of the magnetic yoke lamination, ensuring that each lamination remains precisely concentric with the central positioning post 321 during the stacking process.

[0116] It should also be understood that the workflow is as follows: when the gripping module 4 inserts a lamination into the central positioning post 321 and releases it, the stacking alignment unit is activated. Each radial synchronous telescopic mechanism drives the stacking alignment rod 33 to move radially towards the center simultaneously, slightly touching and pushing the outer edge of the newly placed lamination until its outer contour is completely aligned with the outer contour of the laminations already stacked below, thereby significantly improving stacking coaxiality and overall accuracy. Furthermore, the movement of the stacking alignment rod 33 via the radial synchronous telescopic mechanism also allows for adjustment of the contour size of the surrounding constraint frame formed by the stacking alignment rod 33, thus adapting it to magnetic yoke laminations of different sizes and improving the applicability of the device.

[0117] In a further embodiment, please refer to Figure 5 As shown, the radial synchronous telescopic mechanism includes a rotating plate 341 rotatably sleeved around the outside of the support plate 32. On the upper part of the rotating plate 341, there are multiple slide rail modules 342 arranged in the radial direction and correspondingly stacked and aligned with the rods 33. Each slide rail module 342 is connected to the bottom end of the corresponding stacked and aligned rod 33 through a slide table 343 that slides and cooperates with it.

[0118] Understandably, in this embodiment, the slide rail module 342 refers to a linear motion component with guiding function, which preferably adopts a linear guide rail, and its purpose is to provide precise radial movement path constraints for the stacking alignment rod 33.

[0119] The slide table 343 is a movable component that slides in conjunction with the slide rail module 342, such as a sliding seat. Its purpose is to support and install the stacking alignment rod 33, so that the slide table 343 drives the stacking alignment rod 33 to move radially through the drive of the slide rail module 342.

[0120] Understandably, through the coordinated arrangement of the rotating plate 341, the slide rail module 342, and the slide table 343, the limiting and correction design is achieved when stacking magnetic yoke laminations. Specifically, when the magnetic yoke laminations are placed and stacked, the slide rail module 342 of the radial synchronous telescopic mechanism can drive the slide table 343 to move radially on the rotating plate 341, thereby driving the stacking alignment rods 33 to move radially and gradually closer to the magnetic yoke laminations, so that they fit and abut against the outer edge contour of the laminations, forming a circumferential pre-constraint to ensure that the contact state between the stacking alignment rods 33 and the magnetic yoke laminations is stable and to prevent relative slippage during stacking. At the same time, the rotating plate 341 rotates and synchronously drives all the stacking alignment rods 33 to make circular motion around the center of the laminations. The stacking alignment rod 33 is in contact with the outer edge of the lamination, so the rotating plate 341 can apply a frictional force along the circumferential tangential direction to the magnetic yoke lamination during rotation. Simultaneously, if the magnetic yoke laminations are eccentrically stacked (i.e., stacking deviation occurs), their center of mass and rotation center will not coincide during rotation, resulting in a radial centrifugal inertial force. This centrifugal force, combined with the tangential frictional force of the stacking alignment rod 33, forms a composite corrective force, pushing the laminations towards the rotation center for fine-tuning until the center of mass and rotation center of the magnetic yoke laminations are completely aligned. At this point, the centrifugal force on the laminations is evenly distributed, and radial displacement deviation no longer occurs. This ultimately achieves dynamic coaxiality correction during the magnetic yoke lamination stacking process, significantly improving the structural accuracy and consistency of the magnetic yoke stacking assembly. Furthermore, this solution works organically with components such as the support plate 32, the central positioning column 321, and the stacking alignment rod 33 in the stacking module 3, jointly improving the overall performance and usability of the stacking device.

[0121] As a further embodiment, in Figure 5 and Figure 6 As shown in the figure, an auxiliary wheel 35 is rotatably sleeved outside the stacking alignment rod 33, which contacts the outer edge contour of the magnetic yoke punch. A flexible grinding belt 36 is attached to the outer wheel surface of the auxiliary wheel 35, and a pressure sensor 37 is attached inside the flexible grinding belt 36.

[0122] Understandably, in this embodiment, the auxiliary wheel 35 refers to a wheel-like component that can rotate freely around its center. Its purpose is to transform the sliding friction between the stacking alignment rod 33 and the outer edge of the lamination into rolling contact, so as to protect the lamination surface.

[0123] The flexible grinding belt 36 is a composite material belt attached to the outer surface of the auxiliary wheel 35. Its purpose is to provide flexible cushioning and to trim burrs on the punch during rolling contact.

[0124] The pressure sensor 37 is a sensing element that can detect contact pressure. It can be a thin-film pressure sensor 37 and is ring-shaped inside the flexible grinding belt 36. Its purpose is to monitor the pressure distribution when the auxiliary wheel 35 contacts the punch in real time.

[0125] It should be noted that when the magnetic yoke laminations are aligned by the central positioning post 321 and the stacking alignment rod 33, although uniform centripetal motion has been achieved through the radial synchronous telescopic mechanism, there are still some problems: if the end of the stacking alignment rod 33 is a rigid structure, when it is pushed in radially at high speed and contacts the outer edge of the lamination, it is very easy to scratch the insulating coating on the surface of the lamination due to rigid collision and sliding friction; at the same time, the tiny burrs generated on the outer edge of the lamination due to stamping cannot be processed, affecting the stacking quality.

[0126] Therefore, this embodiment achieves high-precision, non-destructive alignment of the magnetic yoke laminations through the coordinated design of the auxiliary wheel 35, the flexible grinding belt 36, and the pressure sensor 37. Specifically, after the magnetic yoke laminations are placed and stacked, when the stacking alignment rod 33 moves radially toward the center and rotates with the rotating plate 341 while pressing against the outer contour of the magnetic yoke lamination, the auxiliary wheel 35 at its end will first contact the outer contour edge of the lamination and then roll along the lamination contour, thereby avoiding scratches and damage caused by sliding friction between the stacking alignment rod 33 and the outer edge contour of the magnetic yoke lamination. It is also understandable that, since a flexible polishing belt 36 is attached to the outer surface of the auxiliary wheel 35, when the auxiliary wheel 35 rolls, the flexible polishing belt 36 provides flexible cushioning and can also polish and trim the tiny burrs on the outer edge of the magnetic yoke blank during the rolling process through its grinding surface. This eliminates the interference of burrs on the overlapping and bonding surface, ensuring that the edges are tightly fitted when the blanks are subsequently stacked, reducing the stacking thickness deviation caused by burrs, and further adapting to the slight unevenness of the blank outer edge through its elastic properties, avoiding local stress concentration caused by rigid contact and protecting the surface insulating coating. More importantly, the pressure sensor 37 inside the flexible polishing belt 36 can capture the dynamic changes in contact pressure in real time, ensuring that the contact pressure between the flexible polishing belt 36 and the outer edge of the magnetic yoke blank is always consistent. This avoids excessive contact pressure that could cause the blank to warp or deform or the insulating coating to be damaged, and also prevents insufficient pressure that could cause insufficient correction force or blank positioning deviation.

[0127] In a further preferred embodiment, the pressure sensor 37 is signal-connected to the slide rail module 342, so that the real-time contact pressure data it collects can be directly transmitted to the slide rail module 342, enabling it to finely adjust the displacement distance of its slide table 343 in real time based on the pressure data, so as to ensure that the pressure of each stacked alignment rod 33 and the contact point of the magnetic yoke punch edge is balanced, thus avoiding insufficient correction force and preventing punch deformation.

[0128] Therefore, through the above technical solution, this solution not only solves the alignment accuracy problem caused by edge burrs and local unevenness of the magnetic yoke laminations, but also achieves more precise stacking alignment control, significantly improving the usability and stacking effect of the stacking device, ensuring the coaxiality and surface quality of the magnetic yoke laminations during the stacking process, thereby improving the electromagnetic performance and structural stability of the final magnetic yoke assembly.

[0129] Understandably, the slide rail module 342 in this embodiment integrates a corresponding micro-drive unit (e.g., a high-precision servo motor or piezoelectric motor) and a controller. The controller receives real-time contact pressure data from the pressure sensor 37, compares it with a preset pressure threshold, and then controls the micro-drive unit in real-time (closed-loop) to achieve fine-tuning of the displacement of the slide table 343. It should also be noted that the various sensors and components involved in this embodiment are existing technologies, and all can be directly purchased from the market as mature products. Therefore, this embodiment does not elaborate on them, and their structure is not the subject of this application.

[0130] Example 2:

[0131] This embodiment is based on Embodiment 1. To facilitate further understanding of the technical solution of this embodiment, the calculation process of the compensation pressure value in Embodiment 1 is explained here. That is, when the surface flatness data does not meet the allowable threshold for flatness deviation, the adsorption force control unit calculates the compensation pressure value based on the surface flatness data using a linear proportional model. Specifically, the model expression is:

[0132] ,in, The calculated compensation pressure value, For flatness data, This is the preset allowable threshold for flatness deviation.

[0133] Based on this, an accurate linear compensation model is established using the above formula. This model is used to calculate the required pressure compensation value (i.e., the compensation pressure corresponding to the flatness deviation amount that exceeds the allowable range) for the detected flatness data that does not meet the preset allowable flatness deviation threshold. This ensures that the required compensation pressure is proportional to the range by which the flatness data exceeds the allowable threshold, thereby achieving the precise compensation control effect of the adsorption force control unit compensating as needed.

[0134] Example 3:

[0135] To facilitate a further understanding of the technical solution of this embodiment, this embodiment specifically provides a further explanation of the first sub-step of the calculation unit in Embodiment 1 above. In the first sub-step, the surface slot density of the acquired standard image of the magnetic yoke lamination is used to construct a mask matrix. This aims to utilize the standard image of the magnetic yoke lamination (such as a CAD drawing or gold sample image without burrs or reflective interference) as prior knowledge to construct a pixel-level reference map with a resolution completely consistent with the real-time acquired image. The core function of this mask matrix is ​​to pre-define the precise pixel positions of all slots (non-adsorption areas) and solids (adsorption areas) on the surface of the magnetic yoke lamination, thereby providing guiding information for the subsequent contour enhancement-noise separation network and providing a decision basis for adjusting the priority weights of the final adsorption point positioning coordinates. Specifically, the expression for constructing the mask matrix is:

[0136] ,

[0137] In the above formula, For pixel coordinates The generated mask matrix has dimensions that are strictly consistent with the resolution of the images acquired in real time by the subsequent image acquisition device to ensure pixel-level alignment.

[0138] These are the pixel coordinates of the standard image within the mask matrix;

[0139] Let be the set of all entity pixel distribution regions in the standard image, when When the pixel corresponds to the solid part of the magnetic yoke, it is a potential adsorption area for the subsequent vacuum chuck.

[0140] This is the set of all pixel distribution areas with slots in the standard image. When the value is displayed, it indicates that the pixel corresponds to the slot of the magnetic yoke lamination, which is an area that the vacuum chuck must avoid.

[0141] It should be noted that the above and The data source for the set is millimeter-level vector coordinates of slots and outer contours defined in standard images (such as CAD drawings). It is generated using a vector-to-pixel conversion algorithm to map physical dimensions to pixel coordinates. The conversion algorithm formula is expressed as follows:

[0142] ,

[0143] In the above formula, These are the millimeter-level vector coordinates of the slot or contour in the CAD drawing; they represent the raw data. and These are the resolution scaling factors (in pixels per millimeter) for the horizontal and vertical directions, respectively, representing the number of pixels per unit physical length (millimeters). These two factors were obtained in advance through camera calibration experiments to ensure the accuracy of the mapping from the physical world to the pixel world.

[0144] Understandably, the reason for constructing the mask matrix in this scheme is that the edges of the slots, as an inherent structure of the magnetic yoke lamination, are visually (especially in subsequent fusion data) easily confused with the actual outer contour of the workpiece, or with noise interference data such as burrs and scratches, leading to misjudgments in subsequent algorithms. By constructing the mask matrix, this scheme quantifies and encodes the prior information of the standard workpiece's structure, allowing subsequent processing sub-steps to use it as a guiding map to focus on learning and separating unexpected noise contours. At the same time, it provides non-adsorbable areas (slot areas) and adsorbable areas (solid areas) for the final adsorption point selection, thus fundamentally solving the adsorption failure problem caused by slot recognition deviation in existing devices, and providing a key benchmark reference for the accuracy and reliability of subsequent adsorption point positioning.

[0145] Example 4:

[0146] This embodiment further explains the second sub-step in Embodiment 1 above. In the second sub-step, the visible light channel data and near-infrared transmission channel data of the extracted image data are fused into fused data using a dynamic weight allocation algorithm. This aims to solve the problem of high reflectivity and artifact interference on the surface of the magnetic yoke lamination caused by insulating coatings, oil stains, or stamping burrs. Specifically, visible light channel data is good at capturing rich surface textures and edge details, but is easily affected by high light reflection saturation; while near-infrared transmission channel data can effectively penetrate surface coatings and some reflective areas to obtain more stable substrate information, but may lose details. This embodiment uses a dynamic weight allocation algorithm to adaptively adjust the fusion ratio of the two according to the characteristics of different regions of the image (especially reflectivity). Its multimodal fusion expression is:

[0147] ,

[0148] In the above formula, For pixel coordinates The fused data generated at that point, which is the fused image data, will be used as the input for the third sub-step.

[0149] Visible light channel data acquired by the image acquisition device Pixel value at;

[0150] Near-infrared transmission channel data acquired by the image acquisition device in Pixel value at;

[0151] For pixels The dynamic weighting coefficient has a value range of [0,1]. In this embodiment, the coefficient is calculated by using a linear normalized limiting function based on a brightness threshold. This function is based on the brightness value of the visible light channel. The weights are dynamically adjusted, and the calculation formula is as follows:

[0152] ,

[0153] In the above formula, It is a limiting function that limits the calculation result of the expression within parentheses. The function is restricted to the closed interval [0,1]. That is, if the result is less than 0, the function output is 0; if the result is greater than 1, the function output is 1.

[0154] and These are preset low brightness thresholds and high brightness (reflection) thresholds (e.g., in an 8-bit grayscale image). It can be set to 180. (Can be set to 240). These two thresholds define the transition range from fully trusting visible light to fully trusting near-infrared light, and they can be obtained through calibration tests on the reflective properties of typical workpieces.

[0155] Understandably, the reason for using this dynamic weighting algorithm is that when Too high brightness (i.e.) Located in a highly reflective area, higher than )hour, Approaching 0, at this point the data is fused. Near-infrared penetration channel data will be primarily used to leverage its penetrability and avoid reflective interference; while in non-reflective (normal) areas, The brightness is moderate. Approaching 1, at this point The visible light channel data will be primarily used to preserve rich contours and details, thereby producing fused data output in this step. It balances the ability to resist reflective interference with the integrity of details, thus providing high-quality raw data for subsequent contour extraction.

[0156] Example 5:

[0157] This embodiment further explains the third sub-step described in Embodiment 1 above. In this third sub-step, the mask matrix and fused data are input into the contour enhancement-noise separation network to calculate and output high-fidelity contour image data. The core purpose of this step is to utilize the mask matrix constructed in the first sub-step. As prior knowledge, it guides a specially trained deep learning network to process the fused data generated in the second sub-step. In this process, the true contours of the magnetic yoke laminations (outer edges and critical slot edges) are accurately separated, while suppressing noise such as stamping burrs, oil residue, and reflective artifacts. Specifically, the calculation output process expression is as follows:

[0158] ,

[0159] In the above formula, The high-fidelity contour image data output by the network computation, ideally contains only clear, continuous true contour information of the stamping, with pixel values ​​of 1 (contour) or 0 (background), and this data will be used as the input for the fourth sub-step;

[0160] The nonlinear mapping function represented by this contour enhancement-noise separation network can be a convolutional neural network architecture with encoding / decoding structure or attention mechanism, such as U-Net or Mask R-CNN.

[0161] For this network The network parameters (such as weights and biases) are solidified after training with a large amount of labeled data.

[0162] Understandably, traditional edge detection algorithms (such as Sobel and Canny) struggle to distinguish between the burr edges of magnetic yoke laminations and the edges of the workpiece, and are also unable to process fused data. There may be residual complex noise in the network. Through training, the morphological characteristics of the magnetic yoke laminations and the statistical characteristics of noise were learned, and the mask matrix was used to... The introduction of [the technology / mechanism] played a crucial guiding role: the network was trained to trust [the system / mechanism]. Mark the slot boundaries and reinforce them; at the same time, in Marked entity region ( Strong signals (such as burrs and bright spots) that do not conform to the morphological characteristics of the magnetic yoke laminations are identified as noise and suppressed. Therefore, this step can output high-fidelity contour image data with extremely high signal-to-noise ratio. This laid a precise foundation for subsequent calculations of geometric moments.

[0163] Example 6:

[0164] This embodiment further explains the fourth sub-step described in Embodiment 1 above. In this fourth sub-step, the zeroth and first-order geometric moments representing the centroid of the high-fidelity contour image, and the second-order central geometric moment representing the principal axis direction of the contour image, are extracted from the high-fidelity contour image and corrected in real time using the thermal deformation coefficient. This step aims to quantify the geometric characteristics of the contour mathematically and compensate for measurement deviations introduced by physical deformations (such as thermal expansion or slight warping) caused by stamping heat. Specifically, the process consists of two steps:

[0165] The first step is to calculate the original geometric moments, which is based on the output of the third sub-step. Calculate its original Order geometric moments and central moments The calculation formula is as follows:

[0166] ,

[0167] ,in,

[0168] , ,

[0169] In the above formula, and All are non-negative integers. The orders of the moment in the x and y coordinates are defined respectively, and their sums are... That is, the total order of the moment;

[0170] These are the original geometric moments and central moments calculated separately;

[0171] The high-fidelity contour image input for the third sub-step is in coordinates The pixel value at that location (usually 0 or 1);

[0172] These are the original centroid coordinates of the contour image.

[0173] Understandably, the solution unit in this embodiment achieves pose calculation by calculating specific low-order combinations of the aforementioned moments. Specifically, this step requires calculating the following specific moments:

[0174] Zeroth moment (i.e., order) ), representing the total area or total mass of the profile; first-order geometric moments (Right now and )and Combined with the original centroid used to calculate the above ;

[0175] Second-order central moment ( ), ( )and ( Together, they characterize the distribution pattern of the pixels in the contour map relative to the centroid and the direction of the principal axis;

[0176] The set of moments calculated above is the original set of moments. That is to say .

[0177] The second step involves correcting the original moment set by incorporating the thermal deformation coefficient. It is understood that thermal deformation after stamping may cause a deviation between the visual contour of the image and its reference shape at standard temperature, thus affecting the accuracy of the centroid and principal axis directions. Therefore, in this embodiment, a thermal deformation coefficient is introduced to correct the original moment set. Furthermore, this correction process includes two calculation stages:

[0178] In the first stage, the offset is calculated by inputting the thermal deformation coefficient and the original moment set into the correction model to calculate the moment offset caused by thermal deformation. The calculation expression is as follows:

[0179] ,

[0180] In the above formula, The set of moment offsets calculated for this model, its components (such as...) (etc.) and The components are in one-to-one correspondence.

[0181] This is a modified model, which is based on material properties (such as the coefficient of thermal expansion of silicon steel sheets) and the coefficient of thermal deformation;

[0182] The thermal deformation coefficient is obtained in real time, which can be obtained by temperature sensors, thermal imaging, or physical models based on process time, and is used to quantify the current degree of deformation (such as linear expansion rate or warpage vector).

[0183] In the second stage, an offset is applied for correction. The offset of the output moments is subtracted from the corresponding original moments to obtain the corrected set of geometric moments; its calculation expression is:

[0184] ,

[0185] in, , ,

[0186] In the above expression, For the corrected set of geometric moments, that is This will serve as the input for the fifth sub-step;

[0187] It should be noted that, since the thermal deformation (especially warping) of the magnetic yoke lamination is non-uniform, it is very easy for the calculated visual centroid to deviate from its true physical centroid. Therefore, in this embodiment, by calculating the offset and applying the offset for two-stage correction, this systematic deviation caused by thermal deformation can be compensated to ensure that the physical posture of the magnetic yoke lamination can be more realistically reflected, thereby achieving high-precision stacking coaxiality of the magnetic yoke lamination by the subsequent device.

[0188] Example 7:

[0189] This embodiment further explains the fifth sub-step in Embodiment 1 above. In the fifth sub-step, the center coordinates and rotation angle of the magnetic yoke lamination are calculated based on the corrected geometric moments. It can be understood that the purpose of this step is to convert the set of geometric moments output from the fourth sub-step, which has been corrected for thermal deformation, into pose parameters with clear physical meaning required for the grasping and stacking operations of the grasping module. That is, the pose parameters include the center coordinates and rotation angle. Specifically, the calculation formula includes the following:

[0190] , , ,

[0191] In the above formula, The calculated coordinates of the magnetic yoke lamination center (i.e., the corrected coordinates of the centroid) are used as the input for the sixth sub-step.

[0192] The calculated rotation angle of the magnetic yoke lamination, i.e. the angle between the principal axis of the contour and the X-axis of the image coordinate system, is also used as the input for the sixth sub-step.

[0193] These are the sets of geometric moments corrected from the fourth sub-step. The corrected zeroth and first order geometric moments in the data;

[0194] The set of geometric moments corrected from the fourth sub-step The corrected second-order central geometric moment in the model.

[0195] It should be noted that by inputting the corrected geometric moments into the above formula, instead of the original moments, and since the fourth sub-step has already compensated for the influence of thermal deformation on the geometric moments using the thermal deformation coefficient, the center coordinates and rotation angles in the pose parameters calculated from the set of corrected geometric moments in this step can accurately reflect the true physical center and principal axis direction of the magnetic yoke lamination in the current coordinate system, thus providing a benchmark for the accurate mapping of the adsorption point of the subsequent vacuum chuck.

[0196] Example 8:

[0197] This embodiment further explains the sixth sub-step described in Embodiment 1 above. In the sixth sub-step, coordinate mapping is performed based on the corrected center coordinates and rotation angle, combined with the preset array layout of the vacuum chuck on the adsorption plate. The mapped coordinates are then adjusted in priority according to the surface slot density of the magnetic yoke lamination, ultimately generating the adsorption point positioning coordinates of the vacuum chuck. That is, this step is the final output step of the calculation unit, aiming to provide the gripping module with a precise target gripping point for each vacuum chuck, ensuring that the chuck can accurately and reliably adsorb onto the solid area of ​​the lamination, avoiding slots. Specifically, this process includes two stages:

[0198] In the first stage, coordinate mapping is performed to calculate the ideal landing point coordinates of the suction cup. The calculation expression is as follows:

[0199] ,

[0200] in, For the first The preset coordinates of each vacuum suction cup in the adsorption plate coordinate system are the inherent calibration parameters of the device.

[0201] These are the center coordinates and rotation angle input in the fifth sub-step, respectively;

[0202] For the calculated first The ideal landing point coordinates of the suction cup in the image coordinate system, and its coordinate components and These are the X and Y coordinates of the ideal mapped coordinates in the image coordinate system, respectively. It can be understood that the above formula represents the coordinates of the vacuum suction cup in the preset coordinate system of the suction plate. To utilize the lamination pose parameters obtained in the fifth sub-step It then maps the image to the world coordinate system and outputs its ideal landing point coordinates through a 2D rigid transformation (rotation and translation).

[0203] The second stage involves adjusting the priority weights based on slot density. This stage aims to check the ideal landing point coordinates of the output from the first stage. Whether the slot falls on a slot and how to avoid it is understandable, as the surface slot density information in this step comes from the mask matrix constructed in the first sub-step. (in For entities, (For the slot), and the final suction point positioning coordinates of the vacuum suction cup are determined by an optimization function: its expression is:

[0204] ,

[0205] in, for The final suction point positioning coordinates of the first vacuum suction cup. This is the final output of the calculation unit, and these coordinates will be sent to the corresponding motion module to drive the vacuum suction cup into position;

[0206] for A nearby local search area (e.g., with) (a 5x5 pixel window centered on the center).

[0207] Indicates the search area Find the coordinates that minimize the expression within the parentheses. ;

[0208] As a distance penalty term (i.e., the square of the Euclidean distance), ensure as close as possible ;

[0209] This is the adsorption priority term. and These are the weighting coefficients, and (For example ).

[0210] In summary, the logic of the above expression is as follows: if the ideal mapping coordinates calculated in the first stage are located within the solid region marked by the mask matrix, then the final generated adsorption point positioning coordinates will be approximately equal to or very close to these ideal mapping coordinates. However, if these ideal mapping coordinates fall within the slot region marked by the mask matrix, the formula will automatically find a pixel within a nearby local search area that is closest to the ideal mapping coordinates and is marked as a solid region by the mask matrix. This pixel will then be output as the final adsorption point positioning coordinates, thus perfectly solving the problem in existing technologies where the suction cup adsorbs onto the slot, causing the magnetic yoke to fail to adsorb, and ensuring the stability and reliability of the gripping and adsorption.

[0211] It should be noted that the above are merely preferred embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

[0212] Furthermore, the structures, proportions, sizes, etc., illustrated in the accompanying drawings of this specification are all schematic diagrams, intended only to complement the content disclosed in the specification for those skilled in the art to understand and read, and are not intended to limit the conditions under which the present invention can be implemented. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportional relationships, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should still fall within the scope of the technical content disclosed in the present invention.

[0213] Furthermore, the directional terms such as "upper," "lower," "left," "right," and "middle" used in this specification are merely for clarity of description and are not intended to limit the scope of the invention. Any changes or adjustments to their relative relationships, without substantially altering the technical content, should also be considered within the scope of the invention.

Claims

1. A gripping and stacking device for magnetic yoke laminations, the device comprising a base frame, a storage module and a stacking module respectively disposed at both ends of the base frame, and a gripping module disposed above the storage module and the stacking module, the gripping module being connected to a gantry robot mounted on the base frame, characterized in that, The grasping module includes an execution unit, which includes a rotatable adsorption plate and an array of vacuum suction cups arranged around the bottom of the adsorption plate. A sensing and detection unit is also provided on the adsorption plate, which includes a visual positioning component. The visual positioning component includes an image acquisition unit and a calculation unit; the image acquisition unit is used to acquire image data of the magnetic yoke blank when the execution unit grasps and adsorbs it and transmits it to the calculation unit, wherein the image data includes visible light channel data and near-infrared penetrating channel data. The calculation unit calculates and generates the positioning coordinates of the suction point of the vacuum suction cup based on the image data; Each of the vacuum suction cups is connected to the adsorption plate via a moving module, and the moving module is signal-connected to the calculation unit to drive the vacuum suction cup to move to the adsorption point positioning coordinate position of the vacuum suction cup generated by the calculation unit; The specific process by which the calculation unit calculates and generates the positioning coordinates of the suction point of the vacuum suction cup based on image data is as follows: Construct a mask matrix by obtaining the surface slot density of a standard image of a magnetic yoke lamination; The visible light channel data and near-infrared penetration channel data of the image data are extracted, and the dual-channel data are fused using a dynamic weight allocation algorithm to generate fused data. The mask matrix and fused data are input into the contour enhancement-noise separation network to calculate and output high-fidelity contour image data. Extract the zeroth and first order geometric moments representing the centroid of the high-fidelity contour image data, as well as the second order central geometric moment representing the principal axis direction of the contour image, and correct them in real time in conjunction with the thermal deformation coefficient. Based on the corrected geometric moments, the center coordinates and rotation angle of the magnetic yoke lamination are calculated. Based on the corrected center coordinates and rotation angle, coordinate mapping is performed in combination with the preset array layout of the vacuum chuck on the adsorption plate. The priority weight of the mapped coordinates is adjusted according to the surface slot density of the magnetic yoke, and finally the adsorption point positioning coordinates of the vacuum chuck are generated.

2. The gripping and stacking device for magnetic yoke laminations according to claim 1, characterized in that, The sensing and detection unit also includes an adsorption state detection component, which includes a flatness detection module and an adsorption force monitoring module. The flatness detection module is used to detect the surface flatness data of the magnetic yoke punch near the positioning coordinates of the adsorption point before the vacuum suction cup performs the adsorption action. The adsorption force monitoring module is used to monitor the real-time adsorption pressure value parameter during the grasping and stacking process after the vacuum suction cup adsorbs and grasps the magnetic yoke punch.

3. The gripping and stacking device for magnetic yoke laminations according to claim 2, characterized in that, An adsorption force control unit is also provided on the adsorption plate. This control unit is signal-connected to the flatness detection module and, upon receiving flatness data detected and calculated by the flatness detection module, compares this flatness data with a preset allowable flatness deviation threshold to generate an initial adsorption command for the vacuum suction cup. The initial adsorption command is: When the surface flatness data meets the allowable threshold for flatness deviation, the vacuum suction cup is controlled to adsorb the magnetic yoke punch at a first adsorption pressure, where the first adsorption pressure is the standard adsorption pressure threshold. When the surface flatness data does not meet the allowable threshold for flatness deviation, the adsorption force control unit calculates a compensation pressure value based on the surface flatness data, and controls the vacuum suction cup to adsorb the magnetic yoke blank with a second adsorption pressure based on the compensation pressure value.

4. A gripping and stacking device for magnetic yoke laminations according to claim 3, characterized in that, The adsorption force control unit is also signal-connected to the adsorption force monitoring module and is used to generate a dynamic adsorption compensation command for the vacuum suction cup based on a safe adsorption pressure threshold when receiving the real-time adsorption pressure value parameter monitored by the adsorption force monitoring module; the dynamic adsorption compensation command is: If the real-time adsorption pressure value is lower than the safe adsorption pressure threshold, the vacuum suction cup is controlled to increase the adsorption pressure until the real-time adsorption pressure value is not lower than the safe adsorption pressure threshold. The safe adsorption pressure threshold includes a first safe adsorption pressure threshold corresponding to a first adsorption pressure and a second safe adsorption pressure threshold corresponding to a second adsorption pressure, wherein the second safe adsorption pressure threshold is greater than the first safe adsorption pressure threshold.

5. A gripping and stacking device for magnetic yoke laminations according to claim 2, characterized in that, The flatness detection module includes multiple laser distance sensors arrayed around each of the vacuum suction cups, and the adsorption force monitoring module includes vacuum pressure sensors correspondingly located inside each of the vacuum suction cups.

6. A gripping and stacking device for magnetic yoke laminations according to claim 1, characterized in that, The stacking module includes a stacking platform, a support plate is provided in the middle of the upper part of the stacking platform, and a central positioning column is vertically provided in the middle of the support plate.

7. A gripping and stacking device for magnetic yoke laminations according to claim 6, characterized in that, Stacking alignment units are also provided around the support plate. The stacking alignment units include stacking alignment rods arranged in a ring array, and the bottom end of the stacking alignment rods is provided with a radial synchronous telescopic mechanism.

8. A gripping and stacking device for magnetic yoke laminations according to claim 7, characterized in that, The radial synchronous telescopic mechanism includes a rotating plate rotatably sleeved around the outside of the support plate. On the upper part of the rotating plate, there are multiple slide rail modules arranged in the radial direction and correspondingly stacked and aligned with the rods. Each slide rail module is connected to the bottom end of the corresponding stacked and aligned rod through a slide table that slides with it.

9. A gripping and stacking device for magnetic yoke laminations according to claim 7, characterized in that, An auxiliary wheel that rotatably fits into the outside of the stacked alignment rod and contacts the outer edge contour of the magnetic yoke punch is also mounted. A flexible grinding belt is attached to the outer wheel surface of the auxiliary wheel, and a pressure sensor is attached to the inside of the flexible grinding belt.

Images