Automatic assembling method and system for inductor magnetic core

Abstract

The invention provides an automatic assembling method and system for an inductor magnetic core. Wherein continuous images of the surface of the magnetic core are captured through multi-band rapid scanning, and geometric deformation characteristics are extracted; meanwhile, a distributed optical fiber sensing array is adopted to collect pressure data of the contact surface, and a real-time distribution diagram is generated through a three-dimensional field reconstruction algorithm. And then carrying out space coordinate alignment on the edge contour offset of geometric deformation and the pressure gradient, and constructing a comprehensive quality index. And based on the dynamic deviation of the index, a contact angle compensation value and an assembly speed optimization value are solved in real time, and a parameter adjustment instruction is generated. And finally, through parallel calculation of instruction data, a pressure tolerance threshold value and a deformation compensation parameter are synchronously updated, dynamic balance of deformation correction and pressure control in the assembly process is realized, and a self-adaptive closed-loop precision control system is formed. According to the technical scheme, through multi-dimensional sensing data fusion and real-time parameter adjustment, dynamic matching of deformation and stress states in the magnetic core assembling process is achieved.

Description

Technical Field

[0001] This application relates to the field of automatic assembly technology, and in particular to an automatic assembly method and system for inductor cores. Background Technology

[0002] In the field of high-speed assembly of inductor cores, with the surge in demand for miniaturized inductor devices from industries such as 5G communication and new energy vehicles, the production scenario faces three core challenges: First, real-time detection of sub-millimeter-level geometric defects on the surface of the core requires overcoming optical interference from reflective materials and lubricant residues; second, real-time coupling analysis of dynamic stress distribution and core deformation during assembly requires microsecond-level response speed to match the high-speed cycle of hundreds of parts per minute; and third, closed-loop control of multi-physics parameters requires breaking through the traditional static threshold adjustment mechanism to achieve a dynamic balance between deformation compensation and pressure control.

[0003] Current mainstream technologies employ a collaborative control architecture combining machine vision and pressure sensing. High-frame-rate industrial cameras capture the magnetic core contour, edge detection algorithms extract geometric deformation features, and an array of pressure sensors collects contact surface pressure data. The system is based on an LSTM time-series prediction model, and a central controller adjusts the robotic arm's motion parameters to ultimately achieve compensated control of assembly angles and speeds.

[0004] The solution has significant technical bottlenecks: the vision system is prone to misjudging deformation features under complex surface characteristics, which leads to deviations in subsequent control commands; the sampling frequency of the pressure sensing network is mismatched with the high-speed assembly cycle, causing a delay in stress distribution reconstruction; the correlation model between deformation and stress relies on static thresholds and cannot correct the cumulative error of material deformation during the assembly process in real time; the closed-loop control response delay directly affects the assembly accuracy and yield. Summary of the Invention

[0005] This application provides an automated assembly method and system for inductor cores to address the problems in the prior art.

[0006] In a first aspect, this application provides an automatic assembly method for inductor cores, comprising: Continuous image data of minute defects on the surface of the magnetic core are obtained by multi-band rapid scanning, and the continuous image data is decomposed into geometric deformation features based on dynamic feature extraction logic. Pressure change data of the contact surface is captured synchronously by a distributed optical fiber sensor array, and the pressure change data is converted into a real-time pressure distribution map with pressure gradient through a three-dimensional field reconstruction algorithm. Align the edge contour offset in the geometric deformation feature with the pressure gradient in the real-time pressure distribution map in spatial coordinates to generate a comprehensive quality index that includes the relationship between deformation and stress. Based on the deviation of the deformation-stress correlation in the comprehensive quality index, the contact angle adjustment value and assembly speed adjustment value of the pneumatic control device are calculated in real time, and parameter adjustment instructions containing the adjustment value gradient are generated. The parameter adjustment commands are processed quickly by parallel computing, and the pressure standard range and deformation allowable range in the stress distribution model are dynamically adjusted based on the contact angle adjustment value gradient to form an assembly accuracy closed-loop control system.

[0007] Optionally, based on the deviation of the deformation-stress correlation in the comprehensive quality index, the contact angle adjustment value and assembly speed adjustment value of the pneumatic control device are calculated in real time, and a parameter adjustment command containing the adjustment value gradient is generated, including: Extract the deviation of the relationship between deformation and stress from the comprehensive quality index; Based on the sign of the deviation, the contact angle adjustment value is calculated using a preset linear proportional model, and based on the rate of change of the deviation, the assembly speed adjustment value is determined using a preset nonlinear mapping relationship. The contact angle adjustment value and the assembly speed adjustment value are superimposed in time according to the historical data sequence within a preset time window. The changing trend of the contact angle adjustment value and the assembly speed adjustment value in the time dimension is calculated, and a gradient adjustment command containing the trend slope is generated. Based on the trend slope in the gradient adjustment command, the contact angle adjustment value and assembly speed adjustment value are subjected to slope constraint smoothing to eliminate instantaneous fluctuation interference and generate parameter adjustment commands containing continuous gradient changes.

[0008] Optionally, pressure change data of the contact surface is synchronously captured by a distributed fiber optic sensing array, and the pressure change data is converted into a real-time pressure distribution map with pressure gradient using a three-dimensional field reconstruction algorithm, including: A sensor array consisting of multiple optical fibers connected in a preset topology is deployed on the contact surface. Each optical fiber contains multiple equally spaced sensor nodes. The sensor nodes synchronously collect pressure change data at various locations on the contact interface at a fixed frequency. Pressure change data at the same time are mapped into a set of discrete points in three-dimensional space according to a preset topology. Based on the difference in pressure change data of adjacent points in the set of discrete points, a pressure gradient vector is generated along a preset direction in the three-dimensional space. The blank area between adjacent vectors is filled by continuous vector interpolation to form a three-dimensional pressure field covering the entire area of ​​the contact surface. According to preset color rules, the direction of the pressure gradient vector at each coordinate point in the three-dimensional pressure field is mapped to the color distribution direction, and the intensity is mapped to the color depth, so as to generate a real-time pressure distribution map that updates over time.

[0009] Optionally, the edge contour offset in the geometric deformation feature is spatially aligned with the pressure gradient of the real-time pressure distribution map to generate a comprehensive quality index that includes the correlation between deformation and stress, including: The original coordinate points of the magnetic core edge contour are extracted from the geometric deformation features, and the offset direction and offset amount of the original coordinate points relative to the preset reference position are calculated. The pressure gradient direction in the real-time pressure distribution map is superimposed with the offset direction in three-dimensional space to obtain the superimposed vector at each coordinate point. The cosine of the angle between the offset direction and the pressure gradient direction is used as the direction consistency ratio of the superimposed vector, and the ratio of the offset to the pressure intensity is used as the intensity ratio of the superimposed vector. Based on the directional consistency ratio and the intensity ratio, a spatial grid matching algorithm is used to calculate the correlation coefficient of each coordinate point. Based on the deviation of the correlation coefficient from a preset threshold, and based on the distribution density and mean of the deviation, a comprehensive quality index containing the correlation between deformation and stress is generated.

[0010] Optionally, the parameter adjustment commands are processed rapidly through parallel computing, and the pressure standard range and deformation allowable range in the stress distribution model are dynamically adjusted based on the adjustment value gradient to form an assembly accuracy closed-loop control system, including: The contact angle adjustment value gradient and assembly speed adjustment value gradient in the parameter adjustment command are separated into independently calculated task flows. The task flows are then processed synchronously and in parallel by multi-threading to extract the angle adjustment trend slope and the speed compensation trend slope respectively. The sign and absolute value of the trend slope are adjusted according to the angle, and the pressure standard range in the stress distribution model is adjusted frame by frame according to the preset step size. Based on the sign and absolute value of the slope of the velocity compensation trend, the allowable deformation range in the stress distribution model is adjusted in reverse according to a preset ratio; The adjusted pressure standard range and the allowable deformation range are fed back to the comprehensive quality index generation process in real time, forming a closed-loop control system that uses the parameter adjustment command to drive threshold updates and uses the updated threshold to constrain the generation of subsequent parameters.

[0011] Optionally, the contact angle adjustment value and the assembly speed adjustment value are sequentially superimposed according to historical data sequences within a preset time window to calculate the changing trend of the contact angle adjustment value and the assembly speed adjustment value in the time dimension, generating a gradient adjustment command containing the trend slope, including: The contact angle adjustment value and the assembly speed adjustment value are stored in a fixed-length circular data buffer in chronological order to form a continuous sequence containing historical data within a preset time window; The contact angle adjustment value data in the continuous sequence is subjected to linear regression fitting after time-series superposition to obtain the first slope value of the contact angle adjustment trend line. At the same time, the assembly speed adjustment value data is subjected to nonlinear curve fitting after time-series superposition to obtain the second slope value of the speed compensation trend line. The first slope value and the second slope value are vector synthesized according to a preset weight ratio to generate a gradient adjustment instruction containing the trend slope after synthesis. In the gradient adjustment instruction, the slope direction is determined by the direction of the synthesized vector, and the slope intensity is determined by the magnitude of the synthesized vector.

[0012] Optionally, pressure change data at the same time moment are mapped to a set of discrete points in three-dimensional space according to a preset topology. Based on the difference in pressure change data between adjacent points in the set of discrete points, a pressure gradient vector is generated along a preset direction in the three-dimensional space. The blank areas between adjacent vectors are filled by continuous vector interpolation to form a three-dimensional pressure field covering the entire contact surface area, including: Based on the preset topology of the sensor array, the horizontal coordinates of each sensor node are mapped to XY plane coordinates in three-dimensional space, and the numerical values ​​of pressure change data are mapped to Z-axis coordinates, forming a set of discrete points. Calculate the absolute value of the difference between each discrete point and its adjacent points in the Z-axis coordinate in the discrete point set, and generate a pressure gradient vector along a preset radial direction in the XY plane; In the blank area between adjacent pressure gradient vectors, bidirectional continuous interpolation is performed to complete the pressure gradient vectors according to the continuity constraint of the direction of the pressure gradient vectors, so that the completed pressure gradient vectors cover the entire XY plane and form a continuous three-dimensional pressure field.

[0013] Secondly, this application provides an automated assembly system for inductor cores, comprising: The extraction module is used to acquire continuous image data of minute defects on the surface of the magnetic core through multi-band rapid scanning, and decompose the continuous image data into geometric deformation features based on dynamic feature extraction logic; The conversion module is used to synchronously capture pressure change data of the contact surface through a distributed optical fiber sensor array, and convert the pressure change data into a real-time pressure distribution map with pressure gradient through a three-dimensional field reconstruction algorithm. The generation module is used to align the edge contour offset in the geometric deformation feature with the pressure gradient of the real-time pressure distribution map in spatial coordinates, and generate a comprehensive quality index that includes the relationship between deformation and stress. The calculation module is used to calculate the contact angle adjustment value and assembly speed adjustment value of the pneumatic control device in real time based on the deviation of the deformation-stress correlation in the comprehensive quality index, and generate parameter adjustment instructions containing the adjustment value gradient. The forming module is used to quickly process the parameter adjustment instructions through parallel computing and dynamically adjust the pressure standard range and deformation allowable range in the stress distribution model based on the contact angle adjustment value gradient, thereby forming a closed-loop control system for assembly accuracy.

[0014] Thirdly, embodiments of this application provide a computing device, including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are invoked and executed by the processing component to implement an automatic assembly method for inductor cores as described in the first aspect above.

[0015] Fourthly, embodiments of this application provide a computer storage medium storing a computer program, which, when executed by a computer, implements an automatic assembly method for an inductor core as described in the first aspect.

[0016] This application acquires continuous image data of minute defects on the surface of a magnetic core through multi-band rapid scanning. Based on dynamic feature extraction logic, the data is decomposed into geometric deformation features. Its technical effect is to achieve high-precision real-time detection of micro-defects on the magnetic core surface, breaking through the resolution limitations of traditional single-band imaging for minute deformation features. It also synchronously captures contact surface pressure change data through a distributed fiber optic sensor array and generates a real-time pressure distribution map using a three-dimensional field reconstruction algorithm. This technical effect is to establish a full-field dynamic monitoring capability for the pressure gradient distribution on the contact surface, eliminating the lack of spatial correlation caused by single-point pressure sensing. Furthermore, by aligning the geometric deformation edge contour offset with the pressure gradient data in spatial coordinates, it constructs a three-dimensional quantitative model of the coupling effect of deformation and stress, forming a comprehensive evaluation system for the mechanical properties of the assembled contact surface. Finally, it calculates the control parameter adjustment value in real time based on the deviation of the comprehensive quality index, achieving closed-loop feedback control of stress distribution and deformation compensation during the assembly process, and improving the adaptive adjustment capability under complex working conditions. By accelerating processing through parallel computing and dynamically adjusting the parameters of the stress distribution model, the technical effect is to establish a real-time optimization mechanism with a millisecond-level response speed, ensuring the dynamic stability of the assembly accuracy control system.

[0017] Furthermore, by extracting the deviation of the deformation-stress correlation from the comprehensive quality index, calculating the contact angle adjustment value and assembly speed adjustment value based on the preset linear proportional model and nonlinear mapping relationship, and generating gradient adjustment instructions containing trend slopes through time-series superposition analysis, and eliminating instantaneous interference through slope constraint smoothing, the technical effect is to achieve progressive optimization adjustment of control parameters in the time dimension, avoid system oscillation caused by abrupt adjustment, and significantly improve the response speed and anti-disturbance performance of the assembly accuracy closed-loop control system through historical data-driven trend prediction and interference filtering mechanism, ensuring the accurate maintenance of dynamic balance between deformation and stress in complex assembly processes.

[0018] These or other aspects of this application will become more apparent in the following description of the embodiments. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A flowchart of an automatic assembly method for an inductor core provided in this application is shown; Figure 2 This application provides a scenario diagram of an automated assembly system for inductor cores. Figure 3 This invention provides a schematic diagram of the structure of an automatic assembly system for inductor cores. Figure 4 A schematic diagram of the structure of a computing device provided in this application is shown. Detailed Implementation

[0021] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0022] In some of the processes described in the specification, claims, and accompanying drawings of this application, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as 101, 102, etc., are merely used to distinguish different operations and do not themselves represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a chronological order, nor do they limit "first" and "second" to different types.

[0023] Research has revealed significant technical bottlenecks in the collaborative control architecture of machine vision and pressure sensing in existing high-speed assembly technology for inductor cores. First, the high reflectivity of the core surface and the complex optical interference caused by lubricant residue lead to a persistently high false detection rate for sub-millimeter-level geometric deformations using traditional edge detection algorithms. Second, the sampling frequency of array-type pressure sensors differs by orders of magnitude from the high-speed assembly cycle, resulting in a significant delay in dynamic stress distribution reconstruction and hindering real-time, precise matching of robotic arm movements. Finally, deformation compensation mechanisms based on static thresholds struggle to address the cumulative effects of material plastic deformation in continuous assembly, causing lag in the closed-loop control system's response and severely limiting assembly accuracy and yield.

[0024] To address the aforementioned problems, this invention proposes a multi-physics field dynamically coupled method for controlling the assembly precision of magnetic cores. Through multi-band rapid scanning and dynamic feature extraction technology, effective removal of reflective interference and high-precision identification of geometric deformation are achieved under dual-spectral channels. A high-density fiber optic sensing array combined with a three-dimensional field reconstruction algorithm is constructed to achieve real-time dynamic mapping of the pressure gradient at the contact surface. An innovative deformation-stress spatiotemporal correlation model is established, and through coordinate alignment and iterative updates of the dynamic compensation coefficient matrix, the limitations of traditional static threshold control mechanisms are overcome. This scheme employs a parallel computing architecture to achieve rapid fusion of multi-source data and command generation, enabling sub-micron level control precision for assembly angle and speed adjustments. It successfully eliminates deformation accumulation errors in high-speed operations, significantly improves assembly yield, and fundamentally solves the technical challenge of dynamic balance of multi-physics field parameters under complex working conditions.

[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0026] Figure 1A flowchart of an automatic assembly method for an inductor core is provided in this application embodiment, as shown below. Figure 1 As shown, the method includes: 101. Continuous image data of minute defects on the surface of the magnetic core are obtained by multi-band rapid scanning, and the continuous image data is decomposed into geometric deformation features based on dynamic feature extraction logic; In the above scheme, multi-band rapid scanning refers to the simultaneous acquisition of images of the magnetic core surface through multiple spectral channels such as visible light and near-infrared light to eliminate reflective interference. Dynamic feature extraction logic refers to an algorithm based on frequency domain decomposition to separate geometric deformation features from continuous images. Geometric deformation features include quantitative parameters characterizing magnetic core structural anomalies, such as edge contour offset and surface curvature changes.

[0027] In this embodiment, firstly, multi-band rapid scanning technology is used to simultaneously acquire images of the magnetic core surface using visible and near-infrared bands. For example, on a magnetic core assembly line, the visible light band captures reflective areas on the surface, while the near-infrared band penetrates the lubricant residue layer to obtain the underlying structure. Optical interference is eliminated using a dual-spectral difference algorithm, generating continuous image data with a high signal-to-noise ratio. Secondly, based on dynamic feature extraction logic, the continuous images are input into a frequency domain decomposition module to separate low-frequency global deformation components and high-frequency local distortion components. For example, wavelet transform is performed on the edge region of the magnetic core to extract the vector offset of the edge contour and calculate the surface curvature change, ultimately outputting geometric deformation features including edge offset direction, distance, and curvature parameters.

[0028] In practical applications, taking the automated assembly of a certain EE-type magnetic core as an example, its dimensions are 12 mm x 8 mm x 5 mm. After the multi-band rapid scanning system is started, the visible light band with a wavelength range of 450 to 650 nm and the near-infrared band with a wavelength range of 850 to 1050 nm are simultaneously acquired to obtain images of the magnetic core surface. When there is lubricant residue at the edge of the magnetic core, the visible light image is blurred due to reflection interference, and the reflective area accounts for 35%. The near-infrared band penetrates the residue layer and clearly captures the underlying geometry of the magnetic core. Through the dual-spectral difference algorithm, the gray values ​​of corresponding pixels in the visible light and near-infrared images are calculated to eliminate reflection noise and generate a high-quality image sequence with a signal-to-noise ratio of 42 dB. The dynamic feature extraction module performs three-level wavelet decomposition on the image. The low-frequency component reflects the overall deformation of the magnetic core, for example, detecting an average offset of 0.12 mm in the Y-axis direction, while the high-frequency component captures local distortion and identifies three abnormal areas with a curvature radius less than the design value of 0.5 mm. The final output contains a geometric deformation feature set with 12 sets of parameters.

[0029] In step 101 above, multi-band rapid scanning technology is used to simultaneously acquire visible light and near-infrared images, effectively suppressing optical interference from reflective materials and lubricant residues, and significantly improving the detection accuracy of sub-millimeter-level geometric deformation. The dynamic feature extraction logic utilizes a frequency domain decomposition algorithm to accurately separate low-frequency global deformation and high-frequency local distortion features, enabling the quantitative expression of parameters such as core edge contour offset and surface curvature change. This step provides a highly reliable deformation data foundation for subsequent stress coupling analysis, reducing the false detection rate from the source and ensuring robustness of defect detection in high-speed assembly scenarios.

[0030] 102. The pressure change data of the contact surface is synchronously captured by a distributed optical fiber sensor array, and the pressure change data is converted into a real-time pressure distribution map with pressure gradient through a three-dimensional field reconstruction algorithm. Optionally, step 102 includes: 1021. A sensor array consisting of multiple optical fibers connected in a preset topology is deployed on the contact surface. Each optical fiber contains multiple equally spaced sensor nodes. The sensor nodes synchronously collect pressure change data at various locations on the contact interface at a fixed frequency. 1022. The pressure change data at the same time are mapped into a set of discrete points in three-dimensional space according to a preset topology. Based on the difference in pressure change data of adjacent points in the set of discrete points, a pressure gradient vector is generated along a preset direction in the three-dimensional space. The blank area between adjacent vectors is filled by continuous vector interpolation to form a three-dimensional pressure field covering the entire area of ​​the contact surface. Specifically, step 1022 includes: mapping the horizontal coordinates of each sensing node to XY plane coordinates in three-dimensional space according to the preset topology of the sensing array, and mapping the numerical values ​​of pressure change data to Z-axis coordinates to form a set of discrete points; calculating the absolute value of the difference between each discrete point and its adjacent points in the Z-axis coordinates, and generating a pressure gradient vector along a preset radial direction in the XY plane; and performing bidirectional continuous interpolation to complete the pressure gradient vector in the blank area between adjacent pressure gradient vectors according to the continuity constraint of the pressure gradient vector direction, so that the completed pressure gradient vector covers the entire XY plane area, forming a continuous three-dimensional pressure field.

[0031] 1023. Based on preset color rules, the direction of the pressure gradient vector at each coordinate point in the three-dimensional pressure field is mapped to the color distribution direction, and the intensity is mapped to the color depth, so as to generate a real-time pressure distribution map that updates over time.

[0032] In the above scheme, the distributed fiber optic sensor array refers to a sensor network composed of multiple optical fibers arranged according to a preset topology rule. The preset topology refers to the spatial arrangement of sensors on the contact surface, such as a grid or ring. The sensing node refers to a miniature pressure sensing unit evenly distributed in the optical fiber. The pressure change data refers to the dynamic values ​​of local pressure on the contact surface collected in real time by the sensing node. The discrete point set in three-dimensional space refers to the three-dimensional data point set that maps the horizontal coordinates of the sensing node to the XY plane position and the pressure value to the Z-axis height. The pressure gradient vector refers to a vector representing the pressure change trend generated along a preset radial direction with the pressure difference as the modulus. Vector continuous interpolation refers to a bidirectional interpolation completion algorithm based on the continuity of adjacent vector directions. The three-dimensional pressure field refers to a continuous pressure gradient distribution model covering the entire contact surface generated by interpolation. The preset color rule refers to a visualization mapping method that maps the pressure gradient direction to color hue and intensity to color brightness. The real-time pressure distribution map refers to a dynamically updated visualization image that represents the pressure field change with color gradient.

[0033] In this embodiment, firstly, a distributed optical fiber sensor array is deployed on the contact surface between the magnetic core and the assembly fixture in step 1021. Specifically, multiple optical fibers are arranged in a honeycomb topology, with micro-sensor nodes embedded in each fiber at equal intervals. Each node synchronously collects pressure change data of the contact surface at a fixed frequency. For example, on a magnetic core assembly line, optical fibers are embedded in the fixture contact surface according to a preset grid topology, with sensor nodes spaced 5 mm apart on each fiber. Dynamic pressure change data of each tiny area of ​​the contact surface is synchronously collected at a frequency of 100 Hz and transmitted in real time to the central processing unit through the optical fiber network to ensure the spatiotemporal synchronization of the pressure data.

[0034] Secondly, a three-dimensional field reconstruction is performed in step 1022. Based on the preset topological relationship of the sensing nodes, the horizontal coordinates of each sensing node are mapped to discrete points in three-dimensional space, and the pressure change data is used as the Z-axis height value to form a set of three-dimensional discrete points. For example, the contact surface is divided into a 1 cm x 1 cm grid, with each grid intersection corresponding to a sensing node, and its pressure change data is mapped to the Z-axis height to generate a three-dimensional scattered point cloud. Then, the absolute value of the difference in pressure change data between adjacent nodes in the Z-axis direction is calculated, and a pressure gradient vector is generated along the normal direction of the XY plane to characterize the local pressure change trend. For the blank areas not covered by the sensing nodes, based on the consistency constraint of the adjacent gradient vector directions, a cubic spline interpolation algorithm is used for bidirectional completion. First, an intermediate vector is generated by interpolation along the X-axis direction, and then a second interpolation is performed along the Y-axis direction, finally generating a continuous three-dimensional pressure field covering the entire contact surface.

[0035] Finally, visualization mapping is achieved through step 1023. Based on preset color rules, the pressure gradient direction at each coordinate point in the three-dimensional pressure field is mapped to a color wheel: red for positive gradients and blue for negative gradients. Simultaneously, the intensity of pressure change data is mapped to color brightness: high intensity corresponds to dark colors, and low intensity corresponds to light colors. The dynamically generated real-time pressure distribution map is updated at a frequency of 30 frames per second. For example, when the pressure gradient direction on the left contact surface during core assembly is opposite to the assembly direction and the intensity exceeds 50 kPa, this area appears as a dark blue patch on the map, triggering an audible and visual alarm signal to prompt operator intervention.

[0036] In a practical application, for the assembly of a 50mm diameter toroidal magnetic core, a 6x6 honeycomb fiber optic sensor array was deployed on the contact surface of the fixture. Each fiber contained 36 miniature sensor nodes spaced 5mm apart. When the magnetic core was pressed to the target position, the nodes synchronously collected pressure change data at a frequency of 100 Hz. Node number 3 detected a pressure peak of 105 kPa and a fluctuation of 3 kPa during assembly. A 3D field reconstruction algorithm then mapped the node coordinates to an XY plane grid with a grid size of 1 square centimeter and mapped the pressure data to the Z-axis height, forming a 3D discrete point cloud. In the gaps where no nodes were deployed, for example at coordinates 2.5 cm x 3.5 cm, cubic spline interpolation was performed along the X-axis on the pressure gradient vectors of adjacent nodes. The magnitude of the pressure gradient vector was 8.2 kPa per centimeter, and the direction angle was 32 degrees. After generating an intermediate vector, a second interpolation along the Y-axis was performed to fill in the blank areas, ultimately constructing a continuous pressure field covering the entire contact surface. Then, the visualization module maps the pressure gradient direction to a color wheel according to preset rules, with positive gradients represented by red and negative gradients by blue, and intensity mapped to lightness, with darker colors indicating higher pressure. When a blue gradient area with a modulus of 50 kPa per centimeter appears in the lower right quadrant, the system triggers an audible and visual alarm and locates the abnormal area at coordinates of 4.2 cm by 1.8 cm.

[0037] In step 102 above, high-density deployment and synchronous sampling of distributed fiber optic sensor arrays enable millisecond-level real-time acquisition of contact surface pressure change data. A three-dimensional field reconstruction algorithm, combined with interpolation completion technology, maps discrete pressure data into a continuous three-dimensional pressure field, fully covering the entire contact surface. The real-time pressure distribution map visually presents the direction and intensity changes of the pressure gradient through dynamic mapping of hue and brightness. This step overcomes the spatiotemporal delay bottleneck of traditional pressure sensing, providing high-precision input for deformation-stress dynamic correlation modeling, while improving anomaly response efficiency through a visual early warning mechanism.

[0038] 103. Align the edge contour offset in the geometric deformation feature with the pressure gradient in the real-time pressure distribution map in spatial coordinates to generate a comprehensive quality index that includes the relationship between deformation and stress. Optionally, step 103 includes: 1031. Extract the original coordinate points of the magnetic core edge contour from the geometric deformation features, and calculate the offset direction and offset amount of the original coordinate points relative to the preset reference position; 1032. The pressure gradient direction in the real-time pressure distribution map is superimposed with the offset direction in three-dimensional space to obtain the superimposed vector at each coordinate point. The cosine of the angle between the offset direction and the pressure gradient direction is used as the direction consistency ratio of the superimposed vector, and the ratio of the offset to the pressure intensity is used as the intensity ratio of the superimposed vector. 1033. Based on the directional consistency ratio and the strength ratio, the correlation coefficient of each coordinate point is calculated using a spatial grid matching algorithm. Based on the deviation of the correlation coefficient from a preset threshold, and based on the distribution density and mean of the deviation, a comprehensive quality index containing the correlation between deformation and stress is generated.

[0039] In the above scheme, geometric deformation characteristics refer to the quantitative parameters of the deformation of the magnetic core surface, such as edge profile offset. Edge profile offset refers to the vector position deviation of the magnetic core edge point relative to the preset reference position. The original coordinate point refers to the actual detection position coordinates of the magnetic core edge. The preset reference position refers to the theoretical standard position of the magnetic core edge. The offset direction refers to the spatial direction in which the edge point deviates from the reference position. The offset amount refers to the distance of the edge point from the reference position. The pressure gradient direction refers to the spatial direction in which the pressure intensity changes the fastest in the pressure field. The three-dimensional spatial vector superposition refers to the synthesis operation of the offset direction vector and the pressure gradient direction vector. The direction consistency ratio refers to the cosine value of the angle between the offset direction and the pressure gradient direction. The intensity ratio refers to the ratio of the offset amount to the pressure intensity. The spatial grid matching algorithm refers to the method of dividing the contact surface into grids and calculating the correlation between deformation and stress in each grid. The correlation coefficient refers to the quantitative parameter characterizing the degree of matching between deformation and stress. The preset threshold refers to the pre-set qualified standard of the correlation coefficient. The deviation value refers to the degree of difference between the correlation coefficient and the preset threshold. The comprehensive quality index refers to the comprehensive evaluation parameter that integrates the dynamic correlation between deformation and stress.

[0040] In this embodiment, firstly, the original coordinate points of the magnetic core edge contour are extracted from the geometric deformation features in step 1031. Based on the geometric deformation feature data obtained in step 101, the original coordinate point group of the magnetic core edge contour is located, and compared point by point with the theoretical reference position in the CAD model to calculate the offset direction and offset amount of each edge point. For example, detecting the coordinate point group of the upper right edge of the magnetic core reveals that it is offset by an average of 0.15 mm in the positive Y-axis direction, with a maximum offset of 0.3 mm, generating a data matrix of offset direction and offset amount.

[0041] Secondly, vector overlay analysis is performed in step 1032. The pressure gradient direction and edge offset direction in the real-time pressure distribution map are spatially vector synthesized, and the cosine of the angle between them is calculated as the directional consistency ratio, ranging from 0 to 1. For example, if the pressure gradient direction in a certain area is 30 degrees in the negative Z-axis direction and the edge offset direction is 45 degrees in the positive X-axis direction, the calculated cosine of the angle between them is 0.707, indicating moderate directional consistency. Simultaneously, the ratio of the offset to the intensity of the current pressure change data is used as the intensity ratio. For example, the ratio of an offset of 0.15 mm to a pressure intensity of 80 kPa is 0.001875, characterizing the coupling strength between deformation and stress.

[0042] Finally, a comprehensive quality index is generated in step 1033. The contact surface is divided into 1 mm x 1 mm grid cells. Within each grid, the directional consistency ratio and intensity ratio are integrated, and a correlation coefficient is generated through weighted calculation. If the correlation coefficient of a certain grid is lower than a preset threshold of 0.6, it is marked as an abnormal area. The system statistically analyzes the distribution density and standard deviation of deviation values ​​of all abnormal areas to generate a comprehensive quality index. For example, if 28% of the grid cells in the lower right quadrant of the contact surface have a correlation coefficient lower than the threshold and a standard deviation of deviation value of 0.12, the comprehensive quality index determines that this area has the risk of stress concentration and deformation mismatch, triggering subsequent compensation control logic.

[0043] In practical applications, taking the quality inspection of a UU-shaped magnetic core as an example, the window size is 15 mm × 8 mm. First, the system extracts the coordinates of 32 edge points from the geometric deformation features and compares them with the theoretical contour of the CAD model. For example, the theoretical coordinates of point 7 are 7.2 mm × 10.5 mm, but the actual inspection shows that it is offset by 0.25 mm in the positive X-axis direction, with a standard deviation of 0.03 mm. Next, the real-time pressure distribution map shows that the pressure gradient direction in this area is 28 degrees in the negative Z-axis direction, with an angle of 28 degrees with the offset direction, and the direction consistency ratio is calculated to be 0.883 using the cosine value. Subsequently, the system combines the ratio of the offset of 0.25 mm to the pressure intensity of 65 kPa (0.00385) and divides the contact surface into 0.5 square millimeter grid cells in the spatial grid matching algorithm. Ultimately, within a grid of 7.0 to 7.5 mm × 10.3 to 10.8 mm, the weighted correlation coefficient between the orientation consistency weight of 0.6 and the strength ratio weight of 0.4 was 0.72, which was higher than the threshold of 0.6 and was therefore considered acceptable. However, the adjacent grid of 7.5 to 8.0 mm × 10.8 to 11.3 mm triggered anomaly markers due to a correlation coefficient of 0.54, indicating a risk of stress-deformation mismatch in this area.

[0044] In step 103 above, a dynamic correlation model between the deformation offset direction and the pressure gradient direction is established through spatial coordinate alignment and vector superposition analysis. A spatial mesh matching algorithm integrates the directional consistency ratio and the intensity ratio to generate a quantitative correlation coefficient, accurately locating stress concentration and deformation mismatch areas. A comprehensive quality index, based on the deviation value distribution density and standard deviation, enables intelligent hierarchical early warning of assembly anomalies. This step establishes a multi-physics coupling link between deformation and stress, providing a decision-making basis for real-time compensation control and effectively avoiding core damage caused by localized stress overload.

[0045] 104. Based on the deviation of the deformation-stress correlation in the comprehensive quality index, calculate the contact angle adjustment value and assembly speed adjustment value of the pneumatic control device in real time, and generate a parameter adjustment command containing the adjustment value gradient. Optionally, step 104 includes: 1041. Extract the deviation of the deformation-stress correlation from the comprehensive quality index; 1042. Based on the sign of the deviation, calculate the contact angle adjustment value through a preset linear proportional model, and determine the assembly speed adjustment value through a preset nonlinear mapping relationship based on the rate of change of the deviation. 1043. The contact angle adjustment value and the assembly speed adjustment value are superimposed in time according to the historical data sequence within a preset time window, and the changing trend of the contact angle adjustment value and the assembly speed adjustment value in the time dimension is calculated to generate a gradient adjustment command containing the trend slope. Specifically, step 1043 includes: storing the contact angle adjustment value and the assembly speed adjustment value in a fixed-length annular data buffer in chronological order to form a continuous sequence containing historical data within a preset time window; performing linear regression fitting on the contact angle adjustment value data in the continuous sequence after time-series superposition to obtain the first slope value of the contact angle adjustment trend line, and simultaneously performing nonlinear curve fitting on the assembly speed adjustment value data after time-series superposition to obtain the second slope value of the speed compensation trend line; and performing vector synthesis on the first slope value and the second slope value according to a preset weight ratio to generate a gradient adjustment command containing the synthesized trend slope, wherein the slope direction in the gradient adjustment command is determined by the direction of the synthesized vector, and the slope intensity is determined by the magnitude of the synthesized vector.

[0046] 1044. Based on the trend slope in the gradient adjustment command, the contact angle adjustment value and assembly speed adjustment value are subjected to slope constraint smoothing to eliminate instantaneous fluctuation interference and generate parameter adjustment commands containing continuous gradient changes.

[0047] In the above scheme, deviation refers to the degree of deviation in the correlation between deformation and stress in the comprehensive quality index; the preset linear proportional model refers to a mathematical model that linearly calculates the contact angle adjustment value based on the positive and negative signs of the deviation; the preset nonlinear mapping relationship refers to a functional relationship that nonlinearly determines the assembly speed adjustment value based on the rate of change of the deviation; the assembly speed adjustment value refers to the dynamic adjustment amount of the assembly cycle; the circular data buffer refers to a circular storage structure that stores historical data within a preset time window; linear regression fitting refers to a method of extracting the contact angle adjustment trend line through linear regression; nonlinear curve fitting refers to a method of fitting the speed compensation trend line through a nonlinear function; the first slope value refers to the rate of change of the contact angle adjustment trend line; the second slope value refers to the rate of change of the speed compensation trend line; the preset weight ratio refers to the combined weight coefficient of the angle and speed trend slopes; the gradient adjustment command refers to a dynamic control command that includes the combined trend slope; and the slope constraint smoothing processing refers to a data smoothing method that eliminates instantaneous fluctuations through a filtering algorithm. In this embodiment, firstly, step 1041 extracts the deviation of the deformation-stress correlation from the comprehensive quality index. Abnormal area data in the comprehensive quality index is analyzed, and the deviation and its rate of change for each area are extracted. For example, the deviation of the rear contact surface of the magnetic core is positive 0.45, and the rate of change is 0.08 per second, which is determined to be a positive rapid deviation.

[0048] Secondly, the control parameters are calculated in step 1042. Based on the preset linear proportional model, and following the principle that every 0.1 deviation corresponds to 1 degree of angle adjustment, the current contact angle adjustment value is calculated to be +4.5 degrees. Simultaneously, considering the nonlinear characteristics of the deviation rate of change, for example, when the rate exceeds 0.05 per second, an exponential decay function is triggered, and the assembly speed adjustment value is calculated according to the inverse square relationship of the rate; a rate of 0.08 per second corresponds to a 12% speed reduction.

[0049] Next, trend analysis is performed in step 1043. The contact angle adjustment value and assembly speed adjustment value are stored in a circular data buffer in chronological order, storing the most recent 15 seconds of historical data. A linear regression is performed on the angle data sequence to generate a contact angle adjustment trend line, and its slope is extracted as the first slope value, for example, +0.3 degrees per second. An exponential curve is fitted on the speed data sequence to generate a speed compensation trend line, and its slope is extracted as the second slope value, for example, -0.2% per second. The two slope values ​​are synthesized according to a preset weight ratio, with angle accounting for 60% and speed accounting for 40%. The direction of the synthesized vector is determined by the sign of the angle and speed slopes, and the modulus is calculated from the weighted absolute value to be 0.22, generating a gradient adjustment command containing the comprehensive trend intensity.

[0050] Finally, a smoothing command is generated in step 1044. A Kalman filter algorithm is used to eliminate fluctuations caused by sensor noise or transient interference. The angle and velocity parameters in the gradient adjustment command are input into the state-space model. Random fluctuations are suppressed through a prediction-correction loop, resulting in a continuously smooth parameter adjustment command. For example, the original angle adjustment value of ±4.5 degrees plus or minus 0.1 degrees converges to ±4.5 degrees plus or minus 0.03 degrees after filtering, ensuring smooth and shock-free robotic arm movements.

[0051] In practical applications, when the comprehensive quality index detects a deviation of +0.48 from the contact surface of the magnetic core at a rate of 0.09 per second, the system first calculates the contact angle compensation value using a preset linear proportional model: based on the principle of 1 degree adjustment for every 0.1 degree deviation, the angle needs to be adjusted by +4.8 degrees. Simultaneously, the nonlinear module detects a deviation rate exceeding the threshold of 0.05 per second, triggering an exponential decay function to significantly reduce the assembly speed. Subsequently, the circular data buffer stores the adjustment value sequence for the most recent 15 seconds. Linear regression analysis reveals an angle adjustment trend slope of +0.32 degrees / second (R²=0.93) and a speed compensation trend slope of -0.25% / second. Based on this, a comprehensive trend slope of +0.21 is synthesized using a 6:4 weighting. Finally, a Kalman filter algorithm is used to suppress random fluctuations of ±0.15 degrees in the original angle data, outputting a smoothed adjustment value of +4.8 degrees ±0.02 degrees, driving the servo motor to perform precise compensation at an angular velocity of 0.32 degrees / second. In step 104 above, the adjustment values ​​of the contact angle and assembly speed are calculated in real time based on the linear and nonlinear mapping relationship of the deviation; time-series superposition and trend analysis extract the gradient change trend of angle and speed through historical data fitting; the Kalman filter algorithm eliminates instantaneous interference and generates smooth and stable parameter adjustment commands. This step achieves microsecond-level dynamic response and submicron-level control precision, ensuring precise matching between the robotic arm's movements and the assembly cycle, and significantly reducing the cumulative deformation error in high-speed operations.

[0052] 105. The parameter adjustment command is processed quickly through parallel computing, and the pressure standard range and deformation allowable range in the stress distribution model are dynamically adjusted based on the contact angle adjustment value gradient to form an assembly accuracy closed-loop control system.

[0053] Optionally, step 105 includes: 1051. The gradient of the contact angle adjustment value and the gradient of the assembly speed adjustment value in the parameter adjustment command are separated into independent calculation task flows. The task flows are processed synchronously and in parallel by multi-threading to extract the slope of the angle adjustment trend and the slope of the speed compensation trend respectively. 1052. Adjust the sign and absolute value of the trend slope according to the angle, and adjust the pressure standard range in the stress distribution model frame by frame according to the preset step size; 1053. Based on the sign and absolute value of the slope of the speed compensation trend, adjust the allowable deformation range in the stress distribution model in reverse according to a preset ratio; 1054. The adjusted pressure standard range and the allowable deformation range are fed back to the comprehensive quality index generation process in real time, forming a closed-loop control system that uses the parameter adjustment command to drive threshold updates and uses the updated threshold to constrain the generation of subsequent parameters.

[0054] In the above scheme, the contact angle adjustment value gradient refers to the slope of the contact angle adjustment trend, the assembly speed adjustment value gradient refers to the slope of the speed compensation trend, the angle adjustment trend slope refers to the trend strength and direction of the contact angle changing over time, the speed compensation trend slope refers to the trend strength and direction of the assembly speed changing over time, the preset step size refers to the unit amplitude of the pressure standard range adjustment, the pressure standard range refers to the allowable pressure fluctuation threshold in the stress distribution model, the preset ratio refers to the coefficient relationship of the deformation allowable range reverse adjustment, the deformation allowable range refers to the dynamic tolerance boundary of deformation compensation, and the closed-loop control system refers to the dynamic feedback control system in which parameter adjustment commands drive threshold updates and the updated thresholds constrain the generation of subsequent parameters.

[0055] In this embodiment, firstly, step 1051 separates the contact angle adjustment value gradient and assembly speed adjustment value gradient in the parameter adjustment command into independent task flows, such as an angle gradient of +0.3 degrees per second and a speed gradient of -0.2% per second. These are then distributed to two computing cores for parallel processing using a multi-threaded architecture. Each thread independently extracts trend slope data; the angle thread extracts a linear trend of +0.3 degrees per second, and the speed thread extracts an exponential decay trend of -0.2% per second.

[0056] Secondly, in step 1052, based on the positive characteristics of the slope of the angle trend and its absolute value, the upper limit of the pressure standard range in the stress distribution model is gradually widened frame by frame according to a preset step size. For example, the original pressure threshold of 100 kPa is gradually increased to 104.8 kPa after 5 consecutive frames of calculation, expanding the tolerance range for fluctuations in pressure change data and avoiding accidental triggering of compensation actions due to instantaneous over-limit.

[0057] Next, in step 1053, the allowable deformation range is dynamically adjusted according to a preset reverse ratio based on the negative characteristic of the velocity trend slope. For example, a velocity slope of -0.2% per second corresponds to a reduction of 6 micrometers in the allowable deformation range, from the original 15 micrometers to 14.4 micrometers, thereby improving the deformation compensation accuracy and suppressing the cumulative effect of material plastic deformation.

[0058] Finally, in step 1054, the updated pressure standard range of 104.8 kPa and the allowable deformation range of 14.4 micrometers are fed back to the comprehensive quality index generation module in real time. When the system detects that the pressure change data of 9 out of the 12 grid areas originally judged as abnormal has fallen below the new threshold, it automatically updates their status to normal and terminates the redundant compensation action. At the same time, the updated allowable deformation range constrains the generation logic of subsequent parameter adjustment instructions, forming a closed-loop control system of dynamic threshold adjustment, parameter compensation, and feedback verification, ensuring that the assembly accuracy remains stable during high-speed continuous operation.

[0059] In practical applications, under an 8-thread multi-core processor architecture, the system first separates the angle gradient (+0.32 degrees / second) and velocity gradient (-0.25% / second) in the parameter adjustment instructions into independent threads. The angle thread, with a 0.1-second cycle per frame and a step size of 0.8 kPa, dynamically increases the pressure standard range from 100 kPa to 104 kPa within 5 frames, expanding the tolerance for pressure fluctuations. Simultaneously, the velocity thread, based on a slope of -0.25% / second, tightens the allowable deformation range from 15 micrometers to 14.25 micrometers at a ratio of 3 micrometers per 0.1% change. Subsequently, the updated parameters are fed back to the quality inspection module in real time. Nine of the 12 grids originally marked as abnormal have seen their pressure values ​​drop to 103 kPa, below the new threshold of 104 kPa. The system automatically deactivates the alarm and terminates redundant compensation. Ultimately, through 30 dynamic threshold iterations per second, the closed-loop control system stably controls the assembly accuracy error within ±3 micrometers, achieving process consistency under high-speed continuous operation.

[0060] In the scheme of step 105 above, gradient adjustment instructions are processed quickly through a parallel computing architecture, dynamically updating the standard pressure range and allowable deformation range; multi-threaded task flow enables independent optimization of angle and speed parameters, and a real-time feedback mechanism drives adaptive adjustment of thresholds; the closed-loop control system uses the updated thresholds to inversely constrain the comprehensive quality index generation logic, suppressing the vicious cycle of material creep and assembly errors. This step forms a complete closed-loop process of "perception-decision-execution-feedback", maintaining stable assembly yield at a high-speed cycle of hundreds of parts per minute and achieving process consistency control under complex working conditions.

[0061] The following is a complete embodiment for steps 101-105: In the magnetic core assembly line, the multi-band scanning device is activated first: the visible light channel (430~680nm) precisely locks the reflective features of the silver plating layer at the edge of the magnetic core, while the near-infrared channel (850~1050nm) simultaneously penetrates the 0.2 mm lubricant layer on the surface, completely capturing the underlying structure. Through dual-spectral differential fusion, reflective interference is effectively suppressed, generating a high-definition image sequence with a signal-to-noise ratio of 50dB. The dynamic feature extraction module then intervenes, performing wavelet transform analysis on the lower right edge of the magnetic core, detecting an offset of 0.18 mm in the negative X-axis direction, superimposed with a 12% surface curvature anomaly, and the system immediately marks the geometric deformation risk area.

[0062] Simultaneously, a distributed fiber optic sensor array is activated: 256 sensor nodes are densely distributed in a honeycomb topology on the fixture contact surface, achieving high-resolution monitoring with a 5 mm spacing. When the magnetic core contacts the fixture, the peak pressure in the central region reaches 85 kPa, decreasing radially to 22 kPa at the edge. The 3D field reconstruction algorithm responds quickly, mapping the node data into a 3D point cloud, and generating a continuous pressure field model through cubic spline interpolation. Analysis shows that there is a reverse pressure gradient on the left front contact surface, with its direction at a 45-degree angle to the assembly axis, and a peak intensity of 62 kPa / mm. A dark blue patch on the visualization interface provides real-time warning of this anomaly.

[0063] The data fusion stage followed immediately: the system extracted the coordinate group of the upper right edge of the magnetic core, measured the average positive Y-axis offset of 0.15 mm, and performed spatial vector superposition with the corresponding region's negative Z-axis pressure gradient of 30 degrees. In the mesh matching calculation, the cosine similarity between the offset direction and the pressure gradient within cell 12 was only 0.6 (threshold 0.7), and the deformation-to-stress ratio was 0.0018 (0.2 mm / 110 kPa), triggering a mismatch judgment. Further statistics showed that 28% of the mesh cells in the lower right quadrant of the contact surface had abnormal correlation coefficients, with a standard deviation of 0.15, causing the overall quality pass rate to plummet to 89.3%, indicating that the system had entered the compensation and control stage.

[0064] Based on the quality deviation, the dynamic compensation mechanism is immediately triggered: the linear model outputs a positive adjustment command of 4.2 degrees for the contact angle, while the nonlinear function simultaneously calculates a 15% reduction in assembly speed. The time-series analysis module traces 20 seconds of historical data, fitting a linear growth trend of 0.25 degrees / second for the angle and an exponential decay curve of 0.18% / second for the speed. After weighted synthesis and Kalman filtering optimization, the final output is a smooth command: angle 4.2 degrees ± 0.05 degrees, speed reduction 15% ± 0.3%, effectively filtering out sensor noise interference.

[0065] In the final stage of closed-loop optimization, the parallel computing engine splits the task flow, with dual cores processing pressure and deformation parameters synchronously. The upper limit of the pressure standard dynamically widens with the angle trend, increasing from 0.4 kPa per frame to 103.6 kPa; the deformation tolerance shrinks in the opposite direction, tightening to 14.2 micrometers based on the speed trend. After the updated thresholds are fed back to the system in real time, 26 out of 35 abnormal grids have their alarms cleared because the pressure value is lower than the new standard. Through multiple rounds of iterative adjustments, the core assembly position deviation is stabilized at ±4.5 micrometers, a 58% improvement in accuracy compared to traditional control methods, and the defect rate is sharply reduced from 1.2% to 0.18%, fully realizing the technical closed loop from "detection to compensation to verification".

[0066] Figure 3 This application provides a schematic diagram of an automated inductor core assembly system, as shown in the embodiment of the present application. Figure 3 As shown, the system includes: The extraction module is used to acquire continuous image data of minute defects on the surface of the magnetic core through multi-band rapid scanning, and decompose the continuous image data into geometric deformation features based on dynamic feature extraction logic; The conversion module is used to synchronously capture pressure change data of the contact surface through a distributed optical fiber sensor array, and convert the pressure change data into a real-time pressure distribution map with pressure gradient through a three-dimensional field reconstruction algorithm. The generation module is used to align the edge contour offset in the geometric deformation feature with the pressure gradient of the real-time pressure distribution map in spatial coordinates, and generate a comprehensive quality index that includes the relationship between deformation and stress. The calculation module is used to calculate the contact angle adjustment value and assembly speed adjustment value of the pneumatic control device in real time based on the deviation of the deformation-stress correlation in the comprehensive quality index, and generate parameter adjustment instructions containing the adjustment value gradient. The forming module is used to quickly process the parameter adjustment instructions through parallel computing and dynamically adjust the pressure standard range and deformation allowable range in the stress distribution model based on the contact angle adjustment value gradient, thereby forming a closed-loop control system for assembly accuracy.

[0067] Figure 3 The aforementioned automatic assembly system for inductor cores can perform... Figure 1 The implementation principle and technical effects of the automatic inductor core assembly method described in the illustrated embodiment will not be repeated here. The specific methods by which each module and unit of the automatic inductor core assembly system in the above embodiments perform operations have been described in detail in the embodiments related to this method, and will not be elaborated upon here.

[0068] In one possible design, Figure 3 The automated inductor core assembly system of the illustrated embodiment can be implemented as a computing device, such as... Figure 4As shown, the computing device may include a storage component 41 and a processing component 42; The storage component 41 stores one or more computer instructions, wherein the one or more computer instructions are invoked and executed by the processing component 42.

[0069] The processing component 42 is used for the above Figure 1 The embodiment describes an automatic assembly method for inductor cores.

[0070] The processing component 42 may include one or more processors to execute computer instructions to complete all or part of the steps in the above-described method. Alternatively, the processing component may be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-described method.

[0071] Storage component 41 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0072] Of course, computing devices may also include other components, such as input / output interfaces, display components, communication components, etc.

[0073] Input / output interfaces provide interfaces between processing components and peripheral interface modules, which can be output devices, input devices, etc.

[0074] The communication components are configured to facilitate wired or wireless communication between computing devices and other devices.

[0075] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform. In this case, the computing device can refer to a cloud server, and the aforementioned processing components, storage components, etc., can be basic server resources rented or purchased from the cloud computing platform.

[0076] This application also provides a computer storage medium storing a computer program, which, when executed by a computer, can perform the above-described functions. Figure 1 The embodiment shown illustrates an automatic assembly method for inductor cores.

[0077] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0078] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0079] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0080] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. An automatic assembly method for inductor cores, characterized in that, include: Continuous image data of minute defects on the surface of the magnetic core are obtained by multi-band rapid scanning, and the continuous image data is decomposed into geometric deformation features based on dynamic feature extraction logic. Pressure change data of the contact surface is captured synchronously by a distributed fiber optic sensor array, and the pressure change data is converted into a real-time pressure distribution map with pressure gradient through a three-dimensional field reconstruction algorithm. Align the edge contour offset in the geometric deformation feature with the pressure gradient in the real-time pressure distribution map in spatial coordinates to generate a comprehensive quality index that includes the relationship between deformation and stress. Based on the deviation of the deformation-stress correlation in the comprehensive quality index, the contact angle adjustment value and assembly speed adjustment value of the pneumatic control device are calculated in real time, and parameter adjustment instructions containing the adjustment value gradient are generated. The parameter adjustment commands are processed quickly by parallel computing, and the pressure standard range and deformation allowable range in the stress distribution model are dynamically adjusted based on the contact angle adjustment value gradient to form an assembly accuracy closed-loop control system.

2. The method according to claim 1, characterized in that, Based on the deviation of the deformation-stress correlation in the comprehensive quality index, the contact angle adjustment value and assembly speed adjustment value of the pneumatic control device are calculated in real time, and parameter adjustment instructions containing the adjustment value gradient are generated, including: Extract the deviation of the relationship between deformation and stress from the comprehensive quality index; Based on the sign of the deviation, the contact angle adjustment value is calculated using a preset linear proportional model, and based on the rate of change of the deviation, the assembly speed adjustment value is determined using a preset nonlinear mapping relationship. The contact angle adjustment value and the assembly speed adjustment value are superimposed in time according to the historical data sequence within a preset time window. The changing trend of the contact angle adjustment value and the assembly speed adjustment value in the time dimension is calculated, and a gradient adjustment command containing the trend slope is generated. Based on the trend slope in the gradient adjustment command, the contact angle adjustment value and assembly speed adjustment value are subjected to slope constraint smoothing to eliminate instantaneous fluctuation interference and generate parameter adjustment commands containing continuous gradient changes.

3. The method according to claim 1, characterized in that, Pressure change data of the contact surface is synchronously captured by a distributed fiber optic sensor array. This pressure change data is then converted into a real-time pressure distribution map with pressure gradients using a three-dimensional field reconstruction algorithm, including: A sensor array consisting of multiple optical fibers connected in a preset topology is deployed on the contact surface. Each optical fiber contains multiple equally spaced sensor nodes. The sensor nodes synchronously collect pressure change data at various locations on the contact interface at a fixed frequency. Pressure change data at the same time are mapped into a set of discrete points in three-dimensional space according to a preset topology. Based on the difference in pressure change data of adjacent points in the set of discrete points, a pressure gradient vector is generated along a preset direction in the three-dimensional space. The blank area between adjacent vectors is filled by continuous vector interpolation to form a three-dimensional pressure field covering the entire area of ​​the contact surface. According to preset color rules, the direction of the pressure gradient vector at each coordinate point in the three-dimensional pressure field is mapped to the color distribution direction, and the intensity is mapped to the color depth, so as to generate a real-time pressure distribution map that updates over time.

4. The method according to claim 1, characterized in that, The edge contour offset in the geometric deformation feature is aligned with the pressure gradient in the real-time pressure distribution map in spatial coordinates to generate a comprehensive quality index that includes the correlation between deformation and stress, including: The original coordinate points of the magnetic core edge contour are extracted from the geometric deformation features, and the offset direction and offset amount of the original coordinate points relative to the preset reference position are calculated. The pressure gradient direction in the real-time pressure distribution map is superimposed with the offset direction in three-dimensional space to obtain the superimposed vector at each coordinate point. The cosine of the angle between the offset direction and the pressure gradient direction is used as the direction consistency ratio of the superimposed vector, and the ratio of the offset to the pressure intensity is used as the intensity ratio of the superimposed vector. Based on the directional consistency ratio and the intensity ratio, a spatial grid matching algorithm is used to calculate the correlation coefficient of each coordinate point. Based on the deviation of the correlation coefficient from a preset threshold, and based on the distribution density and mean of the deviation, a comprehensive quality index containing the correlation between deformation and stress is generated.

5. The method according to claim 1, characterized in that, The parameter adjustment commands are processed rapidly through parallel computing, and the pressure standard range and deformation allowable range in the stress distribution model are dynamically adjusted based on the contact angle adjustment value gradient to form an assembly accuracy closed-loop control system, including: The contact angle adjustment value gradient and assembly speed adjustment value gradient in the parameter adjustment command are separated into independently calculated task flows. The task flows are then processed synchronously and in parallel by multi-threading to extract the angle adjustment trend slope and the speed compensation trend slope respectively. The sign and absolute value of the trend slope are adjusted according to the angle, and the pressure standard range in the stress distribution model is adjusted frame by frame according to the preset step size. Based on the sign and absolute value of the slope of the velocity compensation trend, the allowable deformation range in the stress distribution model is adjusted in reverse according to a preset ratio; The adjusted pressure standard range and the allowable deformation range are fed back to the comprehensive quality index generation process in real time, forming a closed-loop control system that uses the parameter adjustment command to drive threshold updates and uses the updated threshold to constrain the generation of subsequent parameters.

6. The method according to claim 2, characterized in that, The contact angle adjustment value and the assembly speed adjustment value are sequentially superimposed according to historical data sequences within a preset time window. The changing trends of the contact angle adjustment value and the assembly speed adjustment value in the time dimension are calculated, and a gradient adjustment command containing the trend slope is generated, including: The contact angle adjustment value and the assembly speed adjustment value are stored in a fixed-length circular data buffer in chronological order to form a continuous sequence containing historical data within a preset time window; The contact angle adjustment value data in the continuous sequence is subjected to linear regression fitting after time-series superposition to obtain the first slope value of the contact angle adjustment trend line. At the same time, the assembly speed adjustment value data is subjected to nonlinear curve fitting after time-series superposition to obtain the second slope value of the speed compensation trend line. The first slope value and the second slope value are vector synthesized according to a preset weight ratio to generate a gradient adjustment instruction containing the trend slope after synthesis. In the gradient adjustment instruction, the slope direction is determined by the direction of the synthesized vector, and the slope intensity is determined by the magnitude of the synthesized vector.

7. The method according to claim 3, characterized in that, Pressure change data at the same time are mapped to a set of discrete points in three-dimensional space according to a preset topology. Based on the difference in pressure change data between adjacent points in the set of discrete points, a pressure gradient vector is generated along a preset direction in the three-dimensional space. The blank areas between adjacent vectors are filled by continuous vector interpolation to form a three-dimensional pressure field covering the entire contact surface area, including: Based on the preset topology of the sensor array, the horizontal coordinates of each sensor node are mapped to XY plane coordinates in three-dimensional space, and the numerical values ​​of pressure change data are mapped to Z-axis coordinates, forming a set of discrete points. Calculate the absolute value of the difference between each discrete point and its adjacent points in the Z-axis coordinate in the discrete point set, and generate a pressure gradient vector along a preset radial direction in the XY plane; In the blank area between adjacent pressure gradient vectors, bidirectional continuous interpolation is performed to complete the pressure gradient vectors according to the continuity constraint of the direction of the pressure gradient vectors, so that the completed pressure gradient vectors cover the entire XY plane and form a continuous three-dimensional pressure field.

8. An automated assembly system for inductor cores, characterized in that, include: The extraction module is used to acquire continuous image data of minute defects on the surface of the magnetic core through multi-band rapid scanning, and decompose the continuous image data into geometric deformation features based on dynamic feature extraction logic; The conversion module is used to synchronously capture pressure change data of the contact surface through a distributed optical fiber sensor array, and convert the pressure change data into a real-time pressure distribution map with pressure gradient through a three-dimensional field reconstruction algorithm. The generation module is used to align the edge contour offset in the geometric deformation feature with the pressure gradient of the real-time pressure distribution map in spatial coordinates, and generate a comprehensive quality index that includes the relationship between deformation and stress. The calculation module is used to calculate the contact angle adjustment value and assembly speed adjustment value of the pneumatic control device in real time based on the deviation of the deformation-stress correlation in the comprehensive quality index, and generate parameter adjustment instructions containing the adjustment value gradient. The forming module is used to quickly process the parameter adjustment instructions through parallel computing and dynamically adjust the pressure standard range and deformation allowable range in the stress distribution model based on the contact angle adjustment value gradient, thereby forming a closed-loop control system for assembly accuracy.

9. A computing device, characterized in that, It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are invoked and executed by the processing component to implement the automatic assembly method for inductor cores as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that, The device contains a computer program that, when executed by a computer, implements the automatic assembly method for inductor cores as described in any one of claims 1 to 7.

Images