Automatic process control method and system applied to high-precision special-shaped target material binding

Abstract

The application provides an automatic process control method and system applied to high-precision special-shaped target material binding, relates to the technical field of image processing, and comprises the following steps: acquiring multi-view image data of a target material, generating a three-dimensional geometric structure model of the target material, generating point cloud data of an outer boundary contour and surface topography of the target material, calling a matched temperature and pressure strategy template in a binding process parameter library, realizing fitting clamping, performing heterogeneous material adaptation, and executing a closed-loop control strategy to dynamically adjust heating / pressure output. The application solves the problems in the prior art that due to the complex shape of the target material, heterogeneous material and non-uniform process parameters, the binding process has insufficient temperature and pressure control precision, poor clamping adaptability, uneven thermal stress distribution, and it is difficult to realize high-consistency and high-strength binding of special-shaped target materials, improves the intelligentization and adaptability of the binding process, and realizes precise hot-pressing binding and process closed-loop control between multiple types of target materials and back plates in the differentiated area.

Description

Technical Field

[0001] This application relates to the field of image processing technology, specifically to an automated process control method and system for high-precision bonding of irregularly shaped targets. Background Technology

[0002] As industries such as precision manufacturing, semiconductor sputtering, and high-energy beam processing continue to demand higher requirements for sputtering bonding quality, morphological consistency, and material compatibility, the hot-press bonding process between the sputtering target and the backplane, as a key link affecting the performance of the final product, is receiving increasing attention for its stability, adaptability, and automation.

[0003] Currently, most bonding processes adopt a uniform temperature and pressure strategy, which does not adequately consider the type of material, structural shape, and thermal conductivity characteristics of different regions, making it difficult to meet the precise bonding requirements of irregularly shaped targets. Summary of the Invention

[0004] This application provides an automated process control method and system for high-precision bonding of irregularly shaped targets. It solves the problems of poor adaptability, uneven local stress, and insufficient bonding strength in the bonding process caused by the lack of automatic identification capability of target structural features and multi-segment temperature and pressure coordinated control mechanism in the prior art. It achieves a fine control effect throughout the entire process based on target geometric modeling, automatic generation of hot pressing strategy, zoned temperature and pressure control and closed-loop feedback regulation, thereby significantly improving the stability, consistency and automation level of the bonding process of irregularly shaped targets.

[0005] In view of the above problems, the first aspect of this application provides an automated process control method for high-precision bonding of irregularly shaped targets. The method includes: capturing omnidirectional images of the target material using a multi-angle industrial camera array to obtain multi-view image data; rapidly analyzing the obtained multi-view image data using a CNN image classification model to identify the target material type and material and label and encode it; the multi-view image data also serves as input for subsequent 3D reconstruction; the target material types include rectangular molybdenum targets, ring-shaped copper targets, and irregular titanium targets; calling a 3D reconstruction algorithm based on multi-view geometry to generate a 3D geometric structure model of the target material; performing geometric analysis on the 3D geometric structure model to generate point cloud data of the target material's outer boundary contour and surface morphology; the point cloud data includes data on contour boundaries, feature point coordinates, edge curvature variation areas, and surface abrupt change areas; reading the backplate RFID tag in the bonding preparation area to determine the backplate material type; and, combining the material types of the target material and the backplate, calling a matching temperature and pressure strategy from a bonding process parameter library. The template, comprising a heating zone control module, a pressing module, an atmosphere adjustment module, and a cooling control module, includes data parameters for each module such as: bonding temperature, multi-segment hot zone division method, interval temperature data, pressing pressure distribution method, atmosphere protection strategy, and cooling curve setting. Based on the point cloud data of the outer boundary contour and surface morphology, a multi-degree-of-freedom flexible fixture is controlled to fit and support the actual edge shape of the target material, achieving a fitting clamp. The target material and backplate are fed into the bonding chamber, and the multi-segment temperature control and pressing module heats and presses the target material and backplate according to the strategy, using their thermophysical parameters as input to set differentiated temperature and pressure bonding curves for the central and edge areas, achieving material adaptation. During the bonding process, real-time temperature, pressure, and displacement parameters are collected, and the collected data is compared with the target curve to execute a closed-loop control strategy, dynamically adjusting the heating / pressure output. After bonding is completed, infrared thermal imaging and ultrasonic testing are used to assess whether there are voids or delamination on the bonding surface, and a quality report is generated based on these defects.

[0006] The second aspect of this application provides an automated process control system for high-precision bonding of irregularly shaped targets. The system includes: a graphics data acquisition module, which is used to capture images of the target from all angles using a multi-angle industrial camera array to obtain multi-view image data of the target. The obtained multi-view image data is then rapidly analyzed using an image classification model CNN to identify the type and material of the target and to mark and encode it. The multi-view image data also serves as input for subsequent three-dimensional reconstruction. The types of target materials include rectangular molybdenum targets, ring-shaped copper targets, and irregular titanium targets.

[0007] The point cloud data generation module is used to call a three-dimensional reconstruction algorithm based on multi-view geometry to generate a three-dimensional geometric structure model of the target material, perform geometric analysis on the three-dimensional geometric structure model, and generate point cloud data of the outer boundary contour and surface morphology of the target material. The point cloud data includes data of contour boundary, feature point coordinates, edge curvature change area, and surface concave-convex abrupt change area.

[0008] The temperature and pressure strategy template calling module is used to read the backplate RFID tag in the binding preparation area, determine the backplate material type, and call the matching temperature and pressure strategy template in the binding process parameter library in combination with the material types of the target and the backplate. The temperature and pressure strategy template consists of a heating zone control module, a pressing module, an atmosphere adjustment module, and a cooling control module. The data parameters of each module include: binding temperature, multi-segment hot zone division method, interval temperature data, pressing pressure distribution method, atmosphere protection strategy, and cooling curve setting.

[0009] The fitting and clamping module is used to control a multi-degree-of-freedom flexible clamp to fit and support the actual edge shape of the target material according to the point cloud data of the outer boundary contour and surface morphology, so as to achieve fitting and clamping.

[0010] The dissimilar material adaptation module is used to send the target material and the back plate into the bonding chamber. The multi-segment temperature control and pressing module heats and presses the target material and the back plate in different zones according to a strategy. The thermophysical parameters of the target material and the back plate are used as input to set different temperature and pressure bonding curves for the central and edge zones to achieve dissimilar material adaptation.

[0011] The closed-loop control module is used to collect real-time temperature, pressure, and displacement parameters during the binding process, compare the collected data with the target curve, execute the closed-loop control strategy, and dynamically adjust the heating / pressure output.

[0012] The quality report generation module is used to evaluate whether there are hollow areas or delamination on the bonding surface after the bonding is completed by infrared thermal imaging and ultrasonic detection, and to generate a quality report based on the defects.

[0013] One or more technical solutions provided in this application have at least the following technical effects or advantages: The automated process control method and system for high-precision irregular-shaped target bonding provided in this application relates to the field of graphics processing technology. Combining three-dimensional vision recognition, target structure modeling, temperature and pressure strategy matching, multi-segment collaborative temperature and pressure control and closed-loop feedback control and other technical means, it effectively solves the problems of poor structural adaptability, low process parameter universality, poor hot pressing consistency and insufficient automation in traditional bonding processes.

[0014] By introducing a multi-view geometric reconstruction algorithm to achieve 3D structural modeling of irregularly shaped targets, and using point cloud analysis and edge feature recognition to achieve hot zone division and pressure restriction zone identification, combined with a temperature and pressure strategy template library to automatically call matching hot pressing control parameters, and with the help of a multi-segment heating and servo pressurization system, differentiated heating, heat preservation, cooling, and pressing process control of the bonding center and edge areas can be achieved. In addition, by collecting key parameters such as real-time temperature, pressure, and displacement, and introducing a closed-loop control mechanism to dynamically adjust heating and pressure output, the accuracy, stability, and automatic response capability of the bonding process are significantly improved. It can be widely applied to precision hot pressing integration scenarios of irregularly shaped targets and backplates with various complex structures and high-performance requirements.

[0015] In summary, this application constructs a collaborative framework for three-dimensional geometric modeling and multi-segment temperature and pressure control of irregularly shaped targets. This framework enables integrated binding process management, including automatic identification of target structure, intelligent division of hot zones, dynamic matching of temperature and pressure strategies, and zoned closed-loop control. It reduces process fluctuations caused by traditional manual parameter setting and experience-based operation, and avoids quality problems such as target edge deformation, hollow area fracturing, and thermal stress runaway. It can achieve high consistency and high strength bonding in binding scenarios with targets of different materials and complex geometries, improving the intelligence level and process stability of the binding process. This contributes to improving the manufacturing efficiency of irregularly shaped targets and the integration reliability in high-end application scenarios.

[0016] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating an automated process control method for high-precision bonding of irregularly shaped targets, as provided in an embodiment of this application.

[0019] Figure 2 This is a schematic diagram of the structure of an automated process control system for high-precision bonding of irregularly shaped targets, provided in an embodiment of this application.

[0020] Figure labeling: 10 for graphic data acquisition module, 20 for point cloud data generation module, 30 for temperature and pressure strategy template calling module, 40 for fitting and clamping module, 50 for dissimilar material adaptation module, 60 for closed-loop control module, and 70 for quality report generation module. Detailed Implementation

[0021] This application provides an automated process control method and system for bonding high-precision irregular-shaped targets, which solves the problems in existing bonding processes such as lack of target structure perception, uncontrollable hot pressing process, and inability to adapt temperature and pressure strategies. These problems lead to uneven thermal stress distribution, local pressure overload or insufficient, and inconsistent bonding strength during the bonding process, making it difficult to meet the technical requirements for bonding complex geometries and high-performance targets in multi-material, high-consistency processes. The method and system achieve the technical effects of intelligent target structure recognition, multi-segment differentiated temperature and pressure linkage control, closed-loop dynamic adjustment, and ensuring fine bonding effect.

[0022] 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 a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0023] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices.

[0024] Example 1, as Figure 1 As shown, this application provides an automated process control method for high-precision bonding of irregularly shaped targets, the method comprising:

[0025] S100. The target material is captured from all angles by a multi-angle industrial camera array to obtain multi-view image data of the target material. The obtained multi-view image data is quickly analyzed by an image classification model CNN to identify the type and material of the target material and mark and encode it. The multi-view image data is also used as the input for subsequent three-dimensional reconstruction. The types of target materials include rectangular molybdenum targets, ring copper targets, and irregular titanium targets.

[0026] Specifically, a multi-angle industrial camera array is used to acquire image data of the target material's appearance. An image classification model based on a convolutional neural network (CNN) is then used to automatically identify and label the images, thereby achieving automatic classification and traceability management of the target material types. A multi-view image acquisition array is set up around the target material, including several high-resolution industrial cameras distributed at specific spatial angles above, to the side at an angle, in front of, and diagonally opposite the target material. Preferably, the number of cameras is 4 to 8 to achieve complete coverage of the boundaries and key feature areas of targets with different morphologies. After the target material is transported to the shooting area via a transmission platform, the system initiates shooting through a trigger signal. Each camera simultaneously acquires image frames of the current target material from multiple perspectives, generating a multi-view image sequence.

[0027] The acquired multi-view image sequences are input into a CNN classification model for rapid analysis and recognition. This model employs a deep convolutional network structure, such as ResNet, EfficientNet, or a lightweight MobileNet variant. The features of the input image sequences are fused to output the current target material's category and material label. Specific identifiable target material types include, but are not limited to: rectangular molybdenum targets: regular rectangular plates with straight edges; ring-shaped copper targets: hollow circular configuration with symmetrical outer circles, commonly used in ion sputtering equipment; and irregular titanium targets: with complex edges or polygonal contours and significant differences in surface texture. Once the target material type is identified, the system automatically assigns it a unique code identifier, such as "MOL-001" for a molybdenum target, and records it in the binding task queue for subsequent process parameter calls and binding execution system use. The target material sample to be bound is sent to the imaging station via a transmission device. Industrial cameras in the multi-view array sequentially acquire images, and after recognition by the CNN model, the system outputs the label "RING-CU," indicating that the target material is a ring-shaped copper target. Target material materials include, but are not limited to: molybdenum, copper, titanium, aluminum, and tungsten. Based on this, the method retrieves the ring-shaped copper target bonding strategy template from the bonding process parameter library and completes the initial process settings such as temperature control zone division, pressing strategy setting, and bonding atmosphere configuration. In actual testing, this method achieved an accuracy rate of over 98.7% for identifying the three types of targets, with an average identification time of less than 200ms, significantly improving the efficiency of automatic process configuration and equipment changeover speed of the bonding production line.

[0028] S200: Call a 3D reconstruction algorithm based on multi-view geometry to generate a 3D geometric structure model of the target material, perform geometric analysis on the 3D geometric structure model, and generate point cloud data of the outer boundary contour and surface morphology of the target material. The point cloud data includes data of contour boundary, feature point coordinates, edge curvature change area, and surface concave-convex abrupt change area.

[0029] Specifically, multi-view image data acquired using a multi-angle industrial camera array is used for camera calibration and feature point extraction. The Zhang Zhengyou calibration method is employed to calibrate the intrinsic and extrinsic parameters of the camera array, ensuring that images acquired by each camera can be geometrically calculated within a unified coordinate system. Then, feature point detection and matching algorithms, such as SIFT and ORB, are used to extract key feature points from the images at each viewpoint and establish cross-viewpoint correspondences. Based on these multi-viewpoint correspondences, a sparse reconstruction algorithm is used to solve for the 3D coordinates of the matched feature points, resulting in a sparse point cloud model of the target material. Building upon this, dense point cloud data is generated using dense reconstruction algorithms, such as PatchMatch Stereo or the deep learning-based MVS method, further improving the accuracy of the target material surface morphology reconstruction. Geometric analysis is then performed on the obtained dense point cloud data. The outer boundary contour of the target material is extracted using the α-shape algorithm or the Poisson surface reconstruction method. Based on the variation of the normal vectors of neighboring points in the point cloud data, the local curvature distribution is calculated to identify edge curvature variation regions. By fitting the local surface and statistical residuals, abrupt surface unevenness regions are detected. Simultaneously, the coordinates of key feature points in the point cloud serve as identifier data for the target material's geometric features, used as input for subsequent flexible fixture fitting control and bonding process parameters. The obtained target material point cloud data contains the following information: contour boundary: a complete description of the target material's external dimensions and geometric boundaries; feature point coordinates: including corner points, curvature extrema points, and key nodes of surface morphology; edge curvature variation regions: used to identify bends, cuts, or irregular edge features; and surface unevenness regions: used to determine surface processing defects or local protrusions / depressions. This enables accurate 3D modeling and geometric feature extraction for different types of targets, providing high-precision data support for subsequent flexible fixture fitting control and zoned temperature-pressure bonding processes.

[0030] Furthermore, step S200 in this embodiment of the application also includes:

[0031] S201: Using the SIFT image feature extraction algorithm, local key points are extracted from multi-view image data, forming feature descriptor subsets. The KNN algorithm is used to match the feature descriptor subsets between image pairs, and the RANSAC algorithm is used to remove mismatched pairs. S202: The Structure-from-Motion method is used to globally optimize the matching results, estimating the camera pose and spatial coordinates of some feature points in the 3D scene corresponding to each image, forming a sparse point cloud structure. The camera pose includes position and orientation. S203: Combining the sparse point cloud structure with multi-view images, the PatchMatch dense reconstruction algorithm is used to generate a dense point cloud covering the entire target surface. A surface reconstruction algorithm is then performed on the dense point cloud to generate a closed 3D triangular mesh model, constituting a complete geometric structure representation of the target material. S204: The edges of the target material in the 3D triangular mesh model are identified. S205: Based on the obtained closed boundary lines, extract the closed boundary lines using boundary tracking and curvature change detection algorithms, and label the direction, closure, and symmetry information of the contours; S206: Based on the obtained closed boundary lines, further extract key geometric points on the 3D triangular mesh model. All key geometric points are output in 3D coordinate form. The key geometric points include points, concave and convex corner points, hole boundary points, maximum and minimum diameter points, and endpoints of the axis of symmetry; S207: Calculate the curvature of each boundary point, identify curvature abrupt change regions, including flanges, notches, and cut edges, for subsequent fixture fitting and local pressure control, and output a curvature distribution map and a point cloud set of key curvature anomaly regions; S208: Based on the surface point cloud, calculate the normal vector of each point and its local elevation gradient, detect abrupt change regions on the target surface, label and output the point cloud boundary of the region, and distinguish between general undulations and abrupt change regions by jointly judging the change in normal vector direction and the magnitude of local elevation gradient.

[0032] It should be understood that during the image acquisition phase, multi-view image sequences of the target material are obtained through industrial cameras arranged at multiple spatial angles. The processing flow for each image includes: Feature point extraction: The Scale Invariant Feature Transform (SIFT) algorithm is used to extract local keypoints in the image, forming a scale- and rotation-invariant feature descriptor set. The feature descriptor set is a set of numerical feature vectors extracted from the local region of each keypoint in the image, used to describe its local image texture, gradient distribution, and other information, facilitating keypoint matching across different viewpoints, scales, and rotations. In the SIFT algorithm: First, several keypoints in the image are detected; then, with each keypoint as the center, a gradient direction histogram is calculated from its neighborhood region; each keypoint generates a 128-dimensional vector, which is the feature descriptor of that point; the descriptors of multiple keypoints are combined to form the feature descriptor set of the image. Feature point matching: Based on the KNN (K-Nearest Neighbors) algorithm, optimal candidate pairs of feature points are found between different image pairs. Mismatch removal: The Random Sample Consensus (RANSAC) algorithm is used to remove mismatched point pairs from the matching pairs, retaining the matching set with good spatial consistency. Structure-from-Motion (SfM) technology is used to perform 3D scene restoration operations on the matching results, including: global optimization of the feature point matching results between images, estimating the camera pose corresponding to each image, including the position of the camera center in 3D space and the rotation matrix of the camera orientation; jointly optimizing the spatial relationship between corresponding points between images, recovering the 3D coordinates of some spatial feature points, and constructing a sparse point cloud structure for the target material; the sparse point cloud is used to initially locate the spatial range and shape configuration of the target material, providing a spatial reference benchmark for subsequent dense modeling. In the SfM (Structure-from-Motion) method, the spatial position coordinates of some feature points in the 3D scene refer to the coordinate information of the sparse feature point set in the real 3D space, calculated through multi-view image feature point matching and triangulation algorithms. After obtaining accurate pose information, the PatchMatch Stereo dense matching algorithm is further used to reconstruct the entire target surface: by utilizing the texture consistency between adjacent images, the depth values ​​of all pixels are estimated to generate a dense point cloud covering the entire target surface; noise removal, point cloud filtering, and normal estimation are performed on the dense point cloud to form a high-density, high-fidelity 3D point set; triangulation and surface reconstruction are performed on the point cloud to generate a closed triangular mesh model representing the complete geometric structure of the target.Based on the generated triangular mesh model, the geometric analysis steps include: First, contour line extraction is performed, identifying closed edges of the model using a boundary tracking algorithm, and extracting abrupt edge points using a curvature change detection algorithm; second, the contour's orientation information, closure, symmetry, and other structural description parameters are output to facilitate fixture fitting and region partitioning; third, key geometric points are extracted, with the system labeling and outputting key geometric points on the model, including: boundary vertices, concave and convex corners, hole boundary points, maximum and minimum diameter points, and the two endpoints of the axis of symmetry. All key points are output in three-dimensional coordinate form for structural identification and boundary localization. Finally, geometric analysis is performed on the boundary and surface structure to identify key areas for process control, including: curvature distribution analysis, calculating the principal curvature values ​​of the contour lines and mesh vertices, marking curvature abrupt change areas, generating curvature distribution maps, and outputting point cloud sets of abnormal areas; normal vector and elevation gradient calculation, calculating the normal vector and gradient direction of each surface segment, and detecting surface abrupt change areas with steps, height differences, and platform edges; marking and outputting the boundary and point cloud of this area for subsequent use in temperature compensation, pressure protection, or fixture avoidance path design. For example, taking a titanium target with an asymmetric flange structure and a countersunk hole on one side as an example, the system acquired a total of 8 images. The processing flow is as follows: SIFT extracts approximately 15,000 feature points, retaining approximately 2,300 valid matching pairs; after SfM modeling, 8 sets of camera poses are restored, generating a sparse point cloud of approximately 6,000 points; PatchMatch generates a dense point cloud of approximately 850,000 points, and the triangular mesh model has approximately 1.3 million faces; 4 contour lines are automatically extracted, 11 key geometric points are labeled, and 3 curvature abrupt change regions are identified; 1 step height difference region is detected and marked as a temperature and pressure compensation control area. The final modeling time is approximately 3.8 seconds, and the modeling accuracy is better than ±0.1mm. It supports subsequent flexible clamping boundary dynamic fitting control and zoned temperature control strategy settings. It achieves accurate, automatic, and high-resolution 3D reconstruction of irregular target structures, significantly improving the target process adaptability and facilitating precise control, parameter adaptation, and risk area avoidance during the bonding process.

[0033] S300: Read the RFID tag on the backplate in the binding preparation area to determine the type of backplate material. Combine the material types of the target and the backplate, and call the matching temperature and pressure strategy template in the binding process parameter library. The temperature and pressure strategy template consists of a heating zone control module, a pressing module, an atmosphere adjustment module, and a cooling control module. The data parameters of each module include: binding temperature, multi-segment hot zone division method, interval temperature data, pressing pressure distribution method, atmosphere protection strategy, and cooling curve setting.

[0034] Specifically, the backplate is placed at the identification position in the bonding preparation area, and the pre-set tag information on the backplate is read using an RFID reader to obtain the material type identifier of the backplate, such as molybdenum, copper, titanium, aluminum, tungsten, etc. The backplate material information is transmitted to the process control unit for subsequent process parameter selection. The target material information, provided by the aforementioned step S100, is obtained, including the target type and specific material. The target material information is combined with the backplate material information, and a matching temperature and pressure strategy template is retrieved from the bonding process parameter library. The process parameter library has pre-established temperature and pressure bonding strategy templates for different target and backplate combinations and is stored in the system.

[0035] Furthermore, step S300 in this embodiment of the application also includes:

[0036] S301: An RFID tag with a unique code is pre-embedded in each backplate. The tag contains key process parameters such as backplate material information, specifications, and surface treatment method. S302: An RFID reader is installed in the binding preparation area. At the start of the binding task, the reader automatically reads the RFID tag content on the backplate, obtains the material type and parameter attributes of the backplate, and registers them in the system. S303: The obtained target material information is combined with the material types of the target and backplate to form a material pair index. S304: The binding process parameter library is queried. Based on the material pair index, a matching temperature and pressure strategy template is automatically retrieved and loaded. The temperature and pressure strategy template consists of a heating zone control module, a pressing module, an atmosphere adjustment module, and a cooling control module. The data parameters of each module include binding temperature, multi-segment hot zone division method, interval temperature data, pressing pressure distribution method, atmosphere protection strategy, and cooling curve settings.

[0037] It should be understood that during the backplate production stage, an RFID tag with a unique code is pre-embedded inside each backplate. The tag stores key process parameters of the backplate, including material type, size specifications, surface treatment method, batch number, and traceability information. The tag is embedded in a pre-defined slot or surface groove of the backplate to ensure a tight bond and prevent it from falling off or being damaged during subsequent heating and pressing. Simultaneously, the embedding position does not affect the uniformity of heating of the backplate or the integrity of the target material bonding interface, ensuring bonding quality. In this way, each backplate has an independent and traceable identification. An RFID reader is installed in the bonding preparation area. When the bonding task begins, the RFID reader automatically reads the tag on the backplate, obtaining parameters such as the material type, specifications, surface treatment method, and batch number, and transmits this information to the system database for registration and filing. This process requires no manual intervention, ensuring the accuracy and efficiency of the reading. The backplate material type, such as molybdenum, copper, titanium, aluminum, tungsten, etc., and the backplate size, specifications, model, surface treatment method, such as electroplating, sandblasting, polishing, etc., batch number, and traceability information are specified. RFID tags are embedded in preset slots or surface grooves of the backplate, ensuring a tight bond between the tag and the backplate and preventing detachment or damage during subsequent heating and pressing. The tag embedding position should ensure that it does not affect the contact surface between the backplate and the target material and the bonding quality, while also facilitating reading by the RFID reader in the bonding preparation area. This process requires no manual intervention, ensuring accuracy and efficiency in reading. The target material information is identified by the visual recognition module before the target material is transported to the bonding station and simultaneously stored in the system database. After registering the backplate parameters, the system combines the backplate information obtained from the RFID tag with the target material identification information in the database to generate a unique "material pair index." The material pair index consists of three parts: "target material type + backplate material and treatment method + batch number," for example, "rectangular molybdenum target – silver-plated copper backplate – batch B202509." The system uses material pair indexes as search criteria to query the bound process parameter library. This parameter library contains multiple temperature and pressure strategy templates corresponding to various material combinations. Each template consists of four modules: a heating zone control module (setting the binding temperature, multi-segment hot zone division method, and interval temperature data); a pressing module (setting the pressing pressure distribution method and multi-stage pressing process parameters); an atmosphere conditioning module (setting the atmosphere protection strategy and gas input and exhaust modes); and a cooling control module (setting the cooling rate, interval cooling temperature, and corresponding curves). Internal parameters of the template include the initial temperature. Target temperature The system automatically retrieves and loads the corresponding template based on the material pair index, and distributes the parameters of each module to the temperature control unit, hydraulic system, atmosphere control interface, and cooling module of the bound system, achieving automated and segmented bound control. For example, when the material pair index is "ring copper target – polished molybdenum backing plate – batch A202509", the system matches the template T-CU-MO-03, and the corresponding control parameters include... Heating Zone Control Module: The binding area is divided into a central zone and an edge zone; the central zone heats up at a rate of 15 ℃ / min, with a target temperature of 430 ℃ and a constant temperature hold for 8 min; the edge zone is preheated to 380 ℃ with a 2 min delay to prevent thermal shock; Pressing Module: The central zone pressure is 12 MPa, and the edge zone pressure is gradually increased to 15 MPa; the pressing process is divided into three stages: pre-pressurization, pressure stabilization, and pressure holding; Atmosphere Regulation Module: Argon protection is activated, and ventilation continues for 10 min, maintaining a pressure of 0.9 atm; the evacuation delay setting is used to prevent cold shrinkage and cracking; Cooling Control Module: The cooling rate is 8 ℃ / min; the central zone is held at 170 ℃ for 5 min, and then the entire zone is naturally cooled; air cooling and a heat-conducting base plate are used for heat dissipation.

[0038] S400: Based on the point cloud data of the outer boundary contour and surface morphology, control the multi-degree-of-freedom flexible fixture to fit and support the actual edge shape of the target material to achieve fitting and clamping.

[0039] Specifically, to achieve reliable clamping and edge fitting support for high-precision, irregularly shaped targets of varying shapes before bonding, a multi-degree-of-freedom flexible clamping mechanism is precisely driven by analyzing the target's boundary contour and surface topography point cloud, achieving non-interference, dynamically adaptive fitting clamping. An industrial camera array captures images of a rectangular molybdenum target from multiple angles, and after 3D reconstruction, a dense point cloud is obtained. The system calls the Poisson surface reconstruction algorithm to fit the outer boundary contour of the rectangle and generate continuous boundary curves. On these four boundary curves, the system selects 16 clamping points according to an equidistant rule. Each edge is assigned 4 points. The surface normal vector N(Ci) of each point is calculated, ensuring the edge surface is approximately perpendicular to the clamping direction. The path planner assigns a target position to each clamping unit and adjusts the orientation of the end-effector clamping surface to ensure its normal vector aligns with the corresponding N(Ci). For example, for a clamping unit located on the long side of a rectangle, the clamping surface normal vector maintains a 0° deviation from the outward normal vector of the long side edge. As the clamping unit approaches the target along the path, the end-effector force / displacement sensor detects the contact status in real time. Taking a point as an example, when the sensor detects a contact force of 15N, it automatically locks the clamping state and maintains it within the set range (15±2N). In the target material's corner area, the curvature detection module identifies two abnormal notch points. The system automatically disables the corresponding... , The clamping points prevent localized stress concentration that could lead to target deformation. All remaining 14 clamping points are in a stable clamping state, and the clamping system switches to constant force holding mode to provide uniform support for the target. The rectangular molybdenum target maintains a positional accuracy of ±0.2mm during clamping, ensuring no displacement during subsequent insertion into the bonding chamber.

[0040] Furthermore, step S400 in this embodiment of the application also includes:

[0041] S401: The flexible clamp consists of several independently controllable clamping units. Each unit can move along the XYZ direction and adjust its attitude angle. A force sensor is mounted at the end to detect the contact state and clamping force. S402: The control module includes a point cloud resolver, a path planner, a motion controller, and a force control and obstacle avoidance module. The point cloud resolver uses the Poisson surface reconstruction algorithm to fit the target edge point cloud, generating a continuous fitting boundary curve. S403: The point cloud resolver converts the boundary point cloud data into a fitting path curve, using the Poisson surface reconstruction algorithm to fit the target edge contour, generating a continuous fitting boundary curve. S404: The path planner initially selects a set of clamping points based on the fitting boundary curve in an equidistant manner. S405: The motion controller calculates the surface normal vector N(Ci) and local curvature of each point, and marks and avoids areas with abnormal curvature, disabling the corresponding points; S406: The motion controller outputs motion commands for each clamping unit based on the clamping point and normal information, causing the end effector to move along the boundary point direction and keep the normal of the clamping surface consistent with N(Ci), so that the bonding direction is perpendicular to the surface; S407: During the motion planning stage, areas with abnormal curvature are avoided and the corresponding points are disabled. When the clamp approaches the target material, the contact state is monitored by the force / displacement sensor, and the clamping force is automatically adjusted to enter the set range; S408: After the flexible bonding is completed, the clamp maintains the preset constant force support state, completing the fitting clamping process.

[0042] It should be understood that the flexible clamping structure comprises several independently controllable clamping units. Each clamping unit has functional components including: an XYZ three-axis moving platform (positioning accuracy ±0.02 mm): used to achieve precise positioning of the end effector gripper in space; an attitude adjustment mechanism (X / Y / Z three-axis rotational degrees of freedom, adjustment accuracy ±0.5°): supporting multi-degree-of-freedom attitude rotation around the X, Y, and Z axes, adjusting the alignment of the gripper surface with the target surface normal; and an end effector force / displacement sensor (force detection accuracy ±0.05 N): used to detect the contact state, clamping force, and contact direction between the clamping unit and the target material in real time. The number of clamping units can be adaptively configured according to the target material size and boundary complexity, typically ranging from 4 to 16 in typical applications. Each clamping unit can move independently under the command of the control module, supporting multi-point distributed bonding. The flexible clamp control module includes the following sub-modules: a point cloud resolver, used to convert target boundary point cloud data into continuous boundary fitting curves. Specifically, it uses point cloud fitting algorithms such as Poisson reconstruction and surface fitting to reconstruct a smooth, continuous curve of the target edge contour, generating a boundary fitting path curve. This fitting path can express the geometry of the target edge, including local features such as edge curvature, cuts, and concavity / convexity, and is the basis for subsequent clamping path planning; and a path planner, used to determine the clamping point position and clamping direction based on boundary curve and normal analysis. Specifically, it automatically generates a set of clamping points and corresponding clamping attitude information based on the fitted path curve and boundary geometric properties. The selection of clamping points considers the principle of equidistant distribution, clamping stability, and the requirement to avoid structurally sensitive areas. For each clamping point, its surface normal vector and local curvature are calculated to generate three-dimensional attitude data (including position + attitude angle) for jaw alignment. The system can call RRT, A*, or other kinematic planning algorithms to generate paths; the motion controller issues positioning and attitude adjustment commands to each gripping unit to achieve synchronous motion control, converting the gripper trajectory and attitude information output by the path planner into control commands for the gripping units. Each gripping unit has XYZ three-axis movement capability and attitude adjustment degrees of freedom, and can move precisely to the specified gripping point according to the target command, and complete the alignment of the gripper direction with the normal vector to ensure that the gripper surface is perpendicular to the target surface; the force control and obstacle avoidance module is used to monitor the gripping force and gripper contact status in real time, automatically avoid abnormal edge areas or void structures, and is linked in real time with the force sensor at the end of the gripper to control the clamping force during the approach or initial contact stage to prevent premature loading or overpressure damage. The module has a closed-loop adjustment function for the clamping force to ensure that each gripper unit applies force stably within the set clamping range.The fitting and clamping process is as follows: First, boundary fitting curve generation: Boundary point cloud data provided by the 3D reconstruction system is received, and the Poisson surface reconstruction algorithm is called to perform edge fitting, generating a continuous fitted boundary curve, which is a smooth expression of the target material's outer edge. Second, clamping points and normal vector generation: k clamping points {C1, C2, ..., Ck} are selected at equal intervals on the fitted boundary curve, and the surface normal vector N(Ci) of each point Ci is calculated, derived from the local fitted surface. The local curvature K(Ci) of each point is calculated to determine whether there is a sudden change in morphology at that location. Third, attitude alignment and motion control: The motion controller sends motion commands to each clamping unit, driving the end effector to move along vector Ci to the target point, synchronously adjusting the clamping surface direction to align its normal vector with N(Ci), thus achieving orthogonal contact between the gripper surface and the target surface. During the contact process, the pose is fine-tuned based on sensor data to avoid rigid interference or overpressure contact. Secondly, contact monitoring and clamping execution: when the end of the clamp approaches the edge of the target material, the end force / displacement sensor collects clamping force data in real time. The controller executes the following logic: initial contact: slight pre-compression; reaching the set clamping force threshold (e.g., 2.5N–4.0N): maintaining clamping; exceeding the safety threshold or abnormal force feedback: automatically releasing and refitting. Finally, obstacle avoidance control and constant force support maintenance: based on curvature changes and point cloud topology analysis results, the force control and obstacle avoidance module marks high-risk structures such as notches, hole edges, and weak areas; prohibits or restricts the clamp from approaching and clamping; automatically shuts down the corresponding clamping unit or puts it into standby mode. After all effective clamping points are properly attached, the clamping system maintains a constant clamping force state for each unit, providing a stable support reference for subsequent bonding heating and pressing. Taking an irregularly shaped titanium target with an asymmetrical "L"-shaped edge contour as an example: approximately 8,200 boundary point clouds were extracted, and closed boundary curves were generated using Poisson reconstruction; eight clamping points could be planned, two of which were located at high curvature corners and were disabled by the obstacle avoidance module; the remaining six clamping ends all achieved normal alignment angle control accuracy better than ±1°; the force control stably maintained a clamping force of 3.2±0.3 N, and the deformation of the entire target after clamping was <0.05 mm. This system achieved automatic fitting and stable support for targets of arbitrary shapes using a flexible clamping system, improving the edge positioning accuracy and clamping consistency of the target, which is beneficial for the quality control of subsequent hot-press bonding.

[0043] S500: The target material and backplate are sent into the bonding chamber. The multi-segment temperature control and pressing module heats and presses the target material and backplate in different zones according to the strategy. The thermophysical parameters of the target material and backplate are used as input to set the temperature and pressure bonding curves of the central area and the edge area to achieve the adaptation of dissimilar materials.

[0044] Specifically, after the target and backplate are transported into the bonding chamber via a transfer platform, the multi-segment heating and pressing module inside the chamber is in standby mode. The system calls the material pair index "ring-shaped copper target – polished molybdenum backplate" generated in the previous steps and automatically retrieves the corresponding temperature and pressure strategy template (template number T-CU-MO-03) from the process parameter library. The multi-segment temperature control unit of the bonding chamber divides the bonding area into a central area and an edge area. The pressing module adopts a differentiated distribution, and the pressing process is divided into three stages: "pre-pressurization – pressure stabilization – pressure holding," ensuring the overall bonding of the target and backplate and avoiding edge warping. The atmosphere adjustment module activates argon protection, and the cooling control module executes the zonal cooling curve. After the entire bonding process is completed, the interface between the target and backplate is continuous and dense, with no obvious gaps.

[0045] Furthermore, step S500 in this embodiment of the application also includes:

[0046] S501: After completing visual recognition and fitting clamping positioning, based on the identified target material, backplate material, and structural model information, a corresponding temperature and pressure strategy template is pre-loaded. This template includes the binding temperature, multi-segment hot zone division method, zone temperature data, pressing pressure distribution method, whether to enable binding atmosphere protection, and cooling curve settings. S502: The target material and backplate assembly are precisely fed into the multi-segment temperature-controlled binding chamber. S503: Multiple independent heating and pressing control units are set within the binding chamber, corresponding to the central and edge areas respectively, to achieve independent local control. S504: Based on the temperature and pressure strategy template parameters, the multi-segment temperature control algorithm and the zoned pressure application strategy are called to drive the heating unit and the servo hydraulic module to perform independent temperature control and zoned pressure adjustment in the region. Before applying pressure, the pressure limit area is identified in combination with the target material structure, and protective adjustment is performed. S505: The material thermophysical parameters are used as input to adapt and optimize the temperature and pressure binding curves of the central area and the edge area to achieve heterogeneous material collaboration. S506: After binding is completed, the atmosphere conversion and heat release are performed first according to the strategy, and then the zoned cooling program is executed. After cooling is completed, the clamp is controlled to slowly retract to prevent warping or tearing of the bonding interface.

[0047] It should be understood that the bonding chamber is equipped with multiple independent heating units and a servo hydraulic pressing module, with control areas divided into central and edge zones to achieve independent temperature and pressure regulation in each zone. The chamber is equipped with an atmosphere conditioning unit (capable of filling with argon or nitrogen) and multiple temperature and pressure sensors for real-time monitoring of temperature, pressing force, and atmosphere status. After visual recognition and fitting clamping positioning of the target and backing plate are completed, the system preloads a temperature and pressure strategy template based on the identified material and structural information. The temperature and pressure strategy template includes the bonding temperature, multi-segment hot zone division method, interval temperature data, pressing pressure distribution method, whether to enable bonding atmosphere protection, and cooling curve settings. The control module distributes the template parameters to each execution unit (heating, pressing, atmosphere, cooling) to establish a closed-loop control sequence for the bonding process. Based on the multi-segment hot zone division in the template, each heating unit in the control chamber independently heats the central and edge zones. The temperature control parameters for each thermal control segment are read from the preset bonding process strategy template, including the initial heating temperature. Target temperature Heating rate r, isothermal holding time Cooling rate Temperature curve type These parameters are configured based on the target material properties, the requirements of the heat-sensitive parts, and the results of historical process optimization, and can be automatically matched and retrieved from the process database. Based on the above parameters, a target temperature curve is constructed for each thermal control zone. This curve typically includes three stages: the heating stage: according to the linear function Tref(t) = The formula is calculated as + r × t, where Tref(t) is the temperature at time t. Let be the initial temperature, r be the heating rate, and t be the time from the start of heating. This implements the process from the initial temperature... The temperature is increased uniformly at a heating rate r; the isothermal section continues until the target temperature is reached. Then, maintain this temperature continuously. Seconds to complete the thermal activation, diffusion, or reaction process; Cooling section: through a set cooling rate This allows the temperature to decrease along a specified path, achieving material stabilization and residual stress release. The target temperature profile can be flexibly adapted to different profile types and parameters. This system enables complex control requirements such as nonlinear heating, slow cooling, or pulsed isothermal control. Finally, the generated target temperature curve Tref(t) is sent to the corresponding heating control units as the reference control command for subsequent temperature adjustment. Closed-loop control is employed, with real-time acquisition of temperature sensor feedback to achieve precise temperature adjustment in each zone, ensuring temperature uniformity and preventing localized overheating. Based on the pressing pressure distribution strategy in the template, the system controls multi-channel hydraulic or servo pressing units in the central and edge zones. The pressing process includes: pre-pressing stage: slowly applying initial pressure; stabilizing stage: increasing to the specified template pressure and maintaining it; and holding stage: maintaining pressure until temperature control is complete. Force sensors monitor the pressure in real time, and the control module automatically adjusts in a closed loop based on feedback to ensure stable pressure that does not exceed the material's allowable range. If the template uses atmosphere protection (such as argon), the control module activates the atmosphere adjustment unit: injecting gas to the preset pressure; controlling the ventilation and exhaust flow rates; and ensuring the atmosphere covers the entire heating, isothermal, and pressing stages. Atmosphere protection prevents oxidation, cracking, or warping of the target or backing plate. The temperature-pressure curve in the template is adaptively adjusted based on the thermophysical parameters of the target and backing plate. The control system automatically optimizes: heating rate; pressing delay and pressure distribution; and regional temperature gradients. This ensures a dense and warp-free interface for different material combinations. The template-specified zoned cooling curves control cooling: the central and edge zones can be cooled simultaneously or in stages; the cooling rate and holding time are executed according to the template settings; air cooling, a heat-conducting base plate, or natural cooling are used. After cooling, the fixture slowly retracts to avoid excessive interface stress leading to tearing or warping. For example, the system transmits template parameters to the temperature and pressing control module, providing guidance for independent control of multiple zones. The target material and backplate assembly are precisely fed into the bonding chamber via a transfer platform, ensuring positional accuracy and alignment during multi-zone heating and pressing. The temperature control module controls the heating rate of the central area to 15 ℃ / min to the target temperature of 430 ℃ based on the template parameters, while the heating rate of the edge area is delayed by 2 minutes and reaches the target temperature of 380 ℃ at 12 ℃ / min. The pressing module applies pressure according to the template zones, stabilizing the central area at 12 MPa and gradually increasing the pressure to 15 MPa in the edge areas. The pressing process is divided into three stages: "pre-pressurization – pressure stabilization – pressure holding". Before pressing, the system limits the pressure based on the edge curvature, pores, and high-risk areas of the target material to avoid local overpressure damage. Multi-segment temperature control and zoned pressing are executed in tandem to adapt to the differences in thermal expansion at the interface of dissimilar materials. The system uses the thermophysical parameters of the target material and the backing plate, including the coefficient of thermal expansion, thermal conductivity, specific heat capacity, and surface treatment layer thickness, as input to optimize and adjust the temperature-pressure curve, ensuring that the thermal stress and shrinkage of different materials are consistent during heating and pressing, thereby improving the density of the bonding interface and the overall flatness.After bonding, the chamber undergoes an atmosphere change, with argon gas being introduced to maintain a pressure of 0.9 atm to prevent target oxidation. The zoned cooling curve controls the central area to be kept at 170 ℃ for 5 minutes before natural cooling, while the edge area is cooled at 8 ℃ / min, with heat dissipation through a heat-conducting base plate and a co-current air-cooling module. After cooling, the fixture is slowly retracted to prevent target warping or interface tearing, ensuring bonding quality.

[0048] Furthermore, step S503 in this embodiment of the application also includes:

[0049] S502-1: The bonding chamber is equipped with multiple independent heating and pressing control units. Each unit corresponds to a different bonding area within the chamber and can achieve local independent control. The bonding area includes at least a central area and an edge area. S502-2: Each bonding area is equipped with a control unit, which includes a heating control unit, a pressing control unit, a temperature and pressure sensor group, a controller, and an interface. S502-3: The central area corresponds to the central part of the main bonding surface of the target material. Structurally, it is a rectangular or elliptical area in the center of the chamber and mainly bears the main bonding force and the main heat input. S502-4: The edge area surrounds the central area and covers the edge, corner, and cut area of ​​the target material. It is used to assist in hot pressing and edge deformation control. S502-5: According to the preset strategy template, the process parameters of the central area and the edge area are controlled separately.

[0050] It should be understood that the bonding chamber is equipped with multiple independent heating and pressing control units, each corresponding to a different bonding area within the chamber, enabling independent local control. The central area, located in the center of the chamber, corresponds to the main bonding surface of the target material and bears the main heat input and pressing force. Structurally, it can be a rectangular or elliptical area, covering the bonding center of the target material. The edge area surrounds the central area, covering the edges, corners, and cut areas of the target material, used to assist in hot pressing and prevent edge warping or local overpressure. For example, for a circular target material with a diameter of 120 mm, the central area can be equipped with an 80 mm diameter circular heating plate and a hydraulic pressing unit, while the edge area surrounds a 10 mm wide annular area, independently heated and pressure-applied to ensure a smooth edge and a firm bond. Each bonding area is equipped with a control unit, including: a heating control unit: controlling the heating rate, target temperature, and constant temperature holding time of the heating element in this area; and a pressing control unit: driving the hydraulic or servo pressing device to achieve zoned pressing pressure control, and setting pre-pressing, stabilizing, and holding pressure stages.

[0051] Temperature and pressure sensor group: Used to collect regional temperature and pressure data in real time, providing feedback for closed-loop control. Controller and interface: Receives commands from the host computer, converts template parameters into heating and pressing execution signals, and links with the temperature and pressure sensors in a closed-loop manner. For example, for a rectangular molybdenum target, the central area control unit is set to a target temperature of 450℃, and the edge area control unit to a target temperature of 420℃; the pressing pressure is set to 12 MPa in the central area and 8 MPa in the edge area. The temperature and pressure sensors sample at a frequency of 10 times per second, ensuring an adjustment accuracy of ±2℃ and ±0.2 MPa. The central area corresponds to the central part of the main bonding surface of the target material, responsible for the main heat input and pressing force transmission. Structural characteristics: Rectangular or elliptical area, occupying 50%~70% of the target material bonding area. Function: Achieves the main bonding force, ensures the density of the bonding interface, and provides the main heat input, ensuring that the target material and backing plate are sufficiently softened or plasticized to form a strong bond. For example, during the bonding process of a ring-shaped copper target, a 60 mm diameter heating plate and hydraulic pressing head are installed in the central area, with a heating rate of 15℃ / min and a holding temperature of 430℃ for 8 minutes; the pressing force in the central area is 12 MPa to ensure uniform stress on the main bonding surface of the target material. The edge area surrounds the central area, covering the edges, corners, and cut areas of the target material. Its functions include: assisting the central area in completing the hot pressing bonding; controlling the deformation of the edge area to prevent warping, tearing, or local overpressure; and providing pressure avoidance and heat regulation for targets with cuts or high curvature areas. For an "L"-shaped titanium target, the edge area is divided into four corner areas, each of which is independently heated and pressurized, with a heating rate of 10℃ / min and an edge area pressure of 5~8 MPa to prevent corner warping. According to the preset strategy template, the system separately controls the temperature curves, pressing pressure curves, and isothermal holding time of the central and edge areas. The control module reads sensor feedback in real time and adjusts the heating rate, pressing pressure, and atmosphere protection status to ensure that materials in different areas are compatible under dissimilar material combinations. Before lamination, protective pressure adjustments are performed on the edge areas based on the pressure-limiting areas identified by the target structure (such as notches and hole edges) to avoid damage. For example, when bonding the annular copper target to the polished molybdenum backing plate: central area temperature curve: heating rate 15℃ / min → target temperature 430℃ → hold temperature for 8 min; edge area temperature curve: heating rate 10℃ / min → target temperature 400℃ → hold temperature for 6 min; central area pressure 12 MPa → maintain stable pressure for 8 min; edge area pressure 8 MPa → avoid the hole area and slowly increase to the target pressure.

[0052] Furthermore, step S504 in this embodiment of the application also includes:

[0053] S504-1: Based on the preset temperature and pressure strategy template parameters, extract the temperature control and pressing parameters of each bonding area, and generate the target temperature curve for each bonding area. The temperature and pressure strategy template parameters include differentiated temperature and pressure curves for the central and edge areas to achieve coordinated adaptation during the bonding of dissimilar materials. S504-2: Real-time acquisition of the actual temperature of each bonding area, calculation of the deviation between the actual temperature and the target temperature as the control error, and driving the power output of the corresponding heating unit through a closed-loop control algorithm to achieve independent temperature control of each bonding area. The closed-loop control algorithm includes PID control, fuzzy adaptive control, or piecewise PI control. Control is used to optimize temperature response time and stability; S504-3: Based on the temperature and pressure strategy template and target structure identification results, the pressure parameters for each binding area are determined, and pressure is applied separately using the servo hydraulic module. Simultaneously, protective adjustments are made to the pressure restriction area, where information on target edges, pores, and high curvature areas is obtained through visual recognition or 3D point cloud analysis. Pressure loading in this area is prohibited or restricted during the pressure application process to prevent localized damage to the target; S504-4: During the protective adjustment process, the heating power and pressure parameters of adjacent binding areas are dynamically adjusted. Pressure is applied to ensure a balanced overall temperature and pressure distribution. Minimum pre-pressure is applied to pressure-limited areas or they are kept on standby to prevent warping or tearing of the bonding interface; S504-4: Infrared thermal imaging and ultrasonic detection are used to obtain bonding surface information. The detection results are analyzed to identify bonding surface defects and output structured defect information, including defect location coordinates, defect category labels, confidence scores, and corresponding image viewpoint labels; S504-5: The structured defect information is associated with the bonding area division. The local heating power or pressure application strategy is adjusted for defect areas to achieve coordinated control of temperature and pressure and defect detection.

[0054] It should be understood that the system reads a preset temperature and pressure strategy template and extracts the temperature control and pressing parameters for each bonding area (including the central and edge areas). Target temperature and pressure curves are generated for the central and edge areas respectively to ensure coordinated temperature gradients and pressure distribution during the bonding of dissimilar materials. For example, for a ring-shaped copper target bonded to a polished molybdenum backplate, the target temperature for the central area is 430℃ with a heating rate of 15℃ / min, and the target temperature for the edge area is 380℃ with a heating rate of 10℃ / min; the pressing force for the central area is 12 MPa, gradually increasing to 15 MPa for the edge area. During the bonding process, the system collects the actual temperature in real time through temperature sensors in each hot zone and calculates the deviation between the target and actual temperatures as the error input for closed-loop control. The control algorithm drives the heating unit's output power through PID, fuzzy adaptive, or piecewise PI control, achieving independent temperature control for each hot zone and ensuring timely and stable temperature response. For example, when the actual temperature in the central area is 420℃, the control algorithm automatically increases the heating power, quickly raising the temperature to 430℃, while the edge area maintains a steady temperature rise to 380℃ to avoid thermal shock. The system combines the temperature and pressure strategy template and the target structure identification results to determine the pressure parameters for each bonding area and applies pressure independently through a servo hydraulic module. Protective adjustments are made to pressure-restricted areas, such as high-curvature edges and porous areas, prohibiting or reducing pressure in these areas to prevent localized damage. For example, for high-curvature areas at irregular titanium target corners, only a minimum preload of 0.5–1 MPa is applied, while the central area maintains a pressure of 10–12 MPa to ensure the target material does not crack. During the protective adjustment process, the system dynamically adjusts the heating power and pressure based on real-time feedback from adjacent areas' temperature and pressure to achieve overall temperature and pressure balance. Pressure-restricted areas remain on standby or receive minimum preload to prevent warping or tearing of the bonding interface. For example, when the central area temperature is slightly higher, the heating power in the edge area is slightly reduced; high-curvature points in the edge area maintain low pressure to ensure uniform overall bonding. The system uses infrared thermal imaging and ultrasonic detection to acquire thermal distribution and structural information of the bonding surface. The detection results are analyzed to identify potential defects and output structured defect information, including location coordinates, defect category, confidence level, and image viewing angle. For example, if a bubble defect is detected at the boundary between the central and edge areas, the output coordinates are (X: 23 mm, Y: 45 mm), category "bubble," confidence level 95%, and image viewing angle directly above. The system associates the structured defect information with the bonding area and dynamically adjusts the local heating power or pressure application strategy based on the defect type and location, achieving closed-loop coordinated control of temperature, pressure, and defect detection.

[0055] Furthermore, step S504-1 in the embodiments of this application also includes:

[0056] S504-1-1: Read the preset temperature and pressure strategy template. This template contains temperature control parameters and pressure parameters for each bound area, as well as differentiated temperature and pressure curves for the central and edge areas. The temperature control parameters include initial temperature, target temperature, heating rate, constant temperature holding time, cooling rate, and temperature curve type. S504-1-2: Based on the temperature and pressure strategy template parameters, extract the temperature control parameters and pressure parameters for each bound area, and generate the target temperature curve and target pressure curve for each bound area. S504-1-3: Assign the target temperature curve for each bound area to the corresponding independent heating control unit, and assign the target pressure curve to the corresponding servo hydraulic module, achieving independent temperature control and zoned pressure application for each area. S504-1-4: Real-time acquisition of the actual temperature and pressure of each bound area, calculation of the control error between the actual and target values, and driving the heating unit and hydraulic module through a closed-loop control algorithm to achieve precise temperature and pressure control. The closed-loop control algorithm includes PID control, fuzzy adaptive control, or piecewise PI control. Control; S504-1-5: Based on the target structure identification results, obtain information on the target edge, pores, and high curvature areas through visual recognition or 3D point cloud analysis methods. Apply protective adjustments to high curvature areas, pores, and weak edge areas to limit local pressure loading. Simultaneously, dynamically adjust the temperature and pressure distribution of adjacent areas to ensure overall bonding temperature and pressure balance. The protective adjustment includes applying minimum pre-pressure treatment to the pressure-limiting area; S504-1-6: Perform multi-segment temperature control and zoned pressure application according to the target temperature curve and pressure curve, including independently controlling the central area and the edge areas surrounding the central area to achieve differential... Coordinated adaptation of material bonding; S504-1-7: During temperature and pressure bonding, the heating power and pressure of adjacent bonding areas are dynamically adjusted in real time to prevent warping or tearing of the bonding interface; S504-1-8: Infrared thermal imaging or ultrasonic detection is used to obtain bonding surface information, the detection results are analyzed to identify structural defects, and the local heating power or pressure strategy is adjusted according to the defect area to achieve coordinated control of temperature and pressure and defect detection; S504-1-9: After bonding is completed, atmosphere conversion and zoned cooling are performed, and the fixture is controlled to slowly retract to ensure that the bonding interface between the target material and the backing plate is flat and does not tear.

[0057] It should be understood that the control system reads a preset temperature and pressure strategy template from the process parameter library. This template, created by R&D personnel for different material combinations, includes temperature control parameters and pressing parameters for each bonding area, and differentiates the temperature and pressure curves for the central and edge areas. Taking a certain dissimilar material combination as an example, the temperature control parameters for the central area are set as follows: initial temperature 150℃, target temperature 450℃, heating rate 8℃ / min, holding time 15min, cooling rate 5℃ / min, and a segmented stepped curve. The temperature control parameters for the edge area are set as follows: initial temperature 150℃, target temperature 420℃, heating rate 6℃ / min, holding time 12min, cooling rate 4℃ / min, and a linear smooth curve. The corresponding pressing parameters include a maximum pressing pressure of 2.5MPa for the central area and 1.8MPa for the edge area, using a segmented loading mode. The parameters for the central and edge areas are extracted to generate target temperature and target pressure curves, respectively. The target temperature curve is generated by an interpolation algorithm using a specific time-temperature reference table, and the target pressure curve is represented by a multi-segment linear function, achieving full-process control from pre-pressurization, pressurization, constant pressure, and depressurization. The target temperature curve for the central area is assigned to the independent control loops of the corresponding resistance heating plate and infrared heating tube, while the target temperature curve for the edge area is assigned to the edge heating unit. Simultaneously, the target pressure curve is assigned to the servo hydraulic module to drive the central pressing unit and the edge pressing unit, enabling independent temperature control and zoned pressurization for each area. During the bonding process, temperature and pressure sensors within the chamber collect real-time values ​​of each bonding area and upload them to the control system. The system employs a closed-loop control algorithm for adjustment, with PID control used in the temperature loop and fuzzy adaptive control in the pressure loop. When a large error occurs, it switches to piecewise PI control to ensure response speed and stability. Combining visual recognition and 3D point cloud analysis, the system identifies the edge contour, pores, and high-curvature areas of the target material. For these vulnerable areas, the control system issues protective adjustment commands, such as setting pressure limiting zones around holes and applying only a minimum pre-pressure of 0.2 MPa; limiting the local heating rate and reducing the pressure loading rate in high curvature areas; and simultaneously adjusting the heating and pressure of adjacent areas to ensure a balanced overall temperature and pressure distribution. The system drives the heating and pressing units, strictly adhering to the target temperature and pressure curves to perform multi-segment temperature control and zoned pressure application. The control of the central and edge zones is independent of each other. For example, the central zone is kept at a constant temperature of 450°C and a pressure of 2.5 MPa, while the edge zone is kept at 420°C and 1.8 MPa, thus achieving coordinated adaptation of dissimilar materials. During the heating and pressing process, the system monitors the temperature and pressure differences of adjacent areas in real time. When the edge zone temperature is too low or the pressing is insufficient, the corresponding heating power is automatically increased or local pressure is applied; when the pressure in the central zone is too high, the pressure output of the central zone servo hydraulic module is dynamically reduced to avoid tearing or warping at the bonding interface.During the bonding process, the infrared thermal imaging module scans the temperature distribution of the bonding surface, identifying temperature anomalies or localized cold spots. Simultaneously, the ultrasonic detection module acquires information on the integrity of the bonding interface, identifying any voids or delamination defects. When a defective area is detected, the control system immediately adjusts the local heating power or pressure strategy, such as increasing the heating energy around the defective area or redistributing the pressing pressure, to achieve synergy between temperature and pressure control and defect repair. After bonding is complete, the system executes an atmosphere conversion procedure, switching the process atmosphere in the chamber from a protective inert gas to an air environment for cooling. Subsequently, the temperature is gradually reduced according to a zoned cooling procedure, with the edge area cooling faster than the center area to offset thermal stress. Finally, the clamps are slowly retracted to ensure that the interface between the target and the backing plate remains flat, preventing warping or tearing. This system enables precise zoned temperature and pressure control during the bonding of dissimilar materials. Combined with a dynamic adjustment mechanism for target structure identification and defect detection, it effectively solves the problems of uneven interface bonding, localized tearing, and warping, improving the reliability and yield of the bonding process.

[0058] Furthermore, step S506 in this embodiment of the application also includes:

[0059] S506-1: Based on the binding structure model and thermal control zoning results, determine the pressure application method and pressure curve of each pressing area; S506-2: The pressing control unit is driven by a servo hydraulic system, supporting independent setting and dynamic adjustment of the pressure in each area; S506-3: Perform structural analysis on the three-dimensional geometric model of the target material, automatically identify local pressure limiting areas, including hollow structural areas, weak corner areas, and protruding feature areas; S506-4: During the pressing process, apply amplitude limiting control to the pressure limiting areas to ensure that the pressure does not exceed the structural bearing capacity, and simultaneously match the target temperature curve of the multi-segment temperature control module in real time to achieve coordinated optimization of pressing force and temperature control, avoiding structural deformation or damage.

[0060] It should be understood that, based on the bonding structure model of the target material and the backing plate, and the preliminary thermal control zoning results, the system determines the pressure application method and corresponding pressure curve for each pressing region. The pressure application methods include constant loading, segmented loading, progressive loading, and pulse loading modes. The pressure curve is generated by combining regional geometric features and temperature control targets, used to achieve differentiated pressing control in different regions. The pressing control unit is driven by a servo hydraulic system. Each pressing region is equipped with an independent servo execution module and pressure sensor, supporting independent setting and real-time dynamic adjustment of region pressure. The control module compares the target pressure curve with sensor feedback data through a closed-loop control algorithm and automatically adjusts the hydraulic output to ensure that the pressure in each region remains stable within the set range. Based on the three-dimensional geometric model of the target material, structural analysis is performed to automatically identify local pressure limiting areas, including hollow structural areas, weak corner areas, and protruding feature areas. This identification process can be achieved through finite element analysis (FEA), boundary curvature calculation, and topological feature analysis, and the limiting areas are marked in the generated pressing control strategy. During the pressing process, the control unit applies amplitude limiting control to the pressure-restricted areas to ensure that the local pressure does not exceed the structure's bearing capacity. Simultaneously, it adjusts the pressing force and temperature distribution in real time based on the target temperature curve output by the multi-segment temperature control module, achieving synergistic optimization. This process effectively prevents local deformation, cracking, or interface damage to the target material structure due to excessive stress. Taking an irregularly shaped titanium target with multiple cuts and holes on its edges as an example: First, the edge cut areas and the areas surrounding the holes are identified as pressure-restricted areas based on the target's three-dimensional geometric model. During the bonding process, a progressive loading mode is used in the central area, with a target pressure set at 4.0 MPa, while a segmented loading mode is used in the edge areas, with a target pressure not exceeding 2.0 MPa. For the restricted areas around the cuts and holes, the servo hydraulic system automatically limits the amplitude, maintaining a pressure ≤ 1.0 MPa, and dynamically matches the corresponding edge area temperature curve, allowing this area to undergo auxiliary pressing at a relatively low temperature. After the pressure and temperature are controlled in a coordinated manner, the interface between the target material and the target material is flat, there are no cracks in the cut area, and the overall bonding strength is increased by 18%, effectively avoiding structural damage.

[0061] In summary, the embodiments of this application have at least the following technical effects:

[0062] This application utilizes multi-angle image recognition and 3D reconstruction algorithms to determine the geometric structure and material information of the target material and backing plate, constructing a structural model and thermal control / pressing zone mapping relationship for bonding control. Through temperature and pressure strategy template matching and a curve adaptive adjustment method driven by thermophysical parameters, it achieves the generation and scheduling of differentiated temperature and pressure curves for each thermal control zone. Combined with a servo hydraulic control system and closed-loop feedback algorithm, it achieves precise coordinated control of temperature and pressure in multiple zones and structural protection of pressure-limited areas. It realizes automated and intelligent execution of the bonding process for dissimilar materials, high strength, and high consistency, significantly improving the bonding quality and process adaptability of the target material.

[0063] Example 2, based on the same inventive concept as the automated process control method applied to the high-precision irregular-shaped target bonding in the foregoing examples, such as... Figure 2 As shown, this application provides an automated process control system for high-precision bonding of irregularly shaped targets. The system and method embodiments in this application are based on the same inventive concept. The system includes:

[0064] The image data acquisition module 10 is used to capture the target material from all angles through a multi-angle industrial camera array, acquire multi-view image data of the target material, and use the image classification model CNN to quickly analyze the acquired multi-view image data, identify the type of target material and mark it with code. The types of target materials include rectangular molybdenum target, ring copper target and irregular titanium target.

[0065] The point cloud data generation module 20 is used to call a three-dimensional reconstruction algorithm based on multi-view geometry to generate a three-dimensional geometric structure model of the target material, perform geometric analysis on the three-dimensional geometric structure model, and generate point cloud data of the outer boundary contour and surface morphology of the target material. The point cloud data includes data of contour boundary, feature point coordinates, edge curvature change area, and surface concave-convex abrupt change area.

[0066] Temperature and pressure strategy template calling module 30 is used to determine the material type of the backplate based on the RFID identification result, and combined with the material types of the target and the backplate, call the temperature and pressure strategy template matched in the binding process parameter library. The data of the temperature and pressure strategy template includes binding temperature, multi-segment hot zone division method, interval temperature data, pressing pressure distribution method, whether to enable binding atmosphere protection, and cooling curve setting.

[0067] Fitting and clamping module 40 is used to control a multi-degree-of-freedom flexible clamp to fit and support the actual edge shape of the target material according to the point cloud data of the outer boundary contour and surface morphology, so as to achieve fitting and clamping.

[0068] The dissimilar material adaptation module 50 is used to send the target material and the backing plate into the bonding chamber, and heat and press them in sections according to a strategy through a multi-segment temperature control and pressing module, and combine the thermophysical properties to set the temperature and pressure bonding curves of the central area and the edge area to perform dissimilar material adaptation.

[0069] Closed-loop control module 60 is used to collect real-time parameters such as temperature, pressure, and displacement during the binding process, and to execute a closed-loop control strategy to dynamically adjust the heating / pressure output.

[0070] The quality report generation module 70 is used to evaluate whether there are defects such as hollowness, delamination, and poor bonding on the bonding surface after the bonding is completed by non-contact methods such as infrared thermal imaging and ultrasonic detection, and generate a quality report.

[0071] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0072] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0073] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. An automated process control method for high-precision bonding of irregularly shaped targets, characterized in that, The method includes: S100. The target material is captured from all angles by a multi-angle industrial camera array to obtain multi-view image data of the target material. The obtained multi-view image data is quickly analyzed by the image classification model CNN to identify the type and material of the target material and mark and encode it. The multi-view image data is also used as the input for subsequent three-dimensional reconstruction. The types of target materials include rectangular molybdenum targets, ring copper targets, and irregular titanium targets. S200. Call the three-dimensional reconstruction algorithm based on multi-view geometry to generate a three-dimensional geometric structure model of the target material. Perform geometric analysis on the three-dimensional geometric structure model to generate point cloud data of the outer boundary contour and surface morphology of the target material. The point cloud data includes data of contour boundary, feature point coordinates, edge curvature change area, and surface concave-convex abrupt change area. S300: Read the RFID tag on the backplate in the binding preparation area to determine the type of backplate material. Combine the material types of the target material and the backplate, and call the matching temperature and pressure strategy template in the binding process parameter library. The temperature and pressure strategy template consists of a heating zone control module, a pressing module, an atmosphere adjustment module, and a cooling control module. The data parameters of each module include: binding temperature, multi-segment hot zone division method, interval temperature data, pressing pressure distribution method, atmosphere protection strategy, and cooling curve setting. S400. Based on the point cloud data of the outer boundary contour and surface morphology, control the multi-degree-of-freedom flexible fixture to fit and support the actual edge shape of the target material to achieve fitting clamping. S500: The target material and backplate are sent into the bonding chamber. The multi-segment temperature control and pressing module heats and presses the target material and backplate according to the strategy. The thermophysical parameters of the target material and backplate are used as input to set the temperature and pressure bonding curves of the center area and the edge area to achieve the adaptation of dissimilar materials. S600: During the binding process, real-time temperature, pressure, and displacement parameters are collected, the collected data are compared with the target curve, and a closed-loop control strategy is executed to dynamically adjust the heating / pressure output. S700 After bonding is completed, infrared thermal imaging and ultrasonic testing are used to assess whether there are hollow areas or delamination on the bonding surface, and a quality report is generated based on the defects.

2. The automated process control method for high-precision bonding of irregularly shaped targets as described in claim 1, characterized in that, The method involves calling a 3D reconstruction algorithm based on multi-view geometry to generate a 3D geometric structure model of the target material. Geometric analysis is then performed on this model to generate point cloud data of the target material's outer boundary contour and surface morphology. This point cloud data includes data on the contour boundary, feature point coordinates, edge curvature variation regions, and surface abrupt change regions. The SIFT image feature extraction algorithm is used to extract local key points from multi-view image data and form feature descriptor subsets. The KNN algorithm is used to match the feature descriptor subsets between image pairs, and the RANSAC algorithm is used to remove mismatched pairs. The Structure-from-Motion method is used to globally optimize the matching results, estimate the camera pose and spatial coordinates of some feature points in the 3D scene corresponding to each image, and form a sparse point cloud structure. The camera pose includes position and orientation. By combining sparse point cloud structure with multi-view images, the PatchMatch dense reconstruction algorithm is used to generate a dense point cloud covering the entire target surface. The surface reconstruction algorithm is then applied to the dense point cloud to generate a closed three-dimensional triangular mesh model, which constitutes the complete geometric structure expression of the target. Identify the edge contour of the target material in the three-dimensional triangular mesh model, extract the closed boundary line through boundary tracking and curvature change detection algorithms, and mark the direction, closure, and symmetry information of the contour; Based on the obtained closed boundary line, key geometric points on the three-dimensional triangular mesh model are further extracted. All key geometric points are output in three-dimensional coordinate form. The key geometric points include vertices, concave and convex corners, hole boundary points, maximum and minimum diameter points, and endpoints of the axis of symmetry. Calculate the curvature of each boundary point, identify curvature abrupt change regions, including flanges, notches, and cutting edges, for subsequent fixture fitting and local pressure control, and output curvature distribution map and point cloud set of key curvature anomaly regions; Based on the normal vector and local elevation gradient of each point in the surface point cloud, abrupt changes in the target surface are detected, the point cloud boundary of the region is marked and output, and the general undulation and abrupt change regions are distinguished by the joint determination of the change in normal vector direction and the magnitude of local elevation gradient.

3. The automated process control method for high-precision bonding of irregularly shaped targets as described in claim 1, characterized in that, The process involves reading the RFID tag on the backplate in the binding preparation area to determine the backplate material type. Based on the material types of the target and the backplate, a matching temperature and pressure strategy template is retrieved from the binding process parameter library. This template comprises a heating zone control module, a pressing module, an atmosphere adjustment module, and a cooling control module. The data parameters for each module include: binding temperature, multi-segment hot zone division method, zone temperature data, pressing pressure distribution method, atmosphere protection strategy, and cooling curve settings. S301. An RFID tag with a unique code is pre-embedded in each backplate. The tag contains information on the backplate material, specifications, and key process parameters of the surface treatment method. S302. An RFID reader is installed in the binding preparation area. When the binding task begins, it automatically reads the contents of the RFID tag on the back panel, obtains the material type and parameter attributes of the back panel, and registers them in the system. S303. Obtain the target material information and combine the material types of the target and the backplate to form a material pair index; S304. Query the binding process parameter library, and automatically retrieve and load the matching temperature and pressure strategy template based on the material pair index. The temperature and pressure strategy template consists of a heating zone control module, a pressing module, an atmosphere adjustment module, and a cooling control module. The data parameters of each module include binding temperature, multi-segment hot zone division method, interval temperature data, pressing pressure distribution method, atmosphere protection strategy, and cooling curve settings.

4. The automated process control method for high-precision bonding of irregularly shaped targets as described in claim 1, characterized in that, The step of controlling a multi-degree-of-freedom flexible clamp to conform to and support the actual edge shape of the target material based on the point cloud data of the outer boundary contour and surface morphology, thereby achieving fitting clamping, includes: The flexible clamp consists of several independently controllable clamping units. Each unit can move along the XYZ direction and adjust its posture angle. The end of the unit is equipped with a force sensor to detect the contact state and clamping force. The control module includes a point cloud resolver, a path planner, an action controller, and a force control and obstacle avoidance module. The point cloud resolver uses the Poisson surface reconstruction algorithm to fit the point cloud of the target edge and generate a continuous fitted boundary curve. The point cloud parser is used to convert boundary point cloud data into a fitting path curve, and uses the Poisson surface reconstruction algorithm to fit the edge contour of the target material to generate a continuous fitting boundary curve. The path planner initially selects a set of clamping points based on the fitted boundary curve using an equidistant method. Calculate the surface normal vector N(Ci) and local curvature of each point, and mark and avoid areas with abnormal curvature, disabling the corresponding points; The motion controller outputs motion commands for each clamping unit based on the clamping point and normal information, causing the end effector to move along the boundary point direction and keeping the normal of the clamping surface consistent with N(Ci), so that the contact direction is perpendicular to the surface; During the motion planning stage, areas with abnormal curvature are avoided and corresponding points are disabled. When the fixture approaches the target, the contact status is monitored by a force / displacement sensor, and the clamping force is automatically adjusted to enter the set range. After the flexible bonding is completed, the clamp maintains a preset constant force support state to complete the fitting and clamping process.

5. The automated process control method for high-precision bonding of irregularly shaped targets as described in claim 1, characterized in that, The process involves sending the target material and backing plate into the bonding chamber, heating and bonding them in zones according to a strategy using a multi-segment temperature control and pressing module, and setting differentiated temperature-pressure bonding curves for the central and edge regions using the thermophysical parameters of the target material and backing plate as inputs to achieve dissimilar material adaptation. After completing visual recognition and fitting clamping positioning, based on the identified target material, backplate material and structural model information, the corresponding temperature and pressure strategy template is preloaded. The temperature and pressure strategy template includes binding temperature, multi-segment hot zone division method, interval temperature data, pressing pressure distribution method, whether to enable binding atmosphere protection and cooling curve settings. The target material and backplate assembly are precisely fed into a multi-segment temperature-controlled bonding chamber. Multiple independent heating and pressing control units are set in the binding chamber, corresponding to the central area and the edge area respectively, to achieve local independent control; Based on the temperature and pressure strategy template parameters, the multi-segment temperature control algorithm and the zoned pressure application strategy are called to drive the heating unit and the servo hydraulic module to perform independent temperature control and zoned pressure adjustment in the area. Before applying pressure, the pressure limit area is identified in combination with the target structure, and protective adjustment is performed. By using material thermophysical parameters as input, the temperature-pressure binding curves of the central and edge regions are adapted and optimized to achieve heterogeneous material synergy. After binding is completed, the atmosphere conversion and heat release are performed first according to the strategy, followed by the partition cooling procedure. After cooling is completed, the clamp is controlled to slowly retract to prevent warping or tearing of the bonding interface.

6. The automated process control method for high-precision bonding of irregularly shaped targets as described in claim 5, characterized in that, The bonding chamber is equipped with multiple independent heating and pressing control units, corresponding to the central and edge areas respectively, to achieve local independent control, including: The bonding chamber is equipped with multiple independent heating and pressing control units. Each unit corresponds to a different bonding area within the chamber, enabling local independent control. The bonding area includes at least a central area and an edge area. Each bonding area is equipped with a control unit, which includes a heating control unit, a pressing control unit, a temperature and pressure sensor group, a controller, and an interface; The central area corresponds to the central part of the main bonding surface of the target material. Structurally, it is a rectangular or elliptical area in the center of the cavity, which mainly bears the main bonding force and the main heat input. The edge region surrounds the central region and covers the edge, corner, and cut area of ​​the target material, serving to assist in hot pressing and edge deformation control; Based on the preset strategy template, the process parameters of the central area and the edge area are controlled separately.

7. The automated process control method for high-precision bonding of irregularly shaped targets as described in claim 5, characterized in that, The method, based on temperature and pressure strategy template parameters, invokes a multi-segment temperature control algorithm and a zoned pressure application strategy to drive the heating unit and servo hydraulic module, performing independent temperature control and zoned pressure adjustment for each area. Before applying pressure, it identifies pressure-limiting zones based on the target material structure and performs protective adjustments, including: Based on the preset temperature and pressure strategy template parameters, the temperature control and pressing parameters of each binding area are extracted, and the target temperature curve of each binding area is generated. The temperature and pressure strategy template parameters include the differentiated temperature and pressure curves of the central area and the edge area to achieve collaborative adaptation when bonding dissimilar materials. The actual temperature of each bound area is collected in real time, the deviation between the actual temperature and the target temperature is calculated as the control error, and the power output of the corresponding heating unit is driven through a closed-loop control algorithm to achieve independent temperature control of each bound area. The closed-loop control algorithm includes PID control, fuzzy adaptive control or piecewise PI control, which is used to optimize temperature response time and stability. Based on the temperature and pressure strategy template and the target structure identification results, the pressure parameters for each binding area are determined, and pressure is applied separately using a servo hydraulic module. At the same time, protective adjustments are made to the pressure restriction area. The pressure restriction area obtains information on the target edge, holes, and high curvature areas through visual recognition or three-dimensional point cloud analysis methods, and pressure loading in this area is prohibited or restricted during the pressure application process to prevent local damage to the target. During the protective adjustment process, the heating power and pressure of the adjacent bonding area are dynamically adjusted to ensure a balanced overall temperature and pressure distribution. Minimum pre-pressure is applied to the pressure-limited area or it is kept in standby mode to prevent warping or tearing of the bonding interface. Infrared thermal imaging and ultrasonic detection are used to obtain bonding surface information. The detection results are analyzed to identify bonding surface defects and output structured defect information, including defect location coordinates, defect category labels, confidence scores and corresponding image viewpoint labels. By associating structured defect information with the bound area division, and adjusting the local heating power or pressure strategy for the defect area, the coordinated control of temperature, pressure and defect detection can be achieved.

8. The automated process control method for high-precision bonding of irregularly shaped targets as described in claim 7, characterized in that, The process involves extracting temperature control and pressing parameters for each bonding region based on preset temperature and pressure strategy template parameters, and generating a target temperature curve for each bonding region. The temperature and pressure strategy template parameters include differentiated temperature and pressure curves for the central and edge regions to achieve coordinated adaptation during the bonding of dissimilar materials. Read the preset temperature and pressure strategy template, which contains temperature control parameters, pressing parameters, and differentiated temperature and pressure curves for the center and edge areas for each binding region. The temperature control parameters include the starting temperature, target temperature, heating rate, constant temperature holding time, cooling rate, and temperature curve type. Based on the temperature and pressure strategy template parameters, extract the temperature control parameters and compression parameters of each binding area, and generate the target temperature curve and target pressure curve for each binding area; The target temperature curve of each bound area is assigned to the corresponding independent heating control unit, and the target pressure curve is assigned to the corresponding servo hydraulic module to achieve independent temperature control and zoned pressure application for each area. The system collects the actual temperature and pressure of each binding area in real time, calculates the control error between the actual value and the target value, and drives the heating unit and hydraulic module through a closed-loop control algorithm to achieve precise control of temperature and pressure. The closed-loop control algorithm includes PID control, fuzzy adaptive control or piecewise PI control. Based on the target structure identification results, information on the target edge, holes and high curvature areas is obtained through visual recognition or three-dimensional point cloud analysis methods. Protective adjustments are applied to the high curvature areas, holes and weak edge areas to limit local pressure loading. At the same time, the temperature and pressure distribution in the adjacent areas is dynamically adjusted to ensure overall bonding temperature and pressure balance. The protective adjustment includes applying minimum pre-pressure treatment to the pressure-limited area. Multi-segment temperature control and zoned pressure application are performed according to the target temperature curve and pressure curve, including independent control of the central area and the edge area surrounding the central area, so as to achieve coordinated adaptation of heterogeneous material combination binding; During the temperature and pressure bonding process, the heating power and pressure of the adjacent bonding area are dynamically adjusted in real time to prevent warping or tearing of the bonding interface. Infrared thermal imaging or ultrasonic detection is used to obtain information about the bonding surface. The detection results are analyzed to identify structural defects. Based on the defect area, the local heating power or pressure strategy is adjusted to achieve coordinated control of temperature, pressure and defect detection. After bonding is completed, perform atmosphere conversion and zoned cooling, and control the jig to slowly retract to ensure that the interface between the target and the backplate is flat and does not tear.

9. The automated process control method for high-precision bonding of irregularly shaped targets as described in claim 5, characterized in that, The method, based on temperature and pressure strategy template parameters, invokes a multi-segment temperature control algorithm and a zoned pressure application strategy to drive the heating unit and servo hydraulic module, performing independent temperature control and zoned pressure adjustment for each area. Before applying pressure, it identifies pressure-limiting zones based on the target material structure and performs protective adjustments, including: Based on the binding structure model and thermal control zoning results, the pressure application method and pressure curve of each pressing zone are determined; The pressing control unit is driven by a servo hydraulic system, which supports independent setting and dynamic adjustment of pressure in each zone; Structural analysis is performed on the three-dimensional geometric model of the target material to automatically identify local pressure-restricted areas, including hollow structural areas, weak corner areas, and protruding feature areas. During the pressing process, the pressure limiting zone is subjected to amplitude control to ensure that the pressure does not exceed the structure's bearing capacity. At the same time, it is matched in real time with the target temperature curve of the multi-segment temperature control module to achieve synergistic optimization of pressing force and temperature control, thereby avoiding structural deformation or damage.

10. An automated process control system for high-precision bonding of irregularly shaped targets, characterized in that, The system includes: The image data acquisition module is used to capture the target material from all angles using a multi-angle industrial camera array, acquire multi-view image data of the target material, and use an image classification model CNN to quickly analyze the acquired multi-view image data, identify the type and material of the target material and mark and encode it. The multi-view image data also serves as the input for subsequent 3D reconstruction. The types of target materials include rectangular molybdenum targets, ring copper targets, and irregular titanium targets. The point cloud data generation module is used to call a three-dimensional reconstruction algorithm based on multi-view geometry to generate a three-dimensional geometric structure model of the target material, perform geometric analysis on the three-dimensional geometric structure model, and generate point cloud data of the outer boundary contour and surface morphology of the target material. The point cloud data includes data of contour boundary, feature point coordinates, edge curvature change area, and surface concave-convex abrupt change area. The temperature and pressure strategy template calling module is used to read the backplate RFID tag in the binding preparation area, determine the backplate material type, and call the matching temperature and pressure strategy template in the binding process parameter library in combination with the material types of the target and the backplate. The temperature and pressure strategy template consists of a heating zone control module, a pressing module, an atmosphere adjustment module, and a cooling control module. The data parameters of each module include: binding temperature, multi-segment hot zone division method, interval temperature data, pressing pressure distribution method, atmosphere protection strategy, and cooling curve setting. The fitting and clamping module is used to control a multi-degree-of-freedom flexible clamp to fit and support the actual edge shape of the target material according to the point cloud data of the outer boundary contour and surface morphology, so as to achieve fitting and clamping. The dissimilar material adaptation module is used to send the target material and the back plate into the bonding chamber. The multi-segment temperature control and pressing module heats and presses the target material and the back plate in different zones according to a strategy. The thermophysical parameters of the target material and the back plate are used as input to set different temperature and pressure bonding curves for the central and edge zones to achieve dissimilar material adaptation. The closed-loop control module is used to collect real-time temperature, pressure, and displacement parameters during the binding process, compare the collected data with the target curve, execute the closed-loop control strategy, and dynamically adjust the heating / pressure output. The quality report generation module is used to evaluate whether there are hollow areas or delamination on the bonding surface after the bonding is completed by infrared thermal imaging and ultrasonic detection, and to generate a quality report based on the defects.

Images