Integrated surface mounting method and system for ultrathin chip
By constructing a global error compensation model and optimizing the mounting path using a nonlinear programming algorithm, the problem of accumulated errors in ultra-thin chip mounting is solved, achieving high-precision and stable automated mounting, which is suitable for surface mounting of ultra-thin chips.
Patent Information
- Application Number
- CN202511819213.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-04
- Publication Date
- 2026-03-06
AI Technical Summary
Existing chip surface mount technology is affected by many factors at each stage, and the cumulative error leads to challenges in the stability and accuracy of ultra-thin chip mounting, especially in complex environments where it is difficult to achieve high precision and stability.
By acquiring multispectral image data and substrate 3D point cloud data, a global error compensation model for a multi-degree-of-freedom motion system is constructed. Combining nonlinear programming algorithms and improved genetic algorithms, the mounting path trajectory is optimized to achieve dynamic compensation, eliminate accumulated errors, and improve mounting accuracy and stability.
It significantly improves the mounting accuracy and stability of ultra-thin chips, enhances the robustness of the system in complex environments, and realizes efficient and automated integrated mounting.
Smart Images
Figure CN121620262A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of semiconductor manufacturing and microelectronic packaging technology, specifically relating to an integrated surface mount method and system for ultra-thin chips. Background Technology
[0002] Ultra-thin chips, typically less than 100 μm thick, significantly reduce the size of electronic devices, leading to their increasingly widespread application in flexible electronics, wearable devices, and other fields. The demands for miniaturization, lightweighting, and high reliability in electronic devices also present multiple challenges to ultra-thin chip surface mount technology: it requires precise control of the total package thickness to fit compact spaces, while also ensuring high mounting accuracy and stable production efficiency, and avoiding yield losses due to the fragile and easily deformable nature of ultra-thin chips.
[0003] Current chip surface mount technology mainly involves chip pickup, positioning, placement, and curing processes. Chip pickup is the first step in chip surface mount technology, and its accuracy directly determines the quality foundation of subsequent processes. With the widespread adoption of ultra-thin chips, the alignment tolerance between the placement machine nozzle and the chip has become extremely small; even slight misalignment can reduce placement accuracy in subsequent processes. In the chip positioning stage, existing placement machines generally use high-precision vision systems for position compensation. The clarity of the optical system and the recognition algorithm have a significant impact on positioning accuracy. In addition, the mechanical structure of the placement machine will experience wear and tear during long-term operation, leading to micron-level cumulative errors in motion positioning. Simultaneously, thermal deformation may occur due to motor heating or ambient temperature fluctuations during operation, all of which affect the final positioning accuracy. In the chip placement stage, "tombstone formation" often occurs due to component placement misalignment, asymmetrical pad design, or uneven solder paste printing, resulting in significant accuracy errors. In the chip curing stage, thermal stress usually causes the chip to move on the molten solder, causing final positional misalignment and affecting placement stability and consistency.
[0004] Therefore, existing chip surface mount technology is affected by numerous factors at each stage, leading to cumulative errors that pose a significant challenge to chip mounting accuracy and consequently impact mounting stability. Achieving integrated mounting of ultra-thin chips with high accuracy, while ensuring stability under complex environments, has become a pressing technical problem to be solved. Summary of the Invention
[0005] To address the aforementioned problems in the prior art, this invention provides an integrated surface mount method and system for ultrathin chips. This invention provides an integrated surface mount method for ultra-thin chips, comprising the following steps: Multispectral image data of the target chip surface is acquired during the mounting process, and three-dimensional point cloud data of the substrate is also acquired. Key alignment marker coordinates are extracted from the multispectral image data, and geometric reference point information is extracted from the three-dimensional point cloud data of the substrate. Based on the coordinates of the key alignment marks and the information of the geometric reference points, a global error compensation model for the multi-degree-of-freedom motion system is constructed, and the vertical compensation amount, the first horizontal compensation amount, the second horizontal compensation amount, and the rotation compensation amount in the chip mounting process are calculated to obtain a four-dimensional dynamic compensation vector. Using the four-dimensional dynamic compensation vector as the target point of the mounting path, and combining mounting accuracy and mounting path smoothness, the ultra-thin chip mounting path optimization is modeled as a nonlinear programming problem, and the nonlinear programming problem is solved to obtain the mounting path trajectory. The mounting path trajectory is discretized into a series of execution action sequences including position, speed, and time. These execution action sequences are sent to the motion controller of the mounting head assembly for surface mounting of the ultra-thin chip.
[0006] In one embodiment of the present invention, extracting key alignment marker coordinates from the multispectral image data includes: The multispectral image data is adaptively filtered to obtain a preprocessed image; The preprocessed image is used to identify alignment marker regions to obtain alignment markers. By calibrating the camera parameters, the alignment marks are converted into actual physical coordinates in the mounting system coordinate system, thus obtaining the coordinates of the key alignment marks.
[0007] In one embodiment of the present invention, extracting geometric reference point information from the three-dimensional point cloud data of the substrate includes: The three-dimensional point cloud data of the substrate is segmented to obtain the geometric reference point region; The geometric reference point region is identified using a clustering algorithm, and the three-dimensional coordinates of the feature points in the geometric reference point region are calculated to obtain the geometric reference point information.
[0008] In one embodiment of the present invention, a global error compensation model for a multi-degree-of-freedom motion system is constructed based on the coordinates of the key alignment mark and the geometric reference point information. The vertical compensation vector, the first horizontal compensation vector, the second horizontal compensation vector, and the rotation compensation vector during the chip mounting process are calculated to obtain a four-dimensional dynamic compensation vector, including: The key alignment mark coordinates of the target chip are matched with the geometric reference point information of the substrate to establish the correspondence between the two and obtain the matching result. Calculate the ideal transformation matrix of the target chip relative to the substrate based on the matching results:
[0009] in, Represents the ideal transformation matrix. This represents the coordinates of the target chip's key alignment markers after transformation. This indicates the original coordinates of the key alignment markers for the target chip. Represents the rotation matrix. Represents the translation vector; A global error compensation model is constructed using the ideal transformation matrix:
[0010] in, Indicates global error. This represents the transformed chip coordinates. Indicates the coordinates of the chip's key alignment markers. Indicates the coordinates of geometric reference points on the substrate. Represents the regularization parameter. Indicates the deviation of the rotation angle. Indicates the weighting coefficient. Indicates the number of markers; The global error compensation model is solved using the singular value decomposition method to obtain the optimal rotation matrix and the optimal translation vector. The vertical compensation amount is obtained through the optimal translation vector. First horizontal direction compensation amount Second horizontal direction compensation amount The rotation compensation amount is obtained through the optimal rotation matrix. The four-dimensional dynamic compensation vector is obtained. .
[0011] In one embodiment of the present invention, the objective function of the nonlinear programming problem is:
[0012] in, Indicates the optimization objective. and All are weighting coefficients. Indicates the first The deviation between the individual mounting points and the four-dimensional dynamic compensation vector Indicates the rate of change of the mounting path speed. Indicates time.
[0013] In one embodiment of the present invention, solving the nonlinear programming problem to obtain the mounting path trajectory includes: An improved genetic algorithm is used to solve the objective function of the nonlinear programming problem to obtain the mounting path trajectory; wherein, the selection probability formula of the improved genetic algorithm is:
[0014] in, Indicates the first The probability of an individual being selected Indicates the fitness value. For population size, This is the attenuation factor.
[0015] In one embodiment of the present invention, the mounting path trajectory is:
[0016] in, Total mounting time This refers to the position of the mounting head in the X-axis direction. This refers to the position of the mounting head in the Y-axis direction. This refers to the position of the mounting head in the Z-axis direction. This indicates the angle at which the mounting head rotates around the Z-axis over time.
[0017] In one embodiment of the present invention, the mounting path trajectory is discretized into a series of execution action sequences including position, speed, and time, including: The mounting path trajectory is continuously sampled according to a fixed control cycle: Generate discrete time series ,in, , , For sampling quantity, To control the cycle; For each sampling time point Calculate the corresponding target location: Generate discrete position sequence ; The velocity function is obtained by differentiating the mounting path trajectory: Then for each sampling time point Perform velocity sampling: Generate discrete velocity sequences .
[0018] Another embodiment of the present invention provides an integrated surface mount system for ultra-thin chips, comprising: The data acquisition module is used to acquire multispectral image data of the target chip surface during the mounting process and to acquire three-dimensional point cloud data of the substrate. The feature extraction module is used to extract key alignment marker coordinates from the multispectral image data and extract geometric reference point information from the three-dimensional point cloud data of the substrate. The compensation acquisition module is used to construct a global error compensation model of the multi-degree-of-freedom motion system based on the coordinates of the key alignment mark and the information of the geometric reference point, and to calculate the vertical compensation amount, the first horizontal compensation amount, the second horizontal compensation amount and the rotation compensation amount in the chip mounting process to obtain a four-dimensional dynamic compensation vector. The mounting path trajectory planning module is used to take the four-dimensional dynamic compensation vector as the target point of the mounting path, and combine the mounting accuracy and mounting path smoothness to model the ultra-thin chip mounting path optimization as a nonlinear programming problem, and solve the nonlinear programming problem to obtain the mounting path trajectory. The path discretization module is used to discretize the mounting path trajectory into a series of execution action sequences including position, speed and time. The execution action sequences are sent to the motion controller of the mounting head assembly for surface mounting of ultra-thin chips.
[0019] Compared with the prior art, the beneficial effects of the present invention are as follows: In this invention, key alignment marker coordinates are extracted using multispectral image data, geometric reference point information is extracted using substrate 3D point cloud data, and a global error compensation model for a multi-degree-of-freedom motion system is constructed. This effectively solves the cumulative error problem caused by traditional step-by-step calibration and significantly improves mounting accuracy. The mounting path optimization is modeled as a nonlinear programming problem and solved using an improved genetic algorithm, achieving a balance between mounting efficiency and stability. Real-time calculation of dynamic compensation parameters is realized, eliminating manual intervention and improving production consistency and reliability. This method can achieve integrated mounting of ultra-thin chips with high mounting accuracy, enhances the robustness of the system in complex production environments, and is suitable for various ultra-thin chip mounting processing scenarios. Attached Figure Description
[0020] Figure 1 This is a schematic flowchart of an integrated surface mount method for ultra-thin chips provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of an integrated surface mount system for ultra-thin chips provided in an embodiment of the present invention. Detailed Implementation
[0021] The present invention will be further described in detail below with reference to specific embodiments, but the implementation of the present invention is not limited thereto.
[0022] Example 1 This embodiment achieves high precision, high efficiency, and high reliability in the ultrathin chip mounting process by introducing a joint sensing mechanism of multispectral imaging and three-dimensional contour scanning, combined with a global error compensation model, nonlinear programming algorithm, and improved genetic algorithm.
[0023] Please see Figure 1 , Figure 1 This is a schematic flowchart of an integrated surface mount method for ultra-thin chips provided in an embodiment of the present invention.
[0024] The integrated surface mount method for ultra-thin chips in this embodiment includes the following steps: S1. Collect multispectral image data of the target chip surface during the mounting process, and collect three-dimensional point cloud data of the substrate.
[0025] Specifically, a multispectral imaging module configured on the placement head assembly and a 3D contour scanning module configured on the substrate platform simultaneously acquire surface feature data of the target chip and substrate topography data during the placement process. The multispectral imaging module uses different wavelengths of light (such as visible and infrared light) to illuminate the chip surface, generating multispectral images that include chip surface texture, alignment marks, and other features. Multispectral imaging effectively enhances the contrast of markings on different material surfaces, overcoming the problem of indistinct features under a single light source. The 3D contour scanning module uses laser triangulation to acquire 3D point cloud data of the substrate. This 3D point cloud data accurately reflects the height changes, warpage, and pad positions of the substrate surface.
[0026] S2. Extract key alignment marker coordinates from multispectral image data and extract geometric reference point information from substrate 3D point cloud data.
[0027] Extracting key alignment marker coordinates from multispectral image data includes: First, adaptive filtering is applied to the multispectral image data to remove image noise and uneven illumination interference, resulting in a preprocessed image. Then, a deep learning model is used to identify alignment marker regions in the preprocessed image, obtaining alignment markers; this deep learning model can be a pre-trained convolutional neural network. Finally, after obtaining the alignment markers from the multispectral image data, camera parameter calibration is used to convert the pixel coordinates of the alignment markers into actual physical coordinates in the mounting system coordinate system, yielding the key alignment marker coordinates. .
[0028] Geometric reference point information is extracted from the 3D point cloud data of the substrate, including: First, the 3D point cloud data of the substrate is preprocessed to segment different regions such as pads, wiring, and alignment marks, obtaining geometric reference point regions. Then, a clustering algorithm is used to identify the geometric reference point regions and calculate the 3D coordinates of feature points within these regions. The geometric reference point information is obtained; the clustering algorithm can employ the K-means method. The calculated geometric reference points are the alignment mark coordinates of the substrate. Position marker coordinates Key alignment marker coordinates of the target chip One-to-one correspondence.
[0029] S3. Construct a global error compensation model for the multi-degree-of-freedom motion system based on the coordinates of the key alignment marks and the information of the geometric reference points, and calculate the vertical compensation amount, the first horizontal compensation amount, the second horizontal compensation amount and the rotation compensation amount in the chip mounting process to obtain a four-dimensional dynamic compensation vector.
[0030] Specifically, the key alignment coordinates of the target chip are first marked. Geometric reference point information of the substrate Perform matching, establish the correspondence between the two, and obtain the matching results. ,in, This represents the number of marker points.
[0031] Then, the ideal transformation matrix of the target chip relative to the substrate is calculated based on the matching results:
[0032] in, Represents the ideal transformation matrix. This represents the coordinates of the target chip's key alignment markers after transformation. This indicates the original coordinates of the key alignment markers for the target chip. Represents the rotation matrix. Represents the translation vector. Rotation matrix. Translation vector Corresponding to multi-degree-of-freedom motion.
[0033] Next, a global error compensation model is constructed based on the ideal transformation matrix:
[0034] in, Indicates global error. This represents the transformed chip coordinates. Indicates the coordinates of the chip's key alignment markers. Indicates the coordinates of geometric reference points on the substrate; This represents the regularization parameter, ranging from 0.01 to 0.1; This indicates the deviation in rotation angle, to prevent over-rotation; This represents a regularization term, which prevents the model from undergoing unrealistically large angle rotations in pursuit of minimal error. This enhances the stability and rationality of the model; This represents the weighting coefficient, which is assigned based on the recognition confidence of the marker points, giving different marker points different importance. For example, marker points located at the corners of the chip receive higher weights because they are more sensitive to rotation and translation errors. Indicates the number of markers.
[0035] Subsequently, based on the idea of weighted least squares, the singular value decomposition method was used to solve the global error compensation model, and the optimal rotation matrix and the optimal translation vector were obtained.
[0036] Specifically, the goal of solving the global error compensation model is to find an optimal transformation matrix. (including rotation matrix) Translation vector This makes the transformed chip points With substrate point The goal is to get as close as possible by minimizing the global error. This is achieved through regularization. In this embodiment, due to the regularization term... To ensure numerical stability and prevent overfitting, a penalty is applied to large rotation angles, guiding the solution towards smaller, more realistic angle adjustments. Therefore, the regularization term is ignored. Construct a system that minimizes the global error The objective function is determined, and then the optimal rotation matrix is obtained through decentralization combined with singular value decomposition. and optimal translation vector Among them, minimizing the global error The objective function is:
[0037] It should be noted that gradient descent can also be used to minimize the global error. The optimal rotation matrix is obtained. and optimal translation vector , where the optimal rotation matrix and optimal translation vector These are the transformation parameters in three-dimensional space, the optimal rotation matrix. Corresponding rotation angle around the Z-axis θ, Optimal translation vector It is a three-dimensional vector ,in, Let X be the translation amount in the horizontal direction. Let Y be the translation amount in the horizontal direction. This represents the translation in the vertical direction of Z. The height is determined by the height of the chip and the substrate. This ensures uniform mounting pressure and prevents crushing of ultra-thin chips.
[0038] Finally, through the optimal translation vector Obtain vertical compensation amount First horizontal direction compensation amount Second horizontal direction compensation amount Through the optimal rotation matrix Obtain rotational compensation amount The four-dimensional dynamic compensation vector is obtained. .
[0039] S4. Using the four-dimensional dynamic compensation vector as the target point of the mounting path, and combining mounting accuracy and mounting path smoothness, the ultra-thin chip mounting path optimization is modeled as a nonlinear programming problem, and the nonlinear programming problem is solved to obtain the mounting path trajectory.
[0040] Specifically, in order to optimize the mounting path parameters, a four-dimensional dynamic compensation vector is used. Using the placement path as the target point, and considering both placement accuracy and path smoothness, the optimization of the ultra-thin chip placement path is modeled as a nonlinear programming problem. The objective function of the nonlinear programming problem is:
[0041] in, Indicates the optimization objective. and All are weighting coefficients. Indicates the first The deviation between the individual mounting points and the four-dimensional dynamic compensation vector Indicates the rate of change of the mounting path speed. This represents time. In the objective function, Requires placement deviation at each point along the entire path By keeping it as small as possible, accuracy is guaranteed. The objective function requires a smooth rate of change in the movement speed of the placement head to avoid sudden stops and starts, reduce vibration, and ensure stability in the placement process. Therefore, this objective function balances placement accuracy and path smoothness to achieve efficient path planning.
[0042] Furthermore, an improved genetic algorithm is used to solve the objective function of the nonlinear programming problem to obtain the mounting path trajectory. The selection probability formula of the improved genetic algorithm is as follows:
[0043] in, Indicates the first The probability of an individual being selected Indicates the fitness value. For population size, is the decay factor. This selection probability formula accelerates convergence through an exponential decay mechanism while avoiding getting trapped in local optima.
[0044] Specifically, the obtained mounting path trajectory is:
[0045] in, Total mounting time This refers to the position of the mounting head in the X-axis direction. This refers to the position of the mounting head in the Y-axis direction. This refers to the position of the mounting head in the Z-axis direction. This indicates the angle at which the mounting head rotates around the Z-axis over time.
[0046] S5. Discretize the mounting path trajectory into a series of execution action sequences containing position, speed and time. The execution action sequences are sent to the motion controller of the mounting head assembly for surface mounting of ultra-thin chips.
[0047] Specifically, the process of discretizing the mounting path trajectory into a sequence of execution actions is as follows: The mounting path trajectory is controlled according to a fixed cycle. Perform continuous sampling: Generate discrete time series ,in, , , For sampling quantity, To control the cycle.
[0048] For each sampling time point Calculate the corresponding target location: Generate discrete position sequence .
[0049] The velocity function is obtained by differentiating the mounting path trajectory: Then for each sampling time point Perform velocity sampling: Generate discrete velocity sequences .
[0050] Furthermore, the sequence of actions is sent to the motion controller of the placement head assembly. The controller drives the motor to strictly follow the sequence instructions, completing precise placement actions with real-time dynamic compensation. The entire process requires no manual intervention, achieving full automation.
[0051] Please see Figure 2 , Figure 2 This is a schematic diagram of an integrated surface mount system for ultra-thin chips provided in an embodiment of the present invention. The system includes a data acquisition module, a feature extraction module, a compensation acquisition module, a mounting path trajectory planning module, and a path discretization module.
[0052] Specifically, the data acquisition module is used to acquire multispectral image data of the target chip surface during the mounting process and to acquire 3D point cloud data of the substrate; the feature extraction module is used to extract the coordinates of key alignment marks from the multispectral image data and to extract geometric reference point information from the 3D point cloud data of the substrate; the compensation acquisition module is used to construct a global error compensation model of the multi-degree-of-freedom motion system based on the coordinates of the key alignment marks and the geometric reference point information, and to calculate the vertical compensation amount, the first horizontal compensation amount, the second horizontal compensation amount, and the rotational compensation amount during the chip mounting process to obtain a four-dimensional dynamic compensation vector; the mounting path trajectory planning module is used to use the four-dimensional dynamic compensation vector as the target point of the mounting path, and to combine the mounting accuracy and the smoothness of the mounting path to optimize the ultra-thin chip mounting path as a nonlinear programming problem, and to solve the nonlinear programming problem to obtain the mounting path trajectory; the path discretization module is used to discretize the mounting path trajectory into a series of execution action sequences containing position, velocity, and time, and the execution action sequences are sent to the motion controller of the mounting head assembly for surface mounting of the ultra-thin chip.
[0053] For the specific execution steps of each module in the system, please refer to the above-mentioned integrated surface mount method for ultra-thin chips, which will not be repeated here.
[0054] This embodiment utilizes multispectral image data to extract key alignment marker coordinates, extracts geometric reference point information from substrate 3D point cloud data, and constructs a global error compensation model for a multi-degree-of-freedom motion system. This effectively solves the cumulative error problem caused by traditional step-by-step calibration, significantly improving mounting accuracy. The mounting path optimization is modeled as a nonlinear programming problem and solved using an improved genetic algorithm, achieving a balance between mounting efficiency and stability. Real-time calculation of dynamic compensation parameters is realized, eliminating manual intervention and improving production consistency and reliability. This method enables integrated mounting of ultra-thin chips with high mounting accuracy, enhances the system's robustness in complex production environments, and boasts high mounting efficiency. It solves the problems of high complexity, low mounting accuracy, and low production efficiency associated with traditional processes, making it suitable for various ultra-thin chip mounting scenarios.
[0055] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.
Claims
1. An integrated surface mount method for ultra-thin chips, characterized by, The method comprises the steps of: Collecting multi-spectral image data of the target chip surface during the mounting process and collecting three-dimensional point cloud data of the substrate; Extracting key alignment mark coordinates from the multi-spectral image data and extracting geometric reference point information from the three-dimensional point cloud data of the substrate; Constructing a global error compensation model of a multi-degree-of-freedom motion system according to the key alignment mark coordinates and the geometric reference point information, and calculating a vertical direction compensation amount, a first horizontal direction compensation amount, a second horizontal direction compensation amount and a rotation compensation amount during the chip mounting process to obtain a four-dimensional dynamic compensation vector; Taking the four-dimensional dynamic compensation vector as a target point of the mounting path, combining mounting accuracy and mounting path smoothness, and optimizing the mounting path of the ultra-thin chip to model a non-linear programming problem, and solving the non-linear programming problem to obtain a mounting path trajectory; Discretizing the mounting path trajectory into a series of execution action sequences containing position, speed and time, and sending the execution action sequences to a motion controller of a mounting head assembly for surface mounting of the ultra-thin chip.
2. The integrated surface mount method for ultra-thin chips according to claim 1, wherein, The method for extracting key alignment mark coordinates from the multi-spectral image data comprises the steps of: Adaptive filtering the multi-spectral image data to obtain a preprocessed image; Using a deep learning model to identify an alignment mark region of the preprocessed image to obtain an alignment mark; Converting the alignment mark into actual physical coordinates in a mounting system coordinate system through camera parameter calibration to obtain the key alignment mark coordinates.
3. The integrated surface mount method for ultra-thin chips of claim 1, wherein, The method for extracting geometric reference point information from the three-dimensional point cloud data of the substrate comprises the steps of: Segmenting the three-dimensional point cloud data of the substrate to obtain a geometric reference point region; Using a clustering algorithm to identify the geometric reference point region and calculating three-dimensional coordinates of feature points of the geometric reference point region to obtain the geometric reference point information.
4. The integrated surface mount method for ultra-thin chips of claim 1, wherein, The method for constructing a global error compensation model of a multi-degree-of-freedom motion system according to the key alignment mark coordinates and the geometric reference point information, and calculating a vertical direction compensation vector, a first horizontal direction compensation vector, a second horizontal direction compensation vector and a rotation compensation vector during the chip mounting process to obtain a four-dimensional dynamic compensation vector comprises the steps of: Matching the key alignment mark coordinates of the target chip with the geometric reference point information of the substrate to establish a corresponding relationship therebetween to obtain a matching result; Calculating an ideal transformation matrix of the target chip relative to the substrate according to the matching result: wherein, denotes an ideal transformation matrix, denotes the coordinate of the target chip key alignment mark after transformation, denotes the original coordinate of the target chip key alignment mark, denotes a rotation matrix, denotes a translation vector; Constructing a global error compensation model using the ideal transformation matrix: wherein, represents a global error, represents a transformed chip coordinate, represents a key alignment mark coordinate of a chip, represents a geometric reference point coordinate on a substrate, represents a regularization parameter, represents a rotation angle deviation, represents a weight coefficient, represents a number of mark points; Solving the global error compensation model using a singular value decomposition method to obtain an optimal rotation matrix and an optimal translation vector. a vertical direction compensation amount is obtained by the optimal translation vector a first horizontal direction compensation amount a second horizontal direction compensation amount a rotation compensation amount is obtained by the optimal rotation matrix the four-dimensional dynamic compensation vector is obtained .
5. The integrated surface mounting method for an ultra-thin chip according to claim 1, wherein: The objective function of the non-linear programming problem is: wherein represents an optimization target, and are weight coefficients, represents the deviation between the th mounting point and the four-dimensional dynamic compensation vector, represents the rate of change of the speed of the mounting path, represents time.
6. The integrated surface mount method for ultra-thin chips according to claim 5, wherein Solving the non-linear programming problem to obtain a mounting path trajectory comprises the steps of: Solving the objective function of the non-linear programming problem using an improved genetic algorithm to obtain a mounting path trajectory; wherein the selection probability formula of the improved genetic algorithm is: wherein, represents the probability that the i-th individual is selected, represents the fitness value, is the population size, is the decay factor. 7. The integrated surface mounting method for an ultra-thin chip according to claim 6, wherein: The mounting path trajectory is: wherein, is the total mounting time, is the position of the mounting head in the X-axis direction, is the position of the mounting head in the Y-axis direction, is the position of the mounting head in the Z-axis direction, denotes the angle of the mounting head around the Z-axis as a function of time.
8. The integrated surface mount method for ultra-thin chips according to claim 7, wherein, discretize the placement path trajectory into a series of execution action sequences containing position, velocity and time, comprising: said pick-and-place path trajectory is continuously sampled according to a fixed control period: , generating a discrete time sequence wherein, , , is the number of samples , is the control period; For each sampling time point , the corresponding target position is calculated: , a discrete position sequence is generated ; The velocity function is obtained by differentiating the mounting path trajectory: Then for each sampling time point Perform velocity sampling: Generate discrete velocity sequences .
9. An integrated surface mount system for ultra-thin chips, characterized by comprising: a data acquisition module, configured to acquire multi-spectral image data of a target chip surface in a placement process, and acquire three-dimensional point cloud data of a substrate; a feature extraction module, configured to extract key alignment mark coordinates from the multi-spectral image data, and extract geometric reference point information from the three-dimensional point cloud data of the substrate; a compensation acquisition module, configured to construct a global error compensation model of a multi-degree-of-freedom motion system according to the key alignment mark coordinates and the geometric reference point information, and calculate a vertical direction compensation amount, a first horizontal direction compensation amount, a second horizontal direction compensation amount and a rotation compensation amount in a chip placement process, to obtain a four-dimensional dynamic compensation vector; a placement path trajectory planning module, configured to use the four-dimensional dynamic compensation vector as a target point of a placement path, combine placement accuracy and placement path smoothness, model the placement path of an ultra-thin chip as a nonlinear programming problem, and solve the nonlinear programming problem to obtain a placement path trajectory; a path discretization module, configured to discretize the placement path trajectory into a series of execution action sequences containing position, velocity and time, and send the execution action sequences to a motion controller of a placement head assembly for surface placement of the ultra-thin chip.