A point gluing assembly method for high-temperature silicon pressure sensor

Abstract

The application discloses a point gluing assembly method of a high-temperature silicon pressure sensor, belongs to super-precision operation in the micro field, is applied to high-precision assembly of a pressure chip and a fixed tube seat in the high-temperature silicon pressure sensor, and particularly relates to camera calibration of a double-microscopic visual system, feature recognition of a point gluing hole of the pressure chip and a metal pin of the fixed tube seat and a high-precision alignment assembly strategy. Different from manual or semi-automatic alignment assembly adopted in the industry, the application calibrates the mapping relationship between an industrial area array camera and a world coordinate system and the mapping relationship between double-camera coordinate systems, carries out coordinate conversion on feature information collected by the microscopic vision, and realizes alignment and assembly of the pressure chip and the fixed tube seat of the high-temperature silicon pressure sensor by using the obtained coordinate information. The method avoids the fact that manual assembly error is large, nanosilver paste used in point gluing assembly overflows, and glue stress after solidification influences pressure measurement precision.

Description

Technical Field

[0001] This invention pertains to ultra-precision operations in the microscopic field and is applied to the high-precision assembly of pressure chips and fixed tubes in high-temperature silicon pressure sensors. Specifically, it involves camera calibration of a dual-microscopic vision system, feature identification of pressure chip dispensing holes and metal pins of the fixed tube, and high-precision alignment and assembly strategies. Background Technology

[0002] High-temperature silicon pressure sensors (hereinafter referred to as sensors) are one of the key core components of intelligent sensing systems. They have significant application value in high-precision fields such as aerospace, energy harvesting and industrial monitoring, automotive technology, and earthquake monitoring, where high accuracy of acceleration sensors is required. They have become an important development direction for micro-sensors.

[0003] It primarily utilizes the characteristic that the resistivity of single-crystal silicon material changes systematically under pressure or tension, and the Wheatstone bridge principle, to measure external pressure. Its pressure measurement principle is as follows: Figure 1 As shown, during the fabrication of the pressure chip, a sensitive diaphragm capable of sensing minute pressure changes is obtained through a wet etching process. Four varistors made of single-crystal silicon are etched onto the sensitive diaphragm. These four varistors are distributed in the stress concentration areas of the sensitive diaphragm and form a Wheatstone bridge structure via metal leads. When the sensitive diaphragm is not subjected to pressure or tension, the four varistors have equal resistances, and the Wheatstone bridge is in a balanced state, as shown in the figure (Vout). - pins and Vout + The potential difference between the pins is 0. When the sensitive diaphragm is subjected to pressure or tension of a certain magnitude and direction, the resistance of the varistor will change, and the resistance change will vary in different directions. At this time, the Wheatstone bridge is in an unbalanced state, as shown in the diagram (Vout). - pins and Vout + A certain amount of electromotive force will be output between the pins. The magnitude of the external pressure can be measured by measuring the magnitude of the output electromotive force.

[0004] At present, the MEMS high-temperature silicon pressure sensor based on silicon-on-insulator technology (SOI) adopts a SiO2 insulation layer to isolate the silicon substrate and the pressure measuring circuit of the sensor chip, which can effectively solve the problem of reverse leakage of P-N junction isolation used in traditional diffusion silicon pressure sensors in high-temperature environment, thereby causing the sensor to fail, and greatly improves the high-temperature resistance of the pressure sensor. However, with the increase of working temperature and the maturation of SOI technology, the main factor restricting the high-temperature resistance of the sensor has gradually become the packaging process used. For this sensor, a leadless packaging process is initially proposed, which can reduce the size of the sensor and avoid the influence of the flash point of the insulating silicone oil on the dynamic response capability and high-temperature resistance of the sensor. In the leadless packaging process, the pressure chip and the fixed tube seat are still generally assembled by manual or semi-automatic dispensing assembly. This assembly method has large assembly error and is easy to cause the overflow of nano silver paste during assembly and the generation of glue stress due to solidification, thereby affecting the pressure measuring accuracy of the sensor. Therefore, it is of great value and necessity to develop a dispensing assembly method for the pressure chip and the fixed tube seat of the high-temperature silicon pressure sensor in the high-end field of our country. SUMMARY

[0005] The purpose of the present application is to provide a high-temperature silicon pressure sensor assembly visual positioning, high-precision recognition and alignment assembly method for pressure chip dispensing hole and fixed tube seat metal pin, which solves the problems of low precision and low efficiency in manual assembly. The purpose of the present application is achieved by the following technical solutions.

[0006] A dispensing assembly method for a high-temperature silicon pressure sensor, comprising:

[0007] A calibration plate is used to calibrate the first mapping relationship between the camera coordinate system and the world coordinate system, the second mapping relationship between the upper and lower camera coordinate systems in the double microscopic vision system, and a template library of dispensing holes on the pressure chip is created;

[0008] A pressure chip is placed at the bottom of the calibration plate, an upper camera acquires an image of the pressure chip and extracts pixel coordinates of four assembly sites in the pressure chip, and the world coordinates of the four assembly sites in the pressure chip are calculated through the first mapping relationship;

[0009] According to the world coordinates of the four assembly sites in the pressure chip, nano silver paste is dispensed into the dispensing holes corresponding to each assembly site;

[0010] A pneumatic gripper carries a fixed tube seat to move directly above a lower camera, acquires an image of the fixed tube seat and extracts pixel coordinates of four assembly sites of the fixed tube seat, calculates the pixel coordinates of the four assembly sites of the fixed tube seat in the upper camera coordinate system through the second mapping relationship, and calculates the world coordinates corresponding to the four assembly sites of the fixed tube seat;

[0011] The four assembly sites of the pressure chip and the four assembly sites of the fixed tube seat are aligned and assembled, and the pressure chip and the fixed tube seat are connected through high-temperature sintering.

[0012] Further, the calibration plate is a planar calibration plate, the calibration plate has 27*27 array of calibration points with a diameter of 0.25 mm, the calibration plate has an outer dimension of 150*150 mm, a thickness of 0.5 mm, a machining precision of the calibration points of 1 um, and a collinearity precision of 1 um.

[0013] Further, the planar calibration plate is a film sheet material, and the thickness is less than the overlapping thickness of the depth of field of the upper and lower cameras.

[0014] Further, the calibration includes angle calibration and distance calibration.

[0015] Further, the angle calibration process is as follows:

[0016] The calibration point images collected by the upper and lower cameras are processed to extract the center coordinates, and the pixel coordinates of the centers of the three circle points in the pixel coordinate system of the upper camera are (x1, y1), (x2, y2), and (x3, y3) in sequence, and the center coordinates of the circle points in the lower camera are (x4, y4), (x5, y5), and (x6, y6); first, the average values of the horizontal coordinates and the vertical coordinates of the points are calculated and

[0017]

[0018] The parameter n represents the number of each group of calibration points selected, X1, Y1, X2, Y2, X3, and Y3 correspond to the above coordinates (x1, y1), (x2, y2), and (x3, y3), which are the coordinates of the same point in the pixel coordinate system of the upper and lower cameras, respectively.

[0019] Then, the coefficient L is calculated XX , L YY , and L XY :

[0020]

[0021] The slope of the regression straight line is:

[0022] Similarly, the slope k2 of the straight line common to the circle center points in the lower camera can be obtained; and the rotation angle of the pixel coordinate system of the upper camera and the pixel coordinate system of the lower camera is equal to:

[0023]

[0024] Further, the distance calibration process is as follows:

[0025] The pixel distance D between the two measurement groups on the calibration board is as shown in equation (6):

[0026] D = d / K P (6)

[0027] In the formula, d is the actual physical distance between the two measurement groups, and K p The pixel ratio coefficient is denoted by ; the coordinates of the center points of the circles in the upper and lower camera pixel coordinate systems are established by geometric relationships using formula (7), and then the analytical expressions for A and B are derived as shown in formula (8):

[0028]

[0029]

[0030] A and B are the distance between the origins of the two coordinate systems, λ is the scaling factor, X5, Y5 and x5, y5 are the coordinates of the same point in the upper and lower camera pixel coordinate systems respectively, and β is the angle between the second set of calibration points in the lower camera's field of view and the pixel coordinate system.

[0031] Furthermore, the feature recognition algorithm used for the pressure chip during the alignment and assembly process is as follows:

[0032] Image preprocessing, including image grayscale conversion and smoothing / denoising;

[0033] ROI region extraction is performed based on the region area and its row and column.

[0034] Perform rectangle fitting on the chip border to obtain the coordinates of the chip center and the four corner points of the border, as well as the rotation angle;

[0035] Import the template library, compare it within the chip border area, search for dispensing holes with a similarity greater than a preset threshold to the images in the template library, and output their location information.

[0036] Furthermore, the feature identification performed on the pressure chip further includes feature verification:

[0037] When only one dispensing hole is identified, the equation of the straight line of the chip frame is calculated by using the coordinates of the four corner points of the chip frame. Based on the distance between the axes of the dispensing holes, the identified dispensing holes are offset by this distance along the normal direction of the straight line of the frame to obtain the coordinates of the other three dispensing holes.

[0038] When two dispensing holes are identified, the equation of the straight line of the chip frame is calculated using the coordinates of the four corner points of the chip frame. Based on the distance between the axes of the dispensing holes, the identified dispensing holes are offset by this distance along the normal direction of the straight line of the frame to obtain the coordinates of the other two dispensing holes.

[0039] When three dispensing holes are identified, the intersection of the two normal lines obtained by connecting the centers of two adjacent dispensing holes is the coordinate of the other dispensing hole.

[0040] If no dispensing hole is detected, the dispensing hole image is extracted and added to the template library, and then it is detected again.

[0041] Furthermore, the feature recognition algorithm used for the fixed tube seat during the alignment and assembly process is as follows:

[0042] Image preprocessing: Channel decomposition and filtering are performed sequentially on the fixed tube image; the three-channel image is decomposed into RGB, the image in R space is taken and mean filtering is performed;

[0043] Feature extraction: Threshold segmentation, region connectivity, and region filtering are performed sequentially on the fixed tube socket image. Based on the elements of the metal pin in the image, the metal pin is extracted.

[0044] Furthermore, the feature identification performed on the fixed tube seat further includes:

[0045] Assembly success rate prediction: After extracting the metal pins, connect the center points of each metal pin in sequence and calculate the length of the center connection line. If the variance of the length of each center connection line is greater than the preset threshold, it indicates that the clamping of the fixing tube seat is tilted. Correct the position of the fixing tube seat and then perform the assembly operation.

[0046] The advantages of this invention are as follows: Unlike the manual or semi-automatic alignment and assembly methods used in the industry, this invention performs coordinate transformation on the feature information acquired by microscopic vision by calibrating the mapping relationship between the industrial area array camera and the world coordinate system, as well as the mapping relationship between the dual-camera coordinate systems. Using the obtained coordinate information, the alignment and assembly of the high-temperature silicon pressure sensor pressure chip and the fixed tube base are achieved. This method avoids the large errors in manual assembly that lead to overflow of the nano-silver paste used in dispensing assembly, and the resulting adhesive stress after curing, which affects the pressure measurement accuracy. Attached Figure Description

[0047] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0048] Figure 1 This is a circuit diagram of a Wheatstone bridge in the existing technology.

[0049] Figure 2 This is a schematic diagram of the nine-point calibration method used in this invention.

[0050] Figure 3 This is a schematic diagram illustrating the coordinate mapping relationship between the dual visual coordinate systems of the present invention.

[0051] Figure 4 This is a schematic diagram of the Mark area captured by the upper and lower cameras of this invention.

[0052] Figure 5 This is a schematic diagram showing the correct placement of the calibration plate of this invention.

[0053] Figure 6 This is a schematic diagram of the pressure chip and fixed tube base structure of the present invention.

[0054] Figure 7 This is a flowchart illustrating the assembly process of the pressure chip and the fixed tube seat of the present invention.

[0055] Figure 8 This is a schematic diagram of the pressure chip dispensing hole template library of the present invention.

[0056] Figure 9 This is a diagram of the pressure chip under the camera's field of view in this invention.

[0057] Figure 10 This is a diagram showing the ROI region extraction effect of the present invention.

[0058] Figure 11 This is a diagram showing the fitting effect of the chip border in this invention.

[0059] Figure 12 This is a schematic diagram of the target features searched on the pressure chip of the present invention.

[0060] Figure 13 This is a schematic diagram illustrating the incomplete target feature recognition of the present invention.

[0061] Figure 14 This is a schematic diagram illustrating the solution for incomplete target feature recognition in this invention.

[0062] Figure 15 This is a schematic diagram of the grayscale conversion (left) of the three-channel image and the R-space image obtained after decomposition according to the present invention.

[0063] Figure 16 This is a schematic diagram of the target features extracted from the fixed tube seat of the present invention.

[0064] Figure 17 This is a schematic diagram illustrating the determination of whether the clamping of the fixed tube seat in this invention exhibits tilting.

[0065] Figure 18 This is a schematic diagram showing the correspondence between the pressure chip and the assembly points on the fixed tube seat of the present invention. Detailed Implementation

[0066] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0067] This invention employs a nine-point calibration method to calibrate the mapping relationship between the camera coordinate system and the world coordinate system. A high-precision planar calibration board is designed and used to obtain the mapping relationship between the upper and lower camera coordinate systems in a dual-microscopic vision system. This invention develops a robust identification scheme for the dispensing holes and metal pin features of the fixing socket in the pressure chip of the high-temperature silicon pressure sensor, and provides a strategy for predicting assembly success rate. Based on grayscale image threshold segmentation, the ROI region is quickly extracted, and image processing algorithms such as line fitting and template matching are used to obtain the coordinates of the assembly site.

[0068] The technical solution of this invention will be developed from the following three aspects:

[0069] 1. Camera calibration

[0070] This invention employs a dual-vision microscopy scheme, requiring sequential calibration of the intrinsic and extrinsic parameters of monocular vision and the coordinate mapping relationship between the dual-vision coordinate systems. Currently available industrial cameras exhibit relatively low distortion, and since this invention is used to acquire two-dimensional planar images, no additional calibration of camera distortion parameters is necessary.

[0071] The calibration of intrinsic and extrinsic parameters in monocular vision is essentially to obtain the transformation relationship from the pixel coordinate system to the spatial coordinate system. Industrially, a nine-point calibration method is generally used. The specific calibration principle is as follows:

[0072] t x t y These represent the distances offset from the origin of the world coordinate system relative to the origin of the camera pixel coordinate system in the u and v directions, respectively. For example, a point [u, v, 1] in the pixel coordinate system corresponds to coordinates [x, y, 1] in the world coordinate system. Given the offset t between the two coordinate systems... x t y Then we have u = x + t x v = y + t y In matrix form, it is:

[0073]

[0074] s x s yThese represent the scaling relationship between the world coordinate system and the camera pixel coordinate system, respectively. For example, a point [u, v, 1] in the pixel coordinate system corresponds to [x, y, 1] in the world coordinate system. Given the scaling relationship s between the two coordinate systems... x s y Then we have u = s x x, v = s y y, written in matrix form, is:

[0075]

[0076] The '°' represents the rotation angle of the camera pixel coordinate system relative to the world coordinate system. For example, the polar coordinates of a point [x, y, 1] in the world coordinate system are... Given the rotation angle between the two coordinate systems, the polar coordinates of the point in the pixel coordinate system corresponding to the coordinates [u, v, 1] are:

[0077] Therefore, we have: u = x cosθ - ysinθ, v = x sinθ + y cosθ, which can be written in matrix form as:

[0078]

[0079] β = Theta represents the shear angle between the camera pixel coordinate system and the world coordinate system. For example, given a point [u, v, 1] in the camera pixel coordinate system, and the shear angle θ between the two coordinate systems, the corresponding coordinates of this point in the world coordinate system are [x, y, 1]. Then we have u = x - ysinβ and v = y cosβ. This can be written in matrix form as follows:

[0080]

[0081] Thus, through the above derivation, for a point C1(u, v) in the existing pixel coordinate system, the coordinates of the corresponding point C2 in the spatial coordinate system are C2(x, y). The relationship between C1 and C2 is only translation, scaling, rotation, and shearing, and the corresponding matrix formulas are as follows:

[0082]

[0083] By multiplying the four matrices above into an affine transformation matrix, the transformation from the pixel coordinate system to the spatial coordinate system can be achieved.

[0084]

[0085] The last matrix is ​​obtained by multiplying the first four matrices, resulting in six parameters: a1, b1, c1, a2, b2, and c2. To obtain these six parameters, at least six different equations are needed. Since each point contains (x, y) coordinates, each point can generate two different equations. Therefore, we only need three points to solve for the final affine transformation matrix. To improve accuracy, we can use nine points to obtain N combinations of results. Based on these results, we use a method similar to averaging to further improve accuracy. Figure 2 As shown.

[0086] Since the upper and lower cameras are arranged independently, it is also necessary to calibrate the coordinate mapping relationship between the two vision coordinate systems. The auxiliary tool used for this calibration is the high-precision planar calibration plate designed in this invention. The calibration plate has a 27×27 array of calibration points with a diameter of 0.25mm. The calibration plate has external dimensions of 150×150mm, a thickness of 0.5mm, a calibration point machining accuracy of 1μm, and a collinearity accuracy of 1μm. (See below) Figure 3 As shown, (a) the pixel coordinate mapping relationship between the upper and lower cameras; (b) the high-precision planar calibration board.

[0087] The coordinate mapping relationship between the two sub-camera coordinate systems mainly involves the calibration of four parameters: the distance A and B between the origins of the two coordinate systems; the angle between the two coordinate systems; and the coordinate mapping relationship between the two sub-camera coordinate systems. Scaling factor λ. The coordinates of a point in the upper camera pixel coordinate system can be transformed to the lower camera coordinate system using formula (1). In the formula, x1 and y1 are the coordinates in the upper camera pixel coordinate system, and x2 and y2 are the coordinates in the lower camera pixel coordinate system.

[0088]

[0089] The high-precision planar calibration plate used is made of film material, with a thickness less than the overlap thickness of the depth of field of the upper and lower cameras, allowing the upper and lower cameras to simultaneously capture the calibration points. Considering assembly accuracy, camera field of view, and the distance between the optical axes of the upper and lower cameras, the size of the calibration plate and the calibration points on it was designed so that the upper left and lower right Mark areas of the calibration plate can simultaneously appear within the field of view of the upper and lower cameras, respectively, as shown in the following diagram. Figure 4 As shown. Each Mark region contains nine calibration points in a 3×3 grid. The distances between the calibration points are known. By extracting the coordinates of the center of the outermost calibration point in the Mark region and fitting a rectangle, it can be used to determine whether the calibration plate is placed flat, thereby evaluating the reliability of the calibration parameters. Figure 5 As shown. The left one is incorrectly placed, while the right one is correctly placed.

[0090] The calibration points have a machining accuracy of 1 μm and a collinearity accuracy of 1 μm, ensuring the parallelism of the lines connecting the calibration points. The middle column of calibration points in the Mark area captured by the upper and lower cameras is taken as the measurement group, and the distance d between the two groups is known. The specific calibration principle includes angle calibration and distance calibration. First, angle calibration is performed by processing the calibration point images acquired by the upper and lower cameras to extract the center coordinates. Let the pixel coordinates of the three center points in the upper camera's pixel coordinate system be (x1, y1), (x2, y2), and (x3, y3), respectively, and the center coordinates in the lower camera be (x4, y4), (x5, y5), and (x6, y6). The straight lines formed by any two groups of calibration points on the planar calibration plate designed in this invention are parallel to each other. Therefore, ideally, the slopes of the two sets of straight lines obtained from image processing should also be equal. Thus, the difference in the actual slopes of the two sets of straight lines can be set as the difference in angles between the two sub-camera coordinate systems. Due to manufacturing errors and circle fitting errors, the three center points are not perfectly collinear. A univariate linear regression equation is established to determine the straight line containing the three points, and the slope of the regression line is calculated. Taking the three points in the camera image above as an example, the mean of the x-coordinate and y-coordinate of the points is first calculated. and

[0091]

[0092] The parameter n represents the number of calibration points selected in each group, which is 3 in this invention. The uppercase X1, X2, X3, Y1, Y2, Y3 all correspond to the coordinates (x1, y1), (x2, y2), and (x3, y3) above, which are the coordinates of the same point in the upper and lower camera pixel coordinate systems, respectively.

[0093] Recalculate coefficient L XX L YY L XY :

[0094]

[0095] The slope of the regression line is:

[0096] Similarly, the slope k2 of the line collinear with the center point of the lower camera can be calculated; then the rotation angle between the upper camera pixel coordinate system and the lower camera pixel coordinate system can be determined. equal:

[0097]

[0098] The distance calibration method is as follows: According to the characteristics of the planar calibration plate, no matter how the pixel coordinate systems of the two cameras are rotated or translated, the pixel distance D between the two measurement groups on the calibration plate remains unchanged, as shown in equation (6):

[0099] D = d / K P (6)

[0100] In the formula, d is the actual physical distance between the two measurement groups, and K p The pixel ratio coefficient is denoted by . The coordinates of the center point of the circle in the upper and lower camera pixel coordinate systems can be established by geometric relationship formula (7), and then the analytical expressions of A and B are derived as shown in formula (8).

[0101]

[0102]

[0103] X5, Y5 and x5, y5 are the coordinates of the same point in the upper and lower camera pixel coordinate systems, respectively. β is the angle between the second set of calibration points in the lower camera's field of view and the pixel coordinate system.

[0104] 2. Identification of pressure chip and fixed tube seat features

[0105] The pressure chip and fixed tube socket structure assembled this time are as follows: Figure 6 As shown.

[0106] First, feature identification is required for the four dispensing holes on the pressure chip and the four metal pins on the mounting base. After dispensing nano-silver paste into the glass holes, alignment and assembly are performed. Through high-temperature sintering, the pressure chip and the mounting base are firmly connected together. The assembly process is as follows: Figure 7 As shown.

[0107] The feature recognition algorithm process for the pressure chip and the fixing socket during assembly is as follows:

[0108] ① Pressure chip

[0109] a. Creating a template library; Due to the complex operating conditions of camera-captured pressure chips, this invention proposes to use a shape-based template matching method for dispensing hole identification. This method uses edge feature localization and is insensitive to many interference factors, such as changes in lighting and image grayscale. It can even support models with missing edges, cluttered scenes, noise, defocus, and slight deformation. However, during the matching process, large rotation or scaling of the target image will affect the search results. In this invention, the pressure chip is fixed using a tooling groove, and the target does not undergo significant rotation or scaling. Figure 8 It is a template library built based on images of dispensing holes captured under different working conditions.

[0110] b. Image preprocessing; including image grayscale conversion, smoothing and denoising, etc.; by Figure 9 It is known that the pressure chip has a lot of noise due to factors such as illumination and surface condition, and this noise will interfere with subsequent image processing.

[0111] c. ROI Region Extraction: To quickly locate the approximate region where the desired features are located, reduce the recognition burden, and avoid interference from unknown features in other regions, it is necessary to quickly extract the ROI region. This invention extracts the ROI region using factors such as region area and row / column location. The extraction results are as follows: Figure 10 As shown.

[0112] d. Chip border fitting: There is a gap between the pressure chip and the tooling groove. The actual placement of the pressure chip may be offset. It is necessary to perform rectangular fitting on the chip border to obtain information such as the coordinates of the chip center and four corner points, rotation angle, etc. Figure 11 This is a diagram showing the completed chip border fitting.

[0113] e. Locate the target feature (dispensing hole); import the template library, compare it within the chip border area, search for target features highly similar to the images in the template library, and output their location information. Figure 12 It is a target feature found in a chip image.

[0114] f. Feature verification; Due to operating conditions, there are instances where the identification of four dispensing holes is incomplete, such as... Figure 13 As shown. When only one dispensing hole is identified, the equation of the straight line containing the chip frame is calculated using the chip frame corner coordinates obtained in step d above. Since the pressure chip manufacturing precision is high, the distance between the centers of the dispensing holes is uniform and known. Using this distance, the identified dispensing holes are offset along the normal direction of the straight line containing the frame by this distance, thus obtaining the coordinates of the other three dispensing holes. When two dispensing holes are identified, the coordinates of the other two dispensing holes can be obtained in the same way. When three dispensing holes are identified, the coordinates of the other dispensing hole are obtained by connecting the centers of two adjacent dispensing holes and drawing the normal to the resulting straight line. The intersection of the two normals is the coordinate of the other dispensing hole. When no dispensing holes are identified, it indicates a worse working condition. The dispensing hole image under this condition is extracted and added to the template library to improve the stability of template matching and recognition, and then identification is performed again. The processing is as follows: Figure 14 As shown.

[0115] ②Fixed pipe seat

[0116] a. Image preprocessing: Channel decomposition and filtering are performed sequentially on the image of the fixed tube socket. Figure 15 It can be seen that after grayscale conversion, the contrast between the metal pins and surrounding pixels in the three-channel image of the fixed tube socket is small, and there is a low grayscale band on the metal pins, making it difficult to segment the target feature (metal pins) from the grayscale image. Decomposing the three-channel image into RGB, taking the image in R space, and applying mean filtering can significantly improve the contrast between the metal pins and surrounding pixels, and significantly improve the low grayscale band.

[0117] b. Feature extraction: By sequentially performing thresholding, region connectivity, and region filtering on the fixed tube socket image, the target features (metal pins) can be extracted, such as... Figure 16 As shown. Based on the unique features of the metal pins in the image, such as area and roundness, they can be quickly extracted.

[0118] c. Assembly Success Rate Prediction: This invention employs a high-alignment pneumatic gripper to achieve the position transfer and assembly of the fixed tube seat. However, errors in the gripper's mechanical structure can affect assembly accuracy; therefore, success rate prediction is necessary before assembly. The specific implementation method is as follows: After extracting the features of the metal pins, connect each center point sequentially and calculate the length of the center connection line. If the deviation of each center connection line length is large, i.e., the variance is large, it indicates that the clamping of the fixed tube seat has become tilted, and the position of the fixed tube seat needs to be corrected immediately before proceeding with the assembly operation. Figure 17 As shown.

[0119] 3. High-precision alignment and assembly strategy

[0120] After identifying the dispensing holes on the pressure chip and the metal pins on the mounting base, a pneumatic gripper rotates around its axis to adjust the posture of the mounting base. During adjustment, the assembly success rate is continuously predicted and calculated to prevent the mounting base from becoming loose during rotation. Once the position is correct, each axis works in tandem according to the corresponding assembly point number to achieve high-precision assembly of the pressure chip and the mounting base. Figure 18 As shown. Detailed implementation:

[0121] 1. Define the mapping relationship R1 between the camera coordinate system and the world coordinate system, and the mapping relationship R2 between the upper and lower camera coordinate systems in the dual-microscopic vision system. Create a template library for dispensing holes on the pressure chip.

[0122] 2. Move the camera directly above the pressure chip to acquire its image and extract the pixel coordinates (u) of assembly sites 1, 2, 3, and 4. 10 v 10 ), (u 20 v 20 ), (u 30 v 30 ), (u 40 v 40 The world coordinates (x, y) of the corresponding assembly point are calculated using the mapping relationship R1. 10 y 10 ), (x 20 y 20 ), (x 30 y 30 ), (x 40 y 40 ).

[0123] 3. Based on the world coordinates of assembly sites 1, 2, 3, and 4, apply nano-silver paste into the dispensing holes corresponding to each assembly site.

[0124] 4. The pneumatic gripper carries the fixed tube seat to a position directly above the lower camera, acquires its image, and extracts the pixel coordinates (u) of assembly points 1, 2, 3, and 4. 11 v 11 ), (u 21 v 21 ), (u 31 v 31 ), (u 41 v 41 Using the mapping relationship R2, the corresponding pixel coordinates (u) in the upper camera pixel coordinate system are calculated. 12 v 12 ), (u 22 v 22 ), (u 32 v 32 ), (u 42 v 42 ), thereby calculating the corresponding world coordinates (x) 11 y 11 ), (x 21 y 21 ), (x 31 y 31 ), (x 41 y 41 ).

[0125] 5. Based on the corresponding information of the assembly point, adjust the posture of the fixing tube seat and place the fixing tube seat on the bottom pressure chip to realize the assembly of the high temperature silicon pressure sensor.

[0126] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A dispensing assembly method of high-temperature silicon pressure sensor, characterized in that, a first mapping relationship between a camera coordinate system and a world coordinate system is calibrated using a calibration board, a second mapping relationship between upper and lower camera coordinate systems in a double-microscopic vision system, and a template library of dispensing holes on a pressure chip is created; a pressure chip is placed on the bottom of the calibration board, an image of the pressure chip is acquired by an upper camera, and pixel coordinates of four assembly sites in the pressure chip are extracted, and world coordinates of the four assembly sites in the pressure chip are calculated through the first mapping relationship; nanosilver paste is dispensed into the dispensing holes corresponding to the four assembly sites according to the world coordinates of the four assembly sites in the pressure chip; a pneumatic gripper carries a fixed tube seat to move directly above the lower camera, an image of the fixed tube seat is acquired, and pixel coordinates of four assembly sites of the fixed tube seat are extracted, and the pixel coordinates of the four assembly sites of the fixed tube seat in the upper camera coordinate system are calculated through the second mapping relationship, so as to calculate the world coordinates corresponding to the four assembly sites of the fixed tube seat; the four assembly sites of the pressure chip and the four assembly sites of the fixed tube seat are aligned and assembled, and the pressure chip and the fixed tube seat are connected through high-temperature sintering. 2.The dispensing assembly method of high-temperature silicon pressure sensor according to claim 1, characterized in that, the calibration board is a planar calibration board, and the calibration board has 27*27 array of calibration points with a diameter of 0.25 mm, an outer dimension of 150*150 mm, a thickness of 0.5 mm, a machining precision of 1 μm, and a collinearity precision of 1 μm. 3.The dispensing assembly method of high-temperature silicon pressure sensor according to claim 1 or 2, characterized in that, the calibration board is a film sheet material with a thickness less than the depth of field overlap of the upper and lower cameras. 4.The dispensing assembly method of high-temperature silicon pressure sensor according to claim 1, characterized in that, the calibration includes angle calibration and distance calibration. 5.The dispensing assembly method of high-temperature silicon pressure sensor according to claim 4, characterized in that, the angle calibration process is as follows: the calibration point images collected by the upper and lower cameras are processed, and the center coordinates are extracted, and the pixel coordinates of the centers of three circles in the pixel coordinate system of the upper camera are (x1, y1), (x2, y2), and (x3, y3) in turn, the parameter n represents the number of each group of calibration points selected, X1, Y1, X2, Y2, X3, and Y3 correspond to the above coordinates (x1, y1), (x2, y2), and (x3, y3), and are the coordinates of the same point in the pixel coordinate systems of the upper and lower cameras respectively. The center coordinates of the camera are (x4, y4), (x5, y5), (x6, y6); first calculate the average of the horizontal and vertical coordinates of the points and 6.The dispensing assembly method of high-temperature silicon pressure sensor according to claim 5, characterized in that, Recalculating the coefficient L XX , L YY , L XY : The slope of the regression line is then: Similarly, the slope k2 of the straight line on which the center points of the circles in the lower camera are collinear can be obtained; the rotation angle of the upper camera pixel coordinate system and the lower camera pixel coordinate system is equal to: is equal to: the distance calibration process is as follows: the pixel distance D between the two measurement groups on the calibration board is as formula (6): A and B are the distance between the origins of the two coordinate systems, λ is the scaling factor, X5, Y5, x5, and y5 are the coordinates of the same point in the pixel coordinate systems of the upper and lower cameras respectively, and β is the angle between the second group of calibration points and the pixel coordinate system in the field of view of the lower camera. D = d / K P (6) where d is the actual physical distance between the two measurement groups, K p is the camera pixel scale factor; the coordinates of the circle center point in the upper camera and lower camera pixel coordinate systems are established by the geometric relationship formula (7), and the analytical expressions of A and B are derived as formula (8): ​ 7. The dispensing assembly method of a high-temperature silicon pressure sensor according to claim 1, characterized in that, in the process of alignment assembly, the feature recognition algorithm adopted for the pressure chip is as follows: image preprocessing, including image graying and smoothing denoising; ROI region extraction through area and row and column; rectangular fitting of the chip frame to obtain the chip center and four frame corner coordinates and the rotation angle; importing the template library, searching for dispensing holes with a similarity greater than a preset threshold to the images in the template library in the chip frame region, and outputting the position information thereof.

8. The dispensing assembly method of a high-temperature silicon pressure sensor according to claim 7, characterized in that, the feature recognition adopted for the pressure chip further includes feature verification: when only one dispensing hole is identified, the straight line equation of the straight line on which the chip frame is located is calculated through the four frame corner coordinates of the chip, the identified dispensing hole is offset by the distance between the axes of the dispensing holes in the normal direction of the straight line on which the chip frame is located, and the coordinates of the other three dispensing holes are calculated; when two dispensing holes are identified, the straight line equation of the straight line on which the chip frame is located is calculated through the four frame corner coordinates of the chip, the identified dispensing hole is offset by the distance between the axes of the dispensing holes in the normal direction of the straight line on which the chip frame is located, and the coordinates of the other two dispensing holes are calculated; when three dispensing holes are identified, the intersection of the normals of the straight lines connecting the centers of two adjacent dispensing holes is the coordinate of the other dispensing hole; when no dispensing hole is identified, the dispensing hole image is taken out and added to the template library, and then identification is performed again.

9. The dispensing assembly method of a high-temperature silicon pressure sensor according to claim 1, characterized in that, in the process of alignment assembly, the feature recognition algorithm adopted for the fixed tube seat is as follows: image preprocessing: sequentially performing channel decomposition and filtering on the fixed tube seat image; performing RGB decomposition on the three-channel image, taking the image in the R space, and performing mean filtering; feature extraction: sequentially performing threshold segmentation, region connection, and region screening on the fixed tube seat image, and extracting the metal pins according to the elements of the metal pins in the image.

10. The dispensing assembly method of a high-temperature silicon pressure sensor according to claim 9, characterized in that, the feature recognition adopted for the fixed tube seat further includes: assembly success rate prediction: after the metal pins are extracted, the center points of the metal pins are sequentially connected, and the center line length is calculated; if the variance of the center line lengths is greater than a preset threshold, it indicates that the clamping of the fixed tube seat is inclined, the position of the fixed tube seat is corrected, and then the assembly operation is performed.

Images