PCBA welding process optimization system based on deep learning

Abstract

The invention discloses a PCBA welding process optimization system based on deep learning, and belongs to the technical field of printed circuit board assembly. According to the method, the number of the pixel points in the local welding spot depth image, the pixel difference value between the pixel points and the pixel points at the positions corresponding to the local welding spot depth image template exceeds the set threshold value, is used as the stripping area size feature information of each welding spot, the copper foil stripping degree near each welding spot can be accurately represented, and the stripping degree can be used as an accurate basis for follow-up parameter optimization; according to the method, the boundary points are ingeniously recognized through the welding spot area in the obtained welding spot depth image, and then the sunken area boundary is formed, so that the size and depth of the sunken area can be accurately represented and can be used as an accurate basis for follow-up parameter optimization; the welding temperature, the welding time and the welding pressure can be accurately adjusted based on the stripping area size feature information and the welding spot depression feature information, and optimization of welding process parameters is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of printed circuit board assembly, and in particular to a PCBA welding process optimization system based on deep learning. Background Art

[0002] PCB, also known as printed circuit board or printed circuit board, is an important electronic component, a support for electronic components, and a carrier for electrical connection of electronic components. It acts as a channel for current transmission and signal transmission, and is a key electronic interconnection part in electronic products. Printed circuit boards are widely used in military, communications, medical, electric power, automobiles, industrial control, smart phones, wearables and other high-tech fields. With the development of science and technology, the application scope of printed circuit boards will continue to expand.

[0003] PCBA stands for Printed Circuit Board Assembly. It is based on PCB (Printed Circuit Board). Through surface mount technology (SMT) and plug-in technology (DIP) and other processes, electronic components (such as resistors, capacitors, inductors, chips, etc.) are installed on PCB, and a series of processes such as welding and testing are carried out to finally form an electronic component with specific functions.

[0004] During the welding process, it is very likely that the welding point will be incomplete and the PCB copper foil will peel off. Here, the incomplete welding point mainly refers to the depression in the middle area of ​​the welding point surface. The PCB copper foil peeling is due to the insufficient adhesion between the copper foil and the PCB substrate, which causes the copper foil to be partially or completely separated from the substrate.

[0005] How to optimize the welding process parameters to avoid the above phenomenon is an urgent problem to be solved. To this end, a PCBA welding process optimization system based on deep learning is proposed. Summary of the invention

[0006] The technical problem to be solved by the present invention is: how to better optimize the welding process parameters to avoid the occurrence of incomplete welding points and PCB copper foil peeling, and provide a PCBA welding process optimization system based on deep learning.

[0007] The present invention solves the above technical problems through the following technical solutions, which include an initial parameter setting module, a copper foil peeling detection module, a solder joint sag detection module and a parameter optimization module; The initial parameter setting module is used to initialize the welding temperature, welding time and welding pressure during welding to obtain the initial maximum allowable welding temperature, the initial maximum allowable welding time and the initial maximum allowable welding pressure; The copper foil peeling detection module is used to detect the copper foil peeling phenomenon after each welding is completed, and obtain the size characteristic information of the peeling area; The solder joint depression detection module is used to detect the solder joint depression phenomenon after each welding is completed to obtain solder joint depression feature information; The parameter optimization module is used to adjust the welding temperature, welding time and welding pressure respectively based on the initial maximum allowable welding temperature, initial maximum allowable welding time and initial maximum allowable welding pressure according to the stripping area size feature information and the solder joint depression feature information, so as to optimize the welding process parameters.

[0008] Furthermore, the copper foil peeling detection module includes an image acquisition unit, a solder joint recognition unit and a peeling area size feature information acquisition unit; the image acquisition unit is used to use the RGB-D depth camera installed on the welding head of the welding equipment to shoot the solder joint below after each welding, obtain the solder joint depth image and the corresponding solder joint RGB image, and perform noise reduction processing on the solder joint depth image and the solder joint RGB image; the solder joint recognition unit is used to use the trained solder joint target detection model to recognize the solder joint in the solder joint RGB image, and perform outer contour detection processing to obtain the solder joint area position information; the peeling area size feature information acquisition unit is used to process according to the solder joint area position information and the solder joint depth image template to obtain the peeling area size feature information F i , where i represents the i-th detection.

[0009] Furthermore, the specific processing process of the solder joint identification unit is as follows: Step S11: using the target detection model to identify the solder joints in the solder joint RGB image, and obtaining the coordinates of the upper left corner point and the lower right corner point of the solder joint detection frame in the RGB image; Step S12: Use the contour detection function in OpenCV to perform outer contour detection on the inside of the solder joint detection frame area, obtain the outer contour line of the solder joint and the coordinates of each point on the outer contour line in the solder joint RGB image, that is, obtain the position information of the solder joint area.

[0010] Furthermore, the specific processing process of the stripping area size feature information acquisition unit is as follows: Step S21: Obtaining the solder joint area in the solder joint depth image according to the solder joint outer contour line and the coordinates of each point on the outer contour line in the solder joint RGB image; Step S22: removing the solder joint area in the solder joint depth image from the solder joint depth image and the solder joint depth image template, and correspondingly obtaining a local solder joint depth image and a local solder joint depth image template, wherein the solder joint depth image template is obtained by photographing at each solder joint position by using an RGB-D depth camera before welding, and the photographing position is the same as the photographing position after welding. Step S23: Read the pixel values ​​of the pixels at the same position in the local solder joint depth image and the local solder joint depth image template respectively, and calculate the pixel difference between the pixels at the same position in the local solder joint depth image and the local solder joint depth image template, recorded as PC ij , where j represents the jth pixel; Step S24: When the pixel difference PC ij If the pixel difference value is greater than or equal to the preset pixel difference threshold PC0, it means that the corresponding pixel point in the local solder joint depth image is a peeled pixel point. The peeled pixel point is the pixel point in the copper foil peeling area. The number of all peeled pixels in the local solder joint depth image is counted and recorded as F i , that is, obtaining the characteristic size information of the peeling area.

[0011] Furthermore, the solder joint dent detection module includes a dent area size feature information acquisition unit and a dent depth feature acquisition unit; the dent area size feature information acquisition unit is used to process the solder joint depth image to obtain the dent area boundary, and then obtain the dent area size feature information As i The concave depth feature acquisition unit is used to obtain the pixel value of the pixel point within the concave area boundary line to obtain the concave depth feature information Ad i .

[0012] Furthermore, the specific processing process of the recessed area size feature information acquisition unit is as follows: Step S31: obtaining a solder joint area in the solder joint depth image in step S21, and cutting out the solder joint area from the solder joint depth image to obtain a solder joint area depth image; Step S32: in the depth image of the solder joint area, the pixel difference between the pixel values ​​of two adjacent pixels is calculated from the outside to the inside. When the pixel difference exceeds the set threshold value Ps, the pixel with the smaller pixel value of the two pixels is defined as a boundary point. Step S33: Acquire all boundary points, and all boundary points form the boundary of the concave area; Step S34: Count the number of all pixels within the concave area boundary, denoted as As i , that is, the size characteristic information of the concave area is obtained.

[0013] Furthermore, the specific processing process of the concave depth feature acquisition unit is as follows: Step S41: reading the pixel values ​​of all pixels within the boundary of the concave area in step S33; Step S42: Record the maximum pixel value as Ad i , that is, the concave depth feature information Ad is obtained i .

[0014] Furthermore, the parameter optimization module adjusts the welding temperature as follows: According to the peeling area size characteristic information F i Search in the preset welding temperature adjustment database to obtain the welding temperature adjustment value Tt for the next welding i+1 In the next welding, the adjustment value Tt is subtracted from the current welding temperature. i+1 , until the stripping area size feature information F i When the value is 0, the adjustment of welding temperature can be stopped. The temperature of the first welding is the initial maximum allowable welding temperature. The stripping area size feature information F i The value of the adjustment value Tt i+1 The numerical value of is positively correlated, and the welding temperature adjustment database is preset with the peeling area size feature information F i With adjustment value Tt i+1 The corresponding relationship.

[0015] Furthermore, the parameter optimization module adjusts the welding time as follows: According to the concave area size characteristic information As i Search in the preset welding time adjustment database to obtain the welding time adjustment value St for the next welding. i+1 In the next welding, the adjustment value St is subtracted from the current welding time. i+1 , until the concave area size characteristic information As i When the value is 0, the adjustment of welding time can be stopped. The time of the first welding is the initial maximum allowable welding time. The size characteristic information of the concave area As i The value of the adjustment value St i+1 The numerical value of is positively correlated, and the welding time adjustment database is preset with the concave area size feature information As i With adjustment value St i+1 The corresponding relationship.

[0016] Furthermore, the parameter optimization module adjusts the welding pressure as follows: According to the concave depth feature information Ad i Search in the preset welding pressure adjustment database to obtain the welding pressure adjustment value Pt for the next welding i+1 In the next welding, the adjustment value Pt is subtracted from the welding pressure of this time. i+1 , until the depression depth feature information Ad i When the value is 0, the welding pressure adjustment can be stopped. The pressure during the first welding is the initial maximum allowable welding pressure. The concave depth characteristic information Ad iThe value of the adjustment value Pt i+1 The numerical value of is positively correlated, and the welding pressure adjustment database is preset with Pt i+1 With adjustment value Pt i+1 The corresponding relationship.

[0017] Compared with the prior art, the present invention has the following advantages: the PCBA welding process optimization system based on deep learning uses the number of pixels in the local solder joint depth image whose pixel difference with the pixel points at the corresponding position of the local solder joint depth image template exceeds the set threshold as the stripping area size feature information of each solder joint, and can accurately characterize the degree of copper foil stripping near each solder joint, which can be used as an accurate basis for subsequent parameter optimization; through the solder joint area in the acquired solder joint depth image, the boundary points are cleverly identified to form the boundary of the recessed area, so that the size and depth of the recessed area can be accurately characterized, which can be used as an accurate basis for subsequent parameter optimization; based on the stripping area size feature information and the solder joint recess feature information, the welding temperature, welding time and welding pressure can be accurately adjusted to achieve the optimization of welding process parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a structural schematic block diagram of a PCBA welding process optimization system based on deep learning in an embodiment of the present invention. DETAILED DESCRIPTION

[0019] The following is a detailed description of an embodiment of the present invention. This embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation method and a specific operation process are given, but the protection scope of the present invention is not limited to the following embodiment.

[0020] like Figure 1 As shown, this embodiment provides a technical solution: a PCBA welding process optimization system based on deep learning, which optimizes the initially set welding process parameters (mainly referring to welding temperature, welding time, and welding pressure) through welding tests before batch welding of various components, so as to avoid the occurrence of incomplete welding points and PCB copper foil peeling when batch welding of various components is performed subsequently, and includes the following modules: an initial parameter setting module, a copper foil peeling detection module, a solder joint dent detection module, and a parameter optimization module; In this embodiment, the initial parameter setting module is used to initialize the welding temperature, welding time, and welding pressure during welding to obtain the initial maximum allowable welding temperature, the initial maximum allowable welding time, and the initial maximum allowable welding pressure; More specifically, the initial maximum allowable soldering temperature is initialized and set according to the type of solder used and the heat resistance of the components. Generally, the initialization setting can be performed with reference to the melting point of the solder and the recommended temperature range of the components.

[0021] More specifically, the initial maximum allowable soldering time is initialized and set according to the type of solder used and the pin size of the component.

[0022] More specifically, the initial maximum allowable welding pressure is initialized and set according to the circuit board type, solder type, and component type. It should be noted that the thickness, material, and layout of the pads of the circuit board will affect the welding pressure. For thick boards or multi-layer boards, the welding pressure needs to be appropriately increased to ensure that the solder can fully fill the gap between the pad and the pin.

[0023] After the initial welding process parameters are set in the initial parameter setting module, the current components are welded by the welding equipment to start the welding test.

[0024] In this embodiment, the copper foil peeling detection module is used to detect the copper foil peeling phenomenon after each welding is completed, and obtain the size characteristic information of the peeling area.

[0025] As more specific, the copper foil peeling detection module includes an image acquisition unit, a solder joint recognition unit and a peeling area size feature information acquisition unit; the image acquisition unit is used to use the RGB-D depth camera installed on the welding head of the welding equipment to shoot the solder joint below after each welding, obtain the solder joint depth image and the corresponding solder joint RGB image, and perform noise reduction processing on the solder joint depth image and the solder joint RGB image; the solder joint recognition unit is used to use the trained solder joint target detection model to identify the solder joint in the solder joint RGB image, and perform outer contour detection processing to obtain the solder joint area position information; the peeling area size feature information acquisition unit is used to process according to the solder joint area position information and the solder joint depth image template to obtain the peeling area size feature information F i , where i represents the i-th inspection, and each solder joint is subjected to a copper foil peeling / solder joint depression inspection.

[0026] It should be noted that the RGB-D depth camera can not only obtain the two-dimensional image information (RGB image) of the object, but also simultaneously obtain the depth information (depth image) of the object, and the specifications of the RGB image and the depth image are consistent. The solder joint depth image and solder joint RGB image in this embodiment both contain a single complete solder joint and the copper foil near the solder joint.

[0027] It should be noted that the solder joint target detection model is obtained based on SSD target detection network training.

[0028] As more specific, the specific processing process of the solder joint identification unit is as follows: Step S11: using the target detection model to identify the solder joints in the solder joint RGB image, and obtaining the coordinates of the upper left corner point and the lower right corner point of the solder joint detection frame in the RGB image; Step S12: Use the contour detection function in OpenCV to perform outer contour detection on the inside of the solder joint detection frame area, obtain the outer contour line of the solder joint and the coordinates of each point on the outer contour line in the solder joint RGB image, that is, obtain the position information of the solder joint area.

[0029] As more specific, the specific processing process of the stripping area size feature information acquisition unit is as follows: Step S21: Obtaining the solder joint area in the solder joint depth image according to the solder joint outer contour line and the coordinates of each point on the outer contour line in the solder joint RGB image; Step S22: removing the solder joint area in the solder joint depth image from the solder joint depth image and the solder joint depth image template, and correspondingly obtaining a local solder joint depth image and a local solder joint depth image template; Step S23: Read the pixel values ​​of the pixels at the same position in the local solder joint depth image and the local solder joint depth image template respectively, and calculate the pixel difference between the pixels at the same position in the local solder joint depth image and the local solder joint depth image template, recorded as PC ij , where j represents the jth pixel; Step S24: When the pixel difference PC ij If the pixel difference value is greater than or equal to the preset pixel difference threshold PC0, it means that the corresponding pixel point in the local solder joint depth image is a peeled pixel point. The peeled pixel point is the pixel point in the copper foil peeling area. The number of all peeled pixels in the local solder joint depth image is counted and recorded as F i , that is, obtaining the characteristic size information of the peeling area.

[0030] It should be noted that, in this embodiment, the solder point depth image template is obtained by photographing with the above-mentioned RGB-D depth camera before soldering at each solder point position, and the shooting position is the same as the shooting position after the welding is completed; and after each welding is completed, the RGB-D depth camera photographs each solder point of the lower PCB at the same set height.

[0031] In the present invention, the number of pixels in the local solder point depth image whose pixel difference with the pixel points at the corresponding position of the local solder point depth image template exceeds the set threshold is used as the characteristic size information of the peeling area of ​​each solder point, which can accurately characterize the degree of copper foil peeling near each solder point and can serve as an accurate basis for subsequent parameter optimization.

[0032] In this embodiment, the solder joint depression detection module is used to detect the solder joint depression phenomenon after each welding is completed to obtain solder joint depression feature information.

[0033] As more specifically, the solder joint dent detection module includes a dent area size feature information acquisition unit and a dent depth feature acquisition unit; the dent area size feature information acquisition unit is used to process the solder joint depth image to obtain the dent area boundary, and then obtain the dent area size feature information As i The concave depth feature acquisition unit is used to obtain the pixel value of the pixel point within the concave area boundary line to obtain the concave depth feature information Ad i .

[0034] As more specific, the specific processing process of the recessed area size feature information acquisition unit is as follows: Step S31: obtaining a solder joint area in the solder joint depth image in step S21, and cutting out the solder joint area from the solder joint depth image to obtain a solder joint area depth image; Step S32: in the depth image of the solder joint area, the pixel difference between the pixel values ​​of two adjacent pixels is calculated from the outside to the inside. When the pixel difference exceeds the set threshold value Ps, the pixel with the smaller pixel value of the two pixels is defined as a boundary point. Step S33: Acquire all boundary points, and all boundary points form the boundary of the concave area; Step S34: Count the number of all pixels within the concave area boundary, denoted as As i , that is, the size characteristic information of the concave area is obtained.

[0035] As more specific, the specific processing process of the depression depth feature acquisition unit is as follows: Step S41: reading the pixel values ​​of all pixels within the boundary of the concave area in step S33; Step S42: Record the maximum pixel value as Ad i , that is, the concave depth feature information Ad is obtained i .

[0036] In the present invention, the boundary points are cleverly identified through the solder joint area in the acquired solder joint depth image to form the boundary of the recessed area, so that the size and depth of the recessed area can be accurately characterized.

[0037] In this embodiment, the parameter optimization module is used to adjust the welding temperature, welding time and welding pressure according to the stripping area size feature information and the welding spot depression feature information, so as to optimize the welding process parameters.

[0038] As more specific, the specific processing process of the parameter optimization module is as follows: According to the peeling area size characteristic information F iSearch in the preset welding temperature adjustment database to obtain the welding temperature adjustment value Tt for the next welding i+1 In the next welding, the adjustment value Tt is subtracted from the current welding temperature. i+1 , until the stripping area size feature information F i When the value is 0, the adjustment of welding temperature can be stopped. The temperature of the first welding is the initial maximum allowable welding temperature. The stripping area size feature information F i The value of the adjustment value Tt i+1 The numerical value of is positively correlated, and the welding temperature adjustment database is preset with the peeling area size feature information F i With adjustment value Tt i+1 The corresponding relationship; According to the concave area size characteristic information As i Search in the preset welding time adjustment database to obtain the welding time adjustment value St for the next welding. i+1 In the next welding, the adjustment value St is subtracted from the current welding time. i+1 , until the concave area size characteristic information As i When the value is 0, the adjustment of welding time can be stopped. The time of the first welding is the initial maximum allowable welding time. The size characteristic information of the concave area As i The value of the adjustment value St i+1 The numerical value of is positively correlated, and the welding time adjustment database is preset with the concave area size feature information As i With adjustment value St i+1 The corresponding relationship; According to the concave depth feature information Ad i Search in the preset welding pressure adjustment database to obtain the welding pressure adjustment value Pt for the next welding i+1 In the next welding, the adjustment value Pt is subtracted from the welding pressure of this time. i+1 , until the depression depth feature information Ad i When the value is 0, the welding pressure adjustment can be stopped. The pressure during the first welding is the initial maximum allowable welding pressure. The concave depth characteristic information Ad i The value of the adjustment value Pt i+1 The numerical value of is positively correlated, and the welding pressure adjustment database is preset with Pt i+1 With adjustment value Pt i+1 The corresponding relationship.

[0039] It should be noted that during the welding process, the main factor causing copper foil peeling is excessive welding temperature, so the temperature is gradually adjusted downward based on the initial maximum allowable welding temperature; similarly, the main factors causing solder joint depression are excessive welding time and excessive welding pressure, among which excessive welding time mainly affects the size of the depressed area. The longer the welding time, the larger the depressed area. Therefore, the temperature is gradually adjusted downward based on the initial maximum allowable welding time. Excessive welding pressure mainly affects the depth of the depressed area. The greater the welding pressure, the deeper the depressed area. Therefore, the temperature is gradually adjusted downward based on the initial maximum allowable welding pressure.

[0040] In this embodiment, for the testing process of a single component, the optimization of the welding process parameters of the current component can be completed through 5-8 rounds of detection and optimization process.

[0041] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A PCBA welding process optimization system based on deep learning, characterized in that: It includes initial parameter setting module, copper foil peeling detection module, solder joint sag detection module and parameter optimization module; The initial parameter setting module is used to initialize the welding temperature, welding time and welding pressure during welding to obtain the initial maximum allowable welding temperature, the initial maximum allowable welding time and the initial maximum allowable welding pressure; The copper foil peeling detection module is used to detect the copper foil peeling phenomenon after each welding is completed, and obtain the size characteristic information of the peeling area; The solder joint depression detection module is used to detect the solder joint depression phenomenon after each welding is completed to obtain solder joint depression feature information; The parameter optimization module is used to adjust the welding temperature, welding time and welding pressure respectively based on the initial maximum allowable welding temperature, initial maximum allowable welding time and initial maximum allowable welding pressure according to the stripping area size feature information and the solder joint depression feature information, so as to optimize the welding process parameters.

2. According to a PCBA welding process optimization system based on deep learning according to claim 1, it is characterized in that: The copper foil peeling detection module includes an image acquisition unit, a solder joint recognition unit and a peeling area size feature information acquisition unit; the image acquisition unit is used to use the RGB-D depth camera installed on the welding head of the welding equipment to shoot the solder joint below after each welding, obtain the solder joint depth image and the corresponding solder joint RGB image, and perform noise reduction processing on the solder joint depth image and the solder joint RGB image; The solder joint recognition unit is used to use the trained solder joint target detection model to recognize the solder joints in the solder joint RGB image, and perform outer contour detection processing to obtain the solder joint area position information; the stripping area size feature information acquisition unit is used to process according to the solder joint area position information and the solder joint depth image template to obtain the stripping area size feature information F i , where i represents the i-th detection.

3. According to a PCBA welding process optimization system based on deep learning according to claim 2, it is characterized in that: The specific processing process of the solder joint identification unit is as follows: Step S11: using the target detection model to identify the solder joints in the solder joint RGB image, and obtaining the coordinates of the upper left corner point and the lower right corner point of the solder joint detection frame in the RGB image; Step S12: Use the contour detection function in OpenCV to perform outer contour detection on the inside of the solder joint detection frame area, obtain the outer contour line of the solder joint and the coordinates of each point on the outer contour line in the solder joint RGB image, that is, obtain the position information of the solder joint area.

4. According to a PCBA welding process optimization system based on deep learning according to claim 3, it is characterized in that: The specific processing process of the stripping area size feature information acquisition unit is as follows: Step S21: Obtaining the solder joint area in the solder joint depth image according to the solder joint outer contour line and the coordinates of each point on the outer contour line in the solder joint RGB image; Step S22: removing the solder joint area in the solder joint depth image from the solder joint depth image and the solder joint depth image template, and correspondingly obtaining a local solder joint depth image and a local solder joint depth image template, wherein the solder joint depth image template is obtained by photographing at each solder joint position by using an RGB-D depth camera before welding, and the photographing position is the same as the photographing position after welding. Step S23: Read the pixel values ​​of the pixels at the same position in the local solder joint depth image and the local solder joint depth image template respectively, and calculate the pixel difference between the pixels at the same position in the local solder joint depth image and the local solder joint depth image template, recorded as PC ij , where j represents the jth pixel; Step S24: When the pixel difference PC ij If the pixel difference value is greater than or equal to the preset pixel difference threshold PC0, it means that the corresponding pixel point in the local solder joint depth image is a peeled pixel point. The peeled pixel point is the pixel point in the copper foil peeling area. The number of all peeled pixels in the local solder joint depth image is counted and recorded as F i , that is, obtaining the characteristic size information of the peeling area.

5. According to a PCBA welding process optimization system based on deep learning according to claim 4, it is characterized in that: The solder joint dent detection module includes a dent area size feature information acquisition unit and a dent depth feature acquisition unit; the dent area size feature information acquisition unit is used to process the solder joint depth image to obtain the dent area boundary, and then obtain the dent area size feature information As i The concave depth feature acquisition unit is used to obtain the pixel value of the pixel point within the concave area boundary line to obtain the concave depth feature information Ad i .

6. A PCBA welding process optimization system based on deep learning according to claim 5, characterized in that: The specific processing process of the recessed area size feature information acquisition unit is as follows: Step S31: obtaining a solder joint area in the solder joint depth image in step S21, and cutting out the solder joint area from the solder joint depth image to obtain a solder joint area depth image; Step S32: in the depth image of the solder joint area, the pixel difference between the pixel values ​​of two adjacent pixels is calculated from the outside to the inside. When the pixel difference exceeds the set threshold value Ps, the pixel with the smaller pixel value of the two pixels is defined as a boundary point. Step S33: Acquire all boundary points, and all boundary points form the boundary of the concave area; Step S34: Count the number of all pixels within the concave area boundary, denoted as As i , that is, the size characteristic information of the concave area is obtained.

7. A PCBA welding process optimization system based on deep learning according to claim 6, characterized in that: The specific processing process of the concave depth feature acquisition unit is as follows: Step S41: reading the pixel values ​​of all pixels within the boundary of the concave area in step S33; Step S42: Record the maximum pixel value as Ad i , that is, the concave depth feature information Ad is obtained i .

8. The PCBA welding process optimization system based on deep learning according to claim 5, characterized in that: The parameter optimization module adjusts the welding temperature as follows: According to the peeling area size characteristic information F i Search in the preset welding temperature adjustment database to obtain the welding temperature adjustment value Tt for the next welding i+1 , in the next welding, subtract the adjustment value Tt from the current welding temperature i+1 , until the stripping area size feature information F i When the value is 0, the adjustment of welding temperature can be stopped. The temperature of the first welding is the initial maximum allowable welding temperature. The stripping area size feature information F i The value of the adjustment value Tt i+1 The numerical value of is positively correlated, and the welding temperature adjustment database is preset with the peeling area size feature information F i With adjustment value Tt i+1 The corresponding relationship.

9. The PCBA welding process optimization system based on deep learning according to claim 5, characterized in that: The parameter optimization module adjusts the welding time as follows: According to the concave area size characteristic information As i Search in the preset welding time adjustment database to obtain the welding time adjustment value St for the next welding. i+1 In the next welding, the adjustment value St is subtracted from the current welding time. i+1 , until the concave area size characteristic information As i When the value is 0, the adjustment of welding time can be stopped. The time of the first welding is the initial maximum allowable welding time. The size characteristic information of the concave area As i The value of the adjustment value St i+1 The numerical value of is positively correlated, and the welding time adjustment database is preset with the concave area size feature information As i With adjustment value St i+1 The corresponding relationship.

10. The PCBA welding process optimization system based on deep learning according to claim 5, characterized in that: The parameter optimization module adjusts the welding pressure as follows: According to the concave depth feature information Ad i Search in the preset welding pressure adjustment database to obtain the welding pressure adjustment value Pt for the next welding i+1 In the next welding, the adjustment value Pt is subtracted from the welding pressure of this time. i+1 , until the depression depth feature information Ad i When the value is 0, the welding pressure adjustment can be stopped. The pressure during the first welding is the initial maximum allowable welding pressure. The concave depth characteristic information Ad i The value of the adjustment value Pt i+1 The numerical value of is positively correlated, and the welding pressure adjustment database is preset with Pt i+1 With adjustment value Pt i+1 The corresponding relationship.

Images