Anti-deviation system for solder mask exposure alignment of intelligent PCB (Printed Circuit Board)

Abstract

The invention discloses an anti-deviation system for solder mask exposure alignment of an intelligent PCB (Printed Circuit Board), which relates to the technical field of printed circuit board manufacturing and comprises a management center, an equipment operation end, an equipment monitoring end, an anti-deviation analysis module and an alignment control module. Conveying the PCB to an exposure machine through an equipment operation end so as to carry out alignment operation between the shading plate and the PCB; a real-time alignment image of alignment operation is obtained in real time through an equipment monitoring end, the real-time alignment image is disassembled into a plurality of alignment image frames for sequential analysis, when it is monitored that alignment deviation exceeds a preset deviation threshold value, an alarm signal is generated and sent to a management center, after an anti-deviation analysis module receives the alarm signal, the real-time alignment image is subjected to data analysis, and the alignment operation is completed. According to the method, the PCB anti-deviation control parameter set is generated, and finally, the alignment control module performs alignment control on the PCB on the current exposure machine according to the PCB anti-deviation control parameter set, so that the accurate alignment of the PCB is improved to a great extent, and the alignment deviation is reduced.

Description

Technical Field

[0001] The invention relates to the technical field of printed circuit board manufacturing, in particular to an anti-deflection system for intelligent PCB board solder mask exposure alignment. Background Art

[0002] In the existing PCB board production and processing technology, especially when performing the alignment operation between the shading plate and the PCB board on the exposure machine, it mainly depends on the experience and skills of the operator. The operator needs to manually place the PCB board at the specified position of the exposure machine and ensure the accurate alignment between the shading plate and the PCB board through visual inspection.

[0003] However, this manual alignment method has certain limitations. For example, factors such as operator fatigue and lack of concentration may lead to inaccurate alignment, thus affecting the production quality of PCB boards. In addition, the efficiency of manual alignment is also low. Therefore, how to propose a system that can efficiently help PCB boards to perform solder mask exposure alignment and anti-deflection has become a top priority. Summary of the invention

[0004] In order to solve the above problems, the purpose of the present invention is to provide an anti-deflection system for intelligent PCB board solder mask exposure alignment.

[0005] The purpose of the present invention can be achieved by the following technical solutions: an anti-deflection system for intelligent PCB board solder mask exposure alignment, comprising a management center, wherein the management center is communicatively connected with a device operation terminal, a device monitoring terminal, an anti-deflection analysis module, and an alignment control module;

[0006] The equipment operation end is used to transport the PCB board to the exposure machine and perform the alignment operation between the light shielding plate and the PCB board on the exposure machine;

[0007] The equipment monitoring end is used to obtain the real-time alignment image of the alignment operation between the shading plate and the PCB board, and disassemble the real-time alignment image into a plurality of alignment image frames, analyze each alignment image frame in turn, and when the alignment deviation is detected to exceed a preset deviation threshold, generate a corresponding alarm signal and send it to the management center;

[0008] The anti-deviation analysis module is used to perform data analysis on the real-time image of the equipment monitoring end after receiving the alarm signal, and then generate a corresponding PCB board anti-deviation control parameter set;

[0009] The alignment control module is used to perform alignment control on the PCB board on the current exposure machine according to the PCB board anti-skew control parameter set.

[0010] Furthermore, the process in which the equipment operation end transports the PCB board to the exposure machine and performs an alignment operation between the light shielding plate and the PCB board on the exposure machine includes:

[0011] The equipment operation end starts the conveyor belt to transport the PCB board from the temporary storage area of ​​the silo to the loading area set at the exposure machine. After the PCB board enters the loading area, the PCB board is fixed to the working table of the exposure machine using the positioning pins, and the placement angle of the PCB board on the working table is adjusted to be in the respective placement areas of the X-axis and the Y-axis;

[0012] A machine vision guidance device is arranged on the work surface. The PCB board on the work surface is moved by the positioning pins, and the machine vision guidance device guides the real-time movement operation of the positioning pins. When the PCB board is located below the shading plate and the PCB board and the shading plate are in the preset rough alignment area, the preliminary alignment operation of the PCB board is completed.

[0013] Furthermore, the device monitoring end obtains the real-time alignment image of the alignment operation between the shading plate and the PCB board, and decomposes the real-time alignment image into a plurality of alignment image frames, including:

[0014] The equipment monitoring end deploys a panoramic imaging device to obtain the real-time alignment image of the alignment operation between the shading plate and the PCB, obtain the image stream duration of the real-time alignment image, set the unit frame duration of the real-time alignment image, and decompose the real-time alignment image into several alignment image frames according to the image stream duration and the unit frame duration;

[0015] A key image frame is set, and the key image frame is set with corresponding image feature information. The image feature information includes geometric distribution features, texture mapping features, color-related features, and luminosity-related features. The image feature information of each aligned image frame is extracted in turn, and the image feature information of each aligned image frame is compared with the image feature information of the key image frame for similar information items in turn, so as to screen out aligned image frames that do not meet the requirements.

[0016] Furthermore, the process of comparing similar information items includes:

[0017] The image feature information corresponding to the key image frame is used as a subject feature information set;

[0018] Taking the image feature information corresponding to the aligned image frame as a mapping feature information set;

[0019] For the subject feature information set and the mapping feature information set, the geometric distribution features, texture mapping features, color-related features, and luminosity-related features included in each are respectively used as respective information items, and the information items of the same type in the subject feature information set and the mapping feature information set are merged to generate different similar information items;

[0020] Through the convolutional neural network technology, a bag-of-words analysis model corresponding to different similar information items is constructed. The information coupling rate of each similar information item is obtained from the bag-of-words analysis model, and then it is determined whether the different similar information items constituted by each aligned image frame and the key image frame are in a coupled state. If so, the corresponding aligned image frame is retained, if not, the corresponding aligned image frame is discarded.

[0021] Furthermore, the process of obtaining the information coupling rate of similar information items by the bag-of-words analysis model includes:

[0022] Performing text preprocessing on two information items in the same category of information items respectively, and then generating a number of text segmentations corresponding to the respective information items, setting a filtering segmentation set, inputting a number of text segmentations corresponding to the two information items into the filtering segmentation set for information filtering, marking the filtered text segmentations as irrelevant segmentations, and marking the retained text segmentations as relevant segmentations;

[0023] Summarize all the relevant participles of two information items of the same type of information item, respectively, and then generate the corresponding word bags to be processed, perform Euclidean distance calculation on the word bags to be processed by the word band analysis model, and obtain the Euclidean distance between several relative relevant participles in the two word bags to be processed, record the logarithm of all relative relevant participles whose Euclidean distance is less than or equal to the preset distance threshold as β, and record the total logarithm of several relative relevant participles in the two word bags to be processed as α;

[0024] It is then concluded that the information coupling rate of the current similar information items is β / α, where α and β are both natural numbers greater than 0.

[0025] Furthermore, each alignment image frame is analyzed in turn, and when the alignment deviation is detected to exceed a preset deviation threshold, a corresponding alarm signal is generated and sent to the management center, including:

[0026] All the alignment image frames that have been screened out and retained are analyzed in turn to obtain the alignment deviation between the PCB board and the light shielding plate in each alignment image frame. The alignment deviation includes a horizontal alignment deviation and a vertical alignment deviation. The deviation distances of the horizontal alignment deviation and the vertical alignment deviation are recorded as HJL and ZJL respectively.

[0027] The deviation threshold includes a lateral deviation threshold and a longitudinal deviation threshold, which are denoted as P and Q respectively;

[0028] When HJL≥P, the generated alarm message is: There is an abnormality in the horizontal alignment between the PCB board and the shading plate. Please adjust the horizontal alignment distance between the PCB board and the shading plate in time, generate the message tag Tag1, and bind the message tag Tag1 to the current alarm message;

[0029] When ZJL≥Q, the generated alarm message is: there is an abnormality in the longitudinal alignment between the PCB board and the shading plate. Please adjust the longitudinal alignment distance between the PCB board and the shading plate in time, generate the message tag Tag2, and bind the message tag Tag2 to the current alarm message;

[0030] When HJL<P, or ZJL<Q, no alarm message is generated;

[0031] All alarm messages bound to message tags are aggregated, corresponding alarm signals are generated and sent to the management center.

[0032] Furthermore, after receiving the alarm signal, the anti-deviation analysis module performs data analysis on the real-time image of the equipment monitoring end, and then generates a corresponding PCB board anti-deviation control parameter set, including:

[0033] The anti-bias analysis module consists of a data processing unit and a data analysis unit;

[0034] When the data processing unit receives the alarm signal, it determines the type of the message tag bound to the alarm message included in the alarm signal. If the message tag is Tag1, the longitudinal area image of the aligned image frame retained after the screening is completed in the aligned live image is intercepted; if the message tag is Tag2, the horizontal area image of the corresponding aligned image frame is intercepted;

[0035] The transverse region image and the longitudinal region image intercepted by the data processing unit are input into the data analysis unit, and the data analysis unit performs data analysis according to the transverse region image, thereby generating alignment control parameters for the alignment operation between the PCB board and the light shielding plate in the transverse direction, and then performs data analysis according to the longitudinal region image, thereby generating alignment control parameters for the alignment operation between the PCB board and the light shielding plate in the longitudinal direction;

[0036] The alignment control parameters of the alignment operation in the transverse direction include a transverse correction distance, a transverse correction speed, and a transverse alignment permissible error, and the alignment control parameters of the alignment operation in the longitudinal direction include a longitudinal correction distance, a longitudinal correction speed, and a longitudinal alignment permissible error;

[0037] All alignment control parameters are summarized to generate the corresponding PCB board anti-skew control parameter set.

[0038] Furthermore, the process of the alignment control module performing alignment control on the PCB board on the current exposure machine according to the PCB board anti-skew control parameter set includes:

[0039] The alignment control module receives the anti-deflection control parameter set of the PCB board. The alignment control module is provided with an instruction area. The instruction area inputs the anti-deflection control parameter set of the PCB board, and generates instruction codes for controlling the PCB board on the exposure machine to perform horizontal alignment or vertical alignment according to the anti-deflection control parameter set of the PCB board. The alignment control module controls the positioning pins on the working table of the exposure machine according to the instruction codes, and the positioning pins adjust the alignment position between the PCB board and the light shielding plate, and obtains the alignment coincidence rate between the PCB board and the light shielding plate in real time;

[0040] When the alignment overlap rate is greater than or equal to 95%, the PCB board successfully completes the alignment with the shading plate;

[0041] When the alignment overlap rate is less than 95%, the positioning pins continue to adjust the position of the PCB board relative to the light shielding plate until the alignment overlap rate meets the requirements, and then the positioning pins are stopped from being controlled.

[0042] Compared with the prior art, the beneficial effects of the present invention are as follows: by acquiring the real-time alignment image through the equipment monitoring end, and disassembling it into several frames of alignment image frames for analysis, the alignment deviation of the PCB board during alignment can be discovered in time; the combination of the anti-bias analysis module and the alignment control module can automatically generate a PCB board anti-bias control parameter set, and perform alignment control on the PCB board based on it, thereby reducing dependence on operators, effectively improving the accuracy of alignment of the PCB board with the shading plate on the work surface of the exposure machine, and also improving the alignment efficiency of the PCB board. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a schematic diagram of the present invention. DETAILED DESCRIPTION

[0044] like Figure 1 As shown, an anti-deflection system for intelligent PCB board solder mask exposure alignment includes a management center, and the management center is communicatively connected with a device operation terminal, a device monitoring terminal, an anti-deflection analysis module, and an alignment control module;

[0045] The equipment operation end is used to transport the PCB board to the exposure machine and perform the alignment operation between the light shielding plate and the PCB board on the exposure machine;

[0046] The equipment monitoring end is used to obtain the real-time alignment image of the alignment operation between the shading plate and the PCB board, and disassemble the real-time alignment image into a plurality of alignment image frames, analyze each alignment image frame in turn, and when the alignment deviation is detected to exceed a preset deviation threshold, generate a corresponding alarm signal and send it to the management center;

[0047] The anti-deviation analysis module is used to perform data analysis on the real-time image of the equipment monitoring end after receiving the alarm signal, and then generate a corresponding PCB board anti-deviation control parameter set;

[0048] The alignment control module is used to perform alignment control on the PCB board on the current exposure machine according to the PCB board anti-skew control parameter set.

[0049] It should be further explained that, in a specific implementation process, the equipment operation end transports the PCB board to the exposure machine, and performs the alignment operation between the light shielding plate and the PCB board on the exposure machine, including:

[0050] Set the exposure machine to standby mode, adjust the safety doors around the exposure machine to the closed state, verify whether the shading plate on the exposure machine is correctly installed, and verify whether the shading plate is in good condition. If the shading plate is correctly installed and in good condition, the equipment operation end successfully completes the initialization, otherwise, reinstall or replace the shading plate;

[0051] The equipment operation end starts the conveyor belt to convey the PCB board from the temporary storage area of ​​the silo to the loading area set at the exposure machine. The layout direction of the PCB board on the conveyor belt is parallel to the movement direction of the conveyor belt. When the PCB board enters the corresponding loading area of ​​the exposure machine, the PCB board is fixed to the working table of the exposure machine using the positioning pins, and the placement angle of the PCB board on the working table is adjusted to be in the respective placement areas of the X-axis and the Y-axis, thereby completing the positioning of the PCB board;

[0052] A machine vision guidance device is arranged on the work surface. The PCB board on the work surface is moved by the positioning pins, and the machine vision guidance device guides the real-time movement operation of the positioning pins. When the PCB board is located below the shading plate and the PCB board and the shading plate are in the preset rough alignment area, the preliminary alignment operation of the PCB board is completed.

[0053] It should be further explained that, in the specific implementation process, the device monitoring end obtains the real-time alignment image of the alignment operation between the shading plate and the PCB board, and the process of decomposing the real-time alignment image into a plurality of alignment image frames includes:

[0054] A panoramic imaging device is deployed at the equipment monitoring end. The panoramic imaging device is used to obtain the real-time alignment image of the alignment operation between the shading plate and the PCB. The real-time alignment image is used to record the entire operation process of the alignment operation between the shading plate and the PCB board.

[0055] Obtain the image stream duration of the real-time image, and record the image stream duration as T;

[0056] Set the unit frame duration of the live image, and record the unit frame duration as t;

[0057] According to the duration of the image stream and the duration of the unit frame, the aligned live image is decomposed into a number of aligned image frames, and the number of the aligned image frames is recorded as N, then N=T / t; wherein, T and t are both real numbers greater than 0, when the value of T / t is a non-integer, the value of T / t is rounded up to an integer, and when the value of T / t is an integer, the corresponding integer is directly taken;

[0058] Setting key image frames, the key image frames are correspondingly provided with corresponding image feature information, the image feature information includes geometric distribution features, texture mapping features, color-related features and luminosity-related features, extracting the image feature information of each alignment image frame in turn, and comparing the image feature information of each alignment image frame with the image feature information of the key image frame in turn for similar information items, thereby filtering out alignment image frames that do not meet the requirements;

[0059] The content of the comparison of similar information items is:

[0060] The image feature information corresponding to the key image frame is used as a main feature information set, and for the main feature information set, the geometric distribution features, texture mapping features, color-related features, and luminosity-related features included in the main feature information set are respectively used as information items of the main feature information set;

[0061] The image feature information corresponding to the aligned image frame is used as a mapping feature information set, and for the mapping feature information set, the geometric distribution feature, texture mapping feature, color-related feature and luminosity-related feature included in the mapping feature information set are respectively used as information items of the mapping feature information set;

[0062] Merging information items of the same type in the subject feature information set and the mapping feature information set to generate different similar information items, where the similar information items include a first similar information item, a second similar information item, a third similar information item, and a fourth similar information item;

[0063] Through the convolutional neural network technology, a bag-of-words analysis model corresponding to different similar information items is constructed. The information coupling rate of each similar information item is obtained from the bag-of-words analysis model, and the coupling threshold of the information coupling rate is set, and the coupling threshold is recorded as Oh;

[0064] The first information item of the same kind, the second information item of the same kind, the third information item of the same kind, and the fourth information item of the same kind are expressed as follows:

[0065] First similar information item =<info-One,info-one> ;

[0066] Second similar information item =<info-Two,info-two> ;

[0067] The third similar information item =<info-Three,info-three> ;

[0068] Fourth similar information item =<info-Four,info-four> ;

[0069] Among them, info-One and info-one are information items of geometric distribution features in the key image frame and the matched image frame, respectively; info-Two and info-two are information items of texture mapping features in the key image frame and the matched image frame, respectively; info-Three and info-three are information items of color-related features in the key image frame and the matched image frame, respectively; info-Four and info-four are information items of luminosity-related features in the key image frame and the matched image frame, respectively;

[0070] Determine whether the first information item of the same kind, the second information item of the same kind, the third information item of the same kind, and the fourth information item of the same kind formed by each aligned image frame and the key image frame are all in a coupled state, if so, retain the corresponding aligned image frame, if not, remove the corresponding aligned image frame;

[0071] The details are as follows:

[0072] The information coupling rates of the first similar information item, the second similar information item, the third similar information item and the fourth similar information item are denoted as oh1, oh2, oh3 and oh4 respectively;

[0073] When oh1≥Oh, it is judged that the first information item of the same type is in a coupled state, and when oh1<Oh, it is judged that the first information item of the same type is not in a coupled state;

[0074] When oh2≥Oh, it is judged that the second information item of the same type is in a coupled state, and when oh2<Oh, it is judged that the second information item of the same type is not in a coupled state;

[0075] When oh3≥Oh, it is judged that the third information item of the same type is in a coupled state, and when oh3<Oh, it is judged that the third information item of the same type is not in a coupled state;

[0076] When oh4≥Oh, it is judged that the fourth information item of the same type is in a coupled state, and when oh4<Oh, it is judged that the fourth information item of the same type is not in a coupled state;

[0077] The bag-of-words analysis model obtains the information coupling rate of similar information items as follows:

[0078] Two information items in the same category are subjected to text preprocessing respectively, and then a number of text segmentation words corresponding to the respective information items are generated, and a filtering segmentation word set is set, wherein a number of irrelevant segmentation words that are not helpful for text analysis are recorded in the filtering segmentation word set;

[0079] Inputting a plurality of text segmentations corresponding to two information items included in the same information item of the same type into the filtering segmentation set for information filtering, marking the filtered text segmentations as irrelevant segmentations, and marking the retained text segmentations as relevant segmentations;

[0080] Summarize all the relevant participles of two information items of the same type of information item, respectively, and then generate the corresponding word bags to be processed, perform Euclidean distance calculation on the word bags to be processed by the word band analysis model, and obtain the Euclidean distance between several relative relevant participles in the two word bags to be processed, record the logarithm of all relative relevant participles whose Euclidean distance is less than or equal to the preset distance threshold as β, and record the total logarithm of several relative relevant participles in the two word bags to be processed as α;

[0081] Then the information coupling rate of the current similar information items is obtained as β / α, where α and β are both natural numbers greater than 0.

[0082] It should be further explained that, in the specific implementation process, each alignment image frame is analyzed in turn, and when the alignment deviation is detected to exceed the preset deviation threshold, a corresponding alarm signal is generated and sent to the management center, and the process includes:

[0083] All the alignment image frames that are retained after screening are analyzed in sequence, and then the alignment deviation between the PCB board and the light shielding plate in each alignment image frame is obtained, and the alignment deviation includes a horizontal alignment deviation and a vertical alignment deviation;

[0084] The deviation distances corresponding to the lateral alignment deviation and the longitudinal alignment deviation are respectively recorded as HJL and ZJL, and the allowable deviation thresholds of the alignment deviation generated when the PCB board and the light shielding plate are aligned are preset, and the deviation thresholds include the lateral deviation threshold and the longitudinal deviation threshold;

[0085] The lateral deviation threshold and the longitudinal deviation threshold are denoted as P and Q respectively;

[0086] When HJL≥P, the generated alarm message is: There is an abnormality in the horizontal alignment between the PCB board and the shading plate. Please adjust the horizontal alignment distance between the PCB board and the shading plate in time, generate the message tag Tag1, and bind the message tag Tag1 to the current alarm message;

[0087] When ZJL≥Q, the generated alarm message is: there is an abnormality in the longitudinal alignment between the PCB board and the shading plate. Please adjust the longitudinal alignment distance between the PCB board and the shading plate in time, generate the message tag Tag2, and bind the message tag Tag2 to the current alarm message;

[0088] When HJL<P, or ZJL<Q, no alarm message is generated;

[0089] All alarm messages bound to message tags are aggregated, corresponding alarm signals are generated, and the alarm signals are sent to the management center.

[0090] It should be further explained that, in the specific implementation process, after receiving the alarm signal, the anti-deviation analysis module performs data analysis on the real-time image of the equipment monitoring end, and then generates the corresponding PCB board anti-deviation control parameter set, and the process includes:

[0091] The anti-bias analysis module consists of a data processing unit and a data analysis unit;

[0092] When the data processing unit receives the alarm signal, it determines the type of the message tag bound to the alarm message included in the alarm signal. If the message tag is Tag1, the longitudinal area image of the aligned image frame retained after the screening is completed in the aligned live image is intercepted; if the message tag is Tag2, the horizontal area image of the corresponding aligned image frame is intercepted;

[0093] The longitudinal area image is the image portion corresponding to the area formed by the shorter side between the PCB board and the light shielding plate;

[0094] The horizontal area image is the image portion corresponding to the area formed by the longer side between the PCB board and the light shielding plate;

[0095] The transverse region image and the longitudinal region image intercepted by the data processing unit are input into the data analysis unit, and the data analysis unit performs data analysis according to the transverse region image, thereby generating alignment control parameters for the alignment operation between the PCB board and the light shielding plate in the transverse direction, and then performs data analysis according to the longitudinal region image, thereby generating alignment control parameters for the alignment operation between the PCB board and the light shielding plate in the longitudinal direction;

[0096] The alignment control parameters of the alignment operation in the transverse direction include a transverse correction distance, a transverse correction speed, and a transverse alignment permissible error, and the alignment control parameters of the alignment operation in the longitudinal direction include a longitudinal correction distance, a longitudinal correction speed, and a longitudinal alignment permissible error;

[0097] All alignment control parameters are summarized to generate the corresponding PCB board anti-skew control parameter set.

[0098] It should be further explained that, in a specific implementation process, the process in which the alignment control module performs alignment control on the PCB board on the current exposure machine according to the PCB board anti-skew control parameter set includes:

[0099] The alignment control module receives the anti-deflection control parameter set of the PCB board. The alignment control module is provided with an instruction area. The instruction area inputs the anti-deflection control parameter set of the PCB board, and generates instruction codes for controlling the PCB board on the exposure machine to perform horizontal alignment or vertical alignment according to the anti-deflection control parameter set of the PCB board. The alignment control module controls the positioning pins on the working table of the exposure machine according to the instruction codes, and the positioning pins adjust the alignment position between the PCB board and the light shielding plate, and obtains the alignment coincidence rate between the PCB board and the light shielding plate in real time;

[0100] When the alignment overlap rate is greater than or equal to 95%, the PCB board successfully completes the alignment with the shading plate;

[0101] When the alignment overlap rate is less than 95%, the positioning pins continue to adjust the position of the PCB board relative to the light shielding plate until the alignment overlap rate meets the requirements, and then the positioning pins are stopped from being controlled.

[0102] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An intelligent PCB board solder mask exposure alignment anti-deflection system, including a management center, characterized in that: The management center is communicatively connected to the equipment operation terminal, the equipment monitoring terminal, the anti-deviation analysis module and the alignment control module; The equipment operation end is used to transport the PCB board to the exposure machine and perform the alignment operation between the light shielding plate and the PCB board on the exposure machine; The equipment monitoring end is used to obtain the real-time alignment image of the alignment operation between the shading plate and the PCB board, and disassemble the real-time alignment image into a plurality of alignment image frames, analyze each alignment image frame in turn, and when the alignment deviation is detected to exceed a preset deviation threshold, generate a corresponding alarm signal and send it to the management center; The anti-deviation analysis module is used to perform data analysis on the real-time image of the equipment monitoring end after receiving the alarm signal, and then generate a corresponding PCB board anti-deviation control parameter set; The alignment control module is used to perform alignment control on the PCB board on the current exposure machine according to the PCB board anti-skew control parameter set.

2. According to claim 1, an intelligent PCB board solder mask exposure alignment anti-deflection system is characterized in that: The process of the equipment operating end conveying the PCB board to the exposure machine and performing the alignment operation between the light shielding plate and the PCB board on the exposure machine includes: The equipment operation end starts the conveyor belt to transport the PCB board from the temporary storage area of ​​the silo to the loading area set at the exposure machine. After the PCB board enters the loading area, the PCB board is fixed to the working table of the exposure machine using the positioning pins, and the placement angle of the PCB board on the working table is adjusted to be in the respective placement areas of the X-axis and the Y-axis; A machine vision guidance device is arranged on the work surface. The PCB board on the work surface is moved by the positioning pins, and the machine vision guidance device guides the real-time movement operation of the positioning pins. When the PCB board is located below the shading plate and the PCB board and the shading plate are in the preset rough alignment area, the preliminary alignment operation of the PCB board is completed.

3. The anti-deflection system for intelligent PCB solder mask exposure alignment according to claim 2, characterized in that: The equipment monitoring end obtains the real-time alignment image of the alignment operation between the shading plate and the PCB board, and decomposes the real-time alignment image into several alignment image frames, including: The equipment monitoring end deploys a panoramic imaging device to obtain the real-time alignment image of the alignment operation between the shading plate and the PCB, obtain the image stream duration of the real-time alignment image, set the unit frame duration of the real-time alignment image, and decompose the real-time alignment image into several alignment image frames according to the image stream duration and the unit frame duration; A key image frame is set, and the key image frame is set with corresponding image feature information. The image feature information includes geometric distribution features, texture mapping features, color-related features, and luminosity-related features. The image feature information of each aligned image frame is extracted in turn, and the image feature information of each aligned image frame is compared with the image feature information of the key image frame for similar information items in turn, so as to screen out aligned image frames that do not meet the requirements.

4. The anti-deflection system for intelligent PCB solder mask exposure alignment according to claim 3, characterized in that: The process of comparing similar information items includes: The image feature information corresponding to the key image frame is used as a subject feature information set; Taking the image feature information corresponding to the aligned image frame as a mapping feature information set; For the subject feature information set and the mapping feature information set, the geometric distribution features, texture mapping features, color-related features, and luminosity-related features included in each are respectively used as respective information items, and the information items of the same type in the subject feature information set and the mapping feature information set are merged to generate different similar information items; Through the convolutional neural network technology, a bag-of-words analysis model corresponding to different similar information items is constructed. The information coupling rate of each similar information item is obtained from the bag-of-words analysis model, and then it is determined whether the different similar information items constituted by each aligned image frame and the key image frame are in a coupled state. If so, the corresponding aligned image frame is retained, if not, the corresponding aligned image frame is discarded.

5. The anti-deflection system for intelligent PCB solder mask exposure alignment according to claim 4, characterized in that: The process of obtaining the information coupling rate of similar information items by the bag-of-words analysis model includes: Performing text preprocessing on two information items in the same category of information items respectively, and then generating a number of text segmentations corresponding to the respective information items, setting a filtering segmentation set, inputting a number of text segmentations corresponding to the two information items into the filtering segmentation set for information filtering, marking the filtered text segmentations as irrelevant segmentations, and marking the retained text segmentations as relevant segmentations; Summarize all the relevant participles of two information items of the same type of information item, respectively, and then generate the corresponding word bags to be processed, perform Euclidean distance calculation on the word bags to be processed by the word band analysis model, and obtain the Euclidean distance between several relative relevant participles in the two word bags to be processed, record the logarithm of all relative relevant participles whose Euclidean distance is less than or equal to the preset distance threshold as β, and record the total logarithm of several relative relevant participles in the two word bags to be processed as α; It is then concluded that the information coupling rate of the current similar information items is β / α, where α and β are both natural numbers greater than 0.

6. The intelligent PCB board solder mask exposure alignment anti-deflection system according to claim 5, characterized in that: The process of analyzing each alignment image frame in turn and generating a corresponding alarm signal and sending it to the management center when the alignment deviation exceeds the preset deviation threshold is detected includes: All the alignment image frames that have been screened out and retained are analyzed in turn to obtain the alignment deviation between the PCB board and the light shielding plate in each alignment image frame. The alignment deviation includes a horizontal alignment deviation and a vertical alignment deviation. The deviation distances of the horizontal alignment deviation and the vertical alignment deviation are recorded as HJL and ZJL respectively. The deviation threshold includes a lateral deviation threshold and a longitudinal deviation threshold, which are denoted as P and Q respectively; When HJL≥P, the generated alarm message is: There is an abnormality in the horizontal alignment between the PCB board and the shading plate. Please adjust the horizontal alignment distance between the PCB board and the shading plate in time, generate the message tag Tag1, and bind the message tag Tag1 to the current alarm message; When ZJL≥Q, the generated alarm message is: there is an abnormality in the longitudinal alignment between the PCB board and the shading plate. Please adjust the longitudinal alignment distance between the PCB board and the shading plate in time, generate the message tag Tag2, and bind the message tag Tag2 to the current alarm message; When HJL<P, or ZJL<Q, no alarm message is generated; All alarm messages bound to message tags are aggregated, corresponding alarm signals are generated and sent to the management center.

7. The intelligent PCB board solder mask exposure alignment anti-deflection system according to claim 6, characterized in that: After receiving the alarm signal, the anti-deviation analysis module performs data analysis on the real-time image of the equipment monitoring end, and then generates the corresponding PCB board anti-deviation control parameter set. The process includes: The anti-bias analysis module consists of a data processing unit and a data analysis unit; When the data processing unit receives the alarm signal, it determines the type of the message tag bound to the alarm message included in the alarm signal. If the message tag is Tag1, the longitudinal area image of the aligned image frame retained after the screening is completed in the aligned live image is intercepted; if the message tag is Tag2, the horizontal area image of the corresponding aligned image frame is intercepted; The transverse region image and the longitudinal region image intercepted by the data processing unit are input into the data analysis unit, and the data analysis unit performs data analysis according to the transverse region image, thereby generating alignment control parameters for the alignment operation between the PCB board and the light shielding plate in the transverse direction, and then performs data analysis according to the longitudinal region image, thereby generating alignment control parameters for the alignment operation between the PCB board and the light shielding plate in the longitudinal direction; The alignment control parameters of the alignment operation in the transverse direction include a transverse correction distance, a transverse correction speed, and a transverse alignment permissible error, and the alignment control parameters of the alignment operation in the longitudinal direction include a longitudinal correction distance, a longitudinal correction speed, and a longitudinal alignment permissible error; All alignment control parameters are summarized to generate the corresponding PCB board anti-skew control parameter set.

8. The anti-deflection system for intelligent PCB solder mask exposure alignment according to claim 7, characterized in that: The process of the alignment control module performing alignment control on the PCB board on the current exposure machine according to the PCB board anti-skew control parameter set includes: The alignment control module receives the anti-deflection control parameter set of the PCB board. The alignment control module is provided with an instruction area. The instruction area inputs the anti-deflection control parameter set of the PCB board, and generates instruction codes for controlling the PCB board on the exposure machine to perform horizontal alignment or vertical alignment according to the anti-deflection control parameter set of the PCB board. The alignment control module controls the positioning pins on the working table of the exposure machine according to the instruction codes, and the positioning pins adjust the alignment position between the PCB board and the light shielding plate, and obtains the alignment coincidence rate between the PCB board and the light shielding plate in real time; When the alignment overlap rate is greater than or equal to 95%, the PCB board successfully completes the alignment with the shading plate; When the alignment overlap rate is less than 95%, the positioning pins continue to adjust the position of the PCB board relative to the light shielding plate until the alignment overlap rate meets the requirements, and then the positioning pins are stopped from being controlled.

Images