A PCB board separation machine intelligent control method and system

Abstract

The present application relates to the technical field of PCB panel separation, and provides an intelligent control method and system for a PCB panel separation machine. By continuously scanning the target PCB image, a set of PCB sub-images is obtained, and the PCB sub-image set is image-fused to generate a PCB fusion image; the PCB fusion image and the target panel separation information are input into the PCB panel separation numerical control model to obtain cutting features, and based on the cutting features, cutting trajectory parameters are generated to execute the panel separation of the target PCB to obtain the target PCB segmentation result. The method solves the technical problems in the prior art that PCB panel separation is highly dependent on manual experience, the operation automation level of the PCB panel separation machine is low, and the control stability of the PCB panel separation and the accuracy of the panel separation processing are low. The method realizes the technical effect of reducing the manual dependence of PCB B panel separation, automatically generating PCB cutting trajectories, and automatically performing PCB panel cutting by the PCB panel separation machine.

Description

Technical Field

[0001] The present application relates to the technical field of PCB board depaneling, and in particular to an intelligent control method and system for a PCB depaneling machine. Background Art

[0002] At present, the conventional method for completing PCB panel separation in China is to use CCD-assisted human eye teaching. However, the CCD teaching method has the defect of easily causing visual fatigue of the operator, resulting in misreading the cutting position points. At the same time, it takes a lot of programming time, which is not conducive to the rapid line change of the assembly line.

[0003] Although the existing PCB depaneling method is assisted by a PCB depaneling machine, it still relies heavily on manual experience to operate and control the PCB depaneling machine, and the actual processing quality control stability of PCB depaneling is relatively weak.

[0004] In summary, the existing technology has the following technical problems: PCB depaneling is highly dependent on manual experience, and the degree of automation of the PCB depaneling machine is low, resulting in low PCB depaneling control stability and low depaneling processing accuracy. Summary of the Invention

[0005] Based on this, it is necessary to address the above technical problems and provide an intelligent control method and system for a PCB depaneling machine that can reduce manual dependence on PCB depaneling, automatically generate PCB cutting trajectories, and automatically perform PCB panel cutting on the PCB depaneling machine.

[0006] A method for intelligently controlling a PCB depaneling machine comprises: obtaining a target PCB and target depaneling information; continuously scanning the target PCB using the visual hardware system to obtain a set of PCB sub-images, wherein each PCB sub-image in the set of PCB sub-images is marked with image acquisition coordinates; performing image fusion on the set of PCB sub-images to generate a PCB fused image; inputting the PCB fused image and the target depaneling information into a pre-built PCB depaneling numerical control model to obtain cutting features, wherein the cutting features include features of positions to be cut and feature points to be cut; generating cutting trajectory parameters based on the cutting features; and sending the cutting trajectory parameters to the PCB depaneling machine to execute depaneling of the target PCB to obtain a target PCB segmentation result, wherein the target PCB segmentation result is a plurality of target PCB single boards.

[0007] An intelligent control system for a PCB depaneling machine, the system comprising: a target information acquisition module for obtaining target PCB and target depaneling information; an image scanning execution module for continuously scanning the target PCB through a visual hardware system to obtain a set of PCB sub-images, wherein each PCB sub-image in the set of PCB sub-images is marked with image acquisition coordinates; an image fusion execution module for performing image fusion on the set of PCB sub-images to generate a PCB fusion image; a cutting feature acquisition module for inputting the PCB fusion image and the target depaneling information into a pre-built PCB depaneling numerical control model to obtain cutting features, wherein the cutting features include to-be-cut position features and to-be-cut feature points; a cutting trajectory generation module for generating cutting trajectory parameters based on the cutting features; and a depaneling cutting execution module for sending the cutting trajectory parameters to the PCB depaneling machine to execute depaneling of the target PCB and obtain a target PCB segmentation result, wherein the target PCB segmentation result is a plurality of target PCB single boards.

[0008] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0009] Get target PCB and target sub-board information;

[0010] Continuously scanning the target PCB through the visual hardware system to obtain a set of PCB sub-images, wherein each PCB sub-image in the set of PCB sub-images is marked with image acquisition coordinates;

[0011] Performing image fusion on the PCB sub-image set to generate a PCB fused image;

[0012] Inputting the PCB fusion image and the target panel separation information into a pre-built PCB panel separation numerical control model to obtain cutting features, wherein the cutting features include to-be-cut position features and to-be-cut feature points;

[0013] generating cutting trajectory parameters based on the cutting features;

[0014] The cutting trajectory parameters are sent to a PCB depaneling machine to perform depaneling of the target PCB, and a target PCB segmentation result is obtained, where the target PCB segmentation result is a plurality of target PCB boards.

[0015] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:

[0016] Get target PCB and target sub-board information;

[0017] Continuously scanning the target PCB through the visual hardware system to obtain a set of PCB sub-images, wherein each PCB sub-image in the set of PCB sub-images is marked with image acquisition coordinates;

[0018] Performing image fusion on the PCB sub-image set to generate a PCB fused image;

[0019] Inputting the PCB fusion image and the target panel separation information into a pre-built PCB panel separation numerical control model to obtain cutting features, wherein the cutting features include to-be-cut position features and to-be-cut feature points;

[0020] generating cutting trajectory parameters based on the cutting features;

[0021] The cutting trajectory parameters are sent to a PCB depaneling machine to perform depaneling of the target PCB, and a target PCB segmentation result is obtained, where the target PCB segmentation result is a plurality of target PCB boards.

[0022] The above-mentioned intelligent control method and system of a PCB panel separator solves the technical problems in the prior art that PCB panel separation is highly dependent on manual experience, the PCB panel separator has a low degree of automation in operation, and thus leads to low PCB panel separation control stability and panel separation processing accuracy. The technical effects of reducing manual dependence on PCB panel separation, automatically generating PCB cutting trajectories, and automatically performing PCB panel cutting by the PCB panel separator are achieved.

[0023] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A schematic flow chart of an intelligent control method for a PCB depaneling machine according to an embodiment;

[0025] Figure 2 A schematic diagram of a process for generating a PCB fusion image in an intelligent control method for a PCB depaneling machine according to an embodiment;

[0026] Figure 3 This is a structural block diagram of an intelligent control system for a PCB board separator in one embodiment;

[0027] Figure 4 is a diagram of the internal structure of a computer device in one embodiment;

[0028] Explanation of the reference numerals: target information acquisition module 1, image scanning execution module 2, image fusion execution module 3, cutting feature acquisition module 4, cutting trajectory generation module 5, panel cutting execution module 6. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0030] like Figure 1 As shown, the present application provides an intelligent control method for a PCB depaneling machine, which is applied to an intelligent control system of a PCB depaneling machine, wherein the system is communicatively connected with a visual hardware system, and the method includes:

[0031] S100: Obtain target PCB and target sub-board information;

[0032] Specifically, it should be understood that in order to improve the manufacturing efficiency of printed circuit boards (PCBs) and reduce the production costs of printed circuit boards, it is usually adopted to produce printed circuit boards by merging multiple printed circuit board boards into a single board to obtain PCB panels, and the PCB panels are separated by a PCB panel separator, so that the PCB panels are divided into multiple PCB boards.

[0033] In this embodiment, the target PCB is a PCB panel that has not been processed by a PCB depaneling machine. The target panel information is information about one or more PCB designs to be produced, including but not limited to PCB dimensions, component layout, and routing information. Based on this target panel information, the PCB outline information can be obtained, allowing the PCB panel segmentation trajectory to be planned and the target PCB to be depaneled.

[0034] S200: Continuously scanning the target PCB through the visual hardware system to obtain a set of PCB sub-images, wherein each PCB sub-image in the set of PCB sub-images is marked with image acquisition coordinates;

[0035] Specifically, in this embodiment, the visual hardware system preferably uses a CCD camera as an image acquisition component and an LED as a high-resolution image acquisition device for fill lighting during the image acquisition process. The visual hardware system is installed on the head of the PCB depaneling machine to achieve image acquisition of the target PCB when the target PCB is placed within the working range of the PCB depaneling machine.

[0036] An XYZ Cartesian mechanical coordinate system is pre-constructed in the operating space of the visual hardware system to enable the visual hardware system to perform effective data information recording of the target PCB image acquisition motion trajectory.

[0037] Specifically, the target PCB is continuously scanned by the visual hardware system using a circular array scan to obtain a set of PCB sub-images. Each single PCB sub-image in the set of PCB sub-images records a local image of the target PCB area. In theory, the complete target PCB image can be obtained by splicing the various PCB sub-images in the set of PCB sub-images.

[0038] Since the motion trajectory of the visual hardware system is accurately recorded by the XYZ Cartesian mechanical coordinate system, the multiple PCB sub-images obtained by scanning the target PCB image based on the visual hardware system all have (X, Y, Z) three-dimensional coordinate identifications. At the same time, since the target PCB is placed on the horizontal plane of the PCB depaneling machine, the visual hardware system moves horizontally during image scanning, and the three-dimensional coordinate identifications of the multiple PCB sub-images can be simplified to (X, Y) two-dimensional coordinate identifications.

[0039] Based on the visual hardware system, a circular array scanning method is adopted to acquire the target PCB image, providing image splicing raw materials for the subsequent acquisition of a complete high-resolution target PCB image.

[0040] S300: performing image fusion on the PCB sub-image set to generate a PCB fused image;

[0041] In one embodiment, Figure 2 As shown, the image fusion is performed on the PCB sub-image set to generate a PCB fused image. The method step S300 provided in this application further includes:

[0042] S310: extracting and obtaining a plurality of PCB sub-images and a plurality of image acquisition coordinates based on the PCB sub-image set, wherein the plurality of PCB sub-images and the plurality of image acquisition coordinates have a corresponding relationship;

[0043] S320: performing image stitching position analysis according to the multiple image acquisition coordinates to obtain an image stitching structure;

[0044] S330: performing stitching of the multiple PCB sub-images based on the image stitching structure to obtain a PCB sub-image stitching result;

[0045] S340: Perform fusion rendering on the stitching results of the PCB sub-images to generate the PCB fused image.

[0046] Specifically, based on step S200, it can be known that in the PCB sub-image set obtained based on the visual hardware system, each PCB sub-image has an image acquisition position coordinate identifier. Therefore, in this embodiment, multiple PCB sub-images and multiple image acquisition coordinates with identifier mappings to the multiple PCB sub-images are extracted based on the PCB sub-image set.

[0047] An image stitching position analysis is performed based on the X-axis and Y-axis numerical relationship of the multiple image acquisition coordinates to determine the approximate position of the image content contained in the PCB sub-image on the target PCB surface, and obtain an image stitching structure. The image stitching structure is a mapping relationship between the image content contained in the multiple PCB sub-images and the actual local image position of the target PCB.

[0048] Based on the image stitching structure, the multiple PCB sub-images are stitched according to the multiple image acquisition coordinates to obtain a PCB sub-image stitching result, in which adjacent PCB sub-images have repeated image content.

[0049] Based on the PCB sub-image stitching results, fusion rendering of adjacent PCB sub-images is performed until the fusion rendering of adjacent stitching positions of all PCB sub-images in the PCB sub-image set is completed to obtain the PCB fusion image, which is a high-resolution complete target PCB image that can truly reflect the overall state of the target PCB.

[0050] This embodiment acquires multiple PCB sub-images containing partial images of the target PCB and performs stitching and fusion of the PCB sub-images, thereby obtaining a PCB fusion image that can accurately represent the complete state of the target PCB, achieving the technical effect of providing a reference image for the subsequent intelligent generation of the target PCB panel depaneling trajectory.

[0051] S400: Inputting the PCB fusion image and the target panel separation information into a pre-built PCB panel separation numerical control model to obtain cutting features, wherein the cutting features include to-be-cut position features and to-be-cut feature points;

[0052] In one embodiment, the PCB fusion image and the target panel separation information are input into a pre-built PCB panel separation numerical control model to obtain cutting features. The method step S400 provided in this application further includes:

[0053] S410: Acquire sample panel images, sample panel information, sample cutting position features, and sample cutting feature points of multiple sample PCB panels to obtain a sample panel image set, a sample panel information set, a sample cutting position feature set, and a sample cutting feature point set;

[0054] S420: Obtaining a sample cutting feature set based on the sample cutting position feature set and the sample cutting feature point set;

[0055] S430: Constructing the PCB panel depaneling numerical control model using the sample panel image set, the sample panel depaneling information set, and the sample cutting feature set;

[0056] S440: Inputting the PCB fusion image and the target panel separation information into the PCB panel separation NC model to obtain the cutting features.

[0057] In one embodiment, the PCB panel depaneling numerical control model is constructed using the sample panel image set, the sample panel depaneling information set, and the sample cutting feature set. The method step S430 provided in this application further includes:

[0058] S431: performing data identification and division on the sample panel image set, the sample panel information set, and the sample cutting feature set to obtain a training set, a validation set, and a test set;

[0059] S432: Constructing the PCB panel depaneling numerical control model based on a deep convolutional neural network;

[0060] S433: Presetting an output accuracy threshold of the PCB panel depaneling numerical control model;

[0061] S434: Perform iterative supervised training, verification, and testing of the PCB depaneling numerical control model based on the training set, the validation set, and the test set, and determine whether the output accuracy of the PCB depaneling numerical control model falls within the output accuracy threshold;

[0062] S435: If the output accuracy of the PCB panel decomposition numerical control model falls within the output accuracy threshold, the training of the PCB panel decomposition numerical control model is stopped.

[0063] Specifically, it should be understood that when the existing technology is used for PCB panel production, the panels generally use V-Cut or stamp holes to connect the designed PCB single boards. Therefore, when the PCB panel splitter is used to cut the PCB panel, the position of the panel connection is determined and the cutting track is generated for the PCB panel cutting process to obtain multiple PCB single boards.

[0064] In order to reduce the dependence of manual experience in identifying PCB panel connection sites and producing cutting curves, this embodiment replaces the method based on manual acquisition of PCB panel cutting trajectories with a data processing model. Specifically, this embodiment constructs a PCB panelization CNC model to identify the position feature area for PCB single board cutting in the target PCB panel and the specific cutting feature points within the position feature area.

[0065] The PCB panel depaneling numerical control model is constructed based on a deep convolutional neural network. The input data is the design information of the PCB panel and the multiple PCB single boards that make up the PCB panel. The output result is the cutting position feature area identifier of the PCB panel and the cutting feature point identifier of the specific cutting within the corresponding cutting position feature.

[0066] Based on big data or historical order-taking and plate-making data of PCB panel manufacturers, sample panel images, sample panel division information, sample cutting position features and sample cutting feature points of multiple sample PCB panels are collected to obtain a sample panel image set, a sample panel division information set, a sample cutting position feature set and a sample cutting feature point set.

[0067] This embodiment uses a single sample PCB panel to interpret the sample panel image, sample panel information, sample cutting position features, and sample cutting feature points. The sample panel image is a high-resolution complete image of the sample PCB panel, corresponding to the PCB fusion image. The sample panel information is the design information of multiple PCB single boards actually existing in the sample PCB panel. The sample cutting position feature is the sample PCB board cutting area position identifier defined by the historical PCB panel separation machine operator when performing the sample PCB panel cutting process according to the sample panel information. The sample cutting feature point is the specific cutting site identifier at the sample PCB board cutting area position defined when the historical PCB panel separation machine operator performed the sample PCB panel cutting process according to the sample panel information.

[0068] It should be understood that after the sample PCB panel is identified and the sample cutting feature position is obtained, the sample cutting feature point is accurately positioned, and after multiple sample cutting feature points are connected to generate a sample PCB panel cutting trajectory, the PCB panel cutting machine cuts the sample PCB panel with reference to the sample PCB panel cutting trajectory, and the sample PCB panel can be cut into multiple PCB boards that are consistent with the sample panel information.

[0069] In this embodiment, a PCB depaneling NC model is constructed based on a deep convolutional neural network, and a preset output accuracy threshold for the PCB depaneling NC model is used to determine if the PCB depaneling NC model training has passed. Based on the stringent requirements for PCB depaneling, the output accuracy threshold is preferably set at 99%. The output accuracy of the PCB depaneling NC model is determined by comparing the overall identification image similarity of the PCB cutting feature identifiers actually output by the PCB depaneling NC model with those of corresponding sample cutting feature identifiers.

[0070] The sample panel image set, the sample panel decomposition information set, and the sample cutting feature set are identified and divided according to an 8:1:1 data volume division method to obtain a training set, a validation set, and a test set; iterative supervised training, validation, and testing of the PCB panel decomposition numerical control model are performed based on the training set, validation set, and test set to determine whether the output accuracy of the PCB panel decomposition numerical control model falls within the output accuracy threshold; if the output accuracy of the PCB panel decomposition numerical control model falls within the output accuracy threshold, the training of the PCB panel decomposition numerical control model is stopped.

[0071] The PCB fusion image and the target panel separation information are input into a trained PCB panel separation numerical control model, and the PCB panel separation numerical control model outputs the cutting features, where the cutting features include features of the position to be cut and feature points to be cut. The feature position to be cut is a feature of the target PCB cutting position area marked on the PCB fusion image, and the feature point to be cut is a feature site marked within the position area to be cut for performing a specific cutting process.

[0072] This embodiment constructs a PCB panel separation numerical control model to identify the PCB panel cutting position and cutting feature points, thereby replacing the manual generation of PCB panel segmentation trajectories, achieving the technical effect of intelligently and stably producing PCB panel segmentation feature positions and segmentation feature point acquisition, and reducing the manual dependence of PCB panel segmentation.

[0073] S500: Generate cutting trajectory parameters based on the cutting features;

[0074] S600: Send the cutting trajectory parameters to a PCB depaneling machine to perform depaneling of the target PCB, and obtain a target PCB segmentation result, where the target PCB segmentation result is a plurality of target PCB boards.

[0075] Specifically, in this embodiment, the cutting features include the to-be-cut position features and the to-be-cut feature points. The local cutting trajectory parameters of the area corresponding to the cutting position features are generated by connecting the multiple to-be-cut feature points within each cutting position feature. The local cutting trajectory parameters are the cutting trajectory information for PCB panel cutting in the area corresponding to the cutting position features in the PCB fusion image.

[0076] The local cutting trajectory parameters within each cutting position feature are generated based on the cutting feature, and the generated multiple local cutting trajectory parameters are combined with the PCB fusion image to perform cutting trajectory connection to obtain the cutting trajectory parameters.

[0077] The cutting trajectory parameters are sent to a PCB depaneling machine, which performs depaneling of the target PCB based on the cutting trajectory parameters to obtain a target PCB segmentation result, where the target PCB segmentation result is a plurality of target PCB boards.

[0078] This embodiment performs target PCB cutting based on cutting trajectory parameters, thereby safely cutting the target PCB into multiple target PCB single boards that are consistent with the target panel information, achieving the technical effect of automatically generating PCB panel cutting trajectories and automatically performing PCB panel cutting by a PCB panel separator.

[0079] In one embodiment, the cutting trajectory parameters are sent to a PCB depaneling machine to perform depaneling of the target PCB to obtain a target PCB segmentation result. Subsequently, the method steps provided in this application further include:

[0080] S710: performing image acquisition on the target PCB segmentation result based on the visual hardware system to obtain a PCB single board image set;

[0081] S720: Collect and obtain a PCB defect feature set;

[0082] S730: Obtaining a PCB qualified detection feature set based on the target panel information;

[0083] S740: Constructing a PCB single board defect detection model based on the PCB defect feature set and the PCB qualified detection feature set;

[0084] S750: Inputting the PCB single board image set into the PCB single board defect detection model to obtain a PCB single board defect detection result;

[0085] S760: Optimize the PCB panel depaneling numerical control model based on the PCB single board defect detection result.

[0086] In one embodiment, the PCB single board defect detection model is constructed based on the PCB defect feature set and the PCB qualified detection feature set. Step S740 of the method provided in this application further includes:

[0087] S741: Constructing a quality control defect identification submodule based on the PCB defect feature set;

[0088] S742: Constructing a production defect identification submodule based on the PCB qualified detection feature set;

[0089] S743: Construct the PCB single board defect detection model based on the quality control defect identification submodule and the production defect identification submodule.

[0090] Specifically, in this embodiment, after the PCB splitting machine splits the target PCB based on the cutting trajectory parameters, a target PCB segmentation result consisting of a plurality of target PCB boards is obtained.

[0091] Based on the visual hardware system installed on the head of the PCB depaneling machine, image acquisition is performed on the target PCB segmentation result to obtain a PCB single board image set, and the PCB single board image set includes images of multiple target PCB single boards. The purpose of performing the secondary image acquisition of the target PCB segmentation result based on the visual hardware system in this embodiment is to determine whether the multiple target PCB single boards actually obtained meet the target depaneling information requirements, and to determine whether the multiple target PCB single boards have production control defects that are unusable and have existed in the production process of the target PCBs.

[0092] Based on big data or historical order and plate-making data of PCB panel manufacturers, a set of PCB defect features is collected and acquired. The PCB defect features are common defects in PCB production, including but not limited to nail head defects, resin contamination defects, burr defects, and tumor defects. The PCB defect feature set is composed of multiple types of production control defect feature images.

[0093] A quality control defect recognition submodule is constructed based on the PCB defect feature set. The quality control defect recognition submodule is used to determine whether any target PCB board has a production quality control defect based on an image defect feature comparison and recognition method.

[0094] The target sub-panel information is information about the design of one or more PCBs to be produced, specifically including but not limited to PCB dimensional data, component layout, and routing information. A set of qualified PCB inspection features is obtained based on the target sub-panel information. The set of qualified PCB inspection features includes one or more PCB outline dimensional features and PCB routing features.

[0095] A production defect recognition submodule is constructed based on the PCB qualified inspection feature set. The production defect recognition submodule is used to determine whether any target PCB board has production quality defects that do not meet the requirements of the target panel information based on image defect feature comparison and recognition. The production quality defects include PCB board outline size defects caused by excessive PCB panel cutting and PCB board routing deviation defects caused by inconsistency between PCB board routing and PCB board routing design in the target panel information.

[0096] An image recognition layer of the PCB single board defect detection model is constructed based on the quality control defect recognition submodule and the production defect recognition submodule. In the image recognition layer, the quality control defect recognition submodule and the production defect recognition submodule are set in parallel, and the overall construction of the PCB single board defect detection model is completed in combination with the input layer and the output layer.

[0097] The PCB single board image set is input into the PCB single board defect detection model via the input layer, and defect identification traversal of the PCB single board image set is performed one by one synchronously through the quality control defect identification submodule and the production defect identification submodule.

[0098] The quality control defect identification submodule outputs a production quality defect identifier for each PCB board image in the PCB board image set, wherein the production quality control defect identifier includes, but is not limited to, common defects such as resin contamination defects, burr defects, and nodular defects. The production defect identification submodule outputs a production quality defect identifier for each PCB board image in the PCB board image set, wherein the production quality defect identifier includes a PCB board outline dimension defect and a PCB board routing deviation defect.

[0099] The production quality defect identifier and the production quality control defect identifier constitute the PCB single board defect detection result. Based on the PCB single board defect detection result, the PCB single board contour size defect is extracted. Based on the PCB single board contour size defect, the PCB panel separation numerical control model is optimized so that the cutting trajectory parameters generated by the cutting features output by the PCB panel separation numerical control model do not cause over-cutting of the target PCB, thereby achieving the technical effect of reducing the PCB single board cutting scrap rate of the target PCB cut by the PCB panel separation machine.

[0100] In one embodiment, before optimizing the PCB panel depaneling numerical control model based on the PCB single board defect detection result, the method steps provided in this application further include:

[0101] S751: Determine whether the PCB single board defect detection result is a PCB sub-board control defect;

[0102] S752: If the PCB single board defect detection result does not belong to the PCB panel control defect, a PCB panel production warning reminder is generated;

[0103] S753: If the PCB single board defect detection result is a PCB panel separation control defect, generating a PCB image cutting optimization feature based on the PCB single board defect detection result;

[0104] S764: Optimize the PCB panel depaneling NC model based on the PCB image cutting optimization feature.

[0105] Specifically, in this embodiment, the PCB single board defect detection result may contain both or either the production quality defect identifier and the production quality defect identifier. A determination is made as to whether the PCB single board defect detection result is a PCB panelization control defect. If the production quality defect identifier is present in the PCB single board defect detection result, indicating that the PCB single board defect detection result is not a PCB panelization control defect, a PCB panelization production warning reminder is generated. Based on the warning reminder, production control is performed at the target PCB production stage to reduce common defects during PCB production.

[0106] When the production control defect identifier exists in the PCB single board defect detection result, the PCB single board defect detection result belongs to a PCB panel separation control defect. Based on the PCB single board defect detection result, the PCB single board contour size defect is extracted, and a PCB image cutting optimization feature is generated based on the PCB single board contour size defect. The PCB panel separation CNC model is optimized based on the PCB image cutting optimization feature, thereby achieving the technical effect of improving the panel separation cutting processing accuracy during the panel separation processing of the PCB panel separation machine and reducing the PCB single board scrap rate caused by the PCB panel separation machine cutting.

[0107] In one embodiment, Figure 3 As shown, an intelligent control system for a PCB depaneling machine is provided, comprising: a target information acquisition module 1, an image scanning execution module 2, an image fusion execution module 3, a cutting feature acquisition module 4, a cutting trajectory generation module 5, and a depaneling and cutting execution module 6, wherein:

[0108] Target information acquisition module 1, used to obtain target PCB and target sub-board information;

[0109] An image scanning execution module 2 is configured to continuously scan the target PCB through a visual hardware system to obtain a set of PCB sub-images, wherein each PCB sub-image in the set of PCB sub-images is marked with image acquisition coordinates;

[0110] An image fusion execution module 3 is configured to perform image fusion on the PCB sub-image set to generate a PCB fused image;

[0111] A cutting feature acquisition module 4 is configured to input the PCB fusion image and the target panel separation information into a pre-built PCB panel separation numerical control model to obtain cutting features, wherein the cutting features include the features of the position to be cut and the feature points to be cut;

[0112] A cutting trajectory generating module 5, configured to generate cutting trajectory parameters based on the cutting features;

[0113] The panel cutting execution module 6 is used to send the cutting trajectory parameters to the PCB panel cutting machine to execute panel cutting of the target PCB and obtain the target PCB segmentation result, which is a plurality of target PCB boards.

[0114] In one embodiment, the system provided by the present application further includes:

[0115] A single board image acquisition unit, configured to acquire an image set of the target PCB segmentation result based on the visual hardware system;

[0116] Production defect collection unit, used to collect and obtain PCB defect feature sets;

[0117] A quality control defect collection unit, configured to obtain a PCB qualified detection feature set based on the target sub-board information;

[0118] A detection model building unit, configured to build a PCB single board defect detection model based on the PCB defect feature set and the PCB qualified detection feature set;

[0119] a detection result output unit, configured to input the PCB single board image set into the PCB single board defect detection model to obtain a PCB single board defect detection result;

[0120] The model optimization execution unit is used to optimize the PCB panel depaneling numerical control model based on the PCB single board defect detection result.

[0121] In one embodiment, the detection model building unit further includes:

[0122] A quality control module construction unit, configured to construct a quality control defect identification submodule based on the PCB defect feature set;

[0123] A production module construction unit, configured to construct a production defect identification submodule based on the PCB qualified detection feature set;

[0124] A model combination generation unit is used to build the PCB single board defect detection model based on the quality control defect identification submodule and the production defect identification submodule.

[0125] In one embodiment, the model optimization execution unit further includes:

[0126] A defect type determination unit, configured to determine whether the PCB single board defect detection result is a PCB sub-board control defect;

[0127] An early warning reminder generating unit is used to generate a PCB panel production early warning reminder if the PCB single board defect detection result does not belong to the PCB panel control defect;

[0128] an optimization feature generating unit, configured to generate a PCB image cutting optimization feature based on the PCB single board defect detection result if the PCB single board defect detection result is a PCB panel separation control defect;

[0129] A model optimization execution unit is used to optimize the PCB panel cutting numerical control model based on the PCB image cutting optimization feature.

[0130] In one embodiment, the image fusion execution module 3 further includes:

[0131] An image information extraction unit is configured to extract and obtain a plurality of PCB sub-images and a plurality of image acquisition coordinates based on the PCB sub-image set, wherein the plurality of PCB sub-images and the plurality of image acquisition coordinates have a corresponding relationship;

[0132] a stitching structure construction unit, configured to perform image stitching position analysis based on the plurality of image acquisition coordinates to obtain an image stitching structure;

[0133] a stitching result obtaining unit, configured to stitch the plurality of PCB sub-images based on the image stitching structure to obtain a PCB sub-image stitching result;

[0134] The fused image obtaining unit is used to perform fusion rendering on the stitching results of the PCB sub-images to generate the PCB fused image.

[0135] In one embodiment, the cutting feature acquisition module 4 further includes:

[0136] A sample data acquisition unit is used to acquire sample panel images, sample panel information, sample cutting position features, and sample cutting feature points of multiple sample PCB panels, and obtain a sample panel image set, a sample panel information set, a sample cutting position feature set, and a sample cutting feature point set;

[0137] A sample feature obtaining unit, configured to obtain a sample cutting feature set based on the sample cutting position feature set and the sample cutting feature point set;

[0138] A numerical control model construction unit, configured to construct the PCB panel separation numerical control model using the sample panel image set, the sample panel separation information set, and the sample cutting feature set;

[0139] The cutting feature obtaining unit is used to input the PCB fusion image and the target panel separation information into the PCB panel separation numerical control model to obtain the cutting feature.

[0140] In one embodiment, the numerical control model building unit further includes:

[0141] A sample data identification unit is used to identify and divide the sample panel image set, the sample panel information set and the sample cutting feature set to obtain a training set, a validation set and a test set;

[0142] A model building execution unit, configured to build the PCB panel depaneling numerical control model based on a deep convolutional neural network;

[0143] An output information setting unit, used to preset an output accuracy threshold of the PCB depaneling numerical control model;

[0144] a model training execution unit, configured to perform iterative supervised training, verification, and testing of the PCB depaneling numerical control model based on the training set, the verification set, and the test set, and determine whether an output accuracy of the PCB depaneling numerical control model falls within an output accuracy threshold;

[0145] The numerical control model generating unit is used to stop the training of the PCB panel separation numerical control model if the output accuracy of the PCB panel separation numerical control model falls within the output accuracy threshold.

[0146] Specific embodiments of an intelligent control system for a PCB depaneling machine can be found in the aforementioned embodiments of an intelligent control method for a PCB depaneling machine, and will not be further elaborated here. Each module in the aforementioned intelligent control device for a PCB depaneling machine can be implemented in whole or in part through software, hardware, or a combination thereof. Each of these modules can be embedded in or independent of a processor within a computer device in hardware form, or stored in a computer device memory in software form, allowing the processor to call and execute the corresponding operations of each module.

[0147] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, a memory and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store news data and data such as time attenuation factors. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a method for intelligent control of a PCB board splitter is realized.

[0148] Those skilled in the art will understand that Figure 4The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0149] In one embodiment, a computer device is provided, comprising a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the following steps when executing the computer program: obtaining target PCB and target panel information; continuously scanning the target PCB through the visual hardware system to obtain a set of PCB sub-images, wherein each PCB sub-image in the set of PCB sub-images is marked with image acquisition coordinates; performing image fusion on the set of PCB sub-images to generate a PCB fused image; inputting the PCB fused image and the target panel information into a pre-built PCB panel depaneling numerical control model to obtain cutting features, wherein the cutting features include position features to be cut and feature points to be cut; generating cutting trajectory parameters based on the cutting features; and sending the cutting trajectory parameters to a PCB panel depaneling machine to execute panel cutting of the target PCB to obtain a target PCB segmentation result, wherein the target PCB segmentation result is a plurality of target PCB single boards.

[0150] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0151] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A PCB board separation machine intelligent control method, characterized in that: The method is applied to an intelligent control system of a PCB depaneling machine, wherein the system is communicatively connected with a visual hardware system, and the method comprises: Get target PCB and target sub-board information; Continuously scanning the target PCB through the visual hardware system to obtain a set of PCB sub-images, wherein each PCB sub-image in the set of PCB sub-images is marked with image acquisition coordinates; Performing image fusion on the PCB sub-image set to generate a PCB fused image; Inputting the PCB fusion image and the target panel separation information into a pre-built PCB panel separation numerical control model to obtain cutting features, wherein the cutting features include to-be-cut position features and to-be-cut feature points; generating cutting trajectory parameters based on the cutting features; Sending the cutting trajectory parameters to a PCB depaneling machine to perform depaneling of the target PCB, and obtaining a target PCB segmentation result, wherein the target PCB segmentation result is a plurality of target PCB boards; After sending the cutting trajectory parameters to a PCB depaneling machine to perform depaneling of the target PCB and obtain a target PCB segmentation result, the method further comprises: Performing image acquisition on the target PCB segmentation result based on the visual hardware system to obtain a PCB single board image set; Collect and obtain PCB defect feature set; Obtaining a PCB qualified detection feature set based on the target sub-board information; Constructing a PCB single board defect detection model based on the PCB defect feature set and the PCB qualified detection feature set; Inputting the PCB single board image set into the PCB single board defect detection model to obtain a PCB single board defect detection result; Optimizing the PCB depaneling numerical control model based on the PCB single board defect detection result; The method further comprises: constructing a PCB single board defect detection model based on the PCB defect feature set and the PCB qualified detection feature set; Building a quality control defect identification submodule based on the PCB defect feature set; Constructing a production defect identification submodule based on the PCB qualified detection feature set; Constructing the PCB single board defect detection model based on the quality control defect identification submodule and the production defect identification submodule; Before optimizing the PCB panel separation numerical control model based on the PCB single board defect detection result, the method further includes: Determine whether the PCB single board defect detection result is a PCB sub-board control defect; If the PCB single board defect detection result does not belong to the PCB panel control defect, a PCB panel production warning reminder is generated; If the PCB single board defect detection result is a PCB panel separation control defect, generating a PCB image cutting optimization feature based on the PCB single board defect detection result; The PCB panel cutting numerical control model is optimized based on the PCB image cutting optimization feature.

2. The method according to claim 1, wherein The method further comprises: performing image fusion on the PCB sub-image set to generate a PCB fused image; Extracting and obtaining a plurality of PCB sub-images and a plurality of image acquisition coordinates based on the PCB sub-image set, wherein the plurality of PCB sub-images and the plurality of image acquisition coordinates have a corresponding relationship; Performing image stitching position analysis according to the multiple image acquisition coordinates to obtain an image stitching structure; performing stitching of the plurality of PCB sub-images based on the image stitching structure to obtain a PCB sub-image stitching result; The PCB sub-image splicing results are fused and rendered to generate the PCB fused image.

3. The method according to claim 1, wherein The method further comprises inputting the PCB fusion image and the target panel separation information into a pre-built PCB panel separation numerical control model to obtain cutting features: Collect and obtain sample panel images, sample panel information, sample cutting position features, and sample cutting feature points of multiple sample PCB panels, and obtain a sample panel image set, a sample panel information set, a sample cutting position feature set, and a sample cutting feature point set; Obtaining a sample cutting feature set based on the sample cutting position feature set and the sample cutting feature point set; Constructing the PCB panel depaneling numerical control model using the sample panel image set, the sample panel depaneling information set, and the sample cutting feature set; The PCB fusion image and the target panel separation information are input into the PCB panel separation numerical control model to obtain the cutting features.

4. The method according to claim 3, wherein The method further comprises: constructing the PCB panel depaneling numerical control model by using the sample panel image set, the sample panel depaneling information set, and the sample cutting feature set; Performing data identification and division on the sample panel image set, the sample panel information set, and the sample cutting feature set to obtain a training set, a validation set, and a test set; Based on a deep convolutional neural network, the PCB panel depaneling numerical control model is constructed; Presetting an output accuracy threshold of the PCB panel depaneling numerical control model; Performing iterative supervised training, verification, and testing of the PCB depaneling numerical control model based on the training set, the validation set, and the test set, and determining whether an output accuracy of the PCB depaneling numerical control model falls within an output accuracy threshold; If the output accuracy of the PCB panel decomposition numerical control model falls within the output accuracy threshold, the training of the PCB panel decomposition numerical control model is stopped.

5. An intelligent control system for a PCB board splitter, characterized in that: The system is used to perform the method according to claims 1 to 4, comprising: Target information acquisition module, used to obtain target PCB and target sub-board information; An image scanning execution module is used to continuously scan the target PCB through a visual hardware system to obtain a set of PCB sub-images, each of which is marked with image acquisition coordinates; An image fusion execution module, configured to perform image fusion on the PCB sub-image set to generate a PCB fused image; A cutting feature acquisition module is used to input the PCB fusion image and the target panel separation information into a pre-built PCB panel separation numerical control model to obtain cutting features, wherein the cutting features include the to-be-cut position features and the to-be-cut feature points; A cutting trajectory generating module, configured to generate cutting trajectory parameters based on the cutting features; The panel cutting execution module is used to send the cutting trajectory parameters to the PCB panel cutting machine to execute panel cutting of the target PCB and obtain the target PCB segmentation result, which is a plurality of target PCB boards.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images