Cutting path planning method and cutting path planning system for PCB (Printed Circuit Board)

Abstract

The invention discloses a cutting path planning method and a cutting path planning system for a PCB (Printed Circuit Board), and relates to the technical field of PCB manufacturing. Image information of the PCB is acquired through a visual identification system and is transmitted to an AI image identification module, an AI algorithm is used for carrying out image identification and intelligent marking on connecting ribs, and a graph is fed to a model for training after confirmation; the method comprises the steps of image recognition and automatic path generation, automatic generation of an optimized cutting path, automatic cutting path issuing and execution module, automatic transmission of recognized path data to a board separator execution unit by a system, no manual debugging is needed, rapid recognition and path planning can be carried out on PCBs of various specifications and various shapes, and the production efficiency is improved. Intelligent switching, real-time correction and a self-learning mechanism of batch processing of different products are supported, the system has error feedback and model self-learning capabilities, and the recognition precision and efficiency are improved along with the use time. The whole system realizes closed-loop control, so that the manual intervention time is greatly reduced, and the system intelligence is improved.

Description

Technical Field

[0001] This invention relates to the field of PCB manufacturing technology, and in particular to a PCB cutting path planning method and a cutting path planning system. Background Technology

[0002] With the rapid development of the electronics industry, printed circuit boards (PCBs) are core components of electronic products, and their processing technology is gradually becoming more automated and precise. PCB splitting machines, as key equipment in the PCB manufacturing process, are mainly used to precisely cut and separate multiple PCB boards according to a preset path.

[0003] Traditional PCB depaneling machines typically employ mechanical blades or laser cutting systems, relying on manual programming to set the cutting path. In practical applications, when cutting PCBs of different specifications, shapes, or layouts, the different positions and arrangements of the connecting ribs necessitate reprogramming and resetting the cutting path to achieve accurate rib cutting. This type of cutting system or cutting path setting method suffers from the following technical drawbacks:

[0004] First, each time a new PCB board is replaced, the cutting path needs to be reset. This relies on manual judgment based on experience to determine a suitable preset cutting path, or on manual design of a preset cutting path based on operational experience. This not only requires high computer skills from the operators, but also results in long adjustment cycles and low efficiency. Furthermore, since the positions and arrangements of the connecting ribs of PCB boards with different specifications, shapes, or layouts can be combined in many ways, it can lead to complex cutting path designs. This makes it easy for errors to be set in the path setting process during manual programming, which can easily damage or even scrap the PCB board. Second, this type of traditional board separation and cutting system has a limited number of programmed path schemes. When faced with irregularly shaped or arranged PCB boards, the processing capacity is insufficient. In particular, with the rise of small-batch, multi-variety production models, the low flexibility of traditional board separation systems has become a production bottleneck. Summary of the Invention

[0005] The purpose of this invention is to address the problem that existing PCB board cutting methods or systems rely on manual programming and debugging based on human experience to plan the cutting path, which can easily lead to errors in path setting and a high error rate. The invention provides a PCB board cutting path planning method and system.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] A method for planning the cutting path of a PCB board includes the following steps:

[0008] S1. Collect sample PCB boards with various cutting conditions, and set the first cutting mark at the connecting rib or pre-reserved position of the PCB board; the cutting conditions include at least one of the following: lighting conditions, angle requirements and PCB surface properties;

[0009] S2. Acquire the first image of the sample PCB board;

[0010] S3. Perform image recognition on multiple first images using an image recognition algorithm to obtain the first position information and first feature information of the first cutting mark;

[0011] S4. Use the data of the first location information and the first feature information as a training dataset, input it into the AI ​​model for recognition training, and then obtain the AI ​​tag recognition model.

[0012] S5. Set a second cutting mark at the connecting rib or pre-reserved position of the PCB board to be cut, and input it into the AI ​​mark recognition model for mark recognition, thereby obtaining the second position information and second feature information of the second cutting mark;

[0013] S6. Based on the second location information and the second feature information, a cutting path matching the PCB board to be cut is generated using an automatic path generation algorithm.

[0014] The PCB cutting path planning method of this invention obtains multiple first images of sample PCBs under various cutting conditions using a mobile phone. This image is then used to train an AI model, resulting in an AI mark recognition model capable of more accurately recognizing second cutting marks on various PCBs under different cutting conditions. After the second cutting mark is applied to the PCB, the image is fed into the AI ​​mark recognition model, enabling image recognition to adapt to various environmental changes. It can accurately recognize images even in low light or overexposure conditions, and is more accurate in recognizing features at different angles. Even with irregularities on the board surface, the mark points are clearer. This allows for the acquisition of second position information and second feature information adapted to different cutting conditions and requirements. This improves the accuracy of obtaining the positions of connecting ribs and cutting points, meeting the needs of diverse board types for board separation. Furthermore, by combining this with an automatic path generation algorithm, the method eliminates the need for manual experience in determining connecting rib positions to generate a cutting path matching the PCB. It also eliminates the need for manual programming to set the cutting path, improving the accuracy of the cutting path and increasing the efficiency of PCB board separation and cutting, thus enhancing production efficiency.

[0015] Preferably, the PCB board cutting path planning method of the present invention includes the following steps:

[0016] S61. Based on the cutting starting point of the second location information, find multiple adjacent points adjacent to the cutting starting point, and calculate the first cost value g from the cutting starting point to the adjacent points; based on the cutting target point of the second location information, calculate the second cost value h from the adjacent points to the cutting target point; calculate the third cost value f of the cutting path, where the third cost value f = g + h;

[0017] S62. Obtain all reachable nodes of the cutting path, filter multiple reachable nodes corresponding to the cutting path with the smallest f, and extend the reachable nodes to the next adjacent node until the extended next node is the cutting target point, thereby obtaining the optimal cutting path.

[0018] As a preferred embodiment of the present invention, the above-mentioned specific automatic path generation algorithm generates a stable and accurate cutting path from the marker point recognition results, and obtains the optimal cutting path more accurately and completely. Under the condition of a large search space, it can quickly find the optimal path and obtain the cutting path with the lowest cost or fastest speed to reach the cutting target point, thereby improving the accuracy of the cutting path and the speed of obtaining the cutting path, and improving the quality and production efficiency of plate cutting.

[0019] Preferably, the PCB board cutting path planning method of the present invention, wherein the automatic path generation algorithm further includes the following steps:

[0020] S621. Add all the reachable nodes to the open list; sort the reachable nodes in ascending order according to the third cost value f;

[0021] S622: Select multiple reachable nodes corresponding to the cutting path with the smallest f from the open list; and move the expanded reachable node and the next node to the closed list.

[0022] By adding the reachable nodes to the open list and sorting them, the order of the reachable nodes with the smallest third cost value f can be clearly obtained. Combined with moving the expanded reachable nodes and the next node to the closed list, the filtered reachable nodes and the next node in the open list can be removed. This makes the order of the reachable nodes with the smallest f in the open list dynamically sorted, with the reachable node with the smallest f at the top, until the target node is reached or all nodes are processed. This achieves dynamic filtering of reachable nodes for the optimal cutting path, reducing the failure rate and improving computational efficiency. Furthermore, the closed list prevents already filtered reachable nodes from being further filtered and expanded, further reducing the error in determining the optimal cutting path and improving its accuracy.

[0023] Preferably, the PCB board cutting path planning method of the present invention includes the following steps:

[0024] Calculate the shortest path from the cutting start point to each adjacent node of the cutting start point; filter the unprocessed nodes with the lowest cost of the current cutting path, expand them, and then update the shortest path until all cutting nodes are processed.

[0025] As a preferred embodiment of the present invention, the shortest or optimal path is found among multiple cutting path options, and the path can be calculated and updated in real time when the path diagram changes dynamically, avoiding path conflicts or errors, thereby further improving the accuracy of cutting path planning and enhancing cutting quality and efficiency.

[0026] Preferably, the PCB board cutting path planning method of the present invention, wherein setting the first cutting mark or setting the second cutting mark specifically includes the following steps:

[0027] Cutting marks are set during the PCB design stage; each cutting mark includes at least one of QR codes and graphic symbols; the contrast of the cutting marks is relatively high compared to the PCB; the cutting marks store the category information of the PCB.

[0028] As a preferred embodiment of the present invention, by setting cutting marks on the PCB board, it is possible to ensure stable recognition under different environments. For example, high-contrast graphic symbols such as rectangles and circles can also be used as marks, which have a certain degree of stability and adaptability and can cope with different lighting conditions. Another example is high-contrast connecting rib marks, which are lines or symbols specifically used to mark the specific positions of connecting ribs. They can be flexibly adjusted according to the actual layout of different PCB boards. As long as the image model is trained well, they can be recognized well.

[0029] Preferably, the PCB board cutting path planning method of the present invention includes the following steps, wherein the cutting mark is set during the PCB board design stage:

[0030] The cutting marks include various shapes and sizes; different layers of the PCB board use different shapes or sizes of the cutting marks to form multi-layer cutting marks. The cutting marks of each layer of the multi-layer cutting marks are prioritized so that critical cutting marks take precedence over non-critical cutting marks.

[0031] The image recognition algorithm sets a marker recognition priority, thereby prioritizing the recognition of the key cutting markers.

[0032] In practical applications, different markings may overlap. As a preferred solution of the present invention, the specific method for setting the cutting markings described above can avoid the overlap of different types of markings, ensure that key cutting markings, such as the location of connecting ribs, are identified and processed first, reduce the probability of failure in the generation of cutting paths, and improve the reliability of cutting path planning.

[0033] Preferably, the PCB board cutting path planning method of the present invention further includes the following steps after S4 and before S5:

[0034] Prepare a verification dataset, input the verification dataset into the AI ​​label recognition model for test training, and then obtain the training error and the verification error; compare the training error and the verification error, and if the verification error is close to the training error, then the recognition training is deemed qualified.

[0035] As a preferred embodiment of the present invention, by testing and training the AI ​​tag recognition model using the above-mentioned verification dataset, the training error and verification error can be obtained, thereby determining whether the recognition training is qualified, further optimizing the recognition training of the AI ​​tag recognition model, and further improving the accuracy and reliability of image recognition in the seven path planning.

[0036] Preferably, in the PCB board cutting path planning method of the present invention, step S3 further includes the following steps:

[0037] The first image is cropped, rotated, or scaled; the pixel values ​​of the first image are normalized to the range of [0,1].

[0038] As a preferred embodiment of the present invention, the above-described processing of the first image can improve the stability of the image recognition model training process, accelerate convergence, reduce numerical errors, and improve the model's compatibility.

[0039] To achieve the objectives of this invention, another technical solution is provided:

[0040] A PCB board cutting path planning system includes: an image acquisition module, a visual recognition module, an AI tag recognition model, and an image processing module;

[0041] The image acquisition module is used to acquire a first image of the sample PCB board; the visual recognition module is used to perform image recognition on the first image, and can obtain first position information and first feature information; the AI ​​mark recognition model, after recognition training, is used to mark and recognize the cutting marks on the PCB board to be cut, and can obtain second position information and second feature information of the PCB board to be cut; the image processing module is equipped with an automatic path generation algorithm, which is used to generate the cutting path of the PCB board to be cut.

[0042] The PCB cutting path planning system described in this invention eliminates the need for traditional manual path programming, enabling immediate cutting upon identification and improving changeover efficiency. Ordinary operators can operate the equipment without requiring programming or engineering background, reducing labor costs and increasing production efficiency. It performs better in multi-variety, small-batch production environments, quickly adapting to different board types and enhancing the flexibility of the cutting system. The introduction of image recognition processing technology promotes the intelligent and automated upgrading of PCB manufacturing equipment, further improving production efficiency.

[0043] Preferably, the PCB board cutting path planning system of the present invention further includes: a board separation control module and a cutting unit; the board separation control module is used to obtain the cutting path from the image processing module and send the cutting path to each cutting unit, thereby controlling the cutting unit to perform board separation cutting; the cutting unit is used to perform cutting operation on the PCB board to be cut according to the cutting path.

[0044] As a preferred embodiment of the present invention, by setting up a PCB board splitting control module and a cutting unit as an integrated PCB board cutting path planning system, the PCB board splitting control module receives cutting path data and controls the cutting execution unit, such as a flying knife or laser head, to cut the PCB board in real time and transport it to the receiving area. The entire process does not require manual intervention in path setting, which further improves the automation and intelligence of PCB board splitting and cutting and improves production efficiency.

[0045] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0046] 1. The cutting path planning method described above collects sample PCBs and feeds training datasets to an AI model for training, obtaining an AI marker recognition model. This enables image recognition to adapt to various environmental changes, accurately recognizing images even in low light or overexposure conditions. It is more accurate in recognizing features marked at different angles, and even with irregular board surfaces, the marker points are clearer. This improves the accuracy of obtaining the positions of connecting ribs and cutting points, meeting the board separation needs of diverse board types. Furthermore, combined with an automatic path generation algorithm, it eliminates the need for manual judgment of connecting rib positions to determine the cutting path, generating a cutting path that matches the PCB to be cut. It also eliminates the need for manual programming to set the cutting path, improving the accuracy of the cutting path and increasing the efficiency of PCB board separation and cutting, thereby improving production efficiency.

[0047] 2. The cutting path planning system, by adopting a PCB board cutting path planning system, eliminates the need for traditional manual path programming, enabling cutting upon recognition and improving changeover efficiency; ordinary operators can operate the equipment without programming or engineering background, reducing labor costs and improving production efficiency; it performs better in multi-variety, small-batch production environments, can quickly adapt to different board types, and enhances the flexibility of the cutting system; the introduction of image recognition processing technology promotes the intelligent and automated upgrading of PCB manufacturing equipment, improving production efficiency. Attached Figure Description

[0048] Figure 1 This is a flowchart illustrating the PCB board cutting path planning method of the present invention;

[0049] Figure 2 This is a schematic diagram of the module connection of the PCB board cutting path planning system of the present invention;

[0050] Figure 3 This is a schematic diagram of the cutting marks on the PCB board of the present invention. Detailed Implementation

[0051] The present invention will now be described in detail with reference to the accompanying drawings.

[0052] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0053] Example 1:

[0054] refer to Figure 1 As shown, the present invention discloses a PCB board cutting path planning method, which includes the following steps:

[0055] S1. Collect sample PCB boards with various cutting conditions, and set the first cutting mark at the connecting rib or pre-reserved position of the sample PCB board; the cutting conditions include at least one of the following: lighting conditions, angle requirements and PCB surface properties;

[0056] S2. Acquire the first image of the sample PCB board;

[0057] S3. Perform image recognition on multiple first images using an image recognition algorithm to obtain the first position information and first feature information of the first cutting mark;

[0058] S4. Use the data of the first location information and the first feature information as a training dataset, input it into the AI ​​model for recognition training, and then obtain the AI ​​tag recognition model;

[0059] S5. Set a second cutting mark at the connection rib or pre-reserved position of the PCB board to be cut, and input it into the AI ​​mark recognition model for mark recognition, thereby obtaining the second position information and second feature information of the second cutting mark;

[0060] S6. Based on the second position information and the second feature information, generate a cutting path that matches the PCB board to be cut using an automatic path generation algorithm.

[0061] The automatic path generation algorithm of this invention can conventionally calculate the cutting path based on the marked positions of the connecting ribs on the PCB board, referring to preset rules or empirical data of historical cutting trajectories. In this invention, the preferred automatic path generation algorithm specifically includes the following steps:

[0062] S61. Based on the cutting starting point of the second location information, find multiple adjacent points adjacent to the cutting starting point, and calculate the first cost value g from the cutting starting point to the adjacent points; based on the cutting target point of the second location information, calculate the second cost value h from the adjacent points to the cutting target point; calculate the third cost value f of the cutting path, where the third cost value f = g + h;

[0063] S62. Obtain all reachable nodes of the cutting path, filter multiple reachable nodes corresponding to the cutting path with the smallest f, and extend the reachable nodes to the next adjacent node until the extended next node is the cutting target point, thereby obtaining the optimal cutting path.

[0064] Based on obtaining the optimal cutting path, the specific automatic path generation algorithm also includes the following steps: S621, add all reachable nodes to the open list; sort the reachable nodes in ascending order according to the third cost value f; S622, select multiple reachable nodes corresponding to the cutting path with the smallest f from the open list; and move the expanded reachable nodes and the next node to the closed list.

[0065] It should be noted that the reachable nodes of the present invention are derived through a predefined collision map or walkable area; the blank area that the cutting head can pass through is free of obstacles and within the map boundary. The path planning module rasterizes the PCB board area and pre-generates an "obstacle map" based on the component layout information. The algorithm starts from the cutting start point and determines the reachability of adjacent nodes by querying the obstacle map, adding only reachable nodes to the priority queue of the open list.

[0066] The automatic path generation algorithm described in this invention can also employ a graph theory algorithm to find the shortest path from the starting point to each node. It expands the path incrementally until the target path is found. Preferably, another automatic path generation algorithm specifically includes the following steps: calculating the shortest path from the cutting starting point to each adjacent node; filtering the unprocessed nodes with the lowest cost of the current cutting path and expanding them, thereby updating the shortest path until all cutting nodes are processed.

[0067] It should be noted that the first and second cut marks mentioned in this invention are not interpreted in terms of specific numerical values. For example, both the first cut mark and the second cut mark are understood as cut marks marked on the PCB board. They are only limited to distinguish the different PCB boards where the cut marks are located. The first cut mark is marked on the sample PCB board, and the second cut mark is marked on the PCB board to be cut.

[0068] The first and second cutting marks in this invention can be set using the same method. For example, setting either the first or the second cutting mark specifically includes the following steps: setting the cutting mark during the PCB design stage; the cutting mark includes at least one of a QR code or a graphic symbol; the contrast of the cutting mark is relatively high compared to the PCB; the cutting mark stores the category information of the PCB.

[0069] For example, in this invention, the marking symbols for the connecting rib positions mainly include the following: QR codes, specific graphic symbols, and high-contrast connecting rib markings. During the PCB design stage, QR codes can be selected as markings on the connecting ribs or pre-reserved locations for sub-boards. As high-contrast graphic markings, QR codes ensure stable recognition in different environments. Specific graphic symbols, such as rectangles and circles, are also high-contrast graphic symbols that offer stability and adaptability, capable of handling different lighting conditions. High-contrast connecting rib markings, such as simple lines or symbols, are specifically used to mark the specific locations of the connecting ribs. The selection of these marking methods can be flexibly adjusted according to the actual layout of different PCBs. Through training the image recognition AI model, the location and feature information of the cutting marks can be clearly identified and obtained. (Reference) Figure 3 As shown, for example, the Figure 3 The ▲ symbol indicates a cutting mark.

[0070] The feature information is understood as preset information corresponding to the shape or pattern of the cutting mark. For example, the QR code can be used not only for the location of the connecting rib mark, but also for the category mark of the PCB board. On PCB boards of multiple categories, the QR code can store category information. Cutting marks of specific shapes, such as rectangles and circles, can also be used as options for category marks. They can be designed according to different board characteristics or production needs.

[0071] For example, QR codes are placed at the locations of the connecting ribs as markers. The surface of the PCB board is coated with a metallized layer to ensure that the QR code markers can be clearly recognized by the camera under different lighting conditions. The image recognition system scans the QR codes to identify the exact location of the connecting ribs. As another example, the cutting marks are rectangular graphic symbols. The size of the graphic is strictly controlled within the width of the connecting rib to avoid overlapping with other design elements. The image recognition system can quickly locate these rectangular symbols and generate the corresponding cutting paths.

[0072] It should be noted that the cutting marks described in this invention avoid the problem of overlapping marks. The size and layout requirements of the marks can be controlled; for example, the size of the connecting rib mark should not exceed the width of the connecting rib and can be controlled between 1mm and 5mm to ensure that the marks do not affect the cutting quality and are easy for the image system to recognize. Regarding the requirements for optical contrast and recognition distance: considering the optical contrast, recognition distance, and adaptability requirements of the marks, it is ensured that the marks can be accurately recognized under different PCB layouts and surface materials. For example, on a PCB with blue paint, cutting marks are made with red paint. In reality, after the PCB is transported to the depaneling machine, it is basically in a two-dimensional plane.

[0073] More specifically, to ensure that the cutting marks and the image recognition of this invention can be adapted to certain complex multi-layer PCB board structures, the cutting marks can be designed as a multi-layer structure. Specifically, the cutting marks are set during the PCB board design stage, including the following steps: the cutting marks include various graphics and sizes; different graphics or different sizes of cutting marks are used for different layers of the PCB board, thereby forming multi-layer cutting marks; each layer of the multi-layer cutting marks is prioritized, so that critical cutting marks take precedence over non-critical cutting marks; and a mark recognition priority is set in the image recognition algorithm, so that critical cutting marks are recognized first.

[0074] To avoid overlapping marks, since board separation is a process after SMT is completed on the PCB, such overlapping of connecting ribs is very rare. If the model cannot determine the position due to overlap, it can directly report an error and be handled manually. With a large amount of data, through deep learning training of the AI ​​mark recognition model, it can automatically identify key overlapping cutting marks.

[0075] In step S4 of this invention, the input AI model is used for recognition training, for example, training a CNN model with data, mainly including the following steps: Step 1, manually annotating the connecting rib markers and their positions in each image, uploading images with different lighting and shooting angles, allowing the algorithm to identify the shape and structure of the PCB and the location of the connecting ribs. The feature data obtained in step 1 includes the marker positions of each image, such as the center position of the QR code. This marker information will be provided to the model as training data, mainly for coordinate and size recognition training. For each image, a corresponding training dataset needs to be generated through image processing; Step 3, feeding the model data for training, initializing the parameters of the CNN model → repeatedly inputting the dataset → processing the model to obtain the coordinates of the connecting ribs → multiple rounds of training to improve accuracy → actual verification; Step 4, adaptive image processing, using more image datasets such as strong light and different angles to train the model, making it more accurate in recognizing the location of connecting ribs in real-world environments; Step 5, solidifying the practical data, after the model is trained, repeatedly recognizing the location of the PCB connecting ribs in actual operation. If there is a deviation, the model can be adjusted according to the actual error. The more data, the more accurate the recognition.

[0076] To further improve the accuracy of the AI ​​label recognition model in identifying the location of connecting ribs or pre-cut positions on PCB boards, specifically, after S4 and before S5, the following steps are also included: preparing a verification dataset, inputting the verification dataset into the AI ​​label recognition model for test training, and thus obtaining the training error and verification error; comparing the training error and verification error, if the verification error is close to the training error, then the recognition training is deemed qualified.

[0077] To achieve stability, accelerate convergence, and reduce numerical errors in the training process of the image recognition model, if the pixel value range is too large, such as 0 to 255, the training of the network may become unstable. Specifically, S3 of the present invention also includes the following steps: cropping, rotating, or scaling the first image; normalizing the pixel values ​​of the first image to the range of [0,1].

[0078] Example 2:

[0079] Based on Example 1, and referring to Figure 2As shown, this embodiment discloses a PCB board cutting path planning system for implementing the PCB board cutting path planning method of Embodiment 1, including: an image acquisition module, a visual recognition module, an AI marker recognition model, and an image processing module; the image acquisition module is used to acquire a first image of the sample PCB board; the visual recognition module is used to perform image recognition on the first image to obtain first position information and first feature information; the AI ​​marker recognition model, after recognition training, is used to perform marker recognition on the cutting marks of the PCB board to be cut to obtain second position information and second feature information of the PCB board to be cut; the image processing module is equipped with an automatic path generation algorithm to generate the cutting path of the PCB board to be cut.

[0080] The AI ​​marker recognition model of this invention can process PCB images using a Convolutional Neural Network (CNN), belonging to deep learning models. Its architecture typically includes: convolutional layers, pooling layers, fully connected layers, and an output layer. The convolutional layers extract local features from the input image, such as edges, corners, and textures, through convolution operations. Then, non-linear features are added through activation functions, enabling the network to learn more complex patterns. Pooling operations in the pooling layers reduce the spatial size of the image, thereby reducing computation and preventing overfitting. Commonly used pooling methods include max pooling and average pooling. The fully connected layers synthesize the features after convolution and pooling for the final classification or regression task. The output layer outputs the recognition results, such as the location, category, or other information of the connecting rib markers, according to the task requirements.

[0081] Specifically, the PCB board cutting path planning system of the present invention further includes: a board separation control module and a cutting unit; the board separation control module is used to obtain the cutting path from the image processing module and send the cutting path to each cutting unit, thereby controlling the cutting unit to perform board separation cutting; the cutting unit is used to perform cutting operation on the PCB board to be cut according to the cutting path.

[0082] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for planning the cutting path of a PCB board, characterized in that, Includes the following steps: S1. Collect sample PCBs with various cutting conditions, and set the first cutting mark at the connecting rib or pre-reserved position of the sample PCB. The cutting conditions include at least one of the following: lighting conditions, angle requirements, and PCB surface properties; S2. Acquire the first image of the sample PCB board; S3. Perform image recognition on multiple first images using an image recognition algorithm to obtain the first position information and first feature information of the first cutting mark; S4. Use the data of the first location information and the first feature information as a training dataset, input it into the AI ​​model for recognition training, and then obtain the AI ​​tag recognition model. S5. Set a second cutting mark at the connecting rib or pre-reserved position of the PCB board to be cut, and input it into the AI ​​mark recognition model for mark recognition, thereby obtaining the second position information and second feature information of the second cutting mark; S6. Based on the second location information and the second feature information, a cutting path matching the PCB board to be cut is generated using an automatic path generation algorithm.

2. The PCB board cutting path planning method according to claim 1, characterized in that, The automatic path generation algorithm specifically includes the following steps: S61. Based on the cutting starting point of the second location information, find multiple adjacent points adjacent to the cutting starting point, and calculate the first cost value g from the cutting starting point to the adjacent points; based on the cutting target point of the second location information, calculate the second cost value h from the adjacent points to the cutting target point; calculate the third cost value f of the cutting path, where the third cost value f = g + h; S62. Obtain all reachable nodes of the cutting path, filter multiple reachable nodes corresponding to the cutting path with the smallest f, and extend the reachable nodes to the next adjacent node until the extended next node is the cutting target point, thereby obtaining the optimal cutting path.

3. The PCB board cutting path planning method according to claim 2, characterized in that, The automatic path generation algorithm also includes the following steps: S621. Add all the reachable nodes to the open list; sort the reachable nodes in ascending order according to the third cost value f; S622: Select multiple reachable nodes corresponding to the cutting path with the smallest f from the open list; and move the expanded reachable node and the next node to the closed list.

4. The PCB board cutting path planning method according to claim 1, characterized in that, The automatic path generation algorithm specifically includes the following steps: Calculate the shortest path from the cutting start point to each adjacent node of the cutting start point; filter the unprocessed nodes with the lowest cost of the current cutting path, expand them, and then update the shortest path until all cutting nodes are processed.

5. The PCB board cutting path planning method according to claim 1, characterized in that, Setting the first cutting mark or the second cutting mark specifically includes the following steps: Cutting marks are set during the PCB design stage; each cutting mark includes at least one of QR codes and graphic symbols; the contrast of the cutting marks is relatively high compared to the PCB; the cutting marks store the category information of the PCB.

6. The PCB board cutting path planning method according to claim 5, characterized in that, Setting the cutting marks during the PCB design phase includes the following steps: The cutting marks include various shapes and sizes; different layers of the PCB board use different shapes or sizes of the cutting marks to form multi-layer cutting marks. The cutting marks of each layer of the multi-layer cutting marks are prioritized so that critical cutting marks take precedence over non-critical cutting marks. The image recognition algorithm sets a marker recognition priority, thereby prioritizing the recognition of the key cutting markers.

7. The PCB board cutting path planning method according to any one of claims 1-6, characterized in that, The steps following S4 and S5 are also included: Prepare a verification dataset, input the verification dataset into the AI ​​label recognition model for test training, and then obtain the training error and the verification error; compare the training error and the verification error, and if the verification error is close to the training error, then the recognition training is deemed qualified.

8. The PCB board cutting path planning method according to any one of claims 1-6, characterized in that, S3 further includes the following steps: The first image is cropped, rotated, or scaled; the pixel values ​​of the first image are normalized to the range of [0,1].

9. A PCB board cutting path planning system, characterized in that, A PCB cutting path planning method for implementing the PCB board as described in any one of claims 1-8 includes: an image acquisition module, a visual recognition module, an AI marker recognition model, and an image processing module; The image acquisition module is used to acquire a first image of the sample PCB board; the visual recognition module is used to perform image recognition on the first image, and can obtain first position information and first feature information; the AI ​​mark recognition model, after recognition training, is used to mark and recognize the cutting marks on the PCB board to be cut, and can obtain second position information and second feature information of the PCB board to be cut; the image processing module is equipped with an automatic path generation algorithm, which is used to generate the cutting path of the PCB board to be cut.

10. The PCB board cutting path planning system according to claim 9, characterized in that, Also includes: Board splitting control module and cutting unit; The board splitting control module is used to obtain the cutting path from the image processing module and send the cutting path to each cutting unit, thereby controlling the cutting unit to perform board splitting cutting; the cutting unit is used to perform cutting operation on the PCB board to be cut according to the cutting path.

Images