Battery pole piece ceramic defect detection method and device, and electronic equipment
By combining different convolution scale operations with the YOLOv5 neural network model on the target detection images of lithium-ion battery electrodes, the detection problem of ceramic edges and coating junctions was solved, achieving higher detection accuracy and precision.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING LUSTER LIGHTTECH
- Filing Date
- 2022-12-21
- Publication Date
- 2026-05-29
AI Technical Summary
Existing technologies are insufficient for accurately measuring the edges at the junction of ceramic and coating and for detecting defects in the ceramic area, resulting in inadequate accuracy and precision in lithium-ion battery electrode testing.
By performing convolution operations at different scales on the target detection image, a binary image of the target is obtained. Combined with the YOLOv5 neural network model, the electrode ceramic region is accurately determined and defects are detected.
It improves the accurate division between the edges of the electrode ceramic and the coating edge, enhances the accuracy and precision of defect detection, reduces the amount of data, and improves detection efficiency.
Smart Images
Figure CN115937167B_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of defect detection technology, and in particular relates to a defect detection method, device and electronic equipment for battery electrode ceramics. Background Technology
[0002] Coating is an important manufacturing process for lithium-ion batteries. The main process of electrode coating is to coat a uniformly stirred slurry to form an electrode. The coating effect has a significant impact on battery capacity, internal resistance, cycle life and safety.
[0003] The ceramic material consists of nano-sized alumina particles, possessing excellent properties such as high thermal stability, chemical stability, corrosion resistance, and high hardness. When coating the positive electrode of a lithium-ion battery, a ceramic layer is added. This ceramic layer reduces burrs generated during electrode die-cutting, lowering the risk of short circuits. Furthermore, the ceramic is typically coated on the tab side, extending beyond the separator. If the separator's coverage is insufficient, the negative electrode tab may come into contact with the positive electrode, or vice versa. The ceramic provides insulation in this case.
[0004] Existing technologies employ threshold segmentation edge detection and traditional defect feature comparison to measure ceramic edges and classify defects on the ceramic. However, threshold segmentation techniques based on traditional vision are insufficient for detecting ceramic edge defects from all angles. Accurately measuring the edge at the interface between the ceramic and coating layers and detecting defects in the ceramic area are critical problems that need to be solved in the inspection process of lithium-ion battery electrodes. Summary of the Invention
[0005] This application aims to address at least one of the technical problems existing in the prior art. To this end, this application proposes a method, apparatus, and electronic equipment for defect detection of battery electrode ceramics, which can improve the accuracy and precision of edge measurement and defect detection of electrode ceramics.
[0006] In a first aspect, this application provides a method for detecting defects in battery electrode ceramics, the method comprising:
[0007] Acquire a target detection image of the battery to be tested, the target detection image including the electrode ceramic of the battery to be tested;
[0008] The target detection image is subjected to convolution operations at different convolution scales to obtain a target binarized image;
[0009] Based on the target binarized image, the electrode ceramic region in the target detection image is determined;
[0010] Based on the image feature information of the electrode ceramic region, the defect detection result of the electrode ceramic is determined.
[0011] According to the defect detection method for battery electrode ceramics of this application, by performing convolution operations at different scales on the target detection image, a target binarized image is obtained, the electrode ceramic region in the target detection image is accurately determined, the defect detection region is accurately divided, and the edges of the electrode ceramic and other coating edges are accurately identified, thereby improving the accuracy and precision of defect detection on the electrode ceramic.
[0012] According to one embodiment of this application, performing convolution operations at different convolution scales on the target detection image to obtain a target binarized image includes:
[0013] The target detection image is input into at least two first convolution kernels with different convolution scales to obtain at least two first binarized images output by the at least two first convolution kernels, and the first binarized images correspond one-to-one with the first convolution kernels;
[0014] One of the at least two first binarized images is determined as the target binarized image.
[0015] According to one embodiment of this application, the first convolution kernel is obtained by performing mathematical operations based on at least two second convolution kernels with different convolution scales.
[0016] According to one embodiment of this application, determining the electrode ceramic region in the target detection image based on the target binarized image includes:
[0017] Obtain the cumulative mean of the pixel values in each column of the target binarized image;
[0018] The column of pixels in the target binarized image whose cumulative mean abruptly changes is identified as the edge of the ceramic electrode.
[0019] The electrode ceramic region is determined based on the edge of the electrode ceramic.
[0020] According to one embodiment of this application, determining the defect detection result of the electrode ceramic based on the image feature information of the electrode ceramic region includes:
[0021] The image of the electrode ceramic region is input into the target defect detection model to obtain the defect detection result of the electrode ceramic output by the target defect detection model;
[0022] The target defect detection model is built based on the YOLOv5 neural network and trained using a training dataset.
[0023] According to one embodiment of this application, the target defect detection model is trained through the following steps:
[0024] The training dataset is input into the target defect detection model to be trained for training, and the first model weight parameters of the target defect detection model are determined.
[0025] Export the weight parameters of the first model as an Open Neural Network Exchange Format file;
[0026] Based on the open neural network exchange format file, the target defect detection model is weighted and the second model weight parameters of the target defect detection model are determined to obtain the trained target defect detection model.
[0027] Secondly, this application provides a defect detection device for battery electrode ceramics, the device comprising:
[0028] An acquisition module is used to acquire a target detection image of the battery to be tested, wherein the target detection image includes the electrode ceramic of the battery to be tested;
[0029] The first processing module is used to perform convolution operations at different convolution scales on the target detection image to obtain a target binarized image;
[0030] The second processing module is used to determine the electrode ceramic region in the target detection image based on the target binarized image;
[0031] The third processing module is used to determine the defect detection result of the electrode ceramic based on the image feature information of the electrode ceramic region.
[0032] According to the defect detection device for battery electrode ceramics of this application, by performing convolution operations at different scales on the target detection image, a target binarized image is obtained, the electrode ceramic region in the target detection image is accurately determined, the defect detection region is accurately divided, and the edges of the electrode ceramic and other coating edges are accurately identified, thereby improving the accuracy and precision of defect detection on the electrode ceramic.
[0033] Thirdly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the defect detection method for battery electrode ceramics as described in the first aspect above.
[0034] Fourthly, this application provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the defect detection method for battery electrode ceramics as described in the first aspect above.
[0035] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the defect detection method for battery electrode ceramics as described in the first aspect above.
[0036] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description
[0037] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:
[0038] Figure 1 This is a flowchart illustrating the defect detection method for battery electrode ceramics provided in the embodiments of this application;
[0039] Figure 2 This is a schematic diagram of the region distribution of the target detection image provided in the embodiments of this application;
[0040] Figure 3 This is a schematic diagram of the convolution operation at different convolution scales provided in the embodiments of this application;
[0041] Figure 4 This is a schematic diagram of the electrode ceramic edge provided in the embodiments of this application;
[0042] Figure 5 This is a schematic diagram of the structure of the Open Neural Network Exchange Format file provided in the embodiments of this application;
[0043] Figure 6 This is a schematic diagram of the structure of the first model weight parameters of the target defect detection model provided in the embodiments of this application;
[0044] Figure 7 This is a schematic diagram of the structure of the second model weight parameters of the target defect detection model provided in the embodiments of this application;
[0045] Figure 8 This is a schematic diagram of the structure of the defect detection device for battery electrode ceramics provided in the embodiments of this application;
[0046] Figure 9 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application;
[0047] Figure 10 This is a hardware schematic diagram of the electronic device provided in the embodiments of this application. Detailed Implementation
[0048] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.
[0049] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.
[0050] The following description, in conjunction with the accompanying drawings, details the defect detection method, defect detection device, electronic device, and readable storage medium for battery electrode ceramics provided in this application, through specific embodiments and application scenarios.
[0051] Among them, the defect detection method for battery electrode ceramics can be applied to the terminal, and can be executed by the hardware or software in the terminal.
[0052] The terminal includes, but is not limited to, portable communication devices such as mobile phones or tablets with touch-sensitive surfaces (e.g., touchscreen displays and / or touchpads). It should also be understood that, in some embodiments, the terminal may not be a portable communication device, but rather a desktop computer with touch-sensitive surfaces (e.g., touchscreen displays and / or touchpads).
[0053] The following embodiments describe a terminal including a display and a touch-sensitive surface. However, it should be understood that the terminal may include one or more other physical user interface devices such as a physical keyboard, mouse, and joystick.
[0054] The defect detection method for battery electrode ceramics provided in this application embodiment can be executed by an electronic device or a functional module or entity in an electronic device that can implement the defect detection method for battery electrode ceramics. The electronic devices mentioned in this application embodiment include, but are not limited to, mobile phones, tablets, computers, cameras and wearable devices. The defect detection method for battery electrode ceramics provided in this application embodiment will be described below using an electronic device as the execution subject.
[0055] like Figure 1 As shown, the defect detection method for the battery electrode ceramic includes steps 110, 120, 130 and 140.
[0056] Step 110: Obtain the target detection image of the battery to be detected.
[0057] The target detection image includes the electrode ceramics of the battery to be detected.
[0058] The battery to be tested can be a lithium-ion battery, and the electrode ceramic is the exposed part of the ceramic layer of the positive electrode of a lithium-ion battery.
[0059] In practice, target detection images of the battery to be inspected are acquired through an industrial vision system. The acquired target detection images include the electrode ceramics of the battery to be inspected.
[0060] Among them, the industrial vision system can be a high-frequency linear array camera, which captures images of lithium-ion batteries and acquires target detection images.
[0061] For example, such as Figure 2 As shown, the coated area represents the region where the lithium-ion battery electrode is located, the ceramic area represents the region where the electrode ceramic is located, and the blank areas are not coated with electrode or electrode ceramic. In this embodiment, the ceramic area is elongated, and defects on the electrode ceramic may be distributed at the edges of the elongated ceramic area.
[0062] Step 120: Perform convolution operations at different convolution scales on the target detection image to obtain a target binarized image.
[0063] In this step, convolution operations are performed on the target detection image. Convolution operations of different scales can be achieved by implementing convolution kernels of different convolution scales.
[0064] It is understandable that convolution operations at different scales produce convolution results that represent different image features. Some convolution results contain both global image features and local detail image features. Performing convolution operations at different scales on target detection images can produce convolution results that highlight the detail image features of the electrode ceramic part.
[0065] Image binarization is the process of setting the grayscale value of pixels in an image to 0 or 255, thus giving the entire image a distinct black and white effect.
[0066] In digital image processing, image binarization plays a very important role. Image binarization greatly reduces the amount of data in the image, thereby highlighting the outline of the target.
[0067] In this embodiment, after performing convolution operations at different convolution scales on the target detection image, a target binarized image is obtained. The obtained target binarized image can highlight the outline of the electrode ceramic part, which helps to accurately segment the image of the area where the electrode ceramic is located for defect detection.
[0068] Step 130: Based on the target binarized image, determine the electrode ceramic region in the target detection image.
[0069] Understandably, the target binarization image can highlight the outline of the electrode ceramic part in the target detection image. Based on the target binarization image, the edge position of the junction between the electrode ceramic and the electrode or blank space can be accurately determined, and the electrode ceramic region in the target detection image can be segmented.
[0070] The electrode ceramic region refers to the area where the electrode ceramic is located in the target detection image. The electrode ceramic region can include the complete electrode ceramic as well as the part of the electrode ceramic or the blank space at the boundary with the electrode ceramic.
[0071] In this embodiment, by using the target binarized image, the electrode ceramic region in the target detection image is determined, and the defect detection region is accurately divided. This can reduce the amount of data when performing defect detection, improve defect detection efficiency, and effectively avoid other regions from affecting the defect detection of the electrode ceramic region.
[0072] Step 140: Based on the image feature information of the electrode ceramic region, determine the defect detection result of the electrode ceramic.
[0073] In this step, image feature information of the electrode ceramic region is acquired. Based on the image feature information of the electrode ceramic region, it is determined whether there are defects in the electrode ceramic region, and the defect detection result of the electrode ceramic of the battery to be tested is obtained.
[0074] In this embodiment, when it is determined that there are no defects in the electrode ceramic of the battery to be tested based on the image feature information of the electrode ceramic area, the defect detection result of the electrode ceramic is no defect, and the battery to be tested can proceed to the next step of testing or be determined as a good product.
[0075] When the image feature information of the electrode ceramic area is used to determine that there is a defect in the electrode ceramic of the battery to be tested, the defect detection result of the electrode ceramic can be obtained as defective. The defect detection result can also include the location of the defect on the electrode ceramic and the type of defect.
[0076] In related technologies, most use traditional visual threshold segmentation techniques to detect ceramic edges. Such techniques have difficulty detecting defects at the ceramic edges from all angles, cannot accurately find the boundary line between the ceramic and the electrode coating, and cannot accurately segment the location of the ceramic.
[0077] In this embodiment, by performing convolution operations at different scales on the target detection image, a target binarized image is obtained. The image features of the electrode ceramic are characterized from different convolution scales. Then, based on the target binarized image, the electrode ceramic region in the target detection image is accurately determined, and the defect detection region is accurately divided. This can reduce the amount of data when performing defect detection, improve defect detection efficiency, and effectively avoid the influence of other regions on the defect detection of the electrode ceramic region.
[0078] According to the defect detection method for battery electrode ceramics provided in the embodiments of this application, by performing convolution operations at different scales on the target detection image, a target binarized image is obtained, the electrode ceramic region in the target detection image is accurately determined, the defect detection region is accurately divided, the edge of the electrode ceramic and other coating edges are accurately distinguished, and the accuracy and precision of defect detection on the electrode ceramic are improved.
[0079] In some embodiments, step 120, performing convolution operations at different convolution scales on the target detection image to obtain a target binarized image, may include:
[0080] The target detection image is input into at least two first convolution kernels with different convolution scales to obtain at least two first binarized images output by at least two first convolution kernels, and the first binarized images correspond one-to-one with the first convolution kernels;
[0081] One of the at least two first binarized images is determined as the target binarized image.
[0082] There can be multiple first convolution kernels, and the convolution scales of each pair of first convolution kernels are different.
[0083] For example, there are two first convolution kernels, first convolution kernel A and first convolution kernel B.
[0084] In this embodiment, the target detection image is input into the first convolution kernel A, and a convolution operation is performed to obtain the first binarized image A output by the first convolution kernel A.
[0085] The target detection image is input into the first convolution kernel B, and a convolution operation is performed to obtain the first binarized image B output by the first convolution kernel B.
[0086] One of the first binarized image A and the first binarized image B is used as the target binarized image to determine the electrode ceramic region.
[0087] It should be noted that the convolution scales of the first convolution kernel A and the first convolution kernel B are different, and the detailed image features of the electrode ceramic part represented by the first binarized image A and the first binarized image B are also different.
[0088] In practice, the binarized image with more detailed image features representing the electrode ceramic part in the first binarized image A and the first binarized image B can be selected as the target binarized image.
[0089] In some embodiments, the first convolution kernel is obtained by performing mathematical operations based on at least two second convolution kernels with different convolution scales.
[0090] The first convolution kernel is obtained by performing mathematical operations on the second convolution kernels with different convolution scales. The second convolution kernel can be a convolution kernel with a fixed convolution size. By performing mathematical operations on the second convolution kernels with different convolution scales, multiple first convolution kernels with different convolution scales can be obtained.
[0091] In practice, mathematical operations on at least two second convolution kernels can include operations such as adding or subtracting convolution kernels.
[0092] For example, the second convolutional kernel can be three: a large convolutional kernel (max_filter), a linear convolutional kernel (line_filter), and a small convolutional kernel (min_filter).
[0093] The first convolutional kernel can be obtained by subtracting the linear convolutional kernel from the large convolutional kernel. The first convolutional kernel is max_filter-line_filter.
[0094] The first convolutional kernel can be obtained by subtracting the smaller convolutional kernel from the larger one; the first convolutional kernel is max_filter - min_filter.
[0095] The following is a specific example.
[0096] like Figure 3 As shown, the second convolutional kernel can be max_filter, line_filter, or min_filter.
[0097] The target detection image is input into max_filter-line_filter for convolution operation to obtain PIC_A, and the target detection image is input into max_filter-min_filter for convolution operation to obtain PIC_B.
[0098] Both PIC_A and PIC_B are images that include detailed features of the electrode ceramic portion. Defect detection of the electrode ceramic can be performed based on either PIC_A or PIC_B.
[0099] In this embodiment, the target binarized image PIC_C = PIC_A OR PIC_B.
[0100] In this embodiment, after performing convolution operations at different convolution scales on the target detection image, a target binarized image PIC_C is obtained. The obtained target binarized image PIC_C can highlight the contour of the electrode ceramic part, which helps to accurately segment the image of the area where the electrode ceramic is located for defect detection.
[0101] In some embodiments, step 130, determining the electrode ceramic region in the target detection image based on the target binarized image, includes:
[0102] Obtain the cumulative mean of the pixel values in each column of the target binarized image;
[0103] The column of pixels in the binarized image of the target is identified as the edge of the ceramic electrode;
[0104] The electrode ceramic region is determined based on the edge of the electrode ceramic.
[0105] In this embodiment, the pixel values of each column of the target binarized image are summed, and the sum is divided by the number of rows in the target binarized image to obtain the cumulative mean of that column.
[0106] For example, if the target binarized image has 1000 rows and 2048 columns, for each column of the target binarized image, there are 2048 pixel values added together to obtain the total pixel value. Dividing this total pixel value by the number of rows 1000 of the target binarized image will give the cumulative average value of each column.
[0107] The difference between the cumulative mean of two adjacent pixel columns can be used to obtain the pixel distance between the two adjacent pixel columns. Pixel columns with abrupt changes in the cumulative mean in the target binarized image and pixel columns with larger pixel distances are the electrode ceramic edges.
[0108] like Figure 4 As shown, traverse all pixel columns in the target binarized image, calculate the cumulative mean of pixel values in each pixel column of the target binarized image, and obtain the pixel column with a sudden change in cumulative mean.
[0109] In practice, the position of the A mutation of the cumulative mean can be set as the starting edge A of the electrode ceramic, and the position of the B mutation of the next cumulative mean can be set as the edge B of the electrode ceramic.
[0110] The area between electrode ceramic edge A and electrode ceramic edge B is the electrode ceramic region.
[0111] The position of the C mutation of the cumulative mean can be set as the starting edge C of the electrode ceramic, and the position of the D mutation of the next cumulative mean mutation can be set as the edge D of the electrode ceramic.
[0112] The area between electrode ceramic edge C and electrode ceramic edge D is the electrode ceramic region.
[0113] It should be noted that, as Figure 2 As shown, the ceramic area is elongated and thin. By matching the areas between the edges of the electrode ceramics using these elongated and thin strips, it can be determined whether the area between two electrode ceramic edges is an electrode ceramic area or another coating area.
[0114] In this embodiment, the edge-finding algorithm, which calculates the cumulative mean of the pixel columns of the target binarized image, accurately determines the electrode ceramic region in the target detection image and accurately divides the defect detection region. This helps to reduce the amount of data when performing defect detection, improves defect detection efficiency, and can also effectively avoid the influence of other regions on the defect detection of the electrode ceramic region.
[0115] In some embodiments, step 140, determining the defect detection result of the electrode ceramic based on the image feature information of the electrode ceramic region, includes:
[0116] The image of the electrode ceramic region is input into the target defect detection model to obtain the defect detection result of the electrode ceramic output by the target defect detection model;
[0117] The target defect detection model is built on the YOLOv5 neural network and trained using a training dataset.
[0118] In this embodiment, the target defect detection model is a deep learning model built and trained based on the YOLOv5 neural network. After accurately dividing the electrode ceramic region, deep learning detection is used to accurately detect the electrode ceramic region and obtain the defect detection result of the electrode ceramic.
[0119] In practice, the defect detection results of the electrode ceramic output by the target defect detection model can be processed by post-processing logic to back-calculate the defect location coordinates into the target detection image, thus completing the defect detection of the target detection image.
[0120] The target defect detection model is built and trained based on a YOLOv5 neural network, which can be one of YOLOv5s, YOLOv5m, YOLOv5l, and YOLOv5x.
[0121] The target defect detection model is trained using a training dataset to optimize the weight parameters in the YOLOv5 neural network. The training dataset can be a training set that includes sample images and their corresponding defects.
[0122] In this embodiment, a target defect detection model is built and trained based on the YOLOv5 neural network. The target defect detection model has high flexibility and detection speed, and the rapid deployment of the target defect detection model also has a strong advantage.
[0123] In some embodiments, the target defect detection model is trained through the following steps:
[0124] The training dataset is input into the target defect detection model to be trained, and the first model weight parameter of the target defect detection model is determined.
[0125] Export the weight parameters of the first model as an Open Neural Network Exchange Format file;
[0126] Based on the Open Neural Network Exchange Format file, the weights of the target defect detection model are pruned to determine the second model weight parameters of the target defect detection model, thus obtaining the trained target defect detection model.
[0127] In this embodiment, the target defect detection model is first trained by inputting the training dataset into it, and the weight parameters of the target defect detection model are updated. After the training dataset is completed, the first model weight parameters of the target defect detection model are obtained.
[0128] Export the weight parameters of the first model as an Open Neural Network Exchange (ONNX) exchange format file.
[0129] Among them, the Open Neural Network Exchange (ONNX) is a standard for representing deep learning models, which enables models to be transferred between different frameworks.
[0130] In this embodiment, the weight parameters of the first model are exported as an Open Neural Network Exchange Format (ONEF) file. The structure of the ONEF file can be as follows: Figure 5 As shown.
[0131] Here, "model" represents the model of the entire Open Neural Network Exchange Format file, which includes information such as graph structure and parser format, opset version and exporter type.
[0132] A model-graph represents a graph structure, which is typically the main structure seen by Netron.
[0133] model-graph-node represents all nodes in the graph. It is an array, such as conv, bn, etc., and the connection relationships between nodes are represented by input and output.
[0134] model-graph-initializer is the location where the weight classes are stored.
[0135] model-graph-input is the storage location for the input of the model in the Open Neural Network Exchange Format file. It indicates which nodes are input nodes and also includes information such as the shape of the input data.
[0136] model-graph-output is the storage location of the model's output in the Open Neural Network Exchange Format file. It indicates which nodes are output nodes and also includes information such as the shape of the output data.
[0137] In this embodiment, constant data of the anchorgrid class can be stored in the model-graph-node, and the storage type can be specified as Constant.
[0138] In this embodiment, after exporting the first model weight parameters as an Open Neural Network Exchange Format (ONE) file, the weights of the target defect detection model are optimized according to the ONE file, i.e., the weights of the target defect detection model are trimmed to obtain the second model weight parameters of the target defect detection model, thus completing the training of the target defect detection model.
[0139] For example, Figure 6 The diagram shown is a schematic representation of the structure of the first model weight parameters of the target defect detection model provided in this application embodiment. Figure 6 As shown, after training with the training dataset, the first model of the target defect detection model has a wide variety of weight parameters and complex connections between them.
[0140] Figure 7 The diagram shown is a schematic representation of the structure of the second model weight parameters of the target defect detection model provided in this application embodiment. Figure 7 As shown, after the weights of the target defect detection model are pruned and optimized according to the Open Neural Network Exchange Format file, A1-A4 are optimized to correspond to E1-E4. A5-A8 in the original weights are pruned and removed, resulting in a reduction in the types of weight parameters of the second model of the target defect detection model, and a simpler connection relationship between the weight parameters.
[0141] In this embodiment, the weights of the target defect detection model are pruned and optimized according to the Open Neural Network Exchange Format file, which can effectively improve the detection speed of the target defect detection model and also help to deploy the target defect detection model on edge devices with limited computing power, thus broadening the application environment of the target defect detection model for detecting electrode ceramic defects.
[0142] In practice, the target defect detection model can be deployed in the industrial vision system according to the weight parameters of the second model to achieve automated detection of defects on the electrode ceramic.
[0143] In related technologies, it is difficult to detect ceramic edge defects from all angles using traditional visual threshold segmentation. First, it is difficult to accurately segment the location of the ceramic at the boundary between the ceramic and other coatings. Second, the contrast between the ceramic and the defect foreground is low, and the defect shape is uncertain, making it difficult to accurately locate and identify the defect.
[0144] In this embodiment, convolution operations at different scales are performed, and the binarized image of the target is combined with an averaging operation to find the location of abrupt changes in the averaging, which is the edge of the electrode ceramic, thus accurately dividing the electrode ceramic region. Then, deep learning is used to detect defects in the electrode ceramic region, which can accurately identify defects with low foreground contrast or uncertain defect shapes, effectively improving the accuracy and precision of electrode ceramic defect detection.
[0145] The defect detection method for battery electrode ceramics provided in this application can be executed by a defect detection device for battery electrode ceramics. This application uses an example of a defect detection device for battery electrode ceramics executing the defect detection method to illustrate the defect detection device for battery electrode ceramics provided in this application.
[0146] This application also provides a defect detection device for battery electrode ceramics.
[0147] like Figure 8 As shown, the defect detection device for the battery electrode ceramic includes:
[0148] The acquisition module 810 is used to acquire a target detection image of the battery to be tested, the target detection image including the electrode ceramic of the battery to be tested;
[0149] The first processing module 820 is used to perform convolution operations at different convolution scales on the target detection image to obtain a target binarized image;
[0150] The second processing module 830 is used to determine the electrode ceramic region in the target detection image based on the target binarized image;
[0151] The third processing module 840 is used to determine the defect detection result of the electrode ceramic based on the image feature information of the electrode ceramic region.
[0152] According to the embodiment of this application, the defect detection device for battery electrode ceramics obtains a target binarized image by performing convolution operations at different scales on the target detection image, accurately determines the electrode ceramic region in the target detection image, accurately divides the defect detection region, accurately identifies the edge of the electrode ceramic and other coating edges, and improves the accuracy and precision of defect detection on the electrode ceramic.
[0153] In some embodiments, the first processing module 820, used to perform convolution operations at different convolution scales on the target detection image to obtain a target binarized image, may include:
[0154] The target detection image is input into at least two first convolution kernels with different convolution scales to obtain at least two first binarized images output by at least two first convolution kernels, and the first binarized images correspond one-to-one with the first convolution kernels;
[0155] One of the at least two first binarized images is determined as the target binarized image.
[0156] In some embodiments, the first convolution kernel is obtained by performing mathematical operations based on at least two second convolution kernels with different convolution scales.
[0157] In some embodiments, the second processing module 830, configured to determine the electrode ceramic region in the target detection image based on the target binarized image, may include:
[0158] Obtain the cumulative mean of the pixel values in each column of the target binarized image;
[0159] The column of pixels in the binarized image of the target is identified as the edge of the ceramic electrode;
[0160] The electrode ceramic region is determined based on the edge of the electrode ceramic.
[0161] In some embodiments, the third processing module 840 is used to determine the defect detection result of the electrode ceramic based on the image feature information of the electrode ceramic region, and may include:
[0162] The image of the electrode ceramic region is input into the target defect detection model to obtain the defect detection result of the electrode ceramic output by the target defect detection model;
[0163] The target defect detection model is built on the YOLOv5 neural network and trained using a training dataset.
[0164] In some embodiments, the target defect detection model is trained through the following steps:
[0165] The training dataset is input into the target defect detection model to be trained, and the first model weight parameter of the target defect detection model is determined.
[0166] Export the weight parameters of the first model as an Open Neural Network Exchange Format file;
[0167] Based on the Open Neural Network Exchange Format file, the weights of the target defect detection model are pruned to determine the second model weight parameters of the target defect detection model, thus obtaining the trained target defect detection model.
[0168] The defect detection device for battery electrode ceramics in this application embodiment can be an electronic device or a component of an electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal or other devices besides a terminal. For example, the electronic device can be a mobile phone, tablet computer, laptop computer, PDA, in-vehicle electronic device, mobile internet device (MID), augmented reality (AR) / virtual reality (VR) device, robot, wearable device, ultra-mobile personal computer (UMPC), netbook, or personal digital assistant (PDA), etc. It can also be a server, network attached storage (NAS), personal computer (PC), television (TV), ATM, or self-service machine, etc. This application embodiment does not specifically limit the device.
[0169] The defect detection device for battery electrode ceramics in this application embodiment can be a device with an operating system. This operating system can be Android, iOS, or other possible operating systems; this application embodiment does not specifically limit it.
[0170] The defect detection device for battery electrode ceramics provided in this application embodiment can achieve... Figures 1 to 7 The various processes implemented in the method implementation examples will not be described again here to avoid repetition.
[0171] In some embodiments, such as Figure 9 As shown, this application embodiment also provides an electronic device 900, including a processor 901, a memory 902, and a computer program stored in the memory 902 and executable on the processor 901. When the program is executed by the processor 901, it implements the various processes of the above-described battery electrode ceramic defect detection method embodiment and can achieve the same technical effect. To avoid repetition, it will not be described again here.
[0172] It should be noted that the electronic devices in the embodiments of this application include the aforementioned mobile electronic devices and non-mobile electronic devices.
[0173] Figure 10 A schematic diagram of the hardware structure of an electronic device to implement an embodiment of this application.
[0174] The electronic device 1000 includes, but is not limited to, components such as: radio frequency unit 1001, network module 1002, audio output unit 1003, input unit 1004, sensor 1005, display unit 1006, user input unit 1007, interface unit 1008, memory 1009, and processor 1010.
[0175] Those skilled in the art will understand that the electronic device 1000 may also include a power supply (such as a battery) for supplying power to various components. The power supply may be logically connected to the processor 1010 through a power management system, thereby enabling functions such as managing charging, discharging, and power consumption through the power management system. Figure 10 The electronic device structure shown does not constitute a limitation on the electronic device. The electronic device may include more or fewer components than shown, or combine certain components, or have different component arrangements, which will not be elaborated here.
[0176] The input unit 1004, in this embodiment of the application, is a camera, used to acquire a target detection image of the battery to be detected, the target detection image including the electrode ceramic of the battery to be detected;
[0177] The processor 1010 is used to perform convolution operations at different convolution scales on the target detection image to obtain a target binarized image;
[0178] Based on the target binarized image, determine the electrode ceramic region in the target detection image;
[0179] Based on the image feature information of the electrode ceramic region, the defect detection results of the electrode ceramic are determined.
[0180] According to the electronic device provided in the embodiments of this application, a target binarized image is obtained by performing convolution operations at different scales on the target detection image, accurately determining the electrode ceramic region in the target detection image, accurately dividing the defect detection region, accurately dividing the edge of the electrode ceramic and other coating edges, and improving the accuracy and precision of defect detection on the electrode ceramic.
[0181] In some embodiments, the processor 1010 is further configured to input the target detection image into at least two first convolution kernels of different convolution scales to obtain at least two first binarized images output by at least two first convolution kernels, wherein the first binarized images correspond one-to-one with the first convolution kernels;
[0182] One of the at least two first binarized images is determined as the target binarized image.
[0183] In some embodiments, the first convolution kernel is obtained by performing mathematical operations based on at least two second convolution kernels with different convolution scales.
[0184] In some embodiments, the processor 1010 is further configured to obtain the cumulative mean of the pixel values of each column of pixels in the target binarized image;
[0185] The column of pixels in the binarized image of the target is identified as the edge of the ceramic electrode;
[0186] The electrode ceramic region is determined based on the edge of the electrode ceramic.
[0187] In some embodiments, the processor 1010 is further configured to input an image of the electrode ceramic region into a target defect detection model to obtain the defect detection result of the electrode ceramic output by the target defect detection model;
[0188] The target defect detection model is built on the YOLOv5 neural network and trained using a training dataset.
[0189] In some embodiments, the target defect detection model is trained through the following steps:
[0190] The training dataset is input into the target defect detection model to be trained, and the first model weight parameter of the target defect detection model is determined.
[0191] Export the weight parameters of the first model as an Open Neural Network Exchange Format file;
[0192] Based on the Open Neural Network Exchange Format file, the weights of the target defect detection model are pruned to determine the second model weight parameters of the target defect detection model, thus obtaining the trained target defect detection model.
[0193] It should be understood that, in this embodiment, the input unit 1004 may include a graphics processing unit (GPU) 10041 and a microphone 10042. The GPU 10041 processes image data of still images or videos obtained by an image capture device (such as a camera) in video capture mode or image capture mode. The display unit 1006 may include a display panel 10061, which may be configured in the form of a liquid crystal display, an organic light-emitting diode, or the like. The user input unit 1007 includes a touch panel 10071 and at least one of other input devices 10072. The touch panel 10071 is also called a touch screen. The touch panel 10071 may include a touch detection device and a touch controller. Other input devices 10072 may include, but are not limited to, physical keyboards, function keys (such as volume control buttons, power buttons, etc.), trackballs, mice, and joysticks, which will not be described in detail here.
[0194] The memory 1009 can be used to store software programs and various data. The memory 1009 may primarily include a first storage area for storing programs or instructions and a second storage area for storing data. The first storage area may store the operating system, application programs or instructions required for at least one function (such as sound playback, image playback, etc.). Furthermore, the memory 1009 may include volatile memory or non-volatile memory, or both. The non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus RAM (DRRAM). The memory 1009 in this embodiment includes, but is not limited to, these and any other suitable types of memory.
[0195] Processor 1010 may include one or more processing units; processor 1010 integrates an application processor and a modem processor, wherein the application processor mainly handles operations involving the operating system, user interface, and applications, and the modem processor mainly handles wireless communication signals, such as a baseband processor. It is understood that the aforementioned modem processor may also not be integrated into processor 1010.
[0196] This application also provides a non-transitory computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the various processes of the above-described battery electrode ceramic defect detection method embodiment and achieves the same technical effect. To avoid repetition, it will not be described again here.
[0197] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.
[0198] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method for detecting defects in battery electrode ceramics.
[0199] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.
[0200] This application embodiment also provides a chip, which includes a processor and a communication interface. The communication interface and the processor are coupled. The processor is used to run programs or instructions to implement the various processes of the above-described battery electrode ceramic defect detection method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be described again here.
[0201] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.
[0202] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.
[0203] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of this application.
[0204] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.
[0205] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.
[0206] Although embodiments of this application have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the claims and their equivalents.
Claims
1. A method for detecting defects in battery electrode ceramics, characterized in that, include: Acquire a target detection image of the battery to be tested, the target detection image including the electrode ceramic of the battery to be tested; The target detection image is subjected to convolution operations at different convolution scales to obtain a target binarized image; Based on the target binarized image, the electrode ceramic region in the target detection image is determined; Based on the image feature information of the electrode ceramic region, the defect detection result of the electrode ceramic is determined; The step of determining the defect detection result of the electrode ceramic based on the image feature information of the electrode ceramic region includes: The image of the electrode ceramic region is input into the target defect detection model to obtain the defect detection result of the electrode ceramic output by the target defect detection model; The target defect detection model is built based on the YOLOv5 neural network and trained using a training dataset. The target defect detection model is trained through the following steps: The training dataset is input into the target defect detection model to be trained for training, and the first model weight parameters of the target defect detection model are determined. Export the weight parameters of the first model as an Open Neural Network Exchange Format file; Based on the open neural network exchange format file, the target defect detection model is weighted and the second model weight parameters of the target defect detection model are determined to obtain the trained target defect detection model.
2. The defect detection method for battery electrode ceramics according to claim 1, characterized in that, The step of performing convolution operations at different convolution scales on the target detection image to obtain a target binarized image includes: The target detection image is input into at least two first convolution kernels with different convolution scales to obtain at least two first binarized images output by the at least two first convolution kernels, and the first binarized images correspond one-to-one with the first convolution kernels; One of the at least two first binarized images is determined as the target binarized image.
3. The defect detection method for battery electrode ceramics according to claim 2, characterized in that, The first convolution kernel is obtained by performing mathematical operations based on at least two second convolution kernels with different convolution scales.
4. The defect detection method for battery electrode ceramics according to claim 1, characterized in that, The step of determining the electrode ceramic region in the target detection image based on the target binarized image includes: Obtain the cumulative mean of the pixel values in each column of the target binarized image; The column of pixels in the target binarized image whose cumulative mean abruptly changes is identified as the edge of the ceramic electrode. The electrode ceramic region is determined based on the edge of the electrode ceramic.
5. A defect detection device for battery electrode ceramics, characterized in that, The method for defect detection of battery electrode ceramics as described in any one of claims 1-4 includes: An acquisition module is used to acquire a target detection image of the battery to be tested, wherein the target detection image includes the electrode ceramic of the battery to be tested; The first processing module is used to perform convolution operations at different convolution scales on the target detection image to obtain a target binarized image; The second processing module is used to determine the electrode ceramic region in the target detection image based on the target binarized image; The third processing module is used to determine the defect detection result of the electrode ceramic based on the image feature information of the electrode ceramic region.
6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the defect detection method for battery electrode ceramics as described in any one of claims 1-4.
7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the defect detection method for battery electrode ceramics as described in any one of claims 1-4.
8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the defect detection method for battery electrode ceramics as described in any one of claims 1-4.