CN113971663A - Pole piece folding detection method and device, detection equipment and storage medium - Google Patents
Pole piece folding detection method and device, detection equipment and storage medium Download PDFInfo
- Publication number
- CN113971663A CN113971663A CN202111254711.8A CN202111254711A CN113971663A CN 113971663 A CN113971663 A CN 113971663A CN 202111254711 A CN202111254711 A CN 202111254711A CN 113971663 A CN113971663 A CN 113971663A
- Authority
- CN
- China
- Prior art keywords
- pole piece
- image
- pole
- folded
- turnover
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
- G06T7/0008—Industrial image inspection checking presence/absence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Abstract
The invention relates to the technical field of batteries, in particular to a pole piece folding detection method, a pole piece folding detection device and a storage medium, wherein the pole piece folding detection method of a battery cell pole group comprises the following steps: acquiring an actual pole piece image of the battery cell pole group; inputting the actual pole piece image into a pre-trained pole piece turnover prediction model to obtain a plurality of initial prediction frames in the actual pole piece image, and determining a final prediction frame corresponding to each pole piece of the battery cell pole group according to the non-maximum suppression score of each initial prediction frame; calculating a confidence value of the final prediction frame corresponding to each pole piece, identifying the category of the final prediction frame corresponding to each pole piece, and detecting the turnover pole piece in the cell pole group when the confidence value is greater than or equal to a preset threshold value and the category is the turnover category. The method can automatically and accurately detect the folded pole pieces in the cell pole group, improve the detection efficiency, reduce the detection cost and improve the overall quality of products.
Description
Technical Field
The invention relates to the technical field of batteries, in particular to a pole piece folding detection method, a pole piece folding detection device, pole piece folding detection equipment and a storage medium.
Background
The power battery is used as a core component of the electric automobile, and if the power battery is unqualified, the performance and driving safety of the electric automobile are directly affected, so that the detection of the power battery before delivery is very important.
The pole piece folding detection is one of important detections for power battery detection, because a power battery pack is composed of a single battery, the single battery is composed of a plurality of cell pole groups, and each cell pole group generally uses a lamination process in a production process, it is necessary to detect whether the state of each pole piece in the cell pole group meets a process standard, that is, whether a folding pole piece exists is detected.
In the correlation technique, whether there is a pole piece that turns over in the detection electric core utmost point group through artifical visual detection usually, however, the artifical pole piece that carries out turns over a mode that detects that not only wastes a large amount of manpower and materials that turns over, and the efficiency that detects is lower, and the accuracy that detects is relatively poor moreover, can't detect accurately whether there is a pole piece that turns over in the electric core utmost point group, the whole quality of greatly reduced product.
Disclosure of Invention
In view of this, the present invention provides a method for detecting a folded pole piece of a battery cell pole group, which can automatically and accurately detect the folded pole piece in the battery cell pole group, improve the detection efficiency, reduce the detection cost, and improve the overall quality of the product.
In order to achieve the purpose, the technical scheme of the invention is realized as follows:
a pole piece folding detection method of a battery cell pole group comprises the following steps:
acquiring an actual pole piece image of the battery cell pole group;
inputting the actual pole piece image into a pre-trained pole piece turnover prediction model to obtain a plurality of initial prediction frames in the actual pole piece image, and determining a final prediction frame corresponding to each pole piece of the battery cell pole group according to the non-maximum suppression score of each initial prediction frame; and
calculating a confidence value of the final prediction frame corresponding to each pole piece, identifying the category of the final prediction frame corresponding to each pole piece, and detecting the turnover pole piece in the cell pole group when the confidence value is greater than or equal to a preset threshold value and the category is the turnover category.
Further, before inputting the actual pole piece image to the pre-trained pole piece folding prediction model, the method further includes:
and converting the actual pole piece image into an actual pole piece image in an RGB image format, and performing undistorted conversion to obtain a preprocessed actual pole piece image which is input into the pole piece folding prediction model.
Further, before inputting the actual pole piece image to the pre-trained pole piece folding prediction model, the method further includes:
collecting a folded pole piece image and a normal pole piece image of the cell pole group in training data;
carrying out image cleaning on the folded pole piece image, marking a folded area in the cleaned folded pole piece image, and generating a marking text of the folded pole piece image according to the coordinate and the size of the folded area;
inputting the folded pole piece image, the normal pole piece image, the label text, the image classification category and the prediction frame parameter for framing the folded area into a target detection model to perform pole piece folding prediction training, and obtaining the pre-trained pole piece folding prediction model after the training is completed.
Further, gather in the training data that the pole piece image and the normal pole piece image of turning over of electric core pole group include:
and preliminarily imaging four corners of the battery cell pole group by using X-ray, and generating a folded pole piece image and a normal pole piece image of the battery cell pole group after image enhancement is carried out by using a charge coupling element and an image enhancer.
Further, still include:
and if the confidence value is smaller than a preset threshold value and the category of the battery cell is a turnover category or the category of the battery cell is a non-turnover category, judging that the turnover pole piece is not detected in the battery cell pole group.
Compared with the prior art, the pole piece folding detection method of the battery cell pole group has the following advantages:
the method for detecting the folding of the pole piece of the battery cell pole group can automatically and accurately detect the folding pole piece in the battery cell pole group, does not need manual visual detection, effectively improves the accuracy of the folding detection of the pole piece of the battery cell pole group, saves the manpower and material resources for detection, greatly reduces the detection cost, effectively improves the detection efficiency, and further can improve the overall quality of a product.
The second objective of the present invention is to provide a device for detecting a folded pole piece of a battery cell pole group, which can automatically and accurately detect the folded pole piece in the battery cell pole group, thereby improving the detection efficiency, reducing the detection cost, and improving the overall quality of the product.
In order to achieve the purpose, the technical scheme of the invention is realized as follows:
the utility model provides a pole piece of electric core pole group turns over a detection device, includes:
the acquisition module is used for acquiring the actual pole piece image of the battery cell pole group;
the prediction module is used for inputting the actual pole piece image into a pre-trained pole piece turnover prediction model to obtain a plurality of initial prediction frames in the actual pole piece image, and determining a final prediction frame corresponding to each pole piece of the battery cell pole group according to the non-maximum suppression score of each initial prediction frame; and
the detection module is used for calculating a confidence value of the final prediction frame corresponding to each pole piece, identifying the category of the final prediction frame corresponding to each pole piece, and detecting the turnover pole piece in the cell pole group when the confidence value is greater than or equal to a preset threshold value and the category is the turnover category.
Further, still include:
the conversion module is used for converting the actual pole piece image into an actual pole piece image in an RGB image format before inputting the actual pole piece image into the pre-trained pole piece turnover prediction model, and carrying out undistorted conversion to obtain a preprocessed actual pole piece image which is input into the pole piece turnover prediction model;
the training module is used for acquiring a folded pole piece image and a normal pole piece image of the cell pole group in training data before inputting the actual pole piece image into the pre-trained pole piece folded prediction model; carrying out image cleaning on the folded pole piece image, marking a folded area in the cleaned folded pole piece image, and generating a marking text of the folded pole piece image according to the coordinate and the size of the folded area; inputting the folded pole piece image, the normal pole piece image, the label text, the image classification category and the prediction frame parameter for framing the folded area into a target detection model to perform pole piece folding prediction training, and obtaining the pre-trained pole piece folding prediction model after the training is completed;
wherein, the pole piece image and the normal pole piece image of turning over of electric core pole group in the training data of gathering includes: and preliminarily imaging four corners of the battery cell pole group by using X-ray, and generating a folded pole piece image and a normal pole piece image of the battery cell pole group after image enhancement is carried out by using a charge coupling element and an image enhancer.
Further, still include:
and the judging module is used for judging that the turnover pole piece is not detected in the cell pole group when the confidence value is smaller than a preset threshold value and the category of the turnover pole piece is a turnover category or the category of the turnover pole piece is a non-turnover category.
Compared with the prior art, the pole piece folding detection method of the battery cell pole group has the same advantages as the pole piece folding detection device of the battery cell pole group, and is not repeated herein.
The third purpose of the present invention is to provide a detection device, which can automatically and accurately detect the folded pole piece in the cell pole group, improve the detection efficiency, reduce the detection cost, and improve the overall quality of the product.
In order to achieve the purpose, the technical scheme of the invention is realized as follows:
a detection apparatus, comprising: the detection method comprises a memory, a processor and a computer program which is stored in the memory and can be run on the processor, wherein the processor executes the program to realize the pole piece folding detection method of the battery cell pole group according to the embodiment.
Compared with the prior art, the detection equipment and the pole piece folding detection method of the battery cell pole group have the same advantages, and are not described again here.
A fourth objective of the present invention is to provide a computer-readable storage medium, which can automatically and accurately detect a folded pole piece in a cell pole group, improve detection efficiency, reduce detection cost, and improve overall quality of a product.
In order to achieve the purpose, the technical scheme of the invention is realized as follows:
a computer-readable storage medium, on which a computer program is stored, where the program is executed by a processor, so as to implement the pole piece folding detection method for a cell pole group according to the foregoing embodiment.
Compared with the prior art, the storage medium and the method for detecting the folding of the pole piece of the battery cell pole group have the same advantages, and are not described again here.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate an embodiment of the invention and, together with the description, serve to explain the invention and not to limit the invention. In the drawings:
fig. 1 is a flowchart of a pole piece folding detection method for a battery cell pole group according to an embodiment of the present invention;
FIG. 2 is an exemplary diagram of a normal pole piece image according to an embodiment of the invention;
FIG. 3 is an exemplary diagram of a folded pole piece image according to an embodiment of the invention;
FIG. 4 is a diagram illustrating an exemplary final prediction block according to a first embodiment of the present invention;
FIG. 5 is a diagram illustrating an exemplary final prediction block according to a second embodiment of the present invention;
FIG. 6 is a diagram illustrating an example of a final prediction block according to a third embodiment of the present invention;
FIG. 7 is an exemplary graph of the over-killing rate of the pole piece folding detection according to the embodiment of the invention;
FIG. 8 is an exemplary diagram of the miss rate of the pole piece folding detection according to the embodiment of the present invention;
FIG. 9 is a flow chart of a pre-training of a pole piece folding prediction model according to an embodiment of the present invention;
FIG. 10 is a diagram illustrating an example of labeling folded pole piece images according to an embodiment of the present invention;
fig. 11 is a schematic block diagram of a pole piece folding detection apparatus of a cell pole group according to an embodiment of the present invention;
fig. 12 is a schematic structural diagram of a detection apparatus according to an embodiment of the present invention.
Detailed Description
It should be noted that the embodiments and features of the embodiments may be combined with each other without conflict.
With the increase of consumption level and the increase of environmental awareness, electric vehicles powered by lithium ion batteries have come to be widely used. The power battery pack of the electric automobile is composed of a single battery, the single battery is composed of a plurality of battery cell pole groups, when the battery cell pole group of each battery is in the production process, a lamination process is taken as an example, whether the state of each pole piece in the battery cell pole group meets the process standard needs to be detected, and if the state of the pole piece is turned over and turned over, the pole piece belongs to a defective product, the pole piece needs to be detected in time and scrapped.
In the related technology, when the battery cell pole piece folding detection is carried out, four corners of a battery cell pole group are imaged by using X-ray, and whether the battery cell pole group meets the process standard or not is judged by analyzing the pole piece image characteristics through images. However, the pole piece turnover detection in the related art is detected through traditional algorithms such as gray threshold segmentation, and because the position shape of the pole piece after turnover is irregular, the image is irregular in gray contour, and finally, because the imaging of the pole piece is complicated, the pole piece turnover abnormality cannot be accurately detected, and whether the turnover pole piece exists in the cell pole group or not can not be effectively identified, the cell pole group with the turnover pole needs to be visually selected completely in the later period of manual work, so that the detection workload is greatly increased, and a large amount of manpower and material resources are wasted.
In order to solve the problem that the conventional algorithm cannot accurately detect the abnormal folding of the pole piece in the battery production process, the invention provides the pole piece folding detection method of the battery cell pole group, which improves the accuracy of the folding detection of the battery cell pole piece, replaces the current manual visual inspection situation, reduces the manual participation and improves the product quality.
The present invention will be described in detail below with reference to the embodiments with reference to the attached drawings.
Fig. 1 is a flowchart of a pole piece folding detection method of a cell pole group according to an embodiment of the present invention.
As shown in fig. 1, the method for detecting the folding of the pole piece of the battery cell pole group according to the embodiment of the present invention includes the following steps:
step S101, collecting actual pole piece images of the cell pole group.
In this embodiment, in order to improve the imaging quality of the cell electrode group, four corners of the cell electrode group may be imaged by using an acquisition mode of an X-ray light, a CCD and an image intensifier, so as to obtain an actual pole piece image.
The pole piece image refers to an image obtained after imaging of a pole piece in the cell pole group, and can be divided into a normal pole piece image as shown in fig. 2 and a folded pole piece image as shown in fig. 3 according to whether the image includes a folded pole piece. As shown in fig. 2, in the battery pole group, the size of the negative electrode is larger than that of the positive electrode, the negative electrode wraps the positive electrode, the imaging rule is that one negative electrode is one positive electrode (one is higher and one is lower), and the folded pole piece is contrary to the shape, as shown in the square in fig. 3.
Step S102, inputting the actual pole piece image into a pole piece turnover prediction model trained in advance to obtain a plurality of initial prediction frames in the actual pole piece image, and determining a final prediction frame corresponding to each pole piece of the battery cell pole group according to the non-maximum suppression score of each initial prediction frame.
It can be understood that, in the embodiment of the present invention, the pre-trained pole piece folding prediction model and the non-maximum suppression mode can be used to determine the final prediction frame corresponding to each pole piece of the battery cell pole group, and the area in the battery cell pole group where the pole piece folding may exist can be automatically selected, so that the pole piece folding detection efficiency is improved, manual visual inspection is not needed, and the detection cost is reduced. The pre-trained pole piece folding prediction model will be explained in the following embodiments, and is not described here to avoid redundancy.
Specifically, the embodiment of the invention can utilize a pre-trained pole piece folding prediction model to predict pole piece folding of an actual pole piece image, output a prediction result, decode the prediction result to obtain a plurality of initial prediction frames, stack the plurality of initial prediction frames, select the prediction frame with the highest score in a certain area by using non-maximum inhibition, and use the prediction frame with the highest score as a final prediction frame. The Non-Maximum Suppression (NMS) refers to suppressing an element that is not a Maximum, and may be understood as a local Maximum search, where the local representation is a neighborhood, and the neighborhood has two variable parameters, namely, a dimension of the neighborhood and a size of the neighborhood.
In this embodiment, before inputting the actual pole piece image into the pole piece folding prediction model trained in advance, the method further includes: and converting the actual pole piece image into an actual pole piece image in an RGB image format, and performing undistorted conversion to obtain the preprocessed actual pole piece image which is input into the pole piece turnover prediction model.
In actual use, the embodiment of the present invention may load an actual pole piece image by using img.open, for example, the resolution of the original image may be 1600 × 1600, and convert the actual pole piece image into an RGB image format by using image.convert, so as to prevent the grayscale image from being incorrectly predicted, and improve the accuracy of prediction; then, the loaded image is transformed without distortion, for example, to 608 × 3, and the transformed image is transformed to a torch format and then input to a pre-trained pole piece flipping prediction model.
Step S103, calculating a confidence value of the final prediction frame corresponding to each pole piece, identifying the category of the final prediction frame corresponding to each pole piece, and detecting the turnover pole piece in the cell pole group when the confidence value is greater than or equal to a preset threshold value and the category is the turnover category.
The preset threshold may be specifically calibrated, and may be set to 0.5, for example, which is not limited in this respect. The confidence value is the confidence level that the boxed box really has the object and the confidence level that the boxed box includes all the characteristics of the whole object;
it can be understood that, after the final prediction frame is obtained, the embodiment of the present invention adjusts the final prediction frame, including XY coordinates, width and height, confidence values, classification categories, and the like of a rectangle of the prediction frame, where the classification categories may include a turnover category and a non-turnover category, and when the confidence value is greater than or equal to a preset threshold and the category is the turnover category, a boundary frame is drawn by using draw. And if the confidence value is smaller than a preset threshold value and the category of the confidence value is a turnover category or the category of the confidence value is a non-turnover category, judging that the turnover pole piece is not detected in the cell pole group.
In a specific application, the embodiment of the present invention may obtain the effect diagrams of the final prediction frame as shown in fig. 4, 5, and 6, where the left side in fig. 4, 5, and 6 is the original diagram of the actual pole piece image, and the right side is the detected effect diagram of the actual pole piece image, and the square frame in the diagram is the final prediction frame.
In this embodiment, the electrode sheet folding image and the electrode sheet normal image can be classified and stored locally for data recall, and a result signal is transmitted to a receiving end in a communication manner, if a PLC/MES system is provided in the device, a folding electrode sheet interception function can be performed in a downstream channel, and a cell electrode group with a folding electrode sheet can be timely discharged.
Further, verification of the missed killing rate and the missed killing rate is also performed in the embodiment of the present invention, where the missed killing rate is (method determines folded pole piece-real folded pole piece)/total detection number, and the missed killing rate is 1- (verification determines correct pole group number/total manual full-detection folded pole piece), and during verification, the manually detected folded pole group is used as a basis, and the method of the embodiment of the present invention is used to search for the cell pole group of the real folded pole piece in the determination result, if it is detected, the determination is accurate, otherwise, the missed killing is performed. As shown in fig. 7 and 8, the over-killing rate of the detection method of the embodiment of the invention is less than or equal to 2%, the killing-missing rate is 0%, and the turnover detection accuracy is high.
In some embodiments, in order to improve the accuracy of model prediction, before inputting an actual pole piece image into a pole piece folding prediction model trained in advance, the embodiment of the present invention may first perform model training of the pole piece folding prediction model before prediction, as shown in fig. 9, the pre-training of the pole piece folding prediction model includes the following steps:
step S201, collecting a folded pole piece image and a normal pole piece image of the cell pole group in the training data.
In this embodiment, the pole piece image and the normal pole piece image of turning over of electric core pole group in the training data of collection includes: and preliminarily imaging four corners of the battery cell pole group by using X-ray, and generating a folded pole piece image and a normal pole piece image of the battery cell pole group after image enhancement is carried out by using a charge coupling element and an image enhancer.
It can be understood that the four corners of the cell pole group can be imaged by using X-ray light, a CCD and an image intensifier, and the folded pole piece images can be collected.
Step S202, image cleaning is carried out on the folded pole piece image, a folded area is marked in the cleaned folded pole piece image, and a marking text of the folded pole piece image is generated according to the coordinate and the size of the folded area.
Specifically, after the effective folded pole piece image is screened out, the image can be selected by using cv2. getlotrationmatrix 2D, and the image imaging directions are ensured to be consistent. Labeling the folded pole piece images by using an image labeling tool labellimg, labeling a corresponding folded area on each image, and generating a labeled text corresponding to the images, as shown in fig. 10, wherein the format of the images may be a jpg format, and the format of the labeled text may be an xml format.
Because the probability of the turnover phenomenon of the pole piece is extremely low, the embodiment of the invention can adopt a data enhancement technology to process the insufficient data quantity of partial forms, including image turnover, rotation, cutting, deformation, scaling, noise, mirror image and the like, thereby solving the problem of insufficient sample quantity of the turnover defect.
Step S203, inputting the folded pole piece image, the normal pole piece image, the label text, the image classification category and the prediction frame parameter for framing the folded area into a target detection model to perform pole piece folding prediction training, and obtaining a pre-trained pole piece folding prediction model after the training is completed.
The target detection model can be a Yolov4 model, and a pytorech framework is used for building a network.
Specifically, an example of the training and parameter adjusting process of the pole piece folding prediction model is as follows:
(1) the classification category is set to two types: ok and edgefold. And generating 2007_ train file of the marked image, wherein the file comprises the position of the training picture, the position coordinate of the turnover defect mark and the classification. Wherein ok represents the non-flap category and edgefold represents the flap category.
(2) Setting a picture training size: 608*608. The size may be selected according to the video memory, 416x416 may be used when the video memory is smaller, and 608x608 may be used when the video memory is larger, which is not particularly limited.
(3) Loading classification categories: ok, edgefold and predictor box parameters: (12, 16), (19, 36), (40, 28), (36, 75), (76, 55), (72, 146), (142, 110), (192, 243) and (459, 401). The prediction frame parameters can be obtained by clustering in the ground truth box of all samples in the training set according to the turnover area and by using a k-means algorithm.
(4) Setting parameters of a training model:
model(input_shape＝(608,608)
learning_rate＝1e-3
Batch_size＝16
Init_Epoch＝0
Freeze_Epoch＝50
Cuda＝True,
normalize＝False
mosaic＝False
Cosine_lr＝False,
smooth_label＝0,
model_path＝'last.pth'
anchors_path＝'anchors.txt'
classs_path＝'classes.txt'
)
wherein, 1) learning _ rate: the learning rate, which determines whether and when the objective function can converge to a local minimum. 2) Batch _ size: the number of samples each batch contains. 3) An Epoch: one Epoch is the process of training all training samples once. 4) Cuda: and the video card assigns value to true and calls the GPU. 5) normalize: normalization is to limit the processed data (by some algorithm) to a certain range that you need. 6) mosaic: and enhancing mosaic data, wherein four pictures can be randomly cut and spliced to one picture to serve as training data. 7) Cosine _ scheduler: cosine annealing learning rate. 8) label _ smoothening: smoothing the label; smoothing the label, making the predicted result less confident of the class to which it belongs, serves to enhance robustness.
When the pole piece folding prediction model is actually used, the model parameters can be set as follows:
model(model_path＝'model_data/yolo4_weights.pth'
anchors_path＝'anchors.txt'
classs_path＝'classes.txt'
model_image_size＝(608,608,3)
Freeze_Epoch＝50
confidence＝0.5
iou＝0.3
cuda＝False
letterbox_image＝False
)
confidence value, confidence degree that the boxed box really has the object and confidence degree that the boxed box selected from the box includes all the characteristics of the whole object; iou (interaction over union): and (4) performing intersection comparison, and calculating the ratio of the intersection and union of the 'predicted frame' and the 'real frame'.
According to the pole piece folding and unfolding detection method of the battery cell pole group, provided by the embodiment of the invention, the folding and unfolding pole pieces in the battery cell pole group can be automatically and accurately detected, manual visual detection is not needed, the accuracy of pole piece folding and unfolding detection of the battery cell pole group is effectively improved, the manpower and material resources for detection are saved, the detection cost is greatly reduced, the detection efficiency is effectively improved, and the overall quality of a product can be further improved.
Further, as shown in fig. 11, an embodiment of the present invention further discloses a device 10 for detecting folding of a pole piece of a battery pole group, including: an acquisition module 100, a prediction module 200, and a detection module 300.
Specifically, as shown in fig. 11, the acquisition module 100 is configured to acquire an actual pole piece image of the cell pole group; the prediction module 200 is configured to input an actual pole piece image to a pole piece folding prediction model trained in advance, obtain a plurality of initial prediction frames in the actual pole piece image, and determine a final prediction frame corresponding to each pole piece of the cell pole group according to the non-maximum suppression score of each initial prediction frame; the detection module 300 is configured to calculate a confidence value of the final prediction frame corresponding to each pole piece, identify a category to which the final prediction frame corresponding to each pole piece belongs, and detect a folded pole piece in the cell pole group when the confidence value is greater than or equal to a preset threshold and the category to which the final prediction frame belongs is the folded category.
Further, still include: the conversion module is used for converting the actual pole piece image into an actual pole piece image in an RGB image format before inputting the actual pole piece image into a pre-trained pole piece turnover prediction model, and carrying out undistorted conversion to obtain a preprocessed actual pole piece image which is input into the pole piece turnover prediction model; the training module is used for acquiring a folded pole piece image and a normal pole piece image of the battery cell pole group in training data before inputting an actual pole piece image into a pre-trained pole piece folded prediction model; carrying out image cleaning on the folded pole piece image, marking a folded area in the cleaned folded pole piece image, and generating a marking text of the folded pole piece image according to the coordinate and the size of the folded area; inputting the folded pole piece image, the normal pole piece image, the label text, the image classification category and the prediction frame parameter for framing the folded area into a target detection model to perform pole piece folding prediction training, and obtaining a pre-trained pole piece folding prediction model after the training is completed;
wherein, the pole piece image and the normal pole piece image of turning over of electric core pole group in the training data of gathering includes: and preliminarily imaging four corners of the battery cell pole group by using X-ray, and generating a folded pole piece image and a normal pole piece image of the battery cell pole group after image enhancement is carried out by using a charge coupling element and an image enhancer.
Further, still include: and the judging module is used for judging that the turnover pole piece is not detected in the cell pole group when the confidence value is smaller than a preset threshold value and the category of the turnover pole piece is a turnover category or the category of the turnover pole piece is a non-turnover category.
It should be noted that, a specific implementation manner of the pole piece folding detection apparatus of the battery cell pole group in the embodiment of the present invention is similar to a specific implementation manner of the pole piece folding detection method of the battery cell pole group, and in order to reduce redundancy, details are not described here.
According to the pole piece folding detection device of the battery cell pole group, provided by the embodiment of the invention, the folding pole piece in the battery cell pole group can be automatically and accurately detected, manual visual detection is not needed, the accuracy of pole piece folding detection of the battery cell pole group is effectively improved, the manpower and material resources for detection are saved, the detection cost is greatly reduced, the detection efficiency is effectively improved, and the overall quality of a product can be further improved.
Fig. 12 is a schematic structural diagram of a detection apparatus provided in an embodiment of the present application. The detection device may include:
a memory 1201, a processor 1202, and a computer program stored on the memory 1201 and executable on the processor 1202.
The processor 1202 implements the method for detecting the pole piece folding of the cell pole group provided in the above embodiment when executing the program.
Further, the detection apparatus further includes:
a communication interface 1203 for communication between the memory 1201 and the processor 1202.
A memory 1201 for storing computer programs executable on the processor 1202.
The memory 1201 may comprise high-speed RAM memory, and may also include non-volatile memory (non-volatile memory), such as at least one disk memory.
If the memory 1201, the processor 1202 and the communication interface 1203 are implemented independently, the communication interface 1203, the memory 1201 and the processor 1202 may be connected to each other through a bus and perform communication with each other. The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, an Extended ISA (EISA) bus, or the like. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is shown in FIG. 12, but this is not intended to represent only one bus or type of bus.
Optionally, in a specific implementation, if the memory 1201, the processor 1202, and the communication interface 1203 are integrated on a chip, the memory 1201, the processor 1202, and the communication interface 1203 may complete mutual communication through an internal interface.
The present embodiment also provides a computer-readable storage medium, on which a computer program is stored, where the computer program is executed by a processor to implement the above method for detecting the folding of the pole piece of the cell pole group.
In the description herein, reference to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the application. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or N embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.
Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present application, "N" means at least two, e.g., two, three, etc., unless specifically limited otherwise.
Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more N executable instructions for implementing steps of a custom logic function or process, and alternate implementations are included within the scope of the preferred embodiment of the present application in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of implementing the embodiments of the present application.
It should be understood that portions of the present application may be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, any one or combination of the following techniques, which are known in the art, may be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application specific integrated circuit having an appropriate combinational logic gate circuit, a Programmable Gate Array (PGA), a Field Programmable Gate Array (FPGA), or the like.
It will be understood by those skilled in the art that all or part of the steps carried by the method for implementing the above embodiments may be implemented by hardware related to instructions of a program, which may be stored in a computer readable storage medium, and when the program is executed, the program includes one or a combination of the steps of the method embodiments.
Claims (10)
1. A pole piece folding detection method of a battery cell pole group is characterized by comprising the following steps:
acquiring an actual pole piece image of the battery cell pole group;
inputting the actual pole piece image into a pre-trained pole piece turnover prediction model to obtain a plurality of initial prediction frames in the actual pole piece image, and determining a final prediction frame corresponding to each pole piece of the battery cell pole group according to the non-maximum suppression score of each initial prediction frame; and
calculating a confidence value of the final prediction frame corresponding to each pole piece, identifying the category of the final prediction frame corresponding to each pole piece, and detecting the turnover pole piece in the cell pole group when the confidence value is greater than or equal to a preset threshold value and the category is the turnover category.
2. The method of claim 1, further comprising, prior to inputting the actual pole piece image to the pre-trained pole piece fold-over prediction model:
and converting the actual pole piece image into an actual pole piece image in an RGB image format, and performing undistorted conversion to obtain a preprocessed actual pole piece image which is input into the pole piece folding prediction model.
3. The method of claim 1, further comprising, prior to inputting the actual pole piece image to the pre-trained pole piece fold-over prediction model:
collecting a folded pole piece image and a normal pole piece image of the cell pole group in training data;
carrying out image cleaning on the folded pole piece image, marking a folded area in the cleaned folded pole piece image, and generating a marking text of the folded pole piece image according to the coordinate and the size of the folded area;
inputting the folded pole piece image, the normal pole piece image, the label text, the image classification category and the prediction frame parameter for framing the folded area into a target detection model to perform pole piece folding prediction training, and obtaining the pre-trained pole piece folding prediction model after the training is completed.
4. The method of claim 3, wherein collecting folded pole piece images and normal pole piece images of the cell pole group in the training data comprises:
and preliminarily imaging four corners of the battery cell pole group by using X-ray, and generating a folded pole piece image and a normal pole piece image of the battery cell pole group after image enhancement is carried out by using a charge coupling element and an image enhancer.
5. The method of claim 1, further comprising:
and if the confidence value is smaller than a preset threshold value and the category of the battery cell is a turnover category or the category of the battery cell is a non-turnover category, judging that the turnover pole piece is not detected in the battery cell pole group.
6. The utility model provides a pole piece of electric core pole group turns over a detection device, its characterized in that includes:
the acquisition module is used for acquiring the actual pole piece image of the battery cell pole group;
the prediction module is used for inputting the actual pole piece image into a pre-trained pole piece turnover prediction model to obtain a plurality of initial prediction frames in the actual pole piece image, and determining a final prediction frame corresponding to each pole piece of the battery cell pole group according to the non-maximum suppression score of each initial prediction frame; and
the detection module is used for calculating a confidence value of the final prediction frame corresponding to each pole piece, identifying the category of the final prediction frame corresponding to each pole piece, and detecting the turnover pole piece in the cell pole group when the confidence value is greater than or equal to a preset threshold value and the category is the turnover category.
7. The apparatus of claim 6, further comprising:
the conversion module is used for converting the actual pole piece image into an actual pole piece image in an RGB image format before inputting the actual pole piece image into the pre-trained pole piece turnover prediction model, and carrying out undistorted conversion to obtain a preprocessed actual pole piece image which is input into the pole piece turnover prediction model;
the training module is used for acquiring a folded pole piece image and a normal pole piece image of the cell pole group in training data before inputting the actual pole piece image into the pre-trained pole piece folded prediction model; carrying out image cleaning on the folded pole piece image, marking a folded area in the cleaned folded pole piece image, and generating a marking text of the folded pole piece image according to the coordinate and the size of the folded area; inputting the folded pole piece image, the normal pole piece image, the label text, the image classification category and the prediction frame parameter for framing the folded area into a target detection model to perform pole piece folding prediction training, and obtaining the pre-trained pole piece folding prediction model after the training is completed;
wherein, the pole piece image and the normal pole piece image of turning over of electric core pole group in the training data of gathering includes: and preliminarily imaging four corners of the battery cell pole group by using X-ray, and generating a folded pole piece image and a normal pole piece image of the battery cell pole group after image enhancement is carried out by using a charge coupling element and an image enhancer.
8. The apparatus of claim 6, further comprising:
and the judging module is used for judging that the turnover pole piece is not detected in the cell pole group when the confidence value is smaller than a preset threshold value and the category of the turnover pole piece is a turnover category or the category of the turnover pole piece is a non-turnover category.
9. A detection apparatus, comprising: the device comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor executes the program to realize the pole piece folding detection method of the battery cell pole group according to any one of claims 1 to 5.
10. A computer-readable storage medium, on which a computer program is stored, wherein the program is executed by a processor to implement the method for detecting the pole piece folding of the cell pole group according to any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111254711.8A CN113971663A (en) | 2021-10-27 | 2021-10-27 | Pole piece folding detection method and device, detection equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111254711.8A CN113971663A (en) | 2021-10-27 | 2021-10-27 | Pole piece folding detection method and device, detection equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113971663A true CN113971663A (en) | 2022-01-25 |
Family
ID=79588580
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111254711.8A Pending CN113971663A (en) | 2021-10-27 | 2021-10-27 | Pole piece folding detection method and device, detection equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113971663A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115829908A (en) * | 2022-04-18 | 2023-03-21 | 宁德时代新能源科技股份有限公司 | Method, device and system for detecting bevel of cathode pole piece of composite material belt |
-
2021
- 2021-10-27 CN CN202111254711.8A patent/CN113971663A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115829908A (en) * | 2022-04-18 | 2023-03-21 | 宁德时代新能源科技股份有限公司 | Method, device and system for detecting bevel of cathode pole piece of composite material belt |
| CN115829908B (en) * | 2022-04-18 | 2023-12-22 | 宁德时代新能源科技股份有限公司 | Method, device and system for detecting folding angle of cathode pole piece of composite material belt |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112241699A (en) | Object defect category identification method and device, computer equipment and storage medium | |
| CN113538430B (en) | Pole piece defect detection method, device, equipment and medium based on difference | |
| CN109685765B (en) | X-ray film pneumonia result prediction device based on convolutional neural network | |
| CN110441316B (en) | Battery surface defect detection method and detection system | |
| CN115294135B (en) | Battery edge curling and sealing quality detection method | |
| CN112465746A (en) | Method for detecting small defects in radiographic film | |
| CN115809982B (en) | Method, device and system for detecting cell crush injury | |
| CN114972258B (en) | Battery surface defect detection method and system based on machine vision and related equipment | |
| CN113971663A (en) | Pole piece folding detection method and device, detection equipment and storage medium | |
| CN114594114A (en) | Full-automatic online nondestructive detection method for lithium battery cell | |
| CN112837308A (en) | Building crack detection method, device, equipment and storage medium | |
| CN110334775B (en) | Unmanned aerial vehicle line fault identification method and device based on width learning | |
| CN110008949B (en) | Image target detection method, system, device and storage medium | |
| CN113487563B (en) | EL image-based self-adaptive detection method for hidden cracks of photovoltaic module | |
| CN111242898A (en) | Train pantograph abrasion detection method and system based on deep neural network | |
| CN116051542A (en) | Defect detection method and defect detection device | |
| CN114581446B (en) | Battery core abnormity detection method and system of laminated battery | |
| CN113686871A (en) | Visual inspection system of pole piece | |
| CN115330677A (en) | Pole piece abnormity detection method, device and system of battery cell and storage medium | |
| JP2008084109A (en) | Eye opening/closing determination device and eye opening/closing determination method | |
| CN114549628A (en) | Power pole inclination detection method, device, equipment and storage medium | |
| CN113870754A (en) | Method and system for judging defects of panel detection electronic signals | |
| CN116448666B (en) | Detection device and detection method for lithium battery box based on machine vision | |
| CN117474915B (en) | Abnormality detection method, electronic equipment and storage medium | |
| CN117058105A (en) | Battery pole piece defect detection method, computing device and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |