CN112270659A - Rapid detection method and system for surface defects of pole piece of power battery - Google Patents
Rapid detection method and system for surface defects of pole piece of power battery Download PDFInfo
- Publication number
- CN112270659A CN112270659A CN202010894239.3A CN202010894239A CN112270659A CN 112270659 A CN112270659 A CN 112270659A CN 202010894239 A CN202010894239 A CN 202010894239A CN 112270659 A CN112270659 A CN 112270659A
- Authority
- CN
- China
- Prior art keywords
- pole piece
- power battery
- image
- defects
- defect
- 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
- 230000007547 defect Effects 0.000 title claims abstract description 173
- 238000001514 detection method Methods 0.000 title claims abstract description 58
- 238000012549 training Methods 0.000 claims abstract description 24
- 238000000034 method Methods 0.000 claims abstract description 23
- 230000004927 fusion Effects 0.000 claims abstract description 18
- 238000000605 extraction Methods 0.000 claims abstract description 13
- 239000013598 vector Substances 0.000 claims description 13
- 238000012360 testing method Methods 0.000 claims description 8
- 238000012634 optical imaging Methods 0.000 claims description 7
- 230000007847 structural defect Effects 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 4
- 230000002708 enhancing effect Effects 0.000 claims description 2
- 230000011218 segmentation Effects 0.000 claims description 2
- 230000001678 irradiating effect Effects 0.000 claims 1
- 238000012031 short term test Methods 0.000 claims 1
- 238000013135 deep learning Methods 0.000 description 11
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 238000002372 labelling Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 239000011800 void material Substances 0.000 description 2
- 206010027146 Melanoderma Diseases 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000000576 coating method Methods 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000008570 general process Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000000135 prohibitive effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
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
-
- 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
- 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
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- 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
-
- 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
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30204—Marker
Abstract
The invention discloses a method and a system for rapidly detecting surface defects of a pole piece of a power battery, wherein the method comprises the following steps: acquiring surface images of the power battery pole pieces with different types of defects, and marking the defects on each image to obtain a detection data set of the surface defects of the power battery pole pieces with different types of defects; respectively extracting the features of each image in a one-to-one correspondence manner by using four parallel feature extraction networks, and then fusing by using a multi-mode fusion network to establish a pole piece surface defect identification model; training a pole piece surface defect identification model; and acquiring an image of the power battery pole piece to be detected, inputting the image into a pole piece surface defect identification model, and judging the defect condition of the pole piece surface through an output result. The invention can detect four images on the upper surface and the lower surface of the pole piece at one time, doubles the detection speed under the condition of unchanging hardware conditions, or reduces the requirement on the hardware performance under the same detection speed, thereby saving the cost.
Description
Technical Field
The invention relates to a method and a system for rapidly detecting surface defects of a pole piece of a power battery.
Background
In the coating process of the pole piece of the power battery, the quality and the service life of the battery are seriously influenced by the defects of cavities, scratches and the like generated on the surface of the pole piece. Machine vision is used for detecting the surface defects of the pole pieces so as to find and control the defects in time, the quality of the battery can be effectively improved, and the cost is reduced.
In recent years, classification and detection methods based on deep learning are gradually used in the field of detection of surface defects of pole pieces of power batteries based on machine vision, and the general processes are as follows: firstly, obtaining a sample image and marking defects, then training a neural network by using the marked sample, enabling the neural network to learn the characteristics of the sample, and then detecting the image to be detected by using the trained neural network. However, due to the fact that the image to be detected is large in sampling data quantity and many types of defects to be detected, the scheme is difficult to meet requirements of real-time performance and universality of defect detection.
Although real-time requirements can be achieved to some extent with high-performance GPUs, the equipment cost is prohibitive. How to improve the detection rate while reducing the cost is an urgent problem to be solved when a deep learning method is applied to detect the surface defects of the pole piece.
Disclosure of Invention
The invention aims to provide a method and a system for rapidly detecting surface defects of a pole piece of a power battery, and aims to solve the problem that the existing pole piece surface defect detection mode based on deep learning cannot give consideration to both detection efficiency and equipment cost.
Therefore, the invention provides a method for rapidly detecting surface defects of a pole piece of a power battery, which comprises the following steps: s1, obtaining surface images of the pole piece of the power battery with different types of defects, and marking the defects on each image to obtain a surface defect detection data set of the pole piece of the power battery with different types of defects; the surface image of the power battery pole piece comprises an upper surface bright domain image, an upper surface dark domain image, a lower surface bright domain image and a lower surface dark domain image of the power battery pole piece; s2, respectively extracting the characteristics of the upper surface bright domain image, the lower surface bright domain image, the upper surface dark domain image and the lower surface dark domain image in a one-to-one correspondence manner by using four parallel characteristic extraction networks, and then fusing by using a multi-mode fusion network to establish a pole piece surface defect identification model; s3, dividing the surface defect detection data set of the power battery pole piece into a training set and a testing set, and training the pole piece surface defect identification model; and S4, acquiring the upper surface bright domain image, the upper surface dark domain image, the lower surface bright domain image and the lower surface dark domain image of the power battery pole piece to be detected at one time, inputting the group of data into the trained pole piece surface defect identification model, and judging the defect condition of the pole piece surface through the output result.
The invention also provides a system for rapidly detecting the surface defects of the pole piece of the power battery, which comprises the following components: the acquisition unit is used for acquiring surface images of the power battery pole piece with different types of defects, wherein the surface images of the power battery pole piece comprise an upper surface bright domain image, an upper surface dark domain image, a lower surface bright domain image and a lower surface dark domain image of the power battery pole piece; the modeling unit is used for marking the defects on each image to obtain a power battery pole piece surface defect detection data set with different types of defects; respectively extracting the characteristics of the upper surface bright domain image, the lower surface bright domain image, the upper surface dark domain image and the lower surface dark domain image by using four parallel characteristic extraction networks, and then fusing by using a multi-mode fusion network to establish a pole piece surface defect identification model; dividing the power battery pole piece surface defect detection data set into a training set and a testing set, and training the pole piece surface defect identification model; and the detection unit is used for acquiring the surface image acquisition of the power battery pole piece to be detected, then inputting the group of data into the trained pole piece surface defect identification model, and judging the defect condition of the pole piece surface through the output result.
According to the method for rapidly detecting the surface defects of the power battery pole piece, the defects are marked by acquiring an upper surface bright domain image set, a lower surface bright domain image set, an upper surface dark domain image set and a lower surface dark domain image set of the power battery pole piece; training a pole piece surface defect recognition model which is constructed in advance and based on multi-mode fusion deep learning by using the image set after defect marking; the pole piece surface defect recognition model based on multi-mode fusion deep learning and used for completing training is used for carrying out on-line recognition on the pole piece surface defects, four images of the upper surface and the lower surface of the pole piece can be detected at one time, the detection efficiency is greatly improved, and the detection cost is reduced.
In addition to the objects, features and advantages described above, other objects, features and advantages of the present invention are also provided. The present invention will be described in further detail below with reference to the drawings.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this application, illustrate embodiments 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 flow chart of a method for rapidly detecting surface defects of a pole piece of a power battery according to the invention;
FIG. 2 is a schematic diagram of a power cell pole piece surface image acquisition according to the present invention; and
FIG. 3 is a structural diagram of a system for rapidly detecting surface defects of pole pieces of a power battery according to the present invention.
Detailed Description
The present invention will be described in detail below with reference to the embodiments with reference to the attached drawings.
The invention provides a method for rapidly detecting surface defects of a pole piece of a power battery, which can detect four images on the upper surface and the lower surface of the pole piece at one time, double the detection speed under the condition of unchanging hardware conditions, or reduce the requirement on hardware performance under the same detection speed, save the cost and solve the problem that the existing pole piece surface defect detection mode based on deep learning can not give consideration to both the detection efficiency and the equipment cost.
The invention provides a method for rapidly detecting surface defects of a pole piece of a power battery, which comprises the following steps of:
s1: and acquiring pole piece surface images with different types of defects to form a training set, and marking the defects on each image to obtain pole piece surface defect detection data sets with different types of defects.
S2: and fusing the upper surface bright domain image, the lower surface bright domain image, the upper surface dark domain image and the lower surface dark domain image by using a feature extraction network and a multi-mode fusion network, and establishing a pole piece surface defect identification model based on multi-mode fusion deep learning.
S3: and dividing the pole piece surface defect detection data set into a training set and a testing set, and training a pole piece surface defect recognition model based on multi-mode fusion deep learning.
S4: and (4) carrying out image acquisition on the surface of the pole piece to be detected by using the method in S1, inputting the data into the pole piece surface defect identification model trained in S3, and judging the defect condition of the surface of the pole piece through outputting a result.
In step S1, the surface image of the pole piece is obtained by two sets of novel optical imaging systems working synchronously, and the two sets of novel optical imaging systems are respectively used for acquiring the upper surface image and the lower surface image of the pole piece.
As shown in fig. 2, the novel optical imaging system includes: novel LED light source and binocular camera shoot device, pole piece surface image acquisition process is as follows:
step 1, setting a light source signal controller to control the on and off of two paths of LEDs according to the type of the surface defect of a power battery pole plate to be detected, and determining an LED light path;
step 2, adjusting the irradiation angle through the irradiation angle conversion device, and adjusting the Fresnel lens, the optical filter and the attenuation sheet by using the light guide device according to the determined LED light path to ensure that the uniformity, the irradiation angle and the brightness of light irradiated on the surface of the polar plate meet the defect detection requirements;
step 3, adjusting the working parameters of the camera, including sensing the light path change through a sensor arranged in the camera, and automatically matching the peak wavelength of the light source; sensing the change of the transmission speed of the polar plate through an encoder arranged on the transmission device, adjusting the scanning frequency of a camera and the like, and ensuring to obtain images with constant quality;
and 4, adjusting the positions of the bright domain and the dark domain of the two cameras through the camera signal controller to ensure that the two cameras work synchronously, and combining a bright domain imaging receiving mode and a dark domain imaging receiving mode to obtain defect details to the maximum extent.
In step S1, the pole piece surface image includes a pole piece upper surface image and a pole piece lower surface image; the pole piece upper surface image comprises an upper surface bright domain image and an upper surface dark domain image, and the pole piece lower surface image comprises a lower surface bright domain image and a lower surface dark domain image.
The defect condition on pole piece surface divide into normally, structural defect and color defect, structural defect includes cavity, mar etc., the color defect includes filth, black spot etc..
The image labeling mode comprises the following steps: and marking 0 when the surface image of the pole piece has no defect, marking 1 when the surface image has a structural defect, and marking 2 when the surface image has a color defect.
Preferably, when the surface image of the pole piece has a void defect, the surface image is marked with 101, when the surface image has a scratch defect, the surface image is marked with 102, and so on. And marking 201 when the surface image of the pole piece has a dirty defect, marking 202 when the surface image of the pole piece has a black point defect, and so on.
The pole piece surface defect detection data set utilizes a novel optical imaging system to acquire images of the same plane position and the same angle (or different angles) of the pole piece surface, namely four images of an upper surface bright domain image, a lower surface bright domain image, an upper surface dark domain image and a lower surface dark domain image are obtained as a group of defect detection data, and the defect condition of the pole piece surface is taken as a label of each image, namely each group of pole piece surface defect detection data comprises an upper surface bright domain image, a lower surface bright domain image, an upper surface dark domain image, a lower surface dark domain image and a defect condition label; and forming a pole piece surface defect detection data set by using the collected pole piece surface defect detection data of all groups.
As shown in fig. 3, the feature extraction network uses four parallel DenseNet-121 networks, and takes the collected upper surface bright domain image, lower surface bright domain image, upper surface dark domain image and lower surface dark domain image of each group of pole pieces as input for extracting defect features, so as to obtain four vectors with length N, and connects the vectors to obtain a vector with length 4N as an extracted feature vector to be output.
The multimode fusion network adopts 2-4 layers of fully-connected networks for defect detection, the output of the previous layer of fully-connected network is used as the input of the next layer of fully-connected network, the input of the first layer of fully-connected network is a feature vector with the length of 4N obtained by the feature extraction network, the output of the last layer of fully-connected network is a feature vector representing defect conditions, the length of the feature vector is equal to the number of types of defect condition labels contained in a group of input detection data, each element of the feature vector respectively represents the probability of each type of defect conditions, and the type with the maximum probability and the probability exceeding a set threshold is determined defect classification.
The feature extraction network further comprises a preprocessing unit and a clipping unit. The preprocessing unit is used for denoising and enhancing the acquired surface image of the pole piece and improving the contrast; the cutting unit is used for cutting the preprocessed pole piece surface image into a preset pixel size, extracting a to-be-detected region from the image in a cutting mode by adopting a threshold segmentation algorithm, zooming the segmented to-be-detected region image and storing the zoomed to-be-detected region image as an image sample.
In step S3, the upper surface bright field image, the lower surface bright field image, the upper surface dark field image and the lower surface dark field image in the training set are respectively input into four parallel DenseNet-121 networks of the pole piece surface defect recognition model established in S2, and the pole piece surface defect recognition model is trained through a back propagation algorithm; and judging the training effect of the pole piece surface defect identification model by using the test set so as to obtain four single-channel pole piece surface defect identification models after training. The multi-mode fusion network fuses the four single-channel pole piece surface defect recognition models, the four DenseNet-121 networks are fused by a concat method and are input into the full connection layer together, so that the correlation characteristics among different modes are learned, and the trained pole piece surface defect recognition model based on the multi-mode fusion deep learning is obtained.
In order to obtain more sample sizes and avoid the influence of insufficient data in the pole piece surface defect detection data set on the model training effect, the training set and the test set can be divided in the extended pole piece surface defect detection data set, the dividing proportion can be set by self, and the default is 8: and 2, the default pole piece surface defect detection data set has sufficient data and does not carry out expansion operation.
The data set expansion method comprises the following steps: randomly performing the same rotation and translation data enhancement operation on any group of data in the pole piece surface defect detection data set obtained in the step S1 to generate a group of new data; and adding all the generated new data into the pole piece surface defect detection data set to obtain the expanded pole piece surface defect detection data set.
When the train shows a descending trend and tends to be stable, the model initialization training is completed.
In step S4, the collected top surface bright domain image, bottom surface bright domain image, top surface dark domain image, and bottom surface dark domain image of the pole piece to be detected are input into the trained pole piece surface defect identification model, and a defect condition label with the highest confidence level in the feature vectors representing the pole piece surface defect condition is given.
The defect condition includes but is not limited to a defect identification and a defect type. The defect identifier is a label value set by the defect labeling mode, for example, the defect identifier of the void defect is 101; the defect types are classified into normal, structural and color defects.
And when the defect condition is identified as a defect, giving an alarm, and supporting alarm modes such as voice broadcast, large-screen popup window and audible and visual alarm.
Optionally, the defect type can be marked by different colors for large screen warning.
Compared with the prior art, the detection method has the following technical effects/advantages.
According to the detection method, the upper surface bright domain image set, the lower surface bright domain image set, the upper surface dark domain image set and the lower surface dark domain image set of the pole piece of the power battery are obtained, so that high-quality pictures of structural defects such as cavities and scratches on the upper surface and the lower surface of the pole piece and color defects such as dirt and black points can be obtained, and therefore defect details are obtained to the maximum extent.
The detection method adopts the DenseNet-121 network to train the single-channel pole piece surface defect recognition model, and utilizes the multi-mode fusion network to fuse the single-channel pole piece surface defect recognition model, so that the multi-defect type recognition capability of the model is improved, the trained pole piece surface defect recognition model based on the multi-mode fusion deep learning can adapt to the detection of various power battery pole piece surface image defects with different defect characteristics and different contrasts, and the application range of the model is widened.
The detection method utilizes the pole piece surface defect recognition model which is trained and based on multi-mode fusion deep learning to perform online recognition on the pole piece surface defects, can detect four images of the upper surface and the lower surface of the pole piece at one time, greatly improves the detection efficiency and reduces the detection cost.
The invention also provides a system for rapidly detecting the surface defects of the pole piece of the power battery, which comprises an acquisition unit, a modeling unit and a detection unit as shown in FIG. 3.
The acquisition unit is used for acquiring surface images of the pole piece of the power battery with different types of defects, wherein the surface images of the pole piece of the power battery comprise an upper surface bright domain image, an upper surface dark domain image, a lower surface bright domain image and a lower surface dark domain image of the pole piece of the power battery.
The modeling unit is used for marking the defects on each image to obtain power battery pole piece surface defect detection data sets with different types of defects, respectively extracting the characteristics of the upper surface bright domain image, the lower surface bright domain image, the upper surface dark domain image and the lower surface dark domain image by using four parallel characteristic extraction networks, then fusing by using a multi-mode fusion network to establish a pole piece surface defect identification model, finally dividing the power battery pole piece surface defect detection data sets into a training set and a testing set, and training the pole piece surface defect identification model.
The detection unit is used for acquiring an upper surface bright domain image, an upper surface dark domain image, a lower surface bright domain image and a lower surface dark domain image of the power battery pole piece to be detected at one time, then inputting the group of data into a pole piece surface defect recognition model after training, and judging the defect condition of the pole piece surface through an output result.
The detection system also comprises an alarm output unit which is used for alarming the detected defect condition of the surface of the pole piece.
The modeling unit and the detection unit in the rapid detection system for the surface defects of the pole piece of the power battery can be integrated in electronic equipment, and the electronic equipment can be a server or a terminal and other equipment.
The server may be an independent physical server, a server cluster or a distributed system formed by a plurality of physical servers, or a cloud server providing basic cloud computing services such as cloud service, a cloud database, cloud computing, a cloud function, cloud storage, Network service, cloud communication, middleware service, domain name service, security service, Network acceleration service (CDN), big data and an artificial intelligence platform.
The terminal can be a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart sound box, a smart watch and the like. The terminal and the server may be directly or indirectly connected by wired or wireless communication.
The prediction method is implemented by a combination of hardware and a computer program, which may be stored in a computer readable storage medium, which may include: read Only Memory (ROM), Random Access Memory (RAM), magnetic or optical disks, and the like.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A method for rapidly detecting surface defects of a pole piece of a power battery is characterized by comprising the following steps:
s1, obtaining surface images of the pole piece of the power battery with different types of defects, and marking the defects on each image to obtain a surface defect detection data set of the pole piece of the power battery with different types of defects; the surface image of the power battery pole piece comprises an upper surface bright domain image, an upper surface dark domain image, a lower surface bright domain image and a lower surface dark domain image of the power battery pole piece;
s2, respectively extracting the characteristics of the upper surface bright domain image, the lower surface bright domain image, the upper surface dark domain image and the lower surface dark domain image in a one-to-one correspondence manner by using four parallel characteristic extraction networks, and then fusing by using a multi-mode fusion network to establish a pole piece surface defect identification model;
s3, dividing the surface defect detection data set of the power battery pole piece into a training set and a testing set, and training the pole piece surface defect identification model; and
s4, collecting the upper surface bright domain image, the upper surface dark domain image, the lower surface bright domain image and the lower surface dark domain image of the power battery pole piece to be detected at one time, inputting the group of data into the trained pole piece surface defect recognition model, and judging the defect condition of the pole piece surface through the output result.
2. The method for rapidly detecting the surface defects of the power battery pole piece according to claim 1, characterized in that two sets of optical imaging systems working synchronously are adopted to acquire the upper surface bright field image, the lower surface bright field image, the upper surface dark field image and the lower surface dark field image of the power battery pole piece at one time.
3. The method for rapidly detecting the surface defects of the power battery pole piece according to claim 2, wherein two sets of optical imaging systems are respectively arranged on two sides of the power battery pole piece, and each set of optical imaging system comprises a light source for irradiating the surface of the power battery pole piece, a first camera for acquiring a bright domain image and a second camera for acquiring a dark domain image.
4. The method for rapidly detecting the surface defects of the power battery pole piece according to claim 3, wherein the first camera and the second camera collect images of the pole piece surface at the same plane and the same angle.
5. The method for rapidly detecting the surface defects of the pole pieces of the power battery according to claim 1, wherein a DenseNet-121 network is used as the feature extraction network, a 2-4-layer fully-connected network is used as the multi-mode fusion network for detecting the defects, the output of the previous layer of fully-connected network is used as the input of the next layer of fully-connected network, the input of the first layer of fully-connected network is a feature vector with the length of 4N obtained by the feature extraction network, the output of the last layer of fully-connected network is a feature vector for representing the defect conditions, the length of the feature vector is equal to the number of types of defect condition labels contained in a group of input detection data, each element of the feature vector respectively represents the probability of each type of defect conditions, and the type with the maximum probability and the probability exceeding a set threshold value is the determined defect classification.
6. The method for rapidly detecting the surface defects of the pole pieces of the power battery as claimed in claim 5, wherein the feature extraction network further comprises a preprocessing unit and a cutting unit, and the preprocessing unit is used for denoising and enhancing the acquired surface images of the pole pieces; the cutting unit is used for cutting the preprocessed pole piece surface image into a preset pixel size, extracting a to-be-detected region from the image in a cutting mode by adopting a threshold segmentation algorithm, zooming the segmented to-be-detected region image and storing the zoomed to-be-detected region image as an image sample.
7. The method for rapidly detecting the surface defects of the pole piece of the power battery as claimed in claim 1, wherein the defect conditions of the pole piece surface are divided into normal, structural defects and color defects, the structural defects comprise holes and scratches, and the color defects comprise dirt and black spots.
8. The method for rapidly detecting the surface defects of the pole piece of the power battery as claimed in claim 1, further comprising the step of alarming the defect condition of the surface of the pole piece.
9. The utility model provides a power battery pole piece surface defect short-term test system which characterized in that includes:
the acquisition unit is used for acquiring surface images of the power battery pole piece with different types of defects, wherein the surface images of the power battery pole piece comprise an upper surface bright domain image, an upper surface dark domain image, a lower surface bright domain image and a lower surface dark domain image of the power battery pole piece;
the modeling unit is used for marking the defects on each image to obtain a power battery pole piece surface defect detection data set with different types of defects; respectively extracting the characteristics of the upper surface bright domain image, the lower surface bright domain image, the upper surface dark domain image and the lower surface dark domain image by using four parallel characteristic extraction networks, and then fusing by using a multi-mode fusion network to establish a pole piece surface defect identification model; dividing the power battery pole piece surface defect detection data set into a training set and a testing set, and training the pole piece surface defect identification model; and
and the detection unit is used for acquiring the surface image acquisition of the power battery pole piece to be detected, inputting the group of data into the trained pole piece surface defect identification model, and judging the defect condition of the pole piece surface through the output result.
10. The system for rapidly detecting the surface defects of the pole pieces of the power battery as claimed in claim 9, further comprising an alarm output unit for alarming the defect condition of the pole piece surface.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010894239.3A CN112270659A (en) | 2020-08-31 | 2020-08-31 | Rapid detection method and system for surface defects of pole piece of power battery |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010894239.3A CN112270659A (en) | 2020-08-31 | 2020-08-31 | Rapid detection method and system for surface defects of pole piece of power battery |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112270659A true CN112270659A (en) | 2021-01-26 |
Family
ID=74349646
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010894239.3A Pending CN112270659A (en) | 2020-08-31 | 2020-08-31 | Rapid detection method and system for surface defects of pole piece of power battery |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112270659A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113658137A (en) * | 2021-08-17 | 2021-11-16 | 浙江工商大学 | Aircraft surface defect detection method and system based on capsule network |
| CN114418963A (en) * | 2021-12-28 | 2022-04-29 | 淮阴工学院 | Battery pole plate defect detection method based on machine vision |
| CN115829906A (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 defects of cathode pole piece of composite material belt |
| CN116630305A (en) * | 2023-07-18 | 2023-08-22 | 北京阿丘机器人科技有限公司 | Battery cover plate appearance detection method, device, equipment and storage medium |
| EP4303807A1 (en) * | 2022-07-05 | 2024-01-10 | Lapp Engineering AG | Method, measuring system and computer program product for colour testing |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN203310772U (en) * | 2013-05-15 | 2013-11-27 | 湖州师范学院 | Double-region large-size steel pipe welding line defect detection device |
| CN103785627A (en) * | 2014-01-27 | 2014-05-14 | 河南科技大学 | On-line lithium battery pole piece surface defect detection sorting system and method |
| CN107741427A (en) * | 2017-11-30 | 2018-02-27 | 南京航空航天大学 | The machine vision detection device and method of automobile engine turbine case surface defect |
| CN108108768A (en) * | 2017-12-29 | 2018-06-01 | 清华大学 | Photovoltaic glass defect classification method and device based on convolutional neural networks |
| CN110006919A (en) * | 2018-12-10 | 2019-07-12 | 浙江大学台州研究院 | A kind of apparatus and method polishing the shallow scratch detection of quartz wafer |
| CN110146515A (en) * | 2019-04-19 | 2019-08-20 | 广州超音速自动化科技股份有限公司 | Plate drilling and surface defect detection apparatus and application method |
| CN110458828A (en) * | 2019-08-12 | 2019-11-15 | 广东工业大学 | A kind of laser welding defect identification method and device based on multi-modal fusion network |
| CN111028204A (en) * | 2019-11-19 | 2020-04-17 | 清华大学 | Cloth defect detection method based on multi-mode fusion deep learning |
-
2020
- 2020-08-31 CN CN202010894239.3A patent/CN112270659A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN203310772U (en) * | 2013-05-15 | 2013-11-27 | 湖州师范学院 | Double-region large-size steel pipe welding line defect detection device |
| CN103785627A (en) * | 2014-01-27 | 2014-05-14 | 河南科技大学 | On-line lithium battery pole piece surface defect detection sorting system and method |
| CN107741427A (en) * | 2017-11-30 | 2018-02-27 | 南京航空航天大学 | The machine vision detection device and method of automobile engine turbine case surface defect |
| CN108108768A (en) * | 2017-12-29 | 2018-06-01 | 清华大学 | Photovoltaic glass defect classification method and device based on convolutional neural networks |
| CN110006919A (en) * | 2018-12-10 | 2019-07-12 | 浙江大学台州研究院 | A kind of apparatus and method polishing the shallow scratch detection of quartz wafer |
| CN110146515A (en) * | 2019-04-19 | 2019-08-20 | 广州超音速自动化科技股份有限公司 | Plate drilling and surface defect detection apparatus and application method |
| CN110458828A (en) * | 2019-08-12 | 2019-11-15 | 广东工业大学 | A kind of laser welding defect identification method and device based on multi-modal fusion network |
| CN111028204A (en) * | 2019-11-19 | 2020-04-17 | 清华大学 | Cloth defect detection method based on multi-mode fusion deep learning |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113658137A (en) * | 2021-08-17 | 2021-11-16 | 浙江工商大学 | Aircraft surface defect detection method and system based on capsule network |
| CN114418963A (en) * | 2021-12-28 | 2022-04-29 | 淮阴工学院 | Battery pole plate defect detection method based on machine vision |
| CN114418963B (en) * | 2021-12-28 | 2023-12-05 | 淮阴工学院 | Battery plate defect detection method based on machine vision |
| CN115829906A (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 defects of cathode pole piece of composite material belt |
| CN115829906B (en) * | 2022-04-18 | 2024-01-30 | 宁德时代新能源科技股份有限公司 | Method, device and system for detecting defects of cathode pole piece of composite material belt |
| EP4303807A1 (en) * | 2022-07-05 | 2024-01-10 | Lapp Engineering AG | Method, measuring system and computer program product for colour testing |
| CN116630305A (en) * | 2023-07-18 | 2023-08-22 | 北京阿丘机器人科技有限公司 | Battery cover plate appearance detection method, device, equipment and storage medium |
| CN116630305B (en) * | 2023-07-18 | 2024-02-13 | 北京阿丘机器人科技有限公司 | Battery cover plate appearance detection method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112270659A (en) | Rapid detection method and system for surface defects of pole piece of power battery | |
| US11380232B2 (en) | Display screen quality detection method, apparatus, electronic device and storage medium | |
| CN111553929B (en) | Mobile phone screen defect segmentation method, device and equipment based on converged network | |
| CN111612763B (en) | Mobile phone screen defect detection method, device and system, computer equipment and medium | |
| KR102229594B1 (en) | Display screen quality detection method, device, electronic device and storage medium | |
| CN106875373B (en) | Mobile phone screen MURA defect detection method based on convolutional neural network pruning algorithm | |
| CN111415329B (en) | Workpiece surface defect detection method based on deep learning | |
| CN107123111B (en) | Deep residual error network construction method for mobile phone screen defect detection | |
| CN111814850A (en) | Defect detection model training method, defect detection method and related device | |
| CN111626112A (en) | Smoke video detection method and system based on lightweight 3D-RDNet model | |
| CN112766218B (en) | Cross-domain pedestrian re-recognition method and device based on asymmetric combined teaching network | |
| CN111339832A (en) | Method and device for detecting face synthetic image | |
| CN112686833A (en) | Industrial product surface defect detecting and classifying device based on convolutional neural network | |
| CN110544231A (en) | lithium battery electrode surface defect detection method based on background standardization and centralized compensation algorithm | |
| CN112446437A (en) | Goods shelf commodity specification identification method based on machine vision | |
| CN111145222A (en) | Fire detection method combining smoke movement trend and textural features | |
| CN111398291A (en) | Flat enameled electromagnetic wire surface flaw detection method based on deep learning | |
| CN116071315A (en) | Product visual defect detection method and system based on machine vision | |
| CN115205247A (en) | Method, device and equipment for detecting defects of battery pole piece and storage medium | |
| CN114612418A (en) | Method, device and system for detecting surface defects of mouse shell and electronic equipment | |
| CN111160100A (en) | Lightweight depth model aerial photography vehicle detection method based on sample generation | |
| CN114549489A (en) | Carved lipstick quality inspection-oriented instance segmentation defect detection method | |
| CN112052730B (en) | 3D dynamic portrait identification monitoring equipment and method | |
| CN116363136B (en) | On-line screening method and system for automatic production of motor vehicle parts | |
| Chen et al. | Real-time defect detection of TFT-LCD displays using a lightweight network architecture |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210126 |