CN114749342B - Lithium battery pole piece coating defect identification method, device and medium - Google Patents
Lithium battery pole piece coating defect identification method, device and medium Download PDFInfo
- Publication number
- CN114749342B CN114749342B CN202210414846.4A CN202210414846A CN114749342B CN 114749342 B CN114749342 B CN 114749342B CN 202210414846 A CN202210414846 A CN 202210414846A CN 114749342 B CN114749342 B CN 114749342B
- Authority
- CN
- China
- Prior art keywords
- image
- lithium battery
- pole piece
- battery pole
- gray
- 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.)
- Active
Links
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B05—SPRAYING OR ATOMISING IN GENERAL; APPLYING FLUENT MATERIALS TO SURFACES, IN GENERAL
- B05C—APPARATUS FOR APPLYING FLUENT MATERIALS TO SURFACES, IN GENERAL
- B05C11/00—Component parts, details or accessories not specifically provided for in groups B05C1/00 - B05C9/00
- B05C11/10—Storage, supply or control of liquid or other fluent material; Recovery of excess liquid or other fluent material
- B05C11/1002—Means for controlling supply, i.e. flow or pressure, of liquid or other fluent material to the applying apparatus, e.g. valves
- B05C11/1005—Means for controlling supply, i.e. flow or pressure, of liquid or other fluent material to the applying apparatus, e.g. valves responsive to condition of liquid or other fluent material already applied to the surface, e.g. coating thickness, weight or pattern
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01M—PROCESSES OR MEANS, e.g. BATTERIES, FOR THE DIRECT CONVERSION OF CHEMICAL ENERGY INTO ELECTRICAL ENERGY
- H01M4/00—Electrodes
- H01M4/02—Electrodes composed of, or comprising, active material
- H01M4/04—Processes of manufacture in general
- H01M4/0402—Methods of deposition of the material
- H01M4/0409—Methods of deposition of the material by a doctor blade method, slip-casting or roller coating
Landscapes
- Engineering & Computer Science (AREA)
- Manufacturing & Machinery (AREA)
- Chemical & Material Sciences (AREA)
- Chemical Kinetics & Catalysis (AREA)
- Electrochemistry (AREA)
- General Chemical & Material Sciences (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
- Image Analysis (AREA)
Abstract
The application discloses a method, a device and a medium for identifying coating defects of a lithium battery pole piece, wherein the method comprises the following steps: collecting a coating image of a lithium battery pole piece; preprocessing the acquired coating image of the lithium battery pole piece; dividing a defect target area in the preprocessed lithium battery pole piece coating image, obtaining the coordinate position of the defect target area, extracting morphological characteristics, gray characteristics and texture characteristics of the defect target area, and carrying out characteristic fusion on the extracted characteristics to obtain a fusion characteristic vector; building a BP neural network, taking the fusion feature vector as the input of the BP neural network, and training the BP neural network; and acquiring an image to be identified, inputting the image to be identified into the trained BP neural network, and outputting an identification result. The application adopts a multi-feature fusion mode comprising morphology, gray scale and texture features, improves the identification accuracy of the coating defects of the lithium battery pole pieces, and can be widely applied to the technical field of machine vision defect detection.
Description
Technical Field
The application relates to the technical field of machine vision defect detection, in particular to a method, a device and a medium for identifying coating defects of a lithium battery pole piece.
Background
In recent years, with the rise of new energy automobiles, intelligent wearing, unmanned aerial vehicles and other emerging industries, the application of lithium batteries is more widely extended. While the lithium battery industry is vigorously developed, the safety accidents of the lithium battery also frequently occur, and the safety of the lithium battery is still an industry pain point. The battery has huge fire and explosion power, and can directly threaten the life and property safety of people.
The detection of the coating defect of the pole piece in lithium battery detection is regarded as a ring of the hardest core because of wide scanning area, high required precision, high detection speed, complex and various surface defect types and the like. How to reduce the miss rate and false detection rate to the greatest extent in a high-speed detection environment and improve the identification and classification accuracy of defects, which not only tests the performance of hardware, but also puts higher requirements on the efficiency of software algorithms.
Although research on lithium battery pole piece detection is not few, the performance of each algorithm is uneven, the object aimed at by the research is difficult to cover comprehensively, and the requirements of high-speed and high-precision quality inspection in the actual industrial field cannot be met.
Disclosure of Invention
In order to solve at least one of the technical problems existing in the prior art to a certain extent, the application aims to provide a method, a device and a medium for identifying coating defects of a lithium battery pole piece.
The technical scheme adopted by the application is as follows:
a lithium battery pole piece coating defect identification method comprises the following steps:
collecting a coating image of a lithium battery pole piece;
preprocessing the acquired coating image of the lithium battery pole piece;
dividing a defect target area in the preprocessed lithium battery pole piece coating image, obtaining the coordinate position of the defect target area, extracting morphological characteristics, gray characteristics and texture characteristics of the defect target area, and carrying out characteristic fusion on the extracted characteristics to obtain a fusion characteristic vector;
building a BP neural network, taking the fusion feature vector as the input of the BP neural network, and training the BP neural network;
and acquiring an image to be identified, preprocessing the image to be identified and dividing the image, inputting the preprocessed image to the trained BP neural network, and outputting an identification result.
Further, the preprocessing of the acquired coating image of the lithium battery pole piece comprises the following steps:
carrying out image smoothing denoising operation and image contrast enhancement operation on the acquired lithium battery pole piece coating image;
in the image smoothing denoising operation, a median filter is used for carrying out median filtering on a lithium battery pole piece coating image, interference information in the image is eliminated, key information in the lithium battery pole piece coating image is reserved, and the key information comprises contour information and edge information;
in the image contrast enhancement operation, the global linear gray scale transformation algorithm is adopted to enhance the coating image of the lithium battery pole piece so as to improve the definition and contrast of the image and strengthen the characteristics of the image target area.
Further, the segmenting the defect target area in the preprocessed lithium battery pole piece coating image to obtain the coordinate position of the defect target area includes:
dividing a defect target area in the preprocessed lithium battery pole piece coating image by adopting a threshold segmentation algorithm, and separating a local defect area from a background area to obtain a defect segmentation image represented by binarization;
adopting a morphological algorithm corrosion operation and an expansion operation to acquire an accurate segmentation target from the defect segmentation image;
and acquiring the coordinate position of the defect target area according to the accurate segmentation target, and marking the defect target area by adopting a rectangular frame.
Further, the morphological features include circularity, eccentricity, and minimum circumscribed rectangular aspect ratio, the gray features include gray mean, gray variance, gray entropy, and gray skew coefficient, and the texture features include energy, local uniformity, correlation, and contrast of the gray co-occurrence matrix.
Further, the calculation formula of the gray average value is:
the gray variance is calculated by the following formula:
the calculation formula of the gray entropy is as follows:
the calculation formula of the gray skew coefficient is as follows:
wherein L represents the gray level number of the target area, r i Represents the i-th gray level in the region, P (r i ) Representing a pixel gray value r in a target region i Probability of P (r) i ) The calculation formula of (2) is as follows:
where i denotes the number of gray levels of the target area, G (i) denotes the number of pixels in the target area with a gray value of i, and Num denotes the total number of pixels in the target area.
Further, the energy calculation formula of the gray level co-occurrence matrix is as follows:
the calculation formula of the local uniformity is as follows:
the calculation formula of the correlation is as follows:
the calculation formula of the contrast ratio is as follows:
wherein m and n respectively represent the gray values of two pixel points; d meterStep length, θ represents the direction, and P (m, n; d, θ) represents the frequency of the values of the gray values (m, n) of two pixels separated by d in the θ direction; mu (mu) 1 Sum mu 2 Representing the average value, sigma, of the rows and columns, respectively, in the target region of the image 1 Sum sigma 2 Representing standard deviations of rows and columns, respectively, in the image target area.
Further, the feature fusion of the extracted features to obtain a fused feature vector includes:
adopting a serial fusion mode to connect feature vectors in morphological features, gray features and texture features end to synthesize a multi-dimensional fusion feature vector;
and carrying out normalization processing on the fusion feature vectors by adopting a maximum and minimum normalization algorithm, so that the feature vectors are unified to the same dimension, and the comparability is further achieved.
Further, the training the BP neural network includes:
and obtaining the optimal initial weight and threshold value of the BP neural network by using a whale optimization algorithm so as to further improve the recognition accuracy of the model.
The application adopts another technical scheme that:
a lithium battery pole piece coating defect identification device, comprising:
at least one processor;
at least one memory for storing at least one program;
the at least one program, when executed by the at least one processor, causes the at least one processor to implement the method described above.
The application adopts another technical scheme that:
a computer readable storage medium, in which a processor executable program is stored, which when executed by a processor is adapted to carry out the method as described above.
The beneficial effects of the application are as follows: the application adopts a multi-feature fusion mode comprising morphology, gray scale and texture features, can effectively improve the identification accuracy of the coating defects of the lithium battery pole pieces and improve the performance level of the current detection of the coating defects of the lithium battery pole pieces.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following description is made with reference to the accompanying drawings of the embodiments of the present application or the related technical solutions in the prior art, and it should be understood that the drawings in the following description are only for convenience and clarity of describing some embodiments in the technical solutions of the present application, and other drawings may be obtained according to these drawings without the need of inventive labor for those skilled in the art.
FIG. 1 is a flow chart of a method for identifying coating defects of a lithium battery pole piece based on a multi-feature fusion strategy in an embodiment of the application;
FIG. 2 is a schematic diagram of a detection imaging system in accordance with an embodiment of the present application;
FIG. 3 is a schematic diagram of a BP neural network based on multi-feature fusion in an embodiment of the application; in the figure, F is the circularity of the defect area, E is the eccentricity of the defect area, r is the minimum circumscribing rectangular length-width ratio of the defect area, m is the gray average value of the defect area, E is the gray entropy of the defect area, cor is the correlation of the defect area, and con is the contrast of the defect area;
fig. 4 is a schematic diagram of a BP neural network parameter tuning flow based on a whale optimization algorithm in an embodiment of the present application;
fig. 5 is a flowchart of steps of a method for identifying a coating defect of a lithium battery pole piece based on a multi-feature fusion strategy in an embodiment of the application.
Detailed Description
Embodiments of the present application are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are illustrative only and are not to be construed as limiting the application. The step numbers in the following embodiments are set for convenience of illustration only, and the order between the steps is not limited in any way, and the execution order of the steps in the embodiments may be adaptively adjusted according to the understanding of those skilled in the art.
In the description of the present application, it should be understood that references to orientation descriptions such as upper, lower, front, rear, left, right, etc. are based on the orientation or positional relationship shown in the drawings, are merely for convenience of description of the present application and to simplify the description, and do not indicate or imply that the apparatus or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus should not be construed as limiting the present application.
In the description of the present application, a number means one or more, a number means two or more, and greater than, less than, exceeding, etc. are understood to not include the present number, and above, below, within, etc. are understood to include the present number. The description of the first and second is for the purpose of distinguishing between technical features only and should not be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated or implicitly indicating the precedence of the technical features indicated.
In the description of the present application, unless explicitly defined otherwise, terms such as arrangement, installation, connection, etc. should be construed broadly and the specific meaning of the terms in the present application can be reasonably determined by a person skilled in the art in combination with the specific contents of the technical scheme.
As shown in fig. 1, the embodiment provides a method for identifying coating defects of a lithium battery pole piece based on a multi-feature fusion strategy, which comprises the following steps:
s101, acquiring a lithium battery pole piece coating image in a detection imaging system shown in fig. 2, and intensively summarizing images containing defects for training and testing a BP neural network classification model.
S102, performing image preprocessing operation on the acquired lithium battery pole piece coating image to remove image noise, enhance image characteristics and improve imaging quality of the image.
S103, after a defect target area in the image is segmented by adopting a threshold segmentation algorithm, the coordinate position of the defect target area is obtained, the morphological, gray level and texture characteristics of the defect target area are extracted, and the extracted characteristics are subjected to characteristic fusion operation.
S104, building a BP neural network, as shown in fig. 3, taking the fusion feature vector as the input of the network, and respectively training and testing the neural network classification model.
S105, in the process of training the neural network classification model, as shown in fig. 4, the whale optimization algorithm is used for obtaining the optimal initial weight and the threshold value of the neural network, so as to further improve the recognition accuracy of the model.
As shown in fig. 5, the embodiment provides a method for identifying coating defects of a lithium battery pole piece based on a multi-feature fusion strategy, which specifically includes the following steps:
s201, collecting a target image.
As shown in FIG. 2, under the drive of the transmission system, the pole piece coiled tape with the width of 200mm moves towards the direction of the coiling machine at the speed of 60m/min, when the pole piece coiled tape passes through the detection area, the sensing system can collect image data with the imaging precision of 0.05mm under the illumination of the linear light source, and the collected data is transmitted into the central control system taking the industrial personal computer as a core through the data transmission line. In the sensing system, the front and back sides of the pole piece are scanned simultaneously in a mode of combining two sets of linear array cameras, so that the function of real-time synchronous detection of the front and back sides of the pole piece can be realized.
After image acquisition, high-quality lithium battery pole piece coating images with the resolution ratio of 5120 multiplied by 4096 pixels can be obtained, and the actual imaging range corresponding to each image is 250mm multiplied by 200mm. The acquired lithium battery pole piece coating image data set comprises 8 common lithium battery pole piece coating defects such as positive electrode scratches, negative electrode scratches, positive electrode aluminum leakage, negative electrode copper leakage, holes, cracks, abnormal stains, decarburization and the like.
S202, preprocessing the image.
After obtaining a real image acquired on site, an image preprocessing technology is used for improving the image quality, and the related image preprocessing technology mainly comprises modes of image smoothing and denoising, image contrast enhancement and the like.
In the image smoothing denoising, a median filter with the size of 7×7 is used for median filtering of the image, interference information in the image is eliminated, and key information such as contours, edges and the like in the image is reserved.
In the image contrast enhancement, the image is enhanced by adopting a global linear gray level transformation algorithm, so that the definition and contrast of the image are improved, and the characteristics of a target area of the image are enhanced. The global linear gray scale transformation is to multiply all pixel gray scales in an image by the same coefficient at the same time and then to do equal scale stretching or compression. The gray mapping formula of the global linear gray transformation is:
where f (i, j) and g (i, j) are pixel gray values of the input image and the output image at coordinates (i, j), respectively, [ a, b ] is a gray range of the input image, [ c, d ] is a gray range of the output image, [ c, d ] typically takes values of [0,255].
S203, dividing and positioning the defect target area.
Aiming at the problems of uneven gray distribution and small defect area occupation ratio of the coating defects of the lithium battery pole piece, the defects are segmented by adopting an adaptive threshold segmentation method, and the local defect areas are separated from the background areas, so that a defect segmentation image represented by binarization is finally obtained. The adaptive thresholding method is a local thresholding method, unlike the global thresholding method, the threshold of the algorithm is not a constant one. In the adaptive threshold segmentation method, each pixel point in an image can determine its own threshold value according to the gray average value of the neighboring pixel points around the pixel point, and the higher the average gray value is, the higher the threshold value is generally. In the OpenCV of the open source software library for computer vision, the adaptive threshold segmentation algorithm may be implemented by an adaptive threshold () function operator.
The target is then further segmented accurately, perfecting the segmentation details, using a morphological algorithm erosion and dilation operation. After the defect target is completely extracted, confirming the geometric center position coordinates of the defect, and marking the defect area by using a red rectangular frame.
S204, extracting defect characteristics.
The multi-thread mechanism is used for improving the detection and identification efficiency of the system, and in parallel secondary threads, the form, gray level and texture characteristic parameters of the defect area are respectively calculated and extracted.
Wherein morphology features include circularity, eccentricity, and minimum circumscribed rectangular aspect ratio.
The circularity can be calculated according to the formula (2):
wherein F is the circularity, S is the area of the region, and C is the circumference of the region.
The eccentricity can be calculated according to formula (3):
wherein M is ij Representing the i+j order center distance of the target region.
The minimum circumscribed rectangular aspect ratio can be calculated according to formula (4):
where L represents the length of the rectangular frame and W represents the width of the rectangular frame.
The gray scale features include a gray scale mean, a gray scale variance, a gray scale entropy, and a gray scale bias factor.
The gray average value can be calculated according to the formula (5):
the gray variance can be calculated according to the formula (6):
the gray entropy can be calculated according to the formula (7):
the gray scale skew factor can be calculated according to equation (8):
in the formulas (5) to (8), L represents the number of gray levels of the target region, r i Represents the i-th gray level in the region, P (r i ) Representing a pixel gray value r in a target region i Probability of P (r) i ) The calculation formula of (2) is as follows:
where i denotes the number of gray levels of the target area, G (i) denotes the number of pixels in the target area with a gray value of i, and Num denotes the total number of pixels in the target area.
The texture features comprise 4 main parameters of energy, local uniformity, correlation, contrast and the like of the gray level co-occurrence matrix.
The energy can be calculated according to the formula (10):
wherein m and n respectively represent gray values of two pixel points; d represents the step length, θ represents the direction, and P (m, n; d, θ) represents the frequency of the two pixel gray values (m, n) separated by d in the θ direction.
The local uniformity can be calculated according to formula (11):
the correlation can be calculated as in equation (12):
wherein: mu (mu) 1 Sum mu 2 Respectively representing the average value of the rows and columns in the image target area, wherein the calculation formulas are respectively shown in the formula (13) and the formula (14); sigma (sigma) 1 Sum sigma 2 The standard deviation of the rows and columns in the image target area is represented by the calculation formulas shown in the formulas (15) and (16), respectively.
The contrast can be calculated according to formula (17):
s205, fusing multiple features.
In the feature fusion strategy, a simple and easy serial fusion mode is adopted, namely, all single feature vectors are connected end to end, and finally an 11-dimensional fusion feature vector is synthesized. The expression of serial fusion is:
γ＝(α,β) (18)
where γ is a feature vector obtained by fusing the feature vector α and the feature vector β.
After feature fusion is implemented, the fusion feature vector is normalized by adopting a maximum and minimum normalization algorithm, so that the feature vector is unified to the same dimension, and the comparability is further achieved. In the maximum and minimum normalization, the final data can be set to fall within the [0,1] interval by the adjustment of the expression (19).
Wherein x is i Represents the characteristic value before normalization, z represents the characteristic value after normalization, max (x i ) And min (x) i ) Respectively represent characteristic value x i And the maximum and minimum of (a) are defined.
S206, designing a BP neural network.
Building a BP neural network, determining the number of network layers, determining the number of neuron nodes of an input layer, an hidden layer and an output layer, determining a learning rate and determining an expected error.
1. Determination of the network layer number: according to the theorem of mapping network existence, the network layer number of the BP neural network is set to be 3 layers. It is clear that the 3-layer BP neural network is sufficient to train a proper classification model and can meet the basic classification requirement.
2. Determining the number of input layer neuron nodes: since the number of dimensions of the input fusion feature vector is 11, the number of neuron nodes in the input layer is set to 11.
3. Determining the number of output layer neuron nodes: since the number of defect types to be identified is 8, the number of neuron nodes in the output layer is set to 8.
4. Determining the number of hidden layer neuron nodes: in the 3-layer BP neural network model, the number of neuron nodes of an implicit layer is usually calculated by an empirical formula, and the empirical formula selected in the embodiment is as follows:
wherein p is the number of neuron nodes of the hidden layer, m is the number of neuron nodes of the input layer, n is the number of neuron nodes of the output layer, and alpha is an adjustment constant between 1 and 10.
After calculation, the number of neuron nodes of the hidden layer is set to 10.
5. Determination of learning rate: the learning rate is related to the weight change rate of each iterative training of the neural network. The initial learning rate of this embodiment is set to 0.001, and in order to increase the training speed, an adaptive learning rate algorithm is used in the training process.
6. Determination of the expected error: the target error value of this embodiment is set to 0.005.
S207, training and optimizing the BP neural network classifier.
In the training of BP neural networks, the initial weights and initial thresholds are particularly important to the network, but are difficult to accurately acquire. If the initial weight and the initial threshold are randomly selected and the gradient descent algorithm is used for optimizing, the model is easily caused to sink into local optimum, and even the problems of network training oscillation, slow convergence speed and the like are likely to be caused. In order to reduce the above problems as much as possible, it is important to select an efficient network parameter tuning algorithm.
The whale optimization algorithm (Whale Optimization Algorithm, WOA) is a meta-heuristic intelligent optimization algorithm for simulating whale prey and prey behaviors, and is simple to operate, few in parameters, fast in speed, strong in stability and commonly used for solving optimization problems in various calculation problems.
After extensive evolution, whales develop a unique predation strategy, i.e. searching for, surrounding and capturing prey through a "spiral bubble network" path, and it is from these three predation steps that the whale optimization algorithm builds up a corresponding mathematical model.
The solution to find a problem can be analogized to the process of whale finding a prey, which is randomly found by whales, which builds the following mathematical model:
D＝|X(t)-CX rand (t)| (21)
X(t+1)＝X rand (t)-AD (22)
wherein t is the iteration number; x is the position vector of whale; x is X rand A whale position vector selected randomly, wherein an optimal solution is included; A. c is a coefficient vector, and the calculation formula is as follows:
A＝2ar 1 -a (23)
C＝2r 2 (24)
wherein r is 1 、r 2 Is interval [0,1]]Random vectors within the range; a is a convergence factor, decreasing linearly from 2 to 0 during the iteration; t (T) max Is the maximum number of iterations.
The whale optimization algorithm will typically start with a series of random solutions, updating the whale position through successive iterations. In the model, A takes a random value in the range of interval [ -a, a ], and the fluctuation range of A is continuously reduced along with the reduction of the convergence factor a, so that a search mechanism can be realized by adjusting the value of a. When the A is more than or equal to 1, the whale can expand the searching range, and a random solution can be selected to update the position vector of the whale; and when |A| < 1, the current optimal solution should be selected to update the whale's position vector.
The best solution to the problem can be analogized to the process of a whale surrounding the target prey, and if the current prey is identified as the best prey, the position vector of the whale needs to be updated, and the corresponding mathematical model is as follows:
D＝|X(t)-CX best (t)| (26)
X(t+1)＝X best (t)-AD (27)
wherein X is best Is the best position vector of the current whale.
Let the current whale position vector be X, and the target hunting object position vector be X * The whale may approach the target prey by adjusting vector a and vector C.
After identifying the prey, whales capture the prey by ascending the spiral shrinkage, and the corresponding mathematical model is as follows:
D′＝|X * (t)-X(t)| (28)
X(t+1)＝X * (t)+D′e bl cos(2πl) (29)
where b is a logarithmic spiral shape constant, l is a random number within the interval [ -1,1], and D' is the distance from the current position of the whale to the position of the prey.
In fact, the two actions of whale shrinkage surrounding and capturing the prey can occur simultaneously, the corresponding mathematical model is as follows:
X(t+1)＝X * (t)-AD,p<0.5 (30)
X(t+1)＝X * (t)+D′e bl cos(2πl),p≥0.5 (31)
in the formulas (30) and (31), p is a random number within the interval [0,1 ].
In this embodiment, the whale optimization algorithm is used to obtain the optimal initial weight and threshold of the neural network, and the training process of the network is completed on the basis of the optimal initial weight and threshold, and the corresponding algorithm flow chart is shown in fig. 4, and the specific steps are as follows:
(1) And (5) data acquisition and preprocessing. And collecting the characteristic data of the defect image, and taking the characteristic data as the input of the BP neural network model after serial fusion and normalization processing.
(2) The whale population is initialized. Setting population scale N and maximum iteration times T max And determining N groups of initial weights and thresholds by adopting a random selection method, and taking each group of initial weights and thresholds as an initial position vector of each whale.
(3) Taking the mean square error loss as the fitness value of each whale and incorporating the calculation, and recording the position generating the optimal fitness value, namely the optimal prey position X best 。
(4) Update parameters A, C, a, p, different location update policies are selected according to the values of p and A: updating the whale position using formula (31) when p is greater than or equal to 0.5; updating the whale position using formula (30) when p < 0.5 and |A| < 1; when p < 0.5 and |A| is not less than 1, the whale position is updated using formula (22).
(5) And (3) iterating the process (2) and the process (3) continuously, terminating the iterative process when the iteration times reach the maximum value or the error loss is reduced to a certain value, and outputting the optimal initial weight and the threshold value to the BP neural network.
140 defect data samples are distributed as training sets to train the neural network, and 60 defect data samples are distributed as test sets to evaluate the accuracy of the algorithm.
In order to study the influence of the feature fusion strategy on the classification performance, in this embodiment, the feature type is used as a variable, and the trained BP neural network model is tested by using a test set, and the result is shown in table 1.
Table 1 model classification accuracy under different features
The data in the table shows that the classification contribution effect of the texture features of the important numbers in the single type of features is most obvious, and the feature fusion strategy also enables the classification accuracy to be remarkably improved, and especially the 97.08% classification accuracy is more satisfactory after the texture features are introduced.
In order to study the excellent performance of the whale optimization algorithm, the experimental results of this example were shown in table 2, which were purposely compared with two outstanding, classical, commonly used parameter optimization algorithms, namely genetic algorithm (Genetic Algorithm, GA) and particle swarm optimization algorithm (Particle Swarm Optimization, PSO).
Table 2 model classification accuracy under different optimization algorithms
It can be seen from the table that, compared with the other two algorithms, for the BP neural network model, the whale optimization algorithm can obtain more excellent classification accuracy performance for the BP neural network model.
S208, performing image recognition by using the classification model.
A high-definition camera installed on a detection imaging system is used for acquiring a new lithium battery pole piece coating image, and the extracted defect characteristics are input into a trained BP neural network classification model after pretreatment and image segmentation, so that the identification of the lithium battery pole piece coating defect is realized.
In summary, compared with the prior art, the method of the application has the advantages and beneficial effects that:
(1) The application can realize accurate segmentation and positioning of the coating defect area of the lithium battery pole piece under the detection high precision of 0.05mm and the detection high speed of 60m/min, and has extremely low omission rate and false detection rate in the operation process.
(2) The application can accurately distinguish the types of coating defects, wherein the distinguishable types comprise 8 common coating defects of lithium battery pole pieces, such as positive electrode scratches, negative electrode scratches, positive electrode aluminum leakage, negative electrode copper leakage, holes, cracks, abnormal stains, decarburization and the like.
(3) The application adopts BP neural network to identify the type of the coating defect of the lithium battery pole piece, compared with other traditional classification algorithms, the neural network algorithm has higher classification accuracy, and well solves the problem of low accuracy of identifying the coating defect of the lithium battery pole piece at present.
(4) The application optimizes the initial weight and the threshold value of the neural network by using a whale optimization algorithm, obviously improves the accuracy of defect identification, and further meets the requirements of practical application.
(5) The application adopts a multi-feature fusion strategy, and practice proves that the multi-feature fusion mode comprising morphological, gray scale and texture features can obviously improve the identification accuracy of the coating defects of the lithium battery pole pieces, so that the application has popularization and practical significance.
The embodiment also provides a lithium battery pole piece coating defect identification device, which comprises:
at least one processor;
at least one memory for storing at least one program;
the at least one program, when executed by the at least one processor, causes the at least one processor to implement the method illustrated in fig. 5.
The device for identifying the coating defects of the lithium battery pole piece can be used for executing any combination implementation steps of the method embodiment, and has the corresponding functions and beneficial effects.
Embodiments of the present application also disclose a computer program product or computer program comprising computer instructions stored in a computer readable storage medium. The computer instructions may be read from a computer-readable storage medium by a processor of a computer device, and executed by the processor, to cause the computer device to perform the method shown in fig. 5.
The embodiment also provides a storage medium which stores instructions or programs capable of executing the method for identifying the coating defects of the lithium battery pole pieces, and when the instructions or programs are operated, the instructions or programs can execute any combination implementation steps of the method embodiments, and the method has corresponding functions and beneficial effects.
In some alternative embodiments, the functions/acts noted in the block diagrams may occur out of the order noted in the operational illustrations. For example, two blocks shown in succession may in fact be executed substantially concurrently or the blocks may sometimes be executed in the reverse order, depending upon the functionality/acts involved. Furthermore, the embodiments presented and described in the flowcharts of the present application are provided by way of example in order to provide a more thorough understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is changed, and in which sub-operations described as part of a larger operation are performed independently.
Furthermore, while the application is described in the context of functional modules, it should be appreciated that, unless otherwise indicated, one or more of the described functions and/or features may be integrated in a single physical device and/or software module or one or more functions and/or features may be implemented in separate physical devices or software modules. It will also be appreciated that a detailed discussion of the actual implementation of each module is not necessary to an understanding of the present application. Rather, the actual implementation of the various functional modules in the apparatus disclosed herein will be apparent to those skilled in the art from consideration of their attributes, functions and internal relationships. Accordingly, one of ordinary skill in the art can implement the application as set forth in the claims without undue experimentation. It is also to be understood that the specific concepts disclosed are merely illustrative and are not intended to be limiting upon the scope of the application, which is to be defined in the appended claims and their full scope of equivalents.
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
Logic and/or steps represented in the flowcharts or otherwise described herein, e.g., a ordered listing of executable instructions for implementing logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.
More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer readable medium may even be paper or other suitable medium on which the program is printed, as the program may be electronically captured, via, for instance, optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.
It is to be understood that portions of the present application may be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, the various steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, may be implemented using any one or combination of the following techniques, as is well known in the art: discrete logic circuits having logic gates for implementing logic functions on data signals, application specific integrated circuits having suitable combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), and the like.
In the foregoing description of the present specification, reference has been made to the terms "one embodiment/example", "another embodiment/example", "certain embodiments/examples", and the like, 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, schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
While embodiments of the present application have been shown and described, it will be understood by those of ordinary skill in the art that: many changes, modifications, substitutions and variations may be made to the embodiments without departing from the spirit and principles of the application, the scope of which is defined by the claims and their equivalents.
While the preferred embodiment of the present application has been described in detail, the present application is not limited to the above embodiments, and various equivalent modifications and substitutions can be made by those skilled in the art without departing from the spirit of the present application, and these equivalent modifications and substitutions are intended to be included in the scope of the present application as defined in the appended claims.
Claims (8)
1. The method for identifying the coating defects of the lithium battery pole piece is characterized by comprising the following steps of:
collecting a coating image of a lithium battery pole piece;
preprocessing the acquired coating image of the lithium battery pole piece;
dividing a defect target area in the preprocessed lithium battery pole piece coating image, obtaining the coordinate position of the defect target area, extracting morphological characteristics, gray characteristics and texture characteristics of the defect target area, and carrying out characteristic fusion on the extracted characteristics to obtain a fusion characteristic vector;
building a BP neural network, taking the fusion feature vector as the input of the BP neural network, and training the BP neural network;
acquiring an image to be identified, preprocessing the image to be identified and dividing the image, inputting the preprocessed image to a trained BP neural network, and outputting an identification result;
the morphological characteristics comprise circularity, eccentricity and minimum circumscribed rectangular length-width ratio, the gray characteristics comprise gray average value, gray variance, gray entropy and gray deflection coefficient, and the texture characteristics comprise energy, local uniformity, correlation and contrast of a gray co-occurrence matrix;
the energy calculation formula of the gray level co-occurrence matrix is as follows:
the calculation formula of the local uniformity is as follows:
the calculation formula of the correlation is as follows:
the calculation formula of the contrast ratio is as follows:
wherein m and n respectively represent the gray values of two pixel points; d represents the step length, θ represents the direction, and P (m, n; d, θ) represents the frequency of the two pixel gray values (m, n) separated by d in the θ direction; mu (mu) 1 Sum mu 2 Representing the average value, sigma, of the rows and columns, respectively, in the target region of the image 1 Sum sigma 2 Representing standard deviations of rows and columns, respectively, in the image target area.
2. The method for identifying coating defects of a lithium battery pole piece according to claim 1, wherein the preprocessing of the acquired coating image of the lithium battery pole piece comprises the following steps:
carrying out image smoothing denoising operation and image contrast enhancement operation on the acquired lithium battery pole piece coating image;
wherein in the image smoothing denoising operation, a median filter is used for carrying out median filtering on the lithium battery pole piece coating image, interference information in the image is eliminated, key information in the lithium battery pole piece coating image is reserved,
the key information comprises contour information and edge information;
in the image contrast enhancement operation, the global linear gray scale transformation algorithm is adopted to enhance the coating image of the lithium battery pole piece so as to improve the definition and contrast of the image and strengthen the characteristics of the image target area.
3. The method for identifying a coating defect of a lithium battery pole piece according to claim 1, wherein the step of dividing a defect target area in the pre-processed coating image of the lithium battery pole piece to obtain a coordinate position of the defect target area comprises the steps of:
dividing a defect target area in the preprocessed lithium battery pole piece coating image by adopting a threshold segmentation algorithm, and separating a local defect area from a background area to obtain a defect segmentation image represented by binarization;
adopting a morphological algorithm corrosion operation and an expansion operation to acquire an accurate segmentation target from the defect segmentation image;
and acquiring the coordinate position of the defect target area according to the accurate segmentation target, and marking the defect target area by adopting a rectangular frame.
4. The method for identifying coating defects of a lithium battery pole piece according to claim 1, wherein the calculation formula of the gray average value is as follows:
the gray variance is calculated by the following formula:
the calculation formula of the gray entropy is as follows:
the calculation formula of the gray skew coefficient is as follows:
wherein L represents the gray level number of the target area, r i Represents the i-th gray level in the region, P (r i ) Representing a pixel gray value r in a target region i Probability of P (r) i ) The calculation formula of (2) is as follows:
where i denotes the number of gray levels of the target area, G (i) denotes the number of pixels in the target area with a gray value of i, and Num denotes the total number of pixels in the target area.
5. The method for identifying coating defects of a lithium battery pole piece according to claim 1, wherein the feature fusion is performed on the extracted features to obtain a fused feature vector, and the method comprises the following steps:
adopting a serial fusion mode to connect feature vectors in morphological features, gray features and texture features end to synthesize a multi-dimensional fusion feature vector;
and carrying out normalization processing on the fusion feature vectors by adopting a maximum and minimum normalization algorithm, so that the feature vectors are unified to the same dimension, and the comparability is further achieved.
6. The method for identifying coating defects of a lithium battery pole piece according to claim 1, wherein the training of the BP neural network comprises:
and obtaining the optimal initial weight and threshold value of the BP neural network by using a whale optimization algorithm so as to further improve the recognition accuracy of the model.
7. The utility model provides a lithium battery pole piece coating defect recognition device which characterized in that includes:
at least one processor;
at least one memory for storing at least one program;
the at least one program, when executed by the at least one processor, causes the at least one processor to implement the method of any one of claims 1-6.
8. A computer readable storage medium, in which a processor executable program is stored, characterized in that the processor executable program is for performing the method according to any of claims 1-6 when being executed by a processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210414846.4A CN114749342B (en) | 2022-04-20 | 2022-04-20 | Lithium battery pole piece coating defect identification method, device and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210414846.4A CN114749342B (en) | 2022-04-20 | 2022-04-20 | Lithium battery pole piece coating defect identification method, device and medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114749342A CN114749342A (en) | 2022-07-15 |
| CN114749342B true CN114749342B (en) | 2023-09-26 |
Family
ID=82332091
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210414846.4A Active CN114749342B (en) | 2022-04-20 | 2022-04-20 | Lithium battery pole piece coating defect identification method, device and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114749342B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116188467B (en) * | 2023-04-26 | 2023-07-25 | 厦门微图软件科技有限公司 | Abnormality detection method for long surface of blade lithium battery shell |
| CN116416268B (en) * | 2023-06-09 | 2023-08-18 | 浙江双元科技股份有限公司 | Method and device for detecting edge position of lithium battery pole piece based on recursion dichotomy |
| CN116523908B (en) * | 2023-06-30 | 2023-09-12 | 湖南隆深氢能科技有限公司 | Safe production method, system, equipment and medium based on coil coating production line |
| CN117058468B (en) * | 2023-10-11 | 2023-12-19 | 青岛金诺德科技有限公司 | Image recognition and classification system for recycling lithium batteries of new energy automobiles |
| CN117152444B (en) * | 2023-10-30 | 2024-01-26 | 山东泰普锂业科技有限公司 | Equipment data acquisition method and system for lithium battery industry |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102289671A (en) * | 2011-09-02 | 2011-12-21 | 北京新媒传信科技有限公司 | Method and device for extracting texture feature of image |
| CN104608799A (en) * | 2014-12-12 | 2015-05-13 | 郑州轻工业学院 | Information fusion technology based train wheel set tread damage online detection and recognition method |
| CN108010069A (en) * | 2017-12-01 | 2018-05-08 | 湖北工业大学 | Optimize the rapid image matching method of algorithm and grey correlation analysis based on whale |
| CN109376728A (en) * | 2018-12-28 | 2019-02-22 | 华南农业大学 | A kind of weeds in paddy field recognition methods and its application based on multiple features fusion and BP neural network |
| CN110544231A (en) * | 2019-07-24 | 2019-12-06 | 华南理工大学 | lithium battery electrode surface defect detection method based on background standardization and centralized compensation algorithm |
| CN111612751A (en) * | 2020-05-13 | 2020-09-01 | 河北工业大学 | Lithium battery defect detection method based on Tiny-yolov3 network embedded with grouping attention module |
| CN111951253A (en) * | 2020-05-19 | 2020-11-17 | 惠州高视科技有限公司 | Method, device and readable storage medium for detecting surface defects of lithium battery |
| CN112016577A (en) * | 2019-05-30 | 2020-12-01 | 华为技术有限公司 | Image processing method and device |
| CN112529798A (en) * | 2020-12-04 | 2021-03-19 | 北京小米松果电子有限公司 | Training data acquisition, neural network training and image correction method and device |
| CN112633459A (en) * | 2019-09-24 | 2021-04-09 | 华为技术有限公司 | Method for training neural network, data processing method and related device |
| CN113011575A (en) * | 2019-12-19 | 2021-06-22 | 华为技术有限公司 | Neural network model updating method, image processing method and device |
| CN114120318A (en) * | 2021-11-29 | 2022-03-01 | 哈尔滨工业大学 | Dark field image target point accurate extraction method based on integrated decision tree |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| IL133326A0 (en) * | 1999-12-06 | 2001-04-30 | Nova Measuring Instr Ltd | Method and system for endpoint detection |
| US20190173128A1 (en) * | 2014-12-02 | 2019-06-06 | Polyplus Battery Company | Making and inspecting a web of vitreous lithium sulfide separator sheet and lithium electrode assemblies and battery cells |
-
2022
- 2022-04-20 CN CN202210414846.4A patent/CN114749342B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102289671A (en) * | 2011-09-02 | 2011-12-21 | 北京新媒传信科技有限公司 | Method and device for extracting texture feature of image |
| CN104608799A (en) * | 2014-12-12 | 2015-05-13 | 郑州轻工业学院 | Information fusion technology based train wheel set tread damage online detection and recognition method |
| CN108010069A (en) * | 2017-12-01 | 2018-05-08 | 湖北工业大学 | Optimize the rapid image matching method of algorithm and grey correlation analysis based on whale |
| CN109376728A (en) * | 2018-12-28 | 2019-02-22 | 华南农业大学 | A kind of weeds in paddy field recognition methods and its application based on multiple features fusion and BP neural network |
| CN112016577A (en) * | 2019-05-30 | 2020-12-01 | 华为技术有限公司 | Image processing method and device |
| CN110544231A (en) * | 2019-07-24 | 2019-12-06 | 华南理工大学 | lithium battery electrode surface defect detection method based on background standardization and centralized compensation algorithm |
| CN112633459A (en) * | 2019-09-24 | 2021-04-09 | 华为技术有限公司 | Method for training neural network, data processing method and related device |
| CN113011575A (en) * | 2019-12-19 | 2021-06-22 | 华为技术有限公司 | Neural network model updating method, image processing method and device |
| CN111612751A (en) * | 2020-05-13 | 2020-09-01 | 河北工业大学 | Lithium battery defect detection method based on Tiny-yolov3 network embedded with grouping attention module |
| CN111951253A (en) * | 2020-05-19 | 2020-11-17 | 惠州高视科技有限公司 | Method, device and readable storage medium for detecting surface defects of lithium battery |
| CN112529798A (en) * | 2020-12-04 | 2021-03-19 | 北京小米松果电子有限公司 | Training data acquisition, neural network training and image correction method and device |
| CN114120318A (en) * | 2021-11-29 | 2022-03-01 | 哈尔滨工业大学 | Dark field image target point accurate extraction method based on integrated decision tree |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114749342A (en) | 2022-07-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114749342B (en) | Lithium battery pole piece coating defect identification method, device and medium | |
| CN111428748B (en) | HOG feature and SVM-based infrared image insulator identification detection method | |
| Rachmadi et al. | Vehicle color recognition using convolutional neural network | |
| CN111178197B (en) | Mass R-CNN and Soft-NMS fusion based group-fed adherent pig example segmentation method | |
| CN109684922B (en) | Multi-model finished dish identification method based on convolutional neural network | |
| CN106340016B (en) | A kind of DNA quantitative analysis method based on microcytoscope image | |
| CN111008959A (en) | Grading ring defect detection method, device, medium and equipment based on aerial image | |
| CN109903282B (en) | Cell counting method, system, device and storage medium | |
| CN113822880A (en) | Crack identification method based on deep learning | |
| CN115797813B (en) | Water environment pollution detection method based on aerial image | |
| CN112528845B (en) | Physical circuit diagram identification method based on deep learning and application thereof | |
| CN114419413A (en) | Method for constructing sensing field self-adaptive transformer substation insulator defect detection neural network | |
| CN113313107A (en) | Intelligent detection and identification method for multiple types of diseases on cable surface of cable-stayed bridge | |
| CN107871315B (en) | Video image motion detection method and device | |
| CN116168240A (en) | Arbitrary-direction dense ship target detection method based on attention enhancement | |
| CN111814852A (en) | Image detection method, image detection device, electronic equipment and computer-readable storage medium | |
| Hacıefendioğlu et al. | Wood construction damage detection and localization using deep convolutional neural network with transfer learning | |
| Mi et al. | Research on steel rail surface defects detection based on improved YOLOv4 network | |
| Burton et al. | RustSEG--Automated segmentation of corrosion using deep learning | |
| CN111368865A (en) | Method and device for detecting remote sensing image oil storage tank, readable storage medium and equipment | |
| CN110349119B (en) | Pavement disease detection method and device based on edge detection neural network | |
| CN116824330A (en) | Small sample cross-domain target detection method based on deep learning | |
| Wang et al. | FPA-DNN: a forward propagation acceleration based deep neural network for ship detection | |
| CN113496159B (en) | Multi-scale convolution and dynamic weight cost function smoke target segmentation method | |
| CN115294322A (en) | Underwater ship bottom suspicious target detection method and device, electronic equipment and readable 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |