Abnormality detection method for lithium battery diaphragm processing production line
By establishing a diaphragm feature library and lighting feature library, and automatically adjusting the detection parameters and algorithms, the problem of unsatisfactory adaptability of the detection system in the existing technology is solved, seamless switching and accurate detection of diaphragm detection of lithium battery are realized, and the production line is reduced The adjustment time and cost of the system have improved the degree of intelligent detection.
Patent Information
- Application Number
- CN202411090334.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-09
- Publication Date
- 2024-11-22
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The inspection system of the existing lithium battery separator processing production line is difficult to adapt to differences in materials, sizes, colors, etc. due to the fixed detection target, which requires adjustment of detection parameters and retraining of the model, which consumes time and effort, reduces production efficiency and increases maintenance costs.
By identifying the diaphragm information of lithium battery diaphragm, establishing a diaphragm feature library, updating an abnormality recognition model, automatically adjusting detection parameters and algorithms, and combining with the lighting feature library to perform intelligent lighting adjustments to achieve seamless switching and accurate detection.
It significantly reduces the adjustment time and cost when replacing the diaphragm type, improves the intelligence of the inspection, promptly detects and deals with defects in the production process, and ensures product quality.
Smart Images

Figure CN119007107A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of abnormality detection for lithium battery diaphragm processing, and specifically is an abnormality detection method for a lithium battery diaphragm processing production line. Background Art
[0002] With the rapid development of the new energy industry, the performance and quality of lithium batteries as its core components are crucial to the entire industry chain. As a key component inside the battery, the quality of lithium battery separators directly affects the safety, cycle life and overall performance of the battery. Therefore, in the lithium battery separator processing production line, the implementation of an efficient anomaly detection system is a key link to ensure product quality and improve production efficiency.
[0003] At present, manual inspection or traditional machine vision inspection is commonly used in the production process of lithium battery separators. Manual inspection has the disadvantages of strong subjectivity, low efficiency, and easy fatigue. Although the traditional machine vision inspection method can improve the inspection efficiency to a certain extent, its inspection target is basically fixed, that is, it can only inspect the preset lithium battery separator. When replacing a new lithium battery separator, due to differences in material, size, color, etc., the existing inspection cannot be applied, especially the need to readjust parameters and retrain models, resulting in unsatisfactory adaptability; it may even be necessary to replace some hardware components to adapt to new inspection requirements. This process is not only time-consuming and labor-intensive, reducing production efficiency, but also increasing the complexity and maintenance costs of the system.
[0004] Based on this, in order to solve the above problems, the present invention provides a method for detecting abnormalities in a lithium battery diaphragm processing production line. Summary of the invention
[0005] In order to solve the problems existing in the above-mentioned solutions, the present invention provides a method for detecting abnormalities in a lithium battery diaphragm processing production line.
[0006] The purpose of the present invention can be achieved through the following technical solutions:
[0007] A method for detecting abnormalities in a lithium battery separator processing production line, the method comprising:
[0008] Step 1: Identify the lithium battery diaphragm information and establish a diaphragm feature library based on the lithium battery diaphragm information;
[0009] Furthermore, the method for establishing the diaphragm feature library includes:
[0010] According to the lithium battery diaphragm information, a standard diaphragm image is set; an abnormal simulation model is established; according to the lithium battery diaphragm information, various abnormal problems are obtained; the diaphragm standard image and various abnormal problems are analyzed by the abnormal simulation model to obtain various abnormal simulation images and image abnormality recognition features corresponding to each abnormal simulation image;
[0011] The abnormal simulation images and image abnormality recognition features are integrated to establish a corresponding diaphragm feature library.
[0012] Step 2: updating and adjusting the preset abnormality recognition model based on the diaphragm feature library;
[0013] Furthermore, the method for updating and adjusting the preset abnormality recognition model based on the diaphragm feature library includes:
[0014] The anomaly recognition model is used to perform anomaly recognition on the grayscale image of the lithium battery diaphragm to determine whether there is an abnormality problem;
[0015] Identify each abnormal simulation image and image abnormality recognition feature in the diaphragm feature library, perform grayscale processing on each abnormal simulation image, and obtain a simulated grayscale image corresponding to each abnormal simulation image;
[0016] Marking an image area corresponding to a corresponding image abnormality recognition feature on the simulated grayscale image, marking the corresponding image area as a feature recognition area;
[0017] Setting grayscale recognition features according to the grayscale values corresponding to the feature recognition areas; establishing a feature recognition library according to each grayscale recognition feature;
[0018] The expression of the anomaly recognition model is:
[0019] Where: s is the input data, the input data is the collected grayscale image corresponding to the collected image; the output data is the abnormal recognition value HF(s);
[0020] Update the feature recognition requirements in the anomaly recognition model based on the feature recognition library.
[0021] Step 3: Obtain a feature recognition library, the feature recognition library is used to store each grayscale recognition feature; perform lighting adjustment according to the feature recognition library;
[0022] Furthermore, the method for adjusting illumination according to the feature recognition library includes:
[0023] Establishing an illumination feature library, wherein the illumination feature library is used to store illumination simulation curves corresponding to each candidate feature, wherein the horizontal axis of the illumination simulation curve is illumination condition, and the vertical axis is recognition rate;
[0024] Matching the corresponding illumination simulation curve from the illumination feature library according to each grayscale recognition feature; evaluating the illumination adjustment range based on each illumination simulation curve to obtain the illumination effect value corresponding to each illumination condition;
[0025] The lighting condition with the largest lighting effect value is selected as the lighting target condition, and the lighting adjustment is performed according to the lighting target condition.
[0026] Furthermore, the method for evaluating the illumination adjustment range based on each illumination simulation curve includes:
[0027] Identify each grayscale recognition feature in the feature recognition library, count the probability of occurrence of abnormal problems corresponding to each grayscale recognition feature, and mark it as abnormal probability;
[0028] Identify the corresponding recognition rate of each illumination simulation curve under the same illumination conditions;
[0029] The recognition rate is marked as SPi, i represents the corresponding grayscale recognition feature, i=1, 2, ..., n, n is a positive integer; the abnormal probability corresponding to each grayscale recognition feature is marked as VRi;
[0030] According to the formula Calculate the corresponding lighting effect value;
[0031] In the formula: KP is the lighting effect value; e is the natural constant.
[0032] Step 4: Acquire the collected image of the lithium battery separator in real time, perform grayscale processing on the collected image to obtain a collected grayscale image, and identify the collected grayscale image through the abnormality recognition model to obtain a corresponding abnormality detection result.
[0033] Furthermore, the method further includes step five: obtaining verification results of each abnormal detection result in real time, the verification results including normal detection and detection error; and performing abnormal detection monitoring according to each of the verification results.
[0034] Furthermore, the method for performing anomaly detection monitoring according to each verification result includes:
[0035] Set the unit time, classify each verification result according to the unit time, and obtain the data of each unit;
[0036] Establish a verification model, the expression of the verification model is:
[0037]
[0038] Where: xj is the input data, the input data is the verification result, j=1, 2, ..., m, m is a positive integer; the output data is the verification value YR(xj);
[0039] According to the formula Calculate the unit value of each unit time;
[0040] Where: DA is the unit value; e is the natural coefficient; YR(xj) is the verification value;
[0041] A monitoring curve is generated according to the unit corresponding to each unit time, wherein the horizontal axis of the monitoring curve is the unit time and the vertical axis is the unit value;
[0042] The monitoring curve is fitted to obtain the monitoring function, which is marked as DW(t), where t is the independent variable;
[0043] According to the formula Calculate the corresponding monitoring value;
[0044] In the formula: KU is the monitoring value; A0 is the standard value;
[0045] When the monitoring value is 1, a monitoring warning is issued to prompt the staff to make abnormal detection adjustments;
[0046] When the monitoring value is 0, no corresponding operation is performed.
[0047] Compared with the prior art, the present invention has the following beneficial effects:
[0048] The present invention can automatically adjust the detection parameters and algorithms when replacing the lithium battery film to be detected, thereby achieving seamless switching and accurate detection. This high degree of flexibility and adaptability significantly reduces the adjustment time and cost when changing the diaphragm type on the production line. At the same time, intelligent lighting adjustments are made according to actual conditions to reduce manual participation and improve the intelligence of detection. Through real-time and efficient abnormality detection, defects in the production process can be discovered and handled in a timely manner, thereby ensuring the product quality of the lithium battery diaphragm. BRIEF DESCRIPTION OF THE DRAWINGS
[0049] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0050] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION
[0051] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0052] like Figure 1 As shown, a method for detecting abnormality in a lithium battery separator processing production line comprises:
[0053] Step 1: Obtain lithium battery diaphragm information, including material, size, color, image, type, model and other related information, and determine various abnormal problems of the lithium battery according to the information, such as surface stains, holes, impurities, scratches, oil spots, bright spots, crystal points, coating leakage, spray leakage and other abnormal problems; set the diaphragm standard image according to the lithium battery diaphragm information, that is, intuitively reflect the lithium battery diaphragm information on the image, which can be set manually; establish an abnormal simulation model based on a neural network such as a CNN network or a DNN network, and establish a corresponding training set for training manually. The training set includes input data and output data. The input data is the diaphragm standard image and various abnormal problems. The output data is each abnormal simulation image generated based on the simulation of each abnormal problem and the corresponding image abnormality recognition feature; that is, the corresponding image abnormality recognition feature can be matched with the abnormal simulation image; set the training set in combination with the image embodiment of various existing lithium battery diaphragms on various abnormal problems; analyze the diaphragm standard image and each abnormal problem through the abnormal simulation model after successful training, obtain each abnormal simulation image and image abnormality recognition feature, and integrate each abnormal simulation image and image abnormality recognition feature to establish a diaphragm feature library;
[0054] Step 2: Update and adjust the preset abnormality recognition model based on the diaphragm feature library so that the abnormality recognition model can adapt to the detection of the updated lithium battery diaphragm;
[0055] The method for updating and adjusting the preset abnormality recognition model based on the diaphragm feature library includes:
[0056] The anomaly recognition model is used to perform anomaly recognition on the grayscale image of the lithium battery diaphragm to determine whether there is an abnormality problem;
[0057] Identify each abnormal simulation image and image abnormality recognition feature in the diaphragm feature library, perform grayscale processing on each abnormal simulation image, and obtain a simulated grayscale image corresponding to each abnormal simulation image;
[0058] Mark the image area corresponding to the corresponding image abnormality recognition feature on the simulated grayscale image, mark the corresponding image area as a feature recognition area, set the grayscale recognition feature according to the difference between the grayscale value corresponding to the feature recognition area and the grayscale value of the surrounding normal area, that is, the feature recognition area corresponding to the grayscale recognition feature on the lithium battery diaphragm can be determined by having the grayscale recognition feature, and then determine the abnormal problem; establish a feature recognition library according to each grayscale recognition feature;
[0059] The expression of the anomaly recognition model is:
[0060] Where: s is the input data, the input data is the collected grayscale image corresponding to the collected image; the output data is the abnormal recognition value HF(s);
[0061] The feature recognition requirements in the abnormality recognition model are updated according to the feature recognition library, and the recognition judgment is performed based on the updated feature recognition requirements; that is, the updated features of the lithium battery separator are used for abnormality recognition.
[0062] Step 3: Obtain a feature recognition library, i.e., the feature recognition library in step 2, and perform lighting adjustment according to the feature recognition library;
[0063] Methods for adjusting lighting based on feature recognition libraries include:
[0064] Identify each grayscale recognition feature in the feature recognition library, count the probability of occurrence of abnormal problems corresponding to each grayscale recognition feature, perform statistics based on historical production data, and mark it as abnormal probability;
[0065] According to the various lithium battery films available on the market, various possible grayscale recognition features are determined and marked as candidate features; illumination simulation is performed on each candidate feature, that is, the feature recognition of the candidate feature under different illumination conditions is simulated to obtain the illumination simulation curve corresponding to each candidate feature, the horizontal axis of the illumination simulation curve is the illumination condition, such as illumination parameters and other illumination-related conditions; the vertical axis is the recognition rate, which is the statistical rate of identifying the candidate feature under the illumination condition; the platform party performs preliminary simulation tests; establishes an illumination feature library according to each candidate feature and the illumination simulation curve; that is, the illumination feature library is pre-established by the platform party;
[0066] Set the lighting adjustment range of the lighting equipment, that is, the lighting conditions of the lighting equipment used in the production line, and form the lighting adjustment range according to each lighting condition:
[0067] According to each grayscale recognition feature, the corresponding lighting simulation curve is matched from the lighting feature library; based on each lighting simulation curve, the lighting adjustment range is evaluated to obtain the lighting effect value corresponding to each lighting condition; the lighting condition with the largest lighting effect value is selected as the lighting target condition, and the lighting adjustment is performed according to the lighting target condition.
[0068] Methods for evaluating the illumination adjustment range based on various illumination simulation curves include:
[0069] Identify the recognition rate of each illumination simulation curve under the same illumination condition, and mark the recognition rate as SPi, where i represents the corresponding grayscale recognition feature, i=1, 2, ..., n, and n is a positive integer; mark the abnormal probability corresponding to each grayscale recognition feature as VRi;
[0070] According to the formula Calculate the corresponding lighting effect value;
[0071] In the formula: KP is the lighting effect value; e is the natural constant.
[0072] Step 4: Acquire the collected image of the lithium battery separator in real time, perform grayscale processing on the collected image to obtain a collected grayscale image, identify the collected grayscale image through an anomaly recognition model, and obtain a corresponding anomaly detection result.
[0073] In one embodiment, step 5 is also included: obtaining the verification results of each abnormal detection result in real time, and performing abnormal detection monitoring according to each verification result. The verification result is to verify after determining the corresponding abnormal detection result. For example, if a lithium battery separator is analyzed to have an abnormal problem, and it is determined to have no problem after sorting, then the verification result is a detection error; if it is determined to have an abnormal problem, then the verification result is a correct detection; the verification result can be determined according to the subsequent processing results of the lithium battery separator.
[0074] Methods for performing anomaly detection monitoring based on various verification results include:
[0075] Set the unit time, such as one hour, one day, etc., according to specific needs; classify each verification result according to the unit time to obtain each unit data, that is, the unit data is composed of each verification result, and is classified according to the detection time corresponding to the verification result;
[0076] Establish a verification model, the expression of the verification model is Where: xj is the input data, the input data is the verification result, j=1, 2, ..., m, m is a positive integer; the output data is the verification value YR(xj);
[0077] According to the formula Calculate the unit value of each unit time;
[0078] Where: DA is the unit value; e is the natural coefficient; YR(xj) is the verification value;
[0079] A monitoring curve is generated according to the unit corresponding to each unit time, wherein the horizontal axis of the monitoring curve is the unit time and the vertical axis is the unit value;
[0080] The monitoring curve is fitted to obtain the monitoring function, which is marked as DW(t), where t is the independent variable, which represents the corresponding unit time;
[0081] According to the formula Calculate the corresponding monitoring value;
[0082] Where: KU is the monitoring value; A0 is the standard value, which is set according to the detection accuracy requirement. Generally, it is set according to the expected accuracy that can be achieved under the current conditions. The corresponding unit value is calculated according to the accuracy requirement, which is used as the standard value; the threshold X1 is set according to the minimum accuracy requirement; the threshold X2 is set according to the cumulative deviation requirement;
[0083] When the monitoring value is 1, a monitoring warning is issued to prompt the staff to make abnormal detection adjustments;
[0084] When the monitoring value is 0, no corresponding operation is performed.
[0085] The above formulas are all calculated by removing dimensions and taking numerical values. The formula is a formula that is closest to the actual situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained by simulating a large amount of data.
[0086] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.
Claims
1. A method for detecting abnormality in a lithium battery separator production line, characterized in that: Methods include: Step 1: Identify the lithium battery diaphragm information and establish a diaphragm feature library based on the lithium battery diaphragm information; Step 2: updating and adjusting the preset abnormality recognition model based on the diaphragm feature library; Step 3: Obtain a feature recognition library, the feature recognition library is used to store each grayscale recognition feature; perform lighting adjustment according to the feature recognition library; Step 4: Acquire the collected image of the lithium battery separator in real time, perform grayscale processing on the collected image to obtain a collected grayscale image, and identify the collected grayscale image through the abnormality recognition model to obtain a corresponding abnormality detection result.
2. The method for detecting abnormality in a lithium battery separator production line according to claim 1, characterized in that: The method for establishing the diaphragm feature library includes: According to the lithium battery diaphragm information, a standard diaphragm image is set; an abnormal simulation model is established; according to the lithium battery diaphragm information, various abnormal problems are obtained; the diaphragm standard image and various abnormal problems are analyzed by the abnormal simulation model to obtain various abnormal simulation images and image abnormality recognition features corresponding to each abnormal simulation image; The abnormal simulation images and image abnormality recognition features are integrated to establish a corresponding diaphragm feature library.
3. The method for detecting abnormality in a lithium battery separator production line according to claim 1, characterized in that: The method for updating and adjusting the preset abnormality recognition model based on the diaphragm feature library includes: The anomaly recognition model is used to perform anomaly recognition on the grayscale image of the lithium battery diaphragm to determine whether there is an abnormality problem; Identify each abnormal simulation image and image abnormality recognition feature in the diaphragm feature library, perform grayscale processing on each abnormal simulation image, and obtain a simulated grayscale image corresponding to each abnormal simulation image; Marking an image area corresponding to a corresponding image abnormality recognition feature on the simulated grayscale image, marking the corresponding image area as a feature recognition area; Setting grayscale recognition features according to the grayscale values corresponding to the feature recognition areas; establishing a feature recognition library according to each grayscale recognition feature; The expression of the anomaly recognition model is: Where: s is the input data, the input data is the collected grayscale image corresponding to the collected image; the output data is the abnormal recognition value HF(s); Update the feature recognition requirements in the anomaly recognition model based on the feature recognition library.
4. The method for detecting abnormality in a lithium battery separator production line according to claim 1, characterized in that: Methods for adjusting lighting based on feature recognition libraries include: Establishing an illumination feature library, wherein the illumination feature library is used to store illumination simulation curves corresponding to each candidate feature, wherein the horizontal axis of the illumination simulation curve is illumination condition, and the vertical axis is recognition rate; Matching the corresponding illumination simulation curve from the illumination feature library according to each grayscale recognition feature; evaluating the illumination adjustment range based on each illumination simulation curve to obtain the illumination effect value corresponding to each illumination condition; The lighting condition with the largest lighting effect value is selected as the lighting target condition, and the lighting adjustment is performed according to the lighting target condition.
5. The method for detecting abnormality in a lithium battery separator production line according to claim 4, characterized in that: Methods for evaluating the illumination adjustment range based on various illumination simulation curves include: Identify each grayscale recognition feature in the feature recognition library, count the probability of occurrence of abnormal problems corresponding to each grayscale recognition feature, and mark it as abnormal probability; Identify the corresponding recognition rate of each illumination simulation curve under the same illumination conditions; The recognition rate is marked as SPi, i represents the corresponding grayscale recognition feature, i=1, 2, ..., n, n is a positive integer; the abnormal probability corresponding to each grayscale recognition feature is marked as VRi; According to the formula Calculate the corresponding lighting effect value; In the formula: KP is the lighting effect value; e is the natural constant.
6. The method for detecting abnormality in a lithium battery separator production line according to claim 1, characterized in that: The method also includes step five: obtaining verification results of each abnormal detection result in real time, the verification results including normal detection and detection error; and performing abnormal detection monitoring according to each of the verification results.
7. The method for detecting abnormality in a lithium battery separator production line according to claim 6, characterized in that: Methods for performing anomaly detection monitoring based on various verification results include: Set the unit time, classify each verification result according to the unit time, and obtain the data of each unit; Establish a verification model, the expression of the verification model is: Where: xj is the input data, the input data is the verification result, j=1, 2, ..., m, m is a positive integer; the output data is the verification value YR(xj); According to the formula Calculate the unit value of each unit time; Where: DA is the unit value; e is the natural coefficient; YR(xj) is the verification value; A monitoring curve is generated according to the unit corresponding to each unit time, wherein the horizontal axis of the monitoring curve is the unit time and the vertical axis is the unit value; The monitoring curve is fitted to obtain the monitoring function, which is marked as DW(t), where t is the independent variable; According to the formula Calculate the corresponding monitoring value; In the formula: KU is the monitoring value; A0 is the standard value; When the monitoring value is 1, a monitoring warning is issued to prompt the staff to make abnormal detection adjustments; When the monitoring value is 0, no corresponding operation is performed.
Citation Information
Patent Citations
Training method of neural network for battery diaphragm anomaly detection
CN112418353A
Light source brightness self-adaptive regulation and control system and method for lithium battery pole piece coating detection
CN115175414A
Lithium battery diaphragm defect detection method and system
CN115641329A
Defect detection data generation and detection method and system based on generative adversarial network
CN115937627A
Model generation device for appearance inspection and appearance inspection device
CN116194954A
Cited By
Lithium battery diaphragm processing production risk situation awareness method
CN121094529A
Lithium battery diaphragm production early warning method and system based on big data analysis
CN121094610A
