A mutual information-based fault detection method for laser welding of electric vehicle battery packs

Abstract

The present invention discloses a mutual information-based laser welding fault detection method for electric vehicle battery packs, which belongs to the field of welding technology and includes: obtaining laser data of weld points of laser welding of electric vehicle battery packs, preprocessing the laser data of weld points of laser welding of electric vehicle battery packs, and obtaining preprocessed data; classifying the preprocessed data to generate classified data; calculating the mutual information between the same type of laser data in the classified data; determining the mutual information threshold range based on a fault-free data set in actual engineering applications; comparing the mutual information between the same type of laser data with the mutual information threshold range to obtain faulty weld point data. The method of the present invention can realize the detection of weld point faults in the process of integrated manufacturing of electric vehicle battery packs.

Description

Technical Field

[0001] The present invention belongs to the technical field of welding, and in particular relates to a mutual information-based laser welding fault detection method for an electric vehicle battery pack. Background Art

[0002] The battery pack is one of the core components of electric vehicles. It typically consists of multiple battery cells (such as lithium-ion cells) connected in series and parallel to meet the vehicle's range and power requirements. The battery pack integration process involves multiple steps, with the connection and fixing of the battery cells being key.

[0003] Laser welding technology is currently the primary method for connecting and securing battery cells. Laser welding utilizes a high-intensity laser beam as a heat source for welding. Laser welding boasts high precision, high speed, and a low heat-affected zone, making it suitable for welding a wide range of materials, including metals, plastics, and composites. Laser welding uses a high-energy-density laser beam to locally heat the material, melting it and forming a weld joint. During this process, identifying weld failures is crucial. Currently, image recognition technology is primarily used to identify weld failures. However, due to the complex environment of industrial production, image acquisition is difficult and its effectiveness is compromised. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention proposes a mutual information-based laser welding fault detection method for electric vehicle battery packs to solve the problems existing in the above-mentioned prior art.

[0005] To achieve the above objectives, the present invention provides a method for detecting faults in laser welding of electric vehicle battery packs based on mutual information, comprising:

[0006] Acquire laser data of weld points of laser welding of an electric vehicle battery pack, and preprocess the laser data of weld points of laser welding of the electric vehicle battery pack to obtain preprocessed data;

[0007] classifying the preprocessed data to generate classified data;

[0008] Calculating mutual information between laser data of the same type in the classified data;

[0009] Determine the mutual information threshold range based on a fault-free data set in actual engineering applications;

[0010] The mutual information between the laser data of the same type is compared with the mutual information threshold range to obtain the faulty weld data.

[0011] Preferably, the laser data of the weld points of the laser welding of the electric vehicle battery pack is preprocessed, and the process of obtaining the preprocessed data includes: extracting and labeling the laser data of the weld points of the laser welding of the electric vehicle battery pack according to the welding point pressure head sequence of the electric vehicle battery pack to generate the preprocessed data.

[0012] Preferably, the classification data includes: infrared light data, laser data, reflected light data and visible light data.

[0013] Preferably, the expression for calculating the mutual information between the laser data of the same type in the classified data is:

[0014] I[X;Y]=H(X)+H(Y)-H(X,Y)

[0015] st

[0016]

[0017] Where Pr[·] represents the probability distribution of a random variable, H[·] and H[·,·] are the Shannon entropy and joint entropy, respectively, X is a discrete random variable with x being the value in its domain, and Y is a discrete random variable with y being the value in its domain.

[0018] Preferably, after calculating the mutual information between the laser data of the same category in the classified data, the method further includes: determining the correlation between the laser data based on the mutual information between the laser data of the same category.

[0019] Preferably, the process of comparing the mutual information between the laser data of the same type with the mutual information threshold range to obtain the faulty weld data includes:

[0020] If the number of the weld spot laser data is small, determining the difference between the mutual information value of the faulty weld spot and the mutual information value of the normal weld spot based on the correlation between the laser data;

[0021] If the amount of the welding spot laser data is large, the calculated mutual information value is used to identify outliers through the IQR-test.

[0022] Preferably, the calculation expression of the IQR-test is:

[0023]

[0024] Among them, Q1 represents the upper quartile, Q3 represents the lower quartile, IQR = Q3-Q1 represents the interquartile range, Q up and Q low Represent the upper and lower limits respectively.

[0025] Preferably, the process of identifying outliers on the calculated mutual information value by IQR-test includes: if the mutual information value exceeds the upper limit and the lower limit, it is an outlier, and the calculation expression is:

[0026]

[0027] Among them, β ij represents the mutual information value of the laser data between the i-th and j-th welding points.

[0028] Compared with the prior art, the present invention has the following advantages and technical effects:

[0029] This invention uses a mutual information-based correlation determination method to detect faulty welds in electric vehicle battery packs without disrupting normal industrial production line layout. By analyzing the correlation between laser data, weld failures can be diagnosed more conveniently and quickly. This method can be used to detect weld failures during the integrated manufacturing process of electric vehicle battery packs. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0031] Figure 1 is a flow chart of a mutual information-based electric vehicle battery pack laser welding fault detection method according to an embodiment of the present invention;

[0032] Figure 2 This is a schematic diagram of the pressure head position of the welding machine according to an embodiment of the present invention;

[0033] Figure 3 Schematic diagram of welding points during the integration of an electric vehicle battery pack according to an embodiment of the present invention;

[0034] Figure 4 is a visible light laser spectrum according to an embodiment of the present invention;

[0035] Figure 5 This is a diagram showing the diagnostic results of the electric vehicle battery pack laser welding fault detection method based on mutual information in a fault-free situation according to an embodiment of the present invention.

[0036] Figure 6 This is a diagram showing the diagnostic results of the electric vehicle battery pack laser welding fault detection method based on mutual information in a fault situation according to an embodiment of the present invention. DETAILED DESCRIPTION

[0037] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0038] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0039] Example 1

[0040] like Figure 1-3 As shown, this embodiment provides a method for detecting laser welding faults of electric vehicle battery packs based on mutual information, including:

[0041] Step S101: Collect laser data of all welding points in the battery pack.

[0042] Preprocess the laser data.

[0043] Step S102: classifying the laser data of each welding point.

[0044] According to the actual engineering application environment, the laser data of each weld point is classified into infrared light, laser light, reflected light, and visible light.

[0045] Step S103: Calculating the mutual information between the different welding points and the same type of laser data.

[0046] Among them, the method for calculating mutual information is:

[0047]

[0048] where Pr[·] represents the probability distribution of the random variable.

[0049] Among them, the above method of calculating mutual information can be simplified as follows:

[0050] I[X;Y]=H(X)+H(Y)-H(X,Y)

[0051] st

[0052]

[0053] where H[·] and H[·,·] are Shannon entropy and joint entropy, respectively. X is a discrete random variable with x being the value in its domain, and Y is a discrete random variable with y being the value in its domain.

[0054] When no faults occur, the laser data for the same type across all welds exhibits strong correlation, and the calculated mutual information value fluctuates around a threshold. When a weld fails, the correlation between the laser data for that weld and the others decreases, which is reflected in an abnormal mutual information value.

[0055] Step S104: performing outlier identification on the calculated mutual information value and locating the faulty solder joint.

[0056] For smaller data volumes, the difference between the mutual information values ​​of faulty solder joints and normal solder joints can be directly observed.

[0057] For large amounts of data, the IQR-test is used to identify outliers in the calculated mutual information values.

[0058] Among them, the method for calculating IQR-test is:

[0059]

[0060] Among them, Q1 represents the upper quartile, Q3 represents the lower quartile, IQR = Q3-Q1 represents the interquartile range, Q up and Q low Represent the upper and lower limits respectively.

[0061] Among them, when the existence value exceeds the upper and lower limits, it is an outlier, and the calculation method is:

[0062]

[0063] Among them, β ij represents the mutual information value of the laser data between the i-th and j-th welds. By analyzing the identified outliers, the weld locations associated with the outliers can be determined, thereby locating the weld fault. Based on engineering application requirements, any of the four laser categories for a weld is considered a weld fault.

[0064] The technical solution of the present invention is further described below with reference to a specific application example.

[0065] Taking visible light as an example, we perform mutual information analysis on the visible light data of five solder joints. We calculate the mutual information between solder joint 1 and solder joint 2, solder joint 1 and solder joint 3, solder joint 1 and solder joint 4, solder joint 1 and solder joint 5, solder joint 2 and solder joint 3, and so on. The calculated mutual information of visible light data between every two solder joints is shown in Table 1:

[0066] Table 1

[0067] Solder point #1 Solder point #2 Solder point #3 Solder point #4 Solder point #5 Solder point #1 / 0.8117 0.8021 0.7690 0.7898 Solder point #2 0.8117 / 0.8267 0.7986 0.8208 Solder point #3 0.8021 0.8267 / 0.7797 0.8085 Solder point #4 0.7690 0.7986 0.7797 / 0.7639 Solder point #5 0.7898 0.8208 0.8085 0.7639 /

[0068] It can be observed that the mutual information values ​​calculated above are all near a threshold, that is, around 0.80. Figure 4 In the case of 10 solder joints, the IQR-test is used to identify outliers on the calculated mutual information values.

[0069] When there is a faulty solder joint, the mutual information of the visible light data between every two solder joints is calculated as shown in Table 2:

[0070] Table 2

[0071] Solder point #1 Solder point #2 Solder point #3 Solder point #4 Solder point #5 Solder point #1 / 0.5287 0.8021 0.7690 0.7898 Solder point #2 0.5287 / 0.5635 0.5635 0.5410 Solder point #3 0.8021 0.5635 / 0.7797 0.8085 Solder point #4 0.7690 0.5551 0.7797 / 0.7639 Solder point #5 0.7898 0.5410 0.8085 0.7639 /

[0072] It can be observed that the mutual information values ​​associated with solder joint #2 are in a different range from those of the other solder joints. Figure 5-6 The case of 10 solder joints is shown, and the IQR-test is used to identify outliers on the calculated mutual information values ​​when there are faults.

[0073] In summary, the present invention employs a mutual information-based correlation determination method to detect faulty welds in electric vehicle battery packs without disrupting normal industrial production line layout. By analyzing the correlation between laser data, faulty welds can be diagnosed more conveniently and rapidly. This method can be used to detect weld failures during the integrated manufacturing process of electric vehicle battery packs.

[0074] The above are merely preferred embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A method for detecting faults in laser welding of electric vehicle battery packs based on mutual information, characterized in that: The following steps are involved: Acquire laser data of welding points of laser welding of battery packs for electric vehicles, and preprocess the laser data of welding points of laser welding of battery packs for electric vehicles to obtain preprocessed data; Classifying the preprocessed data to generate classified data; Calculating mutual information between laser data of the same category in the classified data; Determine the mutual information threshold range based on a fault-free data set in actual engineering applications; Comparing the mutual information between the laser data of the same type with the mutual information threshold range to obtain the faulty welding point data; After calculating the mutual information between the laser data of the same category in the classified data, the method further includes: determining the correlation between the laser data based on the mutual information between the laser data of the same category; The process of comparing the mutual information between the same type of laser data with the mutual information threshold range to obtain the faulty solder joint data includes: If the number of the solder joint laser data is small, the difference between the mutual information value of the fault solder joint and the mutual information value of the normal solder joint is determined based on the correlation between the laser data; If the number of laser data of the welding point is large, the calculated mutual information value is used to identify outliers through IQR-test; The calculation expression of the IQR-test is: Among them, Q1 represents the upper quartile, Q3 represents the lower quartile, IQR = Q3-Q1 represents the interquartile range, Q up and Q low Represent the upper and lower limits respectively; The process of identifying outliers on the calculated mutual information value by IQR-test includes: if the mutual information value exceeds the upper limit and the lower limit, it is an outlier, and the calculation expression is: Among them, β ij Represents the mutual information value of the laser data between the i-th and j-th welding points.

2. The electric vehicle battery pack laser welding fault detection method based on mutual information according to claim 1 is characterized in that: The laser data of the weld points of the laser welding of the electric vehicle battery pack are preprocessed, and the process of obtaining the preprocessed data includes: extracting and labeling the laser data of the weld points of the laser welding of the electric vehicle battery pack according to the welding point pressure head sequence of the electric vehicle battery pack to generate the preprocessed data.

3. The electric vehicle battery pack laser welding fault detection method based on mutual information according to claim 1 is characterized in that: The classified data includes: infrared light data, laser data, reflected light data and visible light data.

4. The electric vehicle battery pack laser welding fault detection method based on mutual information according to claim 1 is characterized in that: The expression for calculating the mutual information between the same type of laser data in the classified data is: I[X;Y]=H(X)+H(Y)-H(X,Y) st Where Pr[·] represents the probability distribution of a random variable, H[·] and H[·,·] are the Shannon entropy and joint entropy respectively, X is a discrete random variable, x is the value in its domain, and Y is a discrete random variable, y is the value in its domain.

Images