Alarm method, program product and storage medium for string welding machine

Abstract

The present application relates to the industrial field, and provides an alarm method, program product and storage medium for a string welding machine. Among them, the alarm method for the string welding machine includes: obtaining the first original data, the first original data includes the visual detection results of the defects of the battery string, the result includes the category and location of the defect, and the category of the defect includes cold welding and hidden cracks; obtaining the first data to be analyzed according to the first original data, the first data to be analyzed includes the secondary classification results obtained by secondary classification of the defects according to the category, location and image of the battery string; matching the first data to be analyzed with the alarm rules in the rule base, if successful, outputting the first alarm information based on the matched alarm rules, the first alarm information includes the fault information and / or maintenance information of the string welding machine. This method is conducive to improving the accuracy of fault alarms and / or maintenance alarms for the string welding machine, and ensuring the quality of the produced battery strings.

Description

Technical Field

[0001] The present invention relates to the field of industrial technology, and in particular to an alarm method, a program product and a storage medium for a string welding machine. Background Art

[0002] At present, the degree of automation in industrial production is getting higher and higher, and a large number of equipment will be involved in the production line of products. For example, in the process of manufacturing photovoltaic modules, equipment such as dicing machines, stringing machines, typesetting machines, stacking machines, and laminating machines are required. Among them, the main function of the stringing machine is to use welding ribbons to weld independent battery cells into battery strings, and then the battery strings are further processed into photovoltaic modules by the subsequent process. Once the stringing machine fails, it will seriously affect the quality of the processed battery strings and even photovoltaic modules. At present, the product quality can be detected by computer vision technology, but there is still a lack of high-accuracy diagnosis or early warning methods for stringing machine failures. Summary of the invention

[0003] The purpose of the embodiments of the present application is to provide an alarm method, program product and storage medium for a string welding machine to improve the above technical problems.

[0004] To achieve the above objectives, this application provides the following technical solutions:

[0005] In a first aspect, an embodiment of the present application provides an alarm method for a string welder, comprising: acquiring first original data, the first original data comprising visual inspection results of defects of a battery string produced by the string welder, the visual inspection results comprising the category of the defect and the location of the defect, the category of the defect comprising cold solder joints and hidden cracks; obtaining first data to be analyzed based on the first original data; wherein the first data to be analyzed comprises: a secondary classification result obtained by secondary classification of the defect according to the category of the defect, the location of the defect and the image of the battery string; matching the first data to be analyzed with an alarm rule in a rule base, and if the match is successful, outputting a first alarm message based on the matched alarm rule; wherein the first alarm message comprises fault information and / or maintenance information of the string welder.

[0006] In a second aspect, an embodiment of the present application provides an alarm device for a string welder, comprising: a data acquisition module, used to acquire first original data, the first original data including visual inspection results of defects of a battery string produced by the string welder, the visual inspection results including the category of the defect and the location of the defect, the category of the defect including cold solder joint and hidden crack; a data processing module, used to obtain first data to be analyzed based on the first original data; wherein the first data to be analyzed includes: a secondary classification result obtained by secondary classification of the defect according to the category of the defect, the location of the defect and the image of the battery string; an alarm module, used to match the first data to be analyzed with an alarm rule in a rule base, and if the match is successful, output a first alarm message based on the matched alarm rule; wherein the first alarm message includes fault information and / or maintenance information of the string welder.

[0007] In a third aspect, an embodiment of the present application provides a computer program product, including computer program instructions, which, when read and executed by a processor, execute the method provided by the first aspect or any one of the implementations of the first aspect.

[0008] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which computer program instructions are stored. When the computer program instructions are read and executed by a processor, the method provided by the first aspect or any one of the implementations of the first aspect is executed.

[0009] In a fifth aspect, an embodiment of the present application provides an electronic device, comprising: a memory and a processor, wherein the memory stores computer program instructions, and when the computer program instructions are read and run by the processor, the method provided by the first aspect or any one of the implementations of the first aspect is executed.

[0010] Since the battery string is produced by the string welding machine, once the string welding machine fails or needs maintenance, it will be reflected in the quality of the battery string. Therefore, in the above technical solution, based on the first original data containing the visual inspection results of the defects of the battery string (representing the quality of the battery string) to issue a fault alarm and / or maintenance alarm to the equipment, the accuracy of the alarm can be improved, which is conducive to timely discovery of the string welding machine failure or timely maintenance of the string welding machine, thereby ensuring the quality of the produced battery string. In addition, the scheme is not directly based on the first original data for alarm, but based on the first data to be analyzed further obtained from the first original data for alarm. Since the first data to be analyzed at least contains the results of secondary classification of the defects of the battery string, especially the secondary classification results of the two important defects of virtual welding and hidden cracks, that is, the correlation between the defects and the failure or maintenance requirements of the string welding machine is more deeply excavated, it is also conducive to improving the accuracy of the alarm. In addition, based on this scheme, not only can a fault alarm be issued when the string welding machine fails, but also predictive maintenance can be performed when the string welding machine does not fail, which is conducive to reducing the probability of failure. In addition, this solution uses the same processing logic for fault alarms and maintenance alarms. For users, whether it is a fault alarm or a maintenance alarm, they only need to configure the corresponding alarm rules in the rule base, which makes it convenient to handle. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0012] Figure 1 A flowchart of an alarm method for a string welding machine provided in an embodiment of the present application;

[0013] Figure 2a~Figure 2g A variety of hidden crack morphology diagrams provided for the embodiments of the present application;

[0014] Figure 3 Schematic diagram of the outline and skeleton of the defect area;

[0015] Figure 4 A flowchart of another alarm method for a string welding machine provided in an embodiment of the present application;

[0016] Figure 5 A module diagram of an alarm device for a string welding machine provided in an embodiment of the present application;

[0017] Figure 6 A structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application. It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings.

[0019] The terms "comprises", "comprising" or any other variation thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element. The terms "first", "second", "third", etc. are used only to distinguish descriptions and are not to be understood as indicating or implying relative importance.

[0020] Figure 1 The flowchart of an alarm method for a string welding machine provided in an embodiment of the present application. The method can be, but is not limited to, Figure 6 The electronic device shown is executed, and the structure of the electronic device can be referred to later. Figure 6 Reference Figure 1 , the alarm methods for string welding machines include:

[0021] Step S110: Acquire first original data.

[0022] The first original data is the basic data used for string welding machine alarm analysis. The so-called "original" is relative to the first data to be analyzed in step S120. The first data to be analyzed may be processed on the basis of the first original data, and therefore is no longer "original".

[0023] The first original data includes at least the quality data of the battery string produced by the string welder. The string welder here is the industrial equipment targeted by the alarm method. The input of the string welder is the battery cell, and the output is the welded battery string (a battery string includes multiple battery cells). Note that the battery cells input by the string welder may be cut, and a dicing machine may be set before the string welder to cut the battery cells produced by the battery cell factory according to the production requirements of the photovoltaic module (for example, cut into two). Of course, some string welders have their own dicing function, and a dicing machine may not be set before these string welders.

[0024] The quality data of the battery string represents the quality of the battery string. The quality data mainly includes visual inspection results obtained by performing defect detection on images of the battery string using a computer vision algorithm, but other data that can reflect the quality of the battery string are not excluded.

[0025] The above-mentioned computer vision algorithm can be a traditional image processing algorithm (for example, edge detection, morphological processing algorithm) or an artificial intelligence algorithm (for example, an algorithm based on a neural network model). These algorithms take the image of the battery string (or a part thereof) as input and output the visual inspection result for the image. The visual inspection result can include the category of the defect and the location of the defect in the image (for example, the coordinates of a detection box), and optionally also includes the detection time of the defect.

[0026] The categories of defects include at least cold solder joints and hidden cracks. Cold solder joints refer to the failure of the soldering ribbon to be effectively welded to the target position of the battery string, which may cause problems such as short circuits. Hidden cracks refer to cracks on the battery cell that are not easily noticeable to the naked eye. Hidden cracks can lead to problems such as reduced power generation efficiency of the battery cell or local overheating.

[0027] Normal stringers usually do not have defects (or the defects are within an acceptable level). However, if a string has defects (or the defects exceed the acceptable level), it can be inferred that the stringer has some abnormality, such as a malfunction or urgent maintenance. Therefore, visual inspection results are of great significance for stringer alarms.

[0028] Depending on the implementation method, the quality data of each battery string produced by the stringer can be obtained through analysis, or the quality data of a batch of battery strings can be obtained through analysis after the stringer produces a batch of battery strings, or the two methods can be combined to obtain quality data at different levels.

[0029] In addition to quality data, the first raw data may also include other data, such as data related to the stringer, and specific examples will be given later. In addition, the acquisition method and timing of other data may be different from the quality data, for example, it may be collected during the process of the stringer producing battery strings.

[0030] For example, on an industrial assembly line, a quality inspection device can be set after the stringer, and the battery strings produced by the stringer can be quality inspected based on different methods to obtain product quality data, and then the quality data can be sent to a remote server, and the alarm method provided by the embodiment of the present application can be executed on the server. For example, a camera can be set to capture an image of the battery string, and then the image can be sent to a near-end computer for visual defect detection to obtain visual inspection results, and then the visual inspection results can be sent to a remote server for alarm analysis. Of course, the server can also receive other data from the stringer itself or external sensors.

[0031] Step S120: obtaining first data to be analyzed according to the first original data.

[0032] The first raw data can be processed to obtain the first data to be analyzed. The processing here can include at least one of the following two categories: the first category is preprocessing, the main purpose of which is to improve data quality; the second category is further processing, the main purpose of which is to convert the first raw data into a form that is more suitable for alarm analysis. In different implementations, all the first raw data can be processed to obtain the corresponding first data to be analyzed; only part of the first raw data can be processed to obtain the corresponding first data to be analyzed, and the remaining first raw data can be directly used as the first data to be analyzed; all the first raw data can also be directly used as the first data to be analyzed (this is equivalent to not processing the first raw data, which is a special case). In addition, even if some of the first raw data have been processed, the first raw data before and after the processing can be used as the first data to be analyzed.

[0033] For preprocessing of the first original data, the processing method may include, but is not limited to, at least one of the following:

[0034] (1) Missing value processing

[0035] Missing values ​​refer to blank data caused by abnormal reasons. For example, missing values ​​in a short period of time can be filled, such as filling with fixed values, using interpolation to fill, using data near the missing values ​​for prediction and filling, etc.; for missing values ​​in a long period of time, this part of the data can be directly discarded to avoid low-quality data affecting the accuracy of alarm analysis.

[0036] "Short time" and "long time" can be defined by the number of consecutive missing values. For example, if there are more than 10 consecutive missing values, it is considered to be a long-term missing situation, and if there are no more than 10 consecutive missing values, it is considered to be a short-term missing situation, and so on.

[0037] (2) Noise processing

[0038] There are different processing methods for different noises. For example, for some short-term noise, moving average can be used to smooth and remove it; for some high-frequency noise, low-pass filter can be used to remove it.

[0039] (3) Standardization

[0040] Standardization refers to converting diverse data into a unified form to facilitate subsequent operations. The "unification" here can refer to the unification of data dimensions, the unification of data value ranges, etc. For example, data can be standardized using the Z-score algorithm.

[0041] Further processing of the first original data can be achieved by feature extraction. For example, feature data of the first original data in one or more dimensions can be extracted. These feature data can have a better correspondence with the stringer failure or the stringer maintenance requirement, so as to facilitate subsequent alarm analysis. Among them, the stringer failure means that the stringer is in an abnormal working state or even cannot work at all, and the stringer maintenance requirement means that the stringer is still working normally, but has shown some bad signs, so maintenance is required, otherwise it may cause failure.

[0042] In particular, since the first original data includes the quality data of the battery string, feature extraction can be performed on the quality data, and the obtained feature data is called quality feature data. Feature extraction can adopt, for example, statistical methods or model-based algorithms.

[0043] For example, the quality data can be feature extracted from at least one of the dimensions of morphology, time, and space to obtain quality feature data of the corresponding dimension. The morphological dimension mainly refers to the subdivision category of defects represented by the quality data, the time dimension mainly refers to the characteristics of the quality data over time, and the space dimension mainly refers to the spatial location characteristics of the defects represented by the quality data. The extraction method of quality feature data of each dimension will be further explained later.

[0044] For example, if the defect category is hidden cracks, the hidden cracks can be sub-classified, and the classification results include at least one of the following:

[0045] (11) Whether the hidden cracks penetrate the battery cell

[0046] A crack that penetrates the cell means that the crack extends from one end of the cell to the other. Category (11) can be further divided into horizontal penetration and vertical penetration. Figure 2a It shows the situation where the hidden cracks penetrate the battery cell horizontally. The black broken line in the white box is the hidden crack, and the hidden cracks cross the edges of the left and right sides of the battery cell.

[0047] (12) Is the hidden crack located on the cut surface or chamfered surface of the cell?

[0048] As mentioned in step S110, the battery cells processed by the stringer can be cut (including the case where the stringer cuts by itself), the chamfered surface can be understood as the original edge of the battery cell, and the cut surface is the edge generated after the battery cell is cut (for the edges on the left and right sides of the battery cell). Figure 2b In the figure, the edge of the battery cell on the left represents the cutting surface, and the edge of the battery cell on the right represents the chamfered surface. If the hidden crack starts from the chamfered surface and extends into the interior of the battery cell, it is a hidden crack on the chamfered surface. Figure 2c In the case of a hidden crack, if the hidden crack starts from the cutting surface and extends into the interior of the battery cell, it is a hidden crack on the cutting surface.

[0049] (13) Is the hidden crack located at the cutting corner of the cell?

[0050] The cutting angle of the cell is the corner of the cutting surface of the cell. Figure 2d shown.

[0051] (14) Are the hidden cracks located at the upper and lower edges of the cell?

[0052] For example, in Figure 2e In the case of a crack extending from the upper edge of the cell to the inside of the cell, it is an upper edge crack. Conversely, a crack extending from the lower edge of the cell to the inside of the cell is a lower edge crack.

[0053] (15) Is the shape of the crack straight, reticular, x-shaped or v-shaped?

[0054] Straight line, mesh, x-shape or v-shape are all geometric shapes of hidden cracks. For example, x-shape indicates that the geometric shape of the hidden cracks is similar to the letter x. Of course, it does not rule out that the hidden cracks can have other geometric shapes.

[0055] (16) Is the hidden crack located on the main grid line of the battery cell?

[0056] If the crack starts from the main grid line of the battery cell and extends to the inside of the battery cell, it is a main grid crack. The main grid crack can be further divided into different forms, such as the main grid fishbone crack. Figure 2f In the figure, the endpoint of the hidden crack is on the main grid line of the battery cell, and two upper and lower branches extend into the battery cell, and the angle between the two branches is approximately 90°, and its shape is similar to a fishbone arrangement (the number of upper and lower branches can also be more than one).

[0057] (17) Whether the hidden crack is a blank hidden crack

[0058] For example, in Figure 2g If the crack does not cover any main grid line of the cell, it is a blank crack, which means it occurs in the blank area of ​​the cell.

[0059] (18) The cell number where the hidden crack is located in the battery string

[0060] The cell ordinal number can refer to the number of cells in the battery string, such as the first cell, the second cell, etc. (each number corresponds to a subdivision category). Alternatively, the cell ordinal number may not be a specific cell number, but a cell number that meets certain characteristics, such as first and last, odd and even, etc. For example:

[0061] If the hidden crack is located on the first battery cell in the battery string, it is considered a hidden crack on the first cell. If the hidden crack is located on the last battery cell in the battery string, it is considered a hidden crack on the last cell (the first cell and the last cell are two subcategories).

[0062] If the hidden crack is located on an odd number of cells in the battery string, it is considered a hidden crack on an odd cell. If the hidden crack is located on an even number of cells in the battery string, it is considered a hidden crack on an even cell (odd cells and even cells are two subcategories).

[0063] If the hidden crack is located on the 1st, 4th, 7th... battery cells in the battery string, it belongs to the first type of hidden crack (1 when divided by 3); if the hidden crack is located on the 2nd, 5th, 8th... battery cells in the battery string, it belongs to the second type of hidden crack (2 when divided by 3); if the hidden crack is located on the 3rd, 6th, 9th... battery cells in the battery string, it belongs to the third type of hidden crack (0 when divided by 3).

[0064] If the defect category is a cold solder joint, the cold solder joint can be sub-classified, and the classification result includes at least one of the following:

[0065] (21) Whether the cold solder joint penetrates the battery cell

[0066] Similar to classification (11), except that the defect is a cold solder joint.

[0067] (22) Is the cold solder joint located on the cut surface or chamfered surface of the cell?

[0068] Similar to category (12), except that the defect is a cold solder joint.

[0069] (23) Is the cold solder joint located at the upper and lower edges of the battery cell?

[0070] Similar to classification (14), except that the defect is a cold solder joint.

[0071] (24) Is the cold solder joint large in area?

[0072] Refers to whether the area of ​​cold solder joint exceeds a certain area threshold. If it exceeds, it is a large-area cold solder joint.

[0073] (25) The cell number of the battery string where the cold solder joint is located

[0074] Similar to classification (18), except that the defect is a cold solder joint.

[0075] (26) Is the cold solder joint accompanied by bright spots?

[0076] It refers to whether the two defect categories of cold solder joints and bright spots appear in the image of the battery string at the same time. Of course, the locations where the two defects appear may also be considered to determine whether there is a "accompanying" relationship.

[0077] Note that the same defect can have several different subcategories at the same time, because some of the above subcategories are not mutually exclusive.

[0078] Since the defects have been classified once during visual defect detection, the above sub-classification is actually a secondary classification of the defects. Therefore, the above sub-classification results can also be called secondary classification results of the defects.

[0079] Step S130: Match the first data to be analyzed with the alarm rules in the rule base. If the match is successful, output the first alarm information based on the matched alarm rules.

[0080] Before executing step S130, the rule base has been established in advance. The rule base is a collection of a large number of alarm rules. These alarm rules can be constructed based on the correlation between the historical data to be analyzed and the actual stringer failure or the actual stringer maintenance requirements. The construction method can be, for example, automatic construction by algorithm, manual analysis construction, or a combination of these two methods. The historical data to be analyzed and the first data to be analyzed have similar meanings and are not explained in detail, but the acquisition time is before the first data to be analyzed. The historical data to be analyzed may come from the stringer to be analyzed (or the battery string it produces), or it may come from the same model stringer to be analyzed (or the battery string it produces).

[0081] The alarm rules may include rules of the time dimension and / or rules of the non-time dimension. The rules of the time dimension are for the first data to be analyzed within a time window. For example, if the frequency of abnormalities in the data to be analyzed within a time window exceeds a certain threshold, it is determined that the stringer has some kind of fault. The time window here can refer to a specific duration, such as 5 minutes, 10 minutes, or a certain number of battery strings, such as every 10 battery strings, every 100 battery strings, or a certain batch of battery strings, such as every batch of battery strings, every two batches of battery strings, and so on (the battery strings in the above examples can also be equivalently replaced by battery cells at the input end of the stringer). In short, the time window does not necessarily correspond directly to the duration, and the number of battery strings can also be regarded as another way of expressing the duration. When the time window is mentioned later, it can be understood in a similar way and will not be repeated. The rules of the non-time dimension are for a single first data to be analyzed. For example, if a first data to be analyzed is greater than a certain threshold, it is determined that the stringer needs some maintenance. Since the quality characteristic data of the battery string (eg, secondary classification results of cold solder joints and hidden cracks) are extracted in step S120, the alarm rules must include rules corresponding to the quality characteristic data. Some specific rule examples will be given later.

[0082] Whether it is a rule in the time dimension or a rule in the non-time dimension, there may be some parameters (such as thresholds), which can be initially given by experience or model prediction. And optionally, these parameters can also be continuously optimized as the alarm method is executed. Specific examples will be given later.

[0083] The alarm rules in the rule base may be fixed or variable. In addition to the parameter updates mentioned above, rules may also be added, modified, or deleted.

[0084] Matching the first data to be analyzed with the alarm rules in the rule base means verifying one by one whether the first data to be analyzed conforms to the content of the alarm rules. Of course, different rule bases can be set for different types of stringer machines, or only the alarm rules related to the current type of stringer machine in the rule base can be matched; for different first data to be analyzed, only the alarm rules related to the first data to be analyzed in the rule base can be matched.

[0085] The alarm rule can record the fault information of the stringer, such as the cause of the fault, such as which component in the stringer caused the fault, or which improper operation or setting of the stringer caused the fault, and the solution to the fault, etc. Therefore, once such an alarm rule is matched, the first alarm information containing the fault information can be output. In addition to the cause of the fault and the fault solution, the first alarm information can also include a fault prompt, that is, prompting the user that the current stringer has a fault.

[0086] The alarm rules may also record the maintenance information of the stringer, such as maintenance suggestions, such as how to perform the stringer maintenance, when to perform the stringer maintenance, and the consequences of not performing maintenance, such as what kind of failures may be caused, etc. Therefore, once such an alarm rule is matched, the first alarm information containing the maintenance information may be output. In addition to the maintenance suggestions and the consequences of not performing maintenance, the first alarm information may also contain maintenance prompts, that is, prompting the user that the current stringer needs maintenance.

[0087] The alarm rules that record fault information are called fault rules, and the alarm rules that record maintenance information are called maintenance rules. The rule base can implement both fault rules and maintenance rules, or only one of the two types of rules can be implemented according to needs.

[0088] There are many ways to output the first alarm information, for example, it can be output to a system interface (for example, a web page or client interface) for display, or it can be output through voice broadcast, or it can be sent to the user through text messages or emails, or it can be output through light alarms, or it can also combine several different output methods at the same time.

[0089] If the first data to be analyzed fails to match any alarm rule, no processing is performed.

[0090] Brief summary Figure 1 In the method, since the battery string is produced by a string welding machine, once the string welding machine fails or needs maintenance, it will be reflected in the quality of the battery string. Therefore, in the above method, based on the first original data containing the quality data of the battery string (for example, the visual defect detection result), the equipment is given a fault alarm and / or a maintenance alarm, which can improve the accuracy of the alarm, and is conducive to timely detection of string welding machine failures or timely maintenance of the string welding machine, thereby ensuring the quality of the produced battery string. In contrast, in the prior art, the visual defect detection results of the battery string are usually only used for the quality assessment of the battery string, and are not associated with the status assessment of the string welding machine.

[0091] In addition, the method does not issue an alarm directly based on the first original data, but issues an alarm based on the first data to be analyzed further obtained from the first original data. Since the first data to be analyzed includes quality feature data obtained by feature extraction from quality data, such as the results of secondary classification of defects in battery strings, especially the secondary classification results of the two important defects of cold solder joints and hidden cracks, it further explores the correlation between defects and stringer failures or maintenance needs, and is therefore also conducive to improving the accuracy of alarms.

[0092] In addition, based on this method, not only can a fault alarm be issued when a fault occurs in the string welding machine, but predictive maintenance can also be performed when the string welding machine has not failed, which is helpful to reduce the probability of failure.

[0093] In addition, the method adopts the same processing logic for the fault alarm and maintenance alarm of the string welding machine. For the user, whether it is a fault alarm or a maintenance alarm, it only needs to configure the corresponding alarm rules in the rule library, which is convenient to handle.

[0094] When introducing step S120 above, it was mentioned that the quality data of the battery string can be feature extracted from at least one dimension of form, time and space to obtain quality feature data of the corresponding dimension. The following introduces possible feature extraction methods for the above three dimensions. It should be understood that the feature extraction methods are not limited to the methods mentioned below:

[0095] A. Morphological Dimension

[0096] The defects are secondary classified according to the defect category, defect location and battery string image, and the secondary classification results are obtained as the quality feature data of the morphological dimension. The following introduces possible secondary classification methods using the two defect categories of hidden cracks and poor solder joints as examples. It should be noted that the logic of the secondary classification corresponding to each different defect category may be different. Therefore, the steps to be performed for the secondary classification must first be determined according to the defect category. For example, if the defect category is hidden cracks, execute the relevant steps in A1, and if the defect category is poor solder joints, execute the relevant steps in A2.

[0097] A1. Secondary classification of hidden cracks

[0098] Step a1: extracting a hidden crack image from the image of the battery string according to the location of the defect.

[0099] The location of the defect in the visual inspection result can be a detection box, such as Figure 2a The white box in the detection box is used to extract the image content in the detection box from the image of the battery string, and the image is used as the hidden crack image in step a1. Of course, the image may not be completely captured according to the detection box. For example, the area of ​​the detection box may be appropriately enlarged (for example, enlarged by 1.2 times) before capturing the image, and so on.

[0100] In step a1, a secondary classification of hidden cracks is performed by intercepting a hidden crack image located in a local part of the original image, thereby reducing the image content to be analyzed, especially some non-hidden crack image content, which is conducive to improving the classification accuracy.

[0101] Step a2: Extract the skeleton of the hidden crack according to the hidden crack image.

[0102] For example, morphological or artificial intelligence algorithms can be used to extract the skeleton of a specific object (such as a crack) from a crack image. The skeleton can be a line with a single pixel width, and the trend of the line describes the morphology of the crack.

[0103] Step a3: Determine the secondary classification result corresponding to the hidden crack according to the hidden crack skeleton, or according to the hidden crack skeleton and the position of the constituent elements of the battery cell.

[0104] Step a3 includes two parallel schemes. Scheme 1 is to determine the secondary classification result corresponding to the hidden crack based on the skeleton of the hidden crack. Scheme 2 is to determine the secondary classification result corresponding to the hidden crack based on the skeleton of the hidden crack and the position of the constituent elements of the battery cell.

[0105] For scheme 1, for example, the characteristic points in the skeleton (for example, the inflection points or end points in the skeleton) can be determined, and then the geometric shape of the skeleton can be determined based on the positional relationship between these characteristic points, and then a secondary classification result related to the geometric shape of the hidden cracks can be given, such as X-type hidden cracks, V-type hidden cracks, etc.

[0106] For scheme 2, the constituent elements of the battery cell refer to the main grid, edge and other constituent elements of the battery cell. The position of each battery cell can be detected in the image of the battery string, and then combined with the layout information of the battery cell, the position of each constituent element of the battery cell in the image of the battery string can be determined. It should be noted that if the layout of the battery cell does not change, and the position of the battery string in its image is almost fixed, the position of each constituent element of the battery cell can be calculated only once, and there is no need to calculate repeatedly. On the one hand, the geometric shape of the hidden crack can be determined based on the skeleton, as described in scheme 1; on the other hand, the skeleton also represents the position of the hidden crack, and it is more accurate than the detection frame in the visual inspection result. The position of the skeleton can be compared with the position of the constituent elements of the battery cell to determine the specific position of the hidden crack relative to the battery cell. Combining these two aspects of information, several refined classifications of the hidden cracks can be achieved. Of course, it is not ruled out that some secondary classifications only need to use the second aspect of information.

[0107] For scheme 2, for example, the intersection of the skeleton extracted in step a2 and the main grid line of the cell can be calculated. If there is no intersection, the main grid fishbone crack is ruled out. If there is an intersection, the skeleton is divided into two upper and lower branches from the intersection, and the slopes of the upper and lower branches are calculated respectively (for example, some pixels can be extracted from the upper and lower branches to calculate the slopes). If the ratio of the slopes of the two branches is close to -1 (the error is within a certain range), it means that the angle between the upper and lower branches is close to 90°, so it can be determined that the secondary classification result is a main grid fishbone crack. For the definition of the main grid fishbone crack, please refer to the previous classification (16).

[0108] In step a3, since the skeleton of the hidden crack fully represents the morphology and precise position of the hidden crack, secondary classification of the hidden crack based on the skeleton of the hidden crack is beneficial to improving the classification accuracy.

[0109] In one implementation, step a2 may further include:

[0110] Step a21: determining at least one defective area according to the hidden crack image.

[0111] The defective area is the image area containing hidden cracks. There may not be only one defective area in the hidden crack image, because the hidden cracks in the image are not necessarily distributed continuously. Figure 2a~Figure 2g It is not difficult to see that the pixel value and shape of the hidden crack are obviously different from the surrounding image content, so the location of the defect area can be determined from the hidden crack image using morphology or artificial intelligence algorithms.

[0112] Step a22: A skeleton extracted from the largest area in at least one defect area is used as a skeleton of a hidden crack.

[0113] The maximum region can be defined in different ways, such as the largest area, or the longest border of the region, etc. A possible algorithm for determining the maximum region is to extract the contours of each defect region, for example, the findcountours algorithm in the OpenCV library can be used for extraction. The extracted contour is a closed curve, so the area of ​​each defect region can be calculated, and then the largest region can be selected, and the remaining defect regions can be ignored.

[0114] When extracting the skeleton, you can only extract it for the largest area, of course, you can also extract the skeleton for all defective areas, but only the skeleton in the largest area will be used. For example, you can use the fire model or the maximum disk method to extract the skeleton. Figure 3 Schematic diagram of the outline and skeleton of the defect area, refer to Figure 3 , the parts surrounded by the two contours represent two defect areas, each defect area includes a skeleton, among which the defect area below is the largest area, and only the skeleton in this area will be used in step a3.

[0115] In the above implementation, only the skeleton of the largest defect area is used for the secondary classification of hidden cracks, while the skeletons of other defect areas are ignored, which is conducive to grasping the main morphological features of hidden cracks, thereby obtaining a more accurate classification result.

[0116] In some alternatives, step a22 may also be implemented as: determining a target region based on at least one defect region, and using a skeleton extracted from the target region as a skeleton of a hidden crack.

[0117] The target area can be an area selected from at least one defect area, such as the maximum area mentioned above, but a defect area can also be selected as the target area according to other rules, such as any area whose area is greater than a certain threshold. Alternatively, the target area can also be an area obtained after processing based on at least one defect area. For example, assuming that five defect areas are obtained in step a21, and the areas of three of the five defect areas are greater than a certain threshold, the three defect areas can be combined as the target area, and so on.

[0118] The above alternative schemes have greater flexibility in determining the target area and can better adapt to the characteristics of different forms of hidden cracks, thereby more accurately extracting the skeleton of the hidden cracks.

[0119] In one implementation, step a21 may further include:

[0120] Step a211: perform a binarization operation on the hidden crack image to obtain a binarized image.

[0121] For example, the hidden crack image can be binarized by setting a threshold. In the binarized image, ideally, the pixel with the first value represents the area where the hidden crack is located, that is, the defect area, and the pixel with the second value represents the image content other than the hidden crack. The first value can be 0 or the like, and the second value can be 1, 255 or the like, or vice versa.

[0122] Step a212: filtering the binary image to filter out interference elements other than hidden cracks in the binary image to obtain a filtered image.

[0123] Step a213: Determine a pixel region formed by pixels of the first value in the filtered image as at least one defect region.

[0124] The above two steps are explained in combination. In step a211, the ideal binary image is mentioned, but in practice, not all pixels with the first value are necessarily pixels that constitute hidden cracks. For example, it is possible that the main grid line, auxiliary grid line or edge of the battery cell also has the first value, which makes it difficult to distinguish hidden cracks from these elements in terms of pixel values. We may call them interference elements. Since the defect area is determined based on the pixels with the first value in step a213, in order to improve the accuracy of the determined defect area, the interference elements in the binary image can be filtered out by a filtering algorithm in step a212. It should be noted that the filtering operation does not guarantee that every interference element can be completely filtered out. Even if the influence of the interference element is only weakened, it is also a filtering operation.

[0125] For example, the filtering of the main grid line can be carried out in the following ways:

[0126] First, the position of the main grid line of the battery cell in the binary image is determined according to the layout of the battery cell and the image of the battery string. The determination method has been described above. Then, for the binary image, within a preset range near the position of the main grid line, a convolution operation is performed using a convolution kernel constructed based on the size of the main grid line to filter out the main grid line in the binary image. For example, if the position of a main grid line is a coordinate, and the main grid line is horizontally distributed (similar to Figure 2a~Figure 2g ), the above preset range may be a range of ±m pixels of the coordinate in the vertical direction (m is a positive integer), and the preset range is greater than the width of the main grid line. The size of the main grid line may include the length and width of the main grid line.

[0127] The filtering of other interference elements can refer to the filtering of the main grid line. Of course, other filtering algorithms can also be used, and convolution is not necessary.

[0128] In the filtered image, each connected domain that only includes pixels whose values ​​are the first pixel values ​​is a defect area, so the positions of all defect areas can be determined.

[0129] In the above implementation, since the binary image is filtered, some non-hidden crack interference elements in the image can be excluded, so that the position of the defect area is determined more accurately, so that the skeleton extracted later is more accurate, and the secondary classification result is more accurate. In particular, since there are many main grid lines in the battery cell, the interference of hidden cracks is also more obvious. In the above implementation, a method for filtering the main grid lines is also specifically given.

[0130] It should be understood that even if the interference elements are not filtered, or only some of the interference elements are filtered, the defect area can be determined, and the hidden cracks can be classified again, but the classification accuracy may be reduced. In addition, the determination of the defect area does not necessarily have to be performed in the binary image. For example, it is also possible to determine it directly by analyzing the pixel values ​​from the hidden crack image (which can be a grayscale image or a color image).

[0131] In addition to steps a1 to a3, other methods may be used to determine the subdivision categories of certain hidden cracks:

[0132] For example, for the classification of hidden cracks (18), it is only necessary to compare the location of the defect (detection frame) with the location of each cell on the battery string without extracting the skeleton.

[0133] For another example, for the classification of hidden cracks (15), a hidden crack image can be first captured from the image of the battery string according to the location of the defect, and then the hidden crack image can be input into a hidden crack classification model to obtain a secondary classification result corresponding to the hidden crack. The hidden crack classification model can be, for example, a neural network model, a support vector machine model, etc., and the hidden crack classification model needs to be trained before use.

[0134] In this example, on the one hand, by intercepting a local hidden crack image for secondary classification of the hidden crack, the image content to be analyzed is reduced, which is conducive to improving the classification accuracy. On the other hand, directly using the model to perform secondary classification of the hidden crack not only has relatively simple processing logic, but also has advantages over traditional algorithms in the recognition of certain hidden crack forms, which is also conducive to improving classification accuracy. It should be understood that in addition to classification (15), it is not excluded that subdivision categories can also be identified in this way.

[0135] For the secondary classification result of hidden cracks, the corresponding alarm rules are as follows: if the secondary classification result of hidden cracks is hidden cracks in the main grid fishbone, then the string welding machine has fault a.

[0136] A2. Secondary classification of cold solder joints

[0137] According to the position of the defect and the position of the constituent elements of the battery cell, the secondary classification result corresponding to the cold solder joint is determined; wherein the position of the constituent elements of the battery cell is determined according to the layout of the battery cell and the image of the battery string.

[0138] As has been described above, the position of the cold solder joint (detection frame) can be compared with the position of the constituent elements of the cell, so as to determine the specific position of the cold solder joint relative to the cell, and then achieve several refined classifications of the cold solder joint. For example, if the cold solder joint detection frame overlaps with the upper edge of the cell, the result of the cold solder joint classification (23) can be that the cold solder joint is located at the upper edge of the cell.

[0139] In the above implementation, secondary classification of cold solder joints can be completed by only using the positions of defects and constituent elements of the battery cell. The classification process is simple and the classification result is accurate.

[0140] In addition, it is also possible to determine the subcategories of certain cold solder joints by other means: for example, for the cold solder joint classification (25), it is only necessary to compare the position of the defect (detection frame) with the position of each battery cell on the battery string, without using the position of the constituent elements of the battery cell. For another example, the secondary classification result corresponding to the cold solder joint can be obtained by combining the categories of other defects except the cold solder joint. For example, for the cold solder joint classification (26), on the basis of knowing that the current category is the cold solder joint, it is also necessary to combine the defect information of the bright spot category to obtain the secondary classification result.

[0141] When the computer vision algorithm performs visual defect detection on the battery string, its detection results may include the defect category, that is, a defect classification is performed once, but the classification result may be relatively rough and fails to correspond well to the cause of the stringer failure or the maintenance needs of the stringer. Therefore, the defects of the battery string are secondary classified, and the classification result is more consistent with the cause of the stringer failure or the maintenance needs of the stringer. This classification result is used as quality feature data in the morphological dimension, which is conducive to improving the accuracy of the alarm.

[0142] B. Time Dimension

[0143] Count the frequency and / or density of defects in the first preset time window, and use the obtained frequency and / or density as the quality characteristic data of the time dimension. Optionally, frequency can be defined as the number of times a defect occurs per unit time, and density can be defined as the number of times a defect occurs consecutively. Of course, other definitions are not excluded. Note that if multiple defects appear on a battery string, the number of times the defect occurs can be counted as 1, indicating that the battery string has defects, or it can be counted as the actual number of defects that occur, or there can be other statistical methods.

[0144] For example, the first preset time window is half an hour. The visual inspection results of the battery strings produced within half an hour are counted. If 5 battery strings have defects, the defect frequency is 10 pieces / hour. The corresponding alarm rules are as follows: if the defect frequency is greater than 8 pieces / hour, the string welding machine has fault b1.

[0145] For another example, the first preset time window is a batch of battery strings. The visual inspection results of a batch of battery strings are counted. If a maximum of 20 continuously produced battery strings have defects, the defect density is 20 (or the maximum value of the defect density is 20). The corresponding alarm rules are, for example: if the density of defects is greater than 15, the string welding machine has fault b2.

[0146] According to requirements, the first preset time window can also be multiple values ​​instead of just one value. For example, the first preset time window can be 5 minutes and half an hour, indicating that defect statistics must be performed every 5 minutes and every 30 minutes, and the statistical results for 30 minutes can be obtained based on the statistical results for 5 minutes.

[0147] It should also be noted that the above examples do not distinguish between defect categories. During implementation, the frequency and / or density can be counted according to the defect category, and it is not necessary to mix all categories of defects together for statistics.

[0148] The battery string defects caused by the failure of the string welding machine are likely not accidental, but densely distributed in the time dimension. The two indicators of frequency and density are both a reflection of the density of defects in the time dimension. Therefore, in the above implementation method, the frequency and / or density of the string welding machine defects are statistically analyzed, and such statistical results are used as the quality feature data of the time dimension, which is conducive to improving the accuracy of the alarm.

[0149] C. Spatial Dimension

[0150] According to the defect positions in the visual inspection results and the positions of the equipment components in the stringer, the correlation between the defects and the equipment components is calculated, and the obtained correlation is used as the quality feature data of the spatial dimension.

[0151] Among them, the position of the defect in the visual inspection result can refer to the position of the detection frame containing the defect in the image of the battery string. After the stringer is installed, the position of the equipment component is known, and the position can also be mapped to the image plane, so that the position of the defect and the position of the equipment component are aligned to a coordinate system, which is convenient for subsequent calculations. Of course, position alignment can also be achieved in other ways, such as aligning the two to the world coordinate system. Among them, the equipment components of the stringer include pressure pins, lamps, tooling, suction cups, columns, etc.

[0152] There are many ways to calculate the correlation between defects and equipment parts. For example, the area overlap between the defect and the equipment part can be calculated based on the defect location and the equipment part location, and the obtained overlap can be used as the correlation between the two. The area of ​​the defect can refer to the area of ​​the detection frame in which it is located. Of course, if the shape of the defect in the detection frame is further accurately identified (for example, the skeleton or outline mentioned above), the defect area can also be calculated more accurately; the area of ​​the equipment part can refer to the area of ​​its projection shape on the image plane, or the area of ​​the circumscribed rectangle of the projection, etc. There are also different ways to define the area overlap between the two. For example, the intersection over union (IoU) of the two areas can be used as the area overlap. The location of the defect and the equipment part is required when calculating the intersection over union.

[0153] For example, the area overlaps between the defect and equipment components c1, c2, and c3 are 0.05, 0.1, and 0.35, respectively, that is, the area overlap with c3 is the largest. Combined with other aspects of the alarm rule matching, it is determined whether the defect is caused by a failure in c3.

[0154] For example, the area overlaps between defects and equipment components c4, c5, and c6 are 0.04, 0.22, and 0.39, respectively, and the area overlap threshold is 0.2, which means that only those with an area overlap greater than the threshold are considered likely to cause defects. Thus, only c5 and c6 may cause defects. Combined with the matching of other alarm rules, it is determined whether the defect is caused by the failure of c5 or c6.

[0155] Since the area overlap between defects and equipment components fully reflects the degree of correlation between the two in spatial position, it is reasonable to use the area overlap as the correlation between defects and equipment components. In addition, both the defect area and the equipment component area are relatively easy to obtain, which helps to reduce the difficulty of calculating the correlation.

[0156] In addition to the area overlap, the position offset between the defect position and the position of the equipment component can also be calculated based on the defect position and the position of the equipment component, and the offset can be used as the correlation between the two. Optionally, to simplify the calculation, the defect position can be represented by one coordinate, such as the center coordinate of the detection frame, and the position of the equipment component can be represented by one coordinate, such as the center coordinate of the circumscribed rectangle of the projection shape of the equipment component on the image plane. Of course, one set of coordinates can also be used to represent the defect position and another set of coordinates can be used to represent the position of the equipment component. Each set of coordinates can include one or more coordinates, and the position offset between the two is defined as the distance between the two sets of coordinates.

[0157] Alternatively, the area overlap and position offset of the defect and the equipment component may be calculated simultaneously, and these two types of information may be combined as the correlation between the defect and the equipment component.

[0158] Alternatively, other methods are used to calculate the correlation between defects and equipment components, which will not be given as examples.

[0159] The defects of some battery strings caused by the failure of the stringer are closely related to the positions of the equipment components in the stringer. For example, the equipment component that causes the defect (most likely the equipment component that fails) is likely to be located near the defect. Therefore, in the above implementation method, the correlation between the defect and the equipment component is calculated using the spatial position, and this correlation is used as the quality feature data of the spatial dimension, which is conducive to improving the accuracy of the alarm.

[0160] Alternatively, the quality feature data of the spatial dimension may not be extracted based on the position of the equipment component. For example, the specific position of the defect in the battery string can be determined based on the defect position in the detection result and the image of the battery string, and the obtained specific position is used as the quality feature data of the spatial dimension. For example, referring to the previous description, the position of the battery cell or each component of the battery cell can be determined in combination with the layout of the battery string, and then the position of the defect and the position of the battery cell or each component of the battery cell can be compared to finally determine the specific position of the defect in the battery string, such as the defect is located in the third battery cell of the battery string, the defect is located at the upper edge of the battery cell, the defect is located at the cutting corner of the battery cell, and so on.

[0161] For example, the position of the defect is (100, 200) (center coordinates of the detection frame), which corresponds to a main grid line of the battery cell, and the main grid line equipment component for processing the battery cell is known, so it can be determined that the equipment component has a fault (it can also be combined with other rules).

[0162] It is not difficult to see that the specific location of the above defects and some of the subcategories of defects have similar meanings. For example, the defect located at the upper edge of the cell can be regarded as both a spatial location information and a category information (located at the upper edge, not in the upper edge category). In fact, the morphological, temporal, and spatial dimensions are just different perspectives on the data, but do not affect the essential content of the data.

[0163] Different positions of the battery string may be processed by different equipment components in different string welding machines. Therefore, defects located at different positions of the battery string in the image of the battery string may correspond to failures of different equipment components. Therefore, in the above implementation method, the defect position in the visual inspection result is further located to a specific position on the battery string in combination with the image of the battery string, and this specific position is used as quality feature data in the spatial dimension, which is conducive to improving the accuracy of the alarm.

[0164] On the basis of the above embodiment, optionally, the first original data may also include process data of the stringer and / or operation status data of the stringer.

[0165] The process data refers to the process parameters used by the stringer to process the battery string, such as welding current, welding pressure, heating temperature, heating power, etc. These data can be collected during the production of the battery string, for example, by sensors built into the stringer or external sensors. The process data can be time series data, that is, the collected process data can be timestamped.

[0166] The operating status data refers to the parameters that characterize the operating status of the stringer itself, such as the temperature of the stringer itself, motor vibration data, motor torque data, noise data emitted by the stringer itself, etc. These data can be collected during the production of battery strings, for example, through sensors built into the stringer or external sensors. The operating status data can be time series data or non-time series data.

[0167] The sensors mentioned above are selected according to the data collection requirements, such as temperature sensors, pressure sensors, vibration sensors, sound sensors, etc. For the sensors that come with the stringer, the stringer can also transmit the real-time data collected by these sensors to the Programmable Logic Controller (PLC) through the input port of the PLC, and the server (execution alarm method) then connects to the PLC to obtain these data. Of course, the solution of the stringer directly connecting to the server is not ruled out. For external sensors, they can also be directly connected to the server without being transferred through the PLC. Of course, the solution of transferring through the PLC is not ruled out.

[0168] If the first original data also includes the process data of the string welding machine and / or the operating status data of the string welding machine, the process data and / or the operating status data may also be processed in step S120. The processing method may include preprocessing, feature extraction of different dimensions, etc. For details, please refer to the processing method for quality data, which will not be repeated.

[0169] In the above implementation, the first original data may further include data related to the stringer, which enriches the data source of the alarm analysis and is therefore helpful to improve the accuracy or comprehensiveness of the alarm.

[0170] On the basis of the above embodiment, optionally, the first original data may also include incoming material data during the production of the string welding machine.

[0171] Incoming material data refers to the upstream incoming materials and auxiliary materials data used in the production of stringer, which may include one or more data such as the name, manufacturer, batch, specification, quality of incoming materials and auxiliary materials. Incoming materials include battery cells, and auxiliary materials include flux, replaceable parts of stringer, etc.

[0172] For example, more defects were detected in the battery strings produced by stringer M1, while fewer defects were detected in the battery strings produced by stringer M2 of the same model. The incoming material data of stringers M1 and M2 can be obtained, and the alarm rule can be: if the incoming materials of stringers M1 and M2 are the same, then stringer M2 has a fault or needs maintenance (of course, it can also be judged in combination with other information).

[0173] For another example, auxiliary material data may include not only auxiliary materials for processing battery strings, but also data on some equipment components of string welding machines (such as some components that need to be replaced regularly). Failures or maintenance of these equipment components may be involved in alarm rules.

[0174] For another example, the incoming material data may include the battery cell pattern. According to the above description, the battery cell pattern can be used for secondary classification of defects and other purposes.

[0175] According to the above example, the incoming material data may be directly used to match the alarm rules, or may be used to form alarm rules, extract features of quality data, etc. Since the incoming material data directly reflects the characteristics of the battery string itself, it is helpful to improve the accuracy or comprehensiveness of the alarm.

[0176] In addition to product quality data, process data of the stringer, operating status data of the stringer, and incoming material data of the stringer, some other data can also be collected for alarm analysis, although these data may not be directly used to match the alarm rules. Here are some such data:

[0177] (1) Historical alarm events of the string welding machine and their handling methods

[0178] For example, it may include the time when the alarm event occurred, the data that caused the alarm, and the specific content of the alarm information (including solutions). These data can be used to construct alarm rules.

[0179] (2) Process formula data

[0180] Process recipe data includes: the raw materials, auxiliary materials and their usage (recipe data) required to produce a certain battery string during the battery string manufacturing process, as well as specific process operation steps and parameter settings (process data). Process recipe data is used to ensure the stability of the production process and the quality of the final battery string, covering all key links from raw material use to finished product production. For example, the quality data of the battery string will be different for different recipes, so the parameters in the alarm rules need to be adjusted accordingly.

[0181] (3) Location data of equipment components

[0182] It can be used to calculate quality characteristic data of morphological or spatial dimensions, as described above.

[0183] Note that the process formula data may include the raw material data of the battery string, but the information is relatively rough, mainly including the name and amount of the raw materials, while the incoming material data may include more detailed raw material information. In addition, the formula data may include the process parameters of the stringer, such as the heating temperature, but the process parameters in the formula data are set values, not real-time collected values, and therefore are different from the process data of the stringer.

[0184] Based on the above embodiments, some possible warning rules are described below by way of example:

[0185] Rule 1

[0186] If the defect category is cold soldering, and the secondary classification result of the defect is welding deviation, and the defect is located on the back of the battery string, and the number of battery strings with the above defect within the first preset time window is not less than the first number, then the stringer has a transfer platform that is not level, and the solution to this fault can be to adjust the position of the transfer platform. If rule 1 is matched, the cause of the fault and the solution in the rule can be output as the first alarm information (the content of the first alarm information targeted by the following rules is similar).

[0187] Among them, welding deviation refers to the fact that the welding strip and the main grid line of the battery cell do not overlap (ideally, they should overlap), that is, an offset occurs. When visually inspecting defects, images of the front and back sides of the battery string can be collected, so it can be determined whether the defect occurs on the front or back of the battery string. The first preset time window can be, for example, half an hour. The first quantity can be, for example, 3 battery strings. The transfer platform in the string welding machine is used to transport battery strings. The welding deviation problem may be caused by the transfer platform not being turned to the correct position when transporting the battery string. Of course, there are other possible reasons for welding deviation, which are simplified here as examples.

[0188] Rule 2

[0189] If the defect category is hidden crack, and the secondary classification result of the defect is x-type hidden crack, and the defect is located in the battery cell at the first position in the battery string, and the number of battery strings with the above defect within the first preset time window is not less than the second number, then the stringer has a problem of tin accumulation on the pressure pins. Optionally, the solution to this fault may be to clean and maintain the pressure pins in the light box of the stringer, for example, by cleaning them with a steel brush.

[0190] Among them, the definition of x-type hidden cracks has been mentioned in the previous text. The first position can be any position in the battery string, such as the third position. The first preset time window can be, for example, half an hour, but it should be noted that the first preset time window in Rule 2 is not necessarily the same as the first preset time window in Rule 1. When introducing the quality characteristic data of the time dimension, it is also mentioned that the first preset time window can take multiple different values ​​to adapt to different statistical needs. The second quantity can be, for example, 3 battery strings. What this means is that among the battery strings produced within half an hour, 3 or more battery strings have x-type hidden cracks at the battery cell at the first position. The cracks of x-type hidden cracks are generally split from the middle to the four corners due to force, which may be caused by abnormal force applied when the pressure needle contacts the battery cell due to tin stacking, and generally appear continuously on the battery cell at the same position. This is the basis for setting Rule 2.

[0191] Rule 3

[0192] If the defect category is cold soldering, and the number of defects occurring on a single battery string is greater than the first threshold, and the number of battery strings with the above defects within the first preset time window is not less than the third number, then by comparing the flux information used in different production orders of the stringer and the corresponding production quality, determine whether there is a problem with the flux currently used by the stringer. Optionally, if Rule 3 is used to determine that there is indeed a problem with the flux, the solution may be to replace the flux, or further test the quality of the flux currently used in the laboratory.

[0193] If the number of defects occurring on a single battery string is greater than a first threshold, it indicates that the defect occurs in large quantities on the battery string, or it can also be defined as a batch defect (a secondary classification category). The first preset time window can be, for example, half an hour. The third number can be, for example, 3 battery strings.

[0194] If it is detected that batch cold soldering occurs frequently in a short period of time, the quality problem of flux is one of the possible reasons, but it may also be an occasional problem, so further verification is needed. One production order corresponds to a certain number of battery strings. Usually, the information such as the flux brand and batch number in a production order is fixed. Therefore, the flux information and corresponding production quality of the current order of the stringer and several historical orders can be collected and compared with each other to determine whether there is a problem with the flux currently used by the stringer.

[0195] For example, a total of 10 orders are collected, of which 5 orders use flux of brand X and batch number Y (including the current order), and 5 orders use flux of brand X and batch number Z. All orders using flux of brand X and batch number Y have batch cold solder joints, and the frequency of occurrence is relatively high, while orders using flux of brand X and batch number Z do not have batch cold solder joints. This indicates that there is a high possibility that the flux of brand X and batch number Y has quality problems.

[0196] For another example, information of 5 orders in total is collected, all of which use flux of brand X and batch number Y (including the current order), but no batch cold soldering problem occurs in the orders except the current order. This means that it is very likely that the flux of brand X and batch number Y does not have quality problems, and the current order is an occasional problem or may be a fault of the string soldering machine rather than a problem with the flux.

[0197] Strictly speaking, problems with flux or other raw materials do not belong to the fault information of the stringer, nor do they belong to the maintenance information of the stringer. However, since the raw materials are used by the stringer to produce battery strings and are closely related to the stringer and battery strings, it is acceptable to classify them as fault information or maintenance information for unified analysis and processing.

[0198] Rule 4

[0199] If the defect category is hidden crack, and the secondary classification result of the defect is hidden crack of the main grid, and the correlation between the defect and the pressing pin of the stringer is greater than the second threshold, and the number of battery strings with the above defect within the first preset time window is not less than the fourth number, then the stringer has a problem of pressing pin jam. Optionally, the solution to this fault can be to clean the pressing mesh tooling of the stringer.

[0200] Among them, the definition of main grid hidden cracks has been mentioned in the previous text. How to calculate the correlation between defects and pressure pins has been mentioned in the previous text. For example, the distance between the skeleton coordinates of the hidden cracks and the projection coordinates of the pressure pins can be calculated. The first preset time window can be, for example, half an hour. The fourth number can be, for example, 2 battery strings. The position corresponding to the pressure pin of the string welding machine on the battery cell is the main grid line of the battery cell. The pressure pin jam refers to the jam when the spring pops out the pressure pin. The pressure pin is generally ejected irregularly. Therefore, the shape of hidden cracks on different battery strings may be inconsistent, and it is not convenient to use the geometric shape of hidden cracks as a basis for judgment. However, the locations where hidden cracks appear are basically on the main grid line. Therefore, frequent main grid hidden cracks may be related to the pressure pin jam. This is the basis for setting rule 4.

[0201] Rule 5

[0202] If the defect category is cold solder joint, and the secondary classification result of the defect is large-area cold solder joint, and the power of the lamp tube of the stringer corresponding to the position of the defect during welding is less than the third threshold, and the number of battery strings with the above defect within the first preset time window is not less than the fifth number, then by comparing the production quality of the lamp tube when it is currently configured with the previous power parameter, it is determined whether the lamp tube has a problem of too low power. Optionally, if it is determined by rule 5 that there is indeed a problem with the lamp tube power, the solution to the fault may be to increase the power parameter of the lamp tube.

[0203] The definition of large-area cold solder joints has been mentioned above. The string welding machine may include multiple lamp tubes, and the lamp tubes corresponding to the defective positions may be determined based on the correlation between the defects and the lamp tubes. How to calculate the correlation has been mentioned above. The power of the lamp tube during welding belongs to the process data of the string welding machine. The first preset time window may be, for example, half an hour. The fifth quantity may be, for example, 3 battery strings.

[0204] If large-area cold solder joints are detected to occur frequently in a short period of time, low lamp power is one of the possible reasons, but there may be other reasons, so further verification is needed. The power parameters of the lamp (i.e. the working power of the lamp) are configurable, but if they are not configured properly, defects such as cold solder joints may occur. If the lamp has large-area cold solder joints under the currently configured power parameters, but this problem did not occur under the last configured power parameters, it means that there is a high probability that the problem is caused by adjusting the power parameter configuration. At this time, it can be determined that the fault is that the lamp power is too low; if the lamp has large-area cold solder joints under the currently configured power parameters, and there is also this problem with the last configured power parameters, it means that there is a high probability that the problem is not caused by adjusting the power parameter configuration, but that the problem already exists, which may be caused by other equipment components of the string welding machine. This is the basis for setting Rule 5.

[0205] Rule 6

[0206] If the defect type is cold soldering, and the defect is located on the front of the battery string, and the height of the light box of the stringer corresponding to the location of the defect is greater than the fourth threshold, and the number of battery strings with the above defect within the first preset time window is not less than the sixth number, then the stringer has a problem of too high light box height. Optionally, the solution to this fault can be to reduce the height of the light box.

[0207] Among them, when visually inspecting defects, images of the front and back sides of the battery string can be collected, so it can be determined whether the defect occurs on the front or back of the battery string. The string welding machine includes a light box, and multiple light tubes can be arranged in the light box. The light box corresponding to the defect position can be determined based on the correlation between the defect and the light box (some defects may not correspond to the light box), and how to calculate the correlation has been mentioned in the previous article. The height of the light box can belong to the position data of the equipment component, which has been mentioned in the previous article. The first preset time window can be, for example, half an hour. The sixth quantity can be, for example, 3 battery strings.

[0208] Six possible alarm rules are listed above. These rules fully combine the defect characteristics in time, shape, and space dimensions, as well as data such as equipment and incoming materials, which is conducive to improving the accuracy of alarms.

[0209] Based on the above embodiments, optionally, Figure 1 The alarm methods in also include:

[0210] Step S140: receiving first user feedback for the first alarm information, and iteratively optimizing the alarm rules in the rule base according to the first user feedback.

[0211] The first user feedback includes feedback on the authenticity of the first alarm information, and may include, but is not limited to, at least one of the following feedback:

[0212] (1) Whether the fault in the first alarm information actually exists;

[0213] (2) Whether the maintenance requirement in the first alarm information actually exists;

[0214] (3) Whether the fault cause in the first fault alarm information is correct. If it is not correct, the correct fault cause can be fed back;

[0215] (4) Whether the maintenance suggestion in the first fault alarm information is correct. If not, the correct maintenance suggestion can be fed back.

[0216] Users can fill in feedback information through a system interface (such as a web page or client interface) and submit it. Of course, other methods can also be used, such as users directly speaking or inputting feedback information in natural language, and the system analyzing the content of the user's feedback through a large language model, etc.

[0217] The alarm rules in the rule base include a series of parameters (for example, thresholds), which can be called rule parameters. Iterative optimization is to continuously optimize the values ​​of these rule parameters through some algorithms, so that the accuracy of the alarm rules is getting higher and higher. For example, the iterative optimization of rule parameters can be performed through the Bayesian optimization algorithm, which specifically includes the following steps:

[0218] Step 1: Initialize the value combination of rule parameters

[0219] For example, the rule parameters can be initialized to some empirical values.

[0220] Step 2: Evaluate the value combination of rule parameters

[0221] According to the first original data obtained, an alarm is issued for the first original data based on the value combination of the rule parameters, a first alarm message is output, and first user feedback on the first alarm message is collected, and the alarm effect of the value combination of the rule parameters is evaluated using the feedback.

[0222] Step 3: Build a proxy model and predict the alarm accuracy

[0223] According to the evaluation result of step 2, Gaussian Process Regression (GPR) in the Bayesian optimization algorithm is used to model the relationship between the rule parameters and the alarm accuracy to obtain the proxy model. Optionally, if the evaluation result of step 2 is good, and a proxy model has been built when step 3 is executed, the proxy model may not be rebuilt, but only some adjustments may be made to the parameters of the proxy model; if the evaluation result of step 2 is poor, even if a proxy model has been built when step 3 is executed, the proxy model may be rebuilt.

[0224] The proxy model is used to predict the alarm accuracy corresponding to the value combination of rule parameters.

[0225] Step 4: Determine the new value combination of the rule parameters

[0226] The Bayesian optimization algorithm determines the next value combination to be tried through the acquisition function. The acquisition function can find the current optimal value combination based on the alarm accuracy and uncertainty predicted in step 3.

[0227] Jump to step 2 to continue execution. Note that the new value combination obtained in step 4 before the jump is the value combination in step 2 after the jump.

[0228] The above iterative optimization steps may be performed for each rule separately, or may be performed for several rules together.

[0229] Iterative optimization can be performed continuously as the battery string is produced, or optimization can be stopped after the rule parameters are optimized to a certain extent (for example, the evaluation result in step 2 is better than a certain standard), and so on.

[0230] In addition, some rule parameters may be given by independent model predictions. During iterative optimization, the parameters of these models can also be optimized so that the optimized models can give more accurate rule parameters.

[0231] In the above implementation, the alarm rules in the rule base can be iteratively optimized based on user feedback, which is conducive to integrating more manual experience into the rules and improving the accuracy of alarms.

[0232] Figure 4 A flowchart of another alarm method for a string welding machine provided in an embodiment of the present application. The method can be, but is not limited to, Figure 6 The electronic device shown is executed, and the structure of the electronic device can be referred to later. Figure 6 's explanation. Figure 1 The method (including its various possible implementations) is mainly aimed at warning after the battery string is produced, because the method needs to use the quality data of the battery string. Figure 4 The method (including its various possible implementation methods) is mainly aimed at the alarm during the production process of the battery string. The method does not need to use the quality data of the battery string because the battery string may still be in production at this time. These two methods can be implemented one by one or at the same time. Figure 4 , the alarm methods for string welding machines include:

[0233] Step S210: during the operation of the stringer, obtaining second original data.

[0234] The second original data includes the process data of the stringer and / or the operating status data of the stringer. As can be seen from the foregoing, the process data of the stringer and / or the operating status data of the stringer can also be used as the first original data, and the acquisition method has been explained, which will not be repeated here. If step S210 is executed first, and then step S110 is executed, and the first original data includes the process data of the stringer and / or the operating status data of the stringer, then the data acquired in step S210 can be used directly without repeated acquisition.

[0235] It should be noted that since the second raw data is obtained during the operation of the stringer, Figure 4 The method can be executed in real time during the operation of the string welding machine. Of course, the method can also be implemented in non-real time, such as first saving the collected second original data, and then performing alarm analysis based on the second original data at a suitable time in the future.

[0236] Step S220: obtaining second data to be analyzed according to the second original data.

[0237] Step S220 may refer to step S120 and will not be described again.

[0238] Step S230: using the first anomaly detection model to perform anomaly detection on the second data to be analyzed to obtain anomaly detection results.

[0239] The input of the first anomaly detection model may include the second data to be analyzed, and the output may include an anomaly detection result. The detection result may be whether an anomaly exists in the second data to be analyzed (which may also be regarded as a binary classification of the anomaly), or a more detailed anomaly classification may be given.

[0240] The first anomaly detection model can be a neural network model, a decision tree model, a support vector machine model, etc., without limitation. The first anomaly detection model can be a single model or include multiple sub-models. If the first anomaly detection model includes multiple sub-models, these sub-models can perform anomaly detection on the second data to be analyzed respectively to obtain corresponding anomaly detection sub-results, and then use a certain integration method to obtain the final anomaly detection result based on these anomaly detection sub-results. Obtaining the final anomaly detection result by integrating the anomaly detection sub-results of each sub-model is conducive to learning from each other's strengths and weaknesses, thereby improving the robustness of the detection result.

[0241] The following examples illustrate possible integration methods:

[0242] (1) Hard voting

[0243] Count the anomaly categories in all anomaly detection sub-results, and take the anomaly category with the most votes as the final anomaly detection result. For example, if the first anomaly detection model includes three sub-models, namely the isolation forest model, the single-class support vector machine model, and the autoencoder model, these three sub-models can be used to perform anomaly detection on the second data to be analyzed, and the detection results are as follows:

[0244] Isolation forest model: The probability of anomaly in class X is 0.6, the probability of anomaly in class Y is 0.2, and the probability of anomaly in class Z is 0.2; the converted vote is 1 vote for anomaly in class X.

[0245] Single-class support vector machine model: the probability of anomaly of class X is 0.5, the probability of anomaly of class Y is 0.4, and the probability of anomaly of class Z is 0.1; the converted vote is 1 vote for anomaly of class X.

[0246] Autoencoder model: The probability of anomaly of class X is 0.45, the probability of anomaly of class Y is 0.5, and the probability of anomaly of class Z is 0.05; the converted vote is 1 vote of anomaly of class Y.

[0247] According to the above three voting results, the X-type anomaly has 2 votes, the Y-type anomaly has 1 vote, and the Z-type anomaly has 0 votes. Since the X-type anomaly has the most votes, the final anomaly detection result is the X-type anomaly.

[0248] (2) Soft voting

[0249] Perform weighted calculation on multiple anomaly detection sub-results and determine the final anomaly detection result based on the weighted calculation result. Using the example in (1):

[0250] Isolation forest model: X-class anomaly probability 0.6, Y-class anomaly probability 0.2, Z-class anomaly probability 0.2; weight is 0.3.

[0251] Single-class support vector machine model: the anomaly probability of class X is 0.5, the anomaly probability of class Y is 0.4, the anomaly probability of class Z is 0.1; the weight is 0.3.

[0252] Autoencoder model: X-class anomaly probability 0.45, Y-class anomaly probability 0.5, Z-class anomaly probability 0.05; weight is 0.4.

[0253] The probability of each type of anomaly is weighted:

[0254] Probability of abnormality of type X = 0.6 × 0.3 + 0.5 × 0.3 + 0.45 × 0.4 = 0.51;

[0255] Probability of abnormality of type Y = 0.2 × 0.3 + 0.4 × 0.3 + 0.5 × 0.4 = 0.38;

[0256] Probability of class Z anomaly = 0.2 × 0.3 + 0.1 × 0.3 + 0.05 × 0.4 = 0.11;

[0257] Since the probability of class X anomalies is the highest, the final anomaly detection result is class X anomalies.

[0258] There are different ways to determine the weights used in the above weighted calculations. For example, they can be determined directly based on experience, or they can be determined based on the performance evaluation results of each sub-model. For example, the weights can be positively correlated with the performance of the sub-model. The performance here can be defined as the accuracy of the sub-model in detecting anomalies. The performance evaluation results of the sub-model can be obtained by testing the sub-model using some test data, or they can be determined directly based on the structural characteristics of the sub-model (for example, for similar models, it can be simply considered that the model performance is positively correlated with the number of model layers), or they can be determined by combining the above two evaluation methods, and so on.

[0259] The weights of the anomaly detection sub-results for weighted calculation are determined based on the performance evaluation results of the sub-models, which fully considers the impact of model performance differences on the accuracy of anomaly detection, thereby facilitating more accurate anomaly detection results and thus improving the accuracy of alarms.

[0260] A corresponding first anomaly detection model may be set for each type of second data to be analyzed, or multiple types of second data to be analyzed may share the same first anomaly detection model.

[0261] The first anomaly detection model is a model obtained by unsupervised training. Unsupervised training is training using unlabeled training samples. The labels here can be understood as the actual anomaly detection results corresponding to the training samples, such as the actual anomaly categories.

[0262] Step S240: If the abnormality detection result indicates that the second data to be analyzed is abnormal, a second alarm message is output based on the abnormality detection result.

[0263] The second warning information may include at least one of the following:

[0264] (1) Data anomaly information: for example, whether there are any anomalies in the second data to be analyzed, and the specific anomaly category.

[0265] (2) Fault information of the stringer associated with data anomalies: The fault information has been explained in the previous article and will not be repeated. In order to output the information in item (2), it is necessary to first establish an association between the data anomaly and the fault information of the stringer, or to form a corresponding mapping rule. For example, if the output power of the lamp tube of the stringer is continuously low within a time window (anomaly is detected), then the stringer has a fault of abnormal lamp heating, and the corresponding solution may be to inspect the inside of the lamp tube. For another example, if the transverse speed of the conveyor belt of the stringer is continuously low within a time window (anomaly is detected), then the transverse servo (a type of motor) of the stringer has a fault, and the corresponding solution may be to inspect the transverse servo, and so on. The above rules may be fixed or optimized over time (including addition, modification, and deletion). For example, the anomaly of a certain data was not associated with the fault of the stringer at the beginning. After the anomaly of the data has occurred repeatedly for many times, after manual analysis, the anomaly of the data is associated with the fault of the stringer.

[0266] (3) Maintenance information of the stringer associated with data anomalies: The maintenance information has been explained in the previous article and will not be repeated. In order to output the information in item (3), it is necessary to first establish an association between the data anomaly and the maintenance information of the stringer, or to form a corresponding mapping rule. For example, if the temperature is abnormal, a certain bearing of the stringer needs to be maintained. The above rules can be fixed or optimized over time (including addition, modification, and deletion). For example, if the temperature is abnormal, it is determined that the bearing is faulty. Based on this rule, further rules for bearing maintenance can be added to try to give early warnings and reduce the situation where the alarm is only issued when the bearing is actually faulty. However, there may be certain differences in the judgment of temperature anomalies for maintenance and temperature anomalies for faults. For example, for the latter, the temperature may be higher before it is judged as abnormal.

[0267] There are many ways to output the second warning information, which can refer to the output method of the first warning information, and will not be repeated. If the abnormality detection result shows that the second data to be analyzed does not have an abnormality, no processing is required.

[0268] Brief summary Figure 4 The method in the method supports anomaly detection based on the acquired second original data during the operation of the stringer, so as to timely discover anomalies in the data and issue an alarm. Since the data anomaly may be caused by a stringer failure or a sign that the stringer needs maintenance, this implementation method is conducive to early detection of stringer failure or maintenance of the stringer (rather than waiting until the battery string is produced). In addition, the anomaly detection uses the first anomaly detection model obtained by unsupervised training, that is, there is no need to collect labeled training samples in advance, which is conducive to performing this anomaly detection in the early stage of production line operation and in the absence of data accumulation. It can also detect stringer failure or maintain the stringer early.

[0269] Continue to refer to Figure 4 , optional, Figure 4 The alarm method may also include:

[0270] Step S250: receiving second user feedback for the second alarm information, and constructing a labeled training sample according to the second user feedback and the second data to be analyzed.

[0271] The second user feedback includes feedback on the authenticity of the second alarm information, and may include, but is not limited to, at least one of the following feedback:

[0272] (1) Whether the data anomaly in the second alarm information actually exists;

[0273] (2) Whether the data anomaly category in the second alarm information is correct. If it is not correct, the correct anomaly category can be fed back;

[0274] (3) Whether the fault in the second alarm information actually exists;

[0275] (4) Whether the maintenance requirement in the second alarm information actually exists;

[0276] (5) Whether the fault cause in the second fault alarm information is correct. If it is not correct, the correct fault cause can be fed back;

[0277] (6) Check whether the maintenance suggestion in the second fault alarm information is correct. If not, the correct maintenance suggestion can be fed back.

[0278] The method for obtaining the second user feedback may refer to the method for obtaining the first user feedback, and will not be elaborated again.

[0279] The label of the second data to be analyzed, such as a real abnormal category label, can be determined according to the second user feedback, and the second data to be analyzed with the label can be used as a training sample in step S260.

[0280] Step S260: When the number of collected training samples meets the training requirements, supervised training is performed using the training samples to obtain a second anomaly detection model.

[0281] The training requirements are set by the user. For example, if the collected training samples exceed the preset number, model training will be started, or if the time for collecting training samples exceeds the preset duration, model training will be started, or if no new labeled training samples are collected within a preset period of time, model training will be started based on the already collected training samples. Of course, model training can also be performed once as long as labeled training samples are collected, and so on.

[0282] The second anomaly detection model is similar to the first anomaly detection model in function. For example, it can be a single model or include multiple sub-models. A corresponding second anomaly detection model can be set for each type of third data to be analyzed, or multiple types of third data to be analyzed can share the same second anomaly detection model. If there are multiple second anomaly detection models, and different second anomaly detection models target different types of third data to be analyzed, then these second anomaly detection models may not necessarily perform supervised training synchronously, because the collection progress of training samples corresponding to different types of third data to be analyzed may not be the same, and the requirements of each second anomaly detection model for the number of training samples may not be the same.

[0283] In addition, the second anomaly detection model may be the same as the first anomaly detection model in terms of model selection (for example, both are decision tree models), or it may be different from the first anomaly detection model. For example, the second anomaly detection model includes two sub-models, namely a random forest model and a long short-term memory network, which are different from the three sub-models that may be included in the first anomaly detection model mentioned above.

[0284] Optionally, after the second anomaly detection model is trained and put into use, labeled training samples may continue to be collected, and these training samples may be used to continue iteratively optimize the second anomaly detection model.

[0285] Step S270: Acquire the third original data.

[0286] Step S280: Obtain third data to be analyzed according to the third original data.

[0287] Steps S270 and S280 are basically the same as steps S210 and S220, and will not be described in detail. The third original data is actually the second original data, but it is the second original data used by the second anomaly detection model after the second anomaly detection model is trained. In order to distinguish it from the second original data before the second anomaly detection model is put into use, it is renamed.

[0288] Steps S290 and S292 are two parallel solutions, and you can choose to implement either one. Among them, step S290 directly replaces the first anomaly detection model with the second anomaly detection model, and the first anomaly detection model is no longer used for anomaly detection in the future. Step S292 continues to retain the first anomaly detection model, and uses the first anomaly detection model and the second anomaly detection model for anomaly detection at the same time.

[0289] Step S290: using the second anomaly detection model, perform anomaly detection on the third data to be analyzed to obtain anomaly detection results.

[0290] Step S290 is similar to step S230 and will not be described again.

[0291] Step S292: Combine the second anomaly detection model and the first anomaly detection model to perform anomaly detection on the third data to be analyzed to obtain anomaly detection results.

[0292] There are many ways to combine the first anomaly detection model and the second anomaly detection model for detection. For example, the detection results of the two can be integrated, or the first anomaly detection model or the second anomaly detection model can be used to perform anomaly detection under different conditions, or the first anomaly detection model and the second anomaly detection model can be used in chronological order to perform anomaly detection, etc. More specific examples will be given later.

[0293] In steps S250-S292, a sufficient number of labeled training samples can be constructed based on user feedback, so as to perform supervised training to obtain a second anomaly detection model. Since the second anomaly detection model absorbs more manual experience, it is likely to be better than the first anomaly detection model obtained based on unsupervised training in terms of performance. Therefore, when performing anomaly detection on the subsequent third original data, the first anomaly detection model can be replaced by the second anomaly detection model to improve the accuracy of anomaly detection. Of course, the first anomaly detection model can also be retained, and the two anomaly detection models can be combined for anomaly detection to obtain more reliable anomaly detection results. For example, for some accidental anomalies, the second anomaly detection model that performs supervised training may perform poorly because they are not included in the labeled training samples. Instead, the first anomaly detection model that does not rely on labels performs better.

[0294] Based on the above embodiments, the following further introduces possible combinations of the two anomaly detection models in step S292:

[0295] Combination method 1

[0296] First, the second anomaly detection model is used to perform anomaly detection on the third data to be analyzed to obtain a preliminary anomaly detection result. If the preliminary anomaly detection result shows that the third data to be analyzed does not have an anomaly, the first anomaly detection model is further used to perform anomaly detection on the third data to be analyzed to obtain a final anomaly detection result; if the preliminary anomaly detection result shows that the third data to be analyzed has an anomaly, the preliminary anomaly detection result is directly determined as the final anomaly detection result.

[0297] Combined with method 1, the second anomaly detection model is used preferentially for anomaly detection. Since the second anomaly detection model is obtained through supervised training, it has a high detection accuracy in most cases. If the second anomaly detection model fails to detect an anomaly, it does not necessarily mean that there is no anomaly. It may just be that the training samples of the second anomaly detection model do not contain anomalies with such labels, resulting in failure to detect them. At this time, the first anomaly detection model obtained through unsupervised training is used for detection. Since the first anomaly detection model has a low reliance on labels during training, it is possible to discover missed anomalies, thereby improving the accuracy of anomaly detection.

[0298] Combination method 2

[0299] First, the first anomaly detection model and the second anomaly detection model are used to perform anomaly detection on the third data to be analyzed. The detection process can be performed serially or in parallel. The detection result obtained by the former is the first anomaly detection result, and the detection result obtained by the latter is the second anomaly detection result. Then, the final anomaly detection result is obtained based on the first anomaly detection result and the second anomaly detection result. For example, the soft voting method introduced above can be used to perform weighted calculation on the first anomaly detection result and the second anomaly detection result to obtain the final anomaly detection result.

[0300] Combination method 2 obtains the final anomaly detection result by integrating the detection results of the first anomaly detection model and the second anomaly detection model, which is equivalent to combining the advantages of supervised learning and unsupervised learning, thereby helping to improve the accuracy of anomaly detection.

[0301] Furthermore, if soft voting is used to implement combination method 2, the weights used in the weighted calculation are dynamically adjusted based on at least one of the following factors:

[0302] (1) Data quality of the third raw data

[0303] For example, when the data quality of the third original data is poor, the weight of the first anomaly detection model can be increased, while the weight of the second anomaly detection model can be reduced (the sum of the weights of the two models can be a fixed value, such as 1); when the data quality of the third original data is good, the weight of the first anomaly detection model can be reduced, while the weight of the second anomaly detection model can be increased. The principle is that when the data quality is poor, the characteristics of the abnormal data are relatively insignificant, so that more reliance can be placed on the detection results of the unsupervised model; when the data quality is good, the characteristics of the abnormal data are relatively significant, so that more reliance can be placed on the detection results of the supervised model.

[0304] As for the data quality of the third original data, it can be judged by using statistical methods, artificial intelligence methods or some rules. For example, if there are many missing values ​​in the third original data, it is judged that the data quality is poor.

[0305] (2) Data type of the third original data

[0306] For example, for temperature data, the first anomaly detection model performs better in detecting anomalies, so its weight can be increased, while the weight of the second anomaly detection model can be reduced; for pressure data, the first anomaly detection model performs poorly in detecting anomalies, so its weight can be reduced, while the weight of the second anomaly detection model can be increased.

[0307] (3) Performance evaluation results of the first anomaly detection model and the second anomaly detection model

[0308] For example, when the second anomaly detection model is first put into use, the first anomaly detection model has better performance and a larger weight, while the second anomaly detection model has a smaller weight. As the second anomaly detection model is continuously optimized, its performance gets better and better, and the corresponding weight also increases.

[0309] As for how to evaluate model performance, we have already introduced it in the previous article when introducing the sub-model of the first anomaly detection model, so we will not repeat it here.

[0310] (4) Time lapse

[0311] For example, when the second anomaly detection model is first put into use, the first anomaly detection model has better performance and a larger weight, while the weight of the second anomaly detection model is smaller. As time goes by, the weight of the second anomaly detection model gradually increases, and the weight of the first anomaly detection model gradually decreases.

[0312] The principle of item (4) is similar to that of item (3), except that the passage of time and the improvement of model performance are not necessarily completely synchronized. When it is difficult to determine the model performance, it can be assumed that the performance of the second anomaly detection model will continue to improve over time (as long as labeled training samples are continuously collected).

[0313] (5) Working scene of string welding machine

[0314] The working scenario here can represent the busyness of the stringer or the external environment, etc. For example, when the stringer is in a high-load scenario, the motor vibration pattern is different from usual, resulting in a decrease in the performance of the second abnormality detection model, so its weight can be reduced at this time, while the weight of the first abnormality detection model can be increased; when the stringer is in a medium-to-low load scenario, the motor vibration pattern is the same as usual, and the performance of the second abnormality detection model is better, so its weight can be increased at this time, while the weight of the first abnormality detection model can be reduced.

[0315] (6) Maintenance status of string welding machine

[0316] For example, there is a big difference in the status of the stringer before and after maintenance. This difference will be reflected in the data related to the stringer, and then affect the results of abnormality detection. Therefore, after each maintenance, the weights of the first abnormality detection model and the second abnormality detection model can be adjusted.

[0317] The above six weight determination factors can also be combined with each other to form a more refined weight determination rule when there is no contradiction, and no more examples are given. Based on these factors, the weights of the weighted calculation of the first anomaly detection result and the second anomaly detection result can be dynamically adjusted, which is conducive to flexible setting of weights in different scenarios, giving full play to the advantages of supervised learning and unsupervised learning in different scenarios, so as to obtain more accurate anomaly detection results.

[0318] Combination Figure 1 and Figure 4 The method (including various possible implementation methods) in the embodiment of the present application provides an alarm method for a string welder, which can provide fault and / or maintenance alarms for the string welder before, during, and after the production of a battery string, thereby helping to comprehensively improve the operational stability of the industrial assembly line and thereby improve the quality of the battery string.

[0319] Figure 5 This is a module diagram of an alarm device 300 for a string welding machine provided in an embodiment of the present application. Figure 5 The alarm device 300 for the string welding machine (hereinafter sometimes referred to as the alarm device 300) includes:

[0320] The data acquisition module 310 is used to acquire first original data, wherein the first original data includes visual inspection results of defects of the battery string produced by the stringer, wherein the visual inspection results include the category of the defect and the location of the defect, wherein the category of the defect includes cold solder joint and hidden crack;

[0321] The data processing module 320 is used to obtain first data to be analyzed according to the first original data; wherein the first data to be analyzed includes: a secondary classification result obtained by secondary classification of the defect according to the type of the defect, the location of the defect and the image of the battery string;

[0322] The alarm module 330 is used to match the first data to be analyzed with the alarm rules in the rule base. If the match is successful, the first alarm information is output based on the matched alarm rules; wherein the first alarm information includes the fault information and / or maintenance information of the string welding machine.

[0323] The alarm device 300 for the string welding machine provided in the embodiment of the present application can be used to implement the alarm method for the string welding machine provided in the embodiment of the present application. Its implementation principle and the technical effects produced have been introduced in the aforementioned method embodiment. Therefore, for the sake of brief description, for matters not mentioned in the device embodiment, please refer to the corresponding content in the method embodiment.

[0324] Figure 6 This is a structural diagram of an electronic device provided in an embodiment of the present application. Figure 6 The electronic device 400 includes: a processor 410, a memory 420 and a communication module 430. These components are interconnected and communicate with each other through a communication bus 440 and / or other forms of connection mechanisms (not shown).

[0325] The processor 410 includes one or more (only one is shown in the figure), which can be an integrated circuit chip with signal processing capability. The above-mentioned processor 410 can be a general-purpose processor, including a central processing unit (CPU), a micro control unit (MCU), a network processor (NP), or other conventional processors; it can also be a dedicated processor, including a graphics processor (GPU), a neural network processor (NPU), a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gates or transistor logic devices, and discrete hardware components. In addition, when there are multiple processors 410, some of them can be general-purpose processors, and the other part can be dedicated processors.

[0326] The memory 420 includes one or more (only one is shown in the figure), which may be, but not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), etc. The processor 410 and other possible components may access the memory 420 and read and / or write data therein.

[0327] In particular, one or more computer program instructions may be stored in the memory 420, and the processor 410 may read and execute these computer program instructions to implement the alarm method for the string welding machine provided in the embodiment of the present application.

[0328] The communication module 430 includes one or more (only one is shown in the figure), which is used for communication between the electronic device 400 and other devices, such as receiving original data from other devices. The communication module 430 can be a wired communication module or a wireless communication module.

[0329] Understandably, Figure 6 The structure shown is for illustration only. The electronic device 400 may also include Figure 6 For example, if the electronic device 400 does not need to communicate with other devices, the electronic device 400 may not include a communication module.

[0330] Figure 6 The components shown in the figure can be implemented by hardware, software or a combination thereof. The electronic device 400 may be a physical device, such as a server, a PC, a laptop, a tablet computer, a mobile phone, etc., or a virtual device, such as a virtual machine, a virtualized container, etc. Moreover, the electronic device 400 is not limited to a single device, but may also be a combination of multiple devices or a cluster consisting of a large number of devices.

[0331] The present application also provides a computer-readable storage medium, which stores computer program instructions. When these computer program instructions are read and executed by a computer processor, the alarm method for the string welding machine provided by the present application is executed. For example, the computer-readable storage medium can be implemented as Figure 6 The memory 420 in the electronic device 400, or a mobile hard disk, optical disk, USB flash disk, etc. independent of the electronic device 400.

[0332] The present application also provides a computer program product, which includes computer program instructions. When these computer program instructions are read and executed by a computer processor, the alarm method for the string welding machine provided by the present application is executed. For example, the computer program product can be stored in Figure 6 The computer program product is stored in the memory 420 in the electronic device 400 and read and executed by the processor 410. Of course, the computer program product can also be stored in a storage medium independent of the electronic device 400, such as a mobile hard disk, an optical disk, or a USB flash drive.

[0333] The above description is only an embodiment of the present application and is not intended to limit the protection scope of the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An alarm method for a string welding machine, characterized in that: include: Acquire first original data, wherein the first original data includes visual inspection results of defects of the battery string produced by the stringer, wherein the visual inspection results include the category of the defect and the location of the defect, and the category of the defect includes cold solder joint and hidden crack; Obtaining first data to be analyzed according to the first original data; wherein the first data to be analyzed includes: a secondary classification result obtained by secondary classification of the defect according to the type of the defect, the location of the defect and the image of the battery string; Matching the first data to be analyzed with the alarm rules in the rule base, and if the match is successful, outputting the first alarm information based on the matched alarm rules; wherein the first alarm information includes the fault information and / or maintenance information of the string welding machine; The defect is secondary classified according to the type of the defect, the location of the defect and the image of the battery string to obtain the secondary classification result, including: If the defect is a cold weld, the secondary classification result may be obtained in one of the following ways: Comparing the position of the defect with the position of the constituent elements of the battery cell to determine the secondary classification result corresponding to the cold solder joint; wherein the position of the constituent elements of the battery cell is determined according to the version of the battery cell and the image of the battery string; Comparing the position of the defect with the position of each battery cell on the battery string to obtain a secondary classification result corresponding to the cold solder joint; Combined with the categories of other defects except cold solder joints, the secondary classification results corresponding to cold solder joints are obtained.

2. The alarm method according to claim 1, characterized in that: If the defect type is hidden crack, the secondary classification result is obtained in one of the following three ways: According to the position of the defect, a hidden crack image is intercepted from the image of the battery string; according to the hidden crack image, a skeleton of the hidden crack is extracted; according to the hidden crack skeleton, or according to the hidden crack skeleton and the position of the constituent elements of the battery cell, a secondary classification result corresponding to the hidden crack is determined; wherein the position of the constituent elements of the battery cell is determined according to the pattern of the battery cell and the image of the battery string; intercepting a hidden crack image from the image of the battery string according to the position of the defect; inputting the hidden crack image into a hidden crack classification model to obtain a secondary classification result corresponding to the hidden crack; The position of the defect is compared with the position of each battery cell on the battery string to obtain a secondary classification result corresponding to the hidden crack.

3. The alarm method according to claim 1, characterized in that: If the defect category is cold solder joint, the secondary classification result includes at least one of the following: Whether the cold solder joint penetrates the battery cell; Whether the cold solder joint is located on the cut surface or chamfered surface of the battery cell; Whether the cold solder joints are located at the upper and lower edges of the battery cell; Whether the cold solder joint is large in area; The sequence number of the battery cell where the cold solder joint is located in the battery string; Whether the cold solder joint is accompanied by bright spots; If the defect type is hidden crack, the secondary classification result includes at least one of the following: Whether the hidden cracks penetrate the battery cell; Whether the hidden crack is located on the cut surface or chamfered surface of the battery cell; Whether the hidden crack is located at the cutting corner of the battery cell; Whether the hidden cracks are located at the upper and lower edges of the battery cell; Whether the shape of the crack is straight, reticular, x-shaped or v-shaped; Whether the hidden crack is located on the main grid line of the battery cell; The ordinal number of the battery cell where the hidden crack is located in the battery string; Whether the hidden crack is a blank hidden crack.

4. The alarm method according to claim 2, characterized in that: The step of extracting a hidden crack skeleton according to the hidden crack image comprises: Performing a binarization operation on the hidden crack image to obtain a binarized image; Determine the main grid line position of the battery cell in the binary image according to the pattern of the battery cell and the image of the battery string; For the binary image, within a preset range near the position of the main grid line, a convolution operation is performed using a convolution kernel constructed based on the size of the main grid line to filter out the main grid line in the binary image to obtain a filtered image; Determine a pixel region formed by pixels of a first value in the filtered image as at least one defect region; wherein the first value is a pixel value representing a defect; A target region is determined according to the at least one defect region, and a skeleton extracted from the target region is used as the skeleton of the hidden crack.

5. The alarm method according to claim 1, characterized in that: The first data to be analyzed also includes: The frequency and / or density of defects occurring within the first preset time window.

6. The alarm method according to claim 1, characterized in that: The first data to be analyzed also includes: The correlation between the defect and the equipment component is calculated according to the location of the defect and the location of the equipment component in the stringer.

7. The alarm method according to claim 1, characterized in that: The first original data also includes at least one of the following: process data of the stringer, operating status data of the stringer, and incoming material data during production of the stringer.

8. The alarm method according to claim 1, characterized in that: The alarm rule includes at least one of the following: If the defect category is cold solder joint, the secondary classification result is welding deviation, the defect is located on the back of the battery string, and the number of the battery strings with the defect within the first preset time window is not less than the first number, then the string welding machine has a fault that the transfer platform is not level; If the defect category is hidden crack, the secondary classification result is x-type hidden crack, the defect is located in the battery cell at the first position in the battery string, and the number of battery strings with the above defect within the first preset time window is not less than the second number, then the stringer has the problem of pin tin accumulation; If the defect category is cold soldering, the number of the defect occurring on the battery string is greater than a first threshold, and the number of the battery strings having the defect within a first preset time window is not less than a third number, then by comparing the flux information used in different production orders of the stringer and the corresponding production quality, it is determined whether there is a problem with the flux currently used by the stringer; If the defect category is hidden crack, the secondary classification result is main grid hidden crack, the correlation between the defect and the pressure pin of the stringer is greater than the second threshold, and the number of the battery strings with the above defect within the first preset time window is not less than the fourth number, then the stringer has a problem of pressure pin jamming; If the category of the defect is a cold solder joint, the secondary classification result is a large-area cold solder joint, the power of the lamp tube of the string welding machine corresponding to the position of the defect during welding is less than a third threshold, and the number of the battery strings with the above defect within the first preset time window is not less than a fifth number, then by comparing the production quality of the lamp tube at present and when it is configured with the previous power parameter, it is determined whether the lamp tube has a problem of too low power; If the defect category is cold soldering, the defect is located on the front of the battery string, the height of the light box of the stringer corresponding to the location of the defect is greater than the fourth threshold, and the number of battery strings with the above defect within the first preset time window is not less than the sixth number, then the stringer has a problem of too high light box height.

9. The alarm method according to claim 1, characterized in that: The method further comprises: Receive first user feedback on the first alarm information, and iteratively optimize the alarm rules in the rule base according to the first user feedback; wherein the first user feedback includes feedback on the authenticity of the first alarm information.

10. The alarm method according to any one of claims 1 to 9, characterized in that: The method further comprises: During the operation of the stringer, second original data is acquired, where the second original data includes process data of the stringer and / or operation status data of the stringer; Obtaining second data to be analyzed according to the second original data; Performing anomaly detection on the second data to be analyzed using a first anomaly detection model to obtain an anomaly detection result; wherein the first anomaly detection model is a model obtained by unsupervised training; If the abnormality detection result indicates that the second data to be analyzed is abnormal, a second alarm message is output based on the abnormality detection result; wherein the second alarm message includes at least one of the following: data abnormality information, fault information of the stringer associated with the data abnormality, and maintenance information of the stringer associated with the data abnormality.

11. The alarm method according to claim 10, characterized in that: The method further comprises: receiving second user feedback for the second alarm information, and constructing a labeled training sample according to the second user feedback and the second data to be analyzed; wherein the second user feedback includes feedback on the authenticity of the second alarm information; When the number of the collected training samples meets the training requirements, supervised training is performed using the training samples to obtain a second anomaly detection model; Acquire third original data, where the third original data is subsequent data of the second original data; Obtaining third data to be analyzed according to the third original data; Using the second anomaly detection model, anomaly detection is performed on the third data to be analyzed to obtain an anomaly detection result; or, combining the second anomaly detection model and the first anomaly detection model, anomaly detection is performed on the third data to be analyzed to obtain an anomaly detection result.

12. The alarm method according to claim 11, characterized in that: The combining the second anomaly detection model and the first anomaly detection model to perform anomaly detection on the third data to be analyzed to obtain an anomaly detection result includes: Using the second anomaly detection model to perform anomaly detection on the third data to be analyzed to obtain a preliminary anomaly detection result; if the preliminary anomaly detection result indicates that the third data to be analyzed does not have an anomaly, using the first anomaly detection model to perform anomaly detection on the third data to be analyzed to obtain an anomaly detection result; otherwise, directly determining the preliminary anomaly detection result as the anomaly detection result; or, Using the first anomaly detection model to perform anomaly detection on the third data to be analyzed to obtain a first anomaly detection result; and using the second anomaly detection model to perform anomaly detection on the third data to be analyzed to obtain a second anomaly detection result; obtaining an anomaly detection result based on the first anomaly detection result and the second anomaly detection result.

13. A computer program product, characterized in that The method comprises computer program instructions, and when the computer program instructions are read and executed by a processor, the method according to any one of claims 1 to 12 is executed.

14. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer program instructions, and when the computer program instructions are read and executed by a processor, the method according to any one of claims 1 to 12 is executed.

Images