Welding detection method and device, equipment, storage medium

Abstract

The embodiment of the application discloses a welding detection method and device, equipment and a storage medium, comprising: obtaining a target data set, which is welding data of a battery obtained by a welding device; performing feature extraction processing on each target data set to obtain a target feature set corresponding to each target data set; determining a first identification model according to the data quantity of the target data set, and performing identification processing on the target feature set corresponding to the target data set through the first identification model to obtain a first identification result corresponding to the target data set, wherein the first identification result includes normal welding or abnormal welding, the first identification model is a statistical model or an abnormality detection model, and the abnormality detection model is obtained by combining a deep learning model and a correction model. The welding quality can be effectively monitored, and the welding efficiency is improved.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to information processing technology, and relate to but are not limited to a welding detection method and device, equipment, and storage medium. BACKGROUND

[0002] The tab welding of a lithium battery cell winding device is mostly ultrasonic welding. The welding quality has an important influence on the quality of the cell product. Since the ultrasonic welding process is affected by various factors, such as welding pressure, amplitude, time, and welding surface state. If these parameters cannot be accurately controlled, it may lead to unstable welding quality. Uneven welding or insufficient strength may cause safety hazards in the use of the battery.

[0003] Therefore, how to effectively monitor the welding quality and thus improve the welding efficiency is a problem to be solved. SUMMARY

[0004] Therefore, the welding detection method and device, equipment, and storage medium provided by the embodiments of the present application can effectively monitor the welding quality and improve the welding efficiency. The welding detection method and device, equipment, and storage medium provided by the embodiments of the present application are implemented as follows:

[0005] In a first aspect, the embodiments of the present application provide a welding detection method, comprising:

[0006] obtaining a target data set, the target data set being welding data of a welding device on a battery obtained, the target data set including at least one;

[0007] performing feature extraction processing on each target data set to obtain a target feature set corresponding to each target data set;

[0008] determining a first identification model according to the data volume of the target data set, and performing identification processing on the target feature set corresponding to the target data set through the first identification model to obtain a first identification result corresponding to the target data set, the first identification result including normal welding or abnormal welding, the first identification model being a statistical model or an anomaly detection model, and the anomaly detection model being obtained by combining a deep learning model and a correction model.

[0009] In some embodiments, the identification processing on the target feature set corresponding to the target data set through the first identification model to obtain the first identification result corresponding to the target data set includes:

[0010] in a case where the data volume of the target data set is less than a data volume threshold, performing identification processing on the target feature set through the statistical model to obtain the first identification result corresponding to the target data set;

[0011] In a case where the data quantity of the target data set is greater than or equal to the data quantity threshold, the target feature set is subjected to identification processing by the anomaly detection model to obtain a first identification result corresponding to the target data set.

[0012] In some embodiments, after the identification result corresponding to the target data set is obtained, the method further comprises:

[0013] performing data stability analysis on the target data set to obtain a stability analysis result;

[0014] performing identification processing on the stability analysis result and the first identification result corresponding to each target data set by a second identification model to obtain a second identification result, the second identification result including whether to replace the welding equipment and / or welding material, the second identification model being a deep learning model.

[0015] In some embodiments, after the second identification result is obtained, the method further comprises:

[0016] in a case where the second identification result indicates that the welding equipment and / or welding material is to be replaced, updating the first identification model to obtain an updated first identification model, and performing identification processing on a next target data set by the updated first identification model.

[0017] In some embodiments, after the first identification result corresponding to the target data set is obtained, the method further comprises:

[0018] in a case where the first identification result is a welding anomaly, determining a cause of the welding anomaly by obtaining production data of the battery during a welding process of the welding equipment, the cause of the welding anomaly including one or more of a welding equipment parameter anomaly, a welding device anomaly, or a human parameter adjustment anomaly.

[0019] In some embodiments, the welding equipment includes a welding machine, the production data includes a setting parameter of the welding equipment and a welding parameter of the welding machine, and the determining of the cause of the welding anomaly by obtaining the production data of the battery during the welding process of the welding equipment includes:

[0020] performing identification processing on the setting parameter by a first sub-identification model to obtain a first sub-identification result, the first sub-identification result including whether the setting parameter of the welding equipment is normal or abnormal, the first sub-identification model being a statistical model;

[0021] The welding parameters of the welding machine are identified by a second sub-identification model to obtain a second sub-identification result, the second sub-identification result including normal welding parameters of the welding machine or abnormal welding parameters of the welding machine, the second sub-identification model being an anomaly detection model, and the anomaly detection model being obtained by combining a deep learning model and a correction model.

[0022] The first sub-identification result and the second sub-identification result are combined to determine whether the welding equipment parameters are abnormal.

[0023] In some embodiments, the welding equipment includes a pressing unit and a welding unit, the production data includes sensor data collected during operation of the pressing unit and welding parameters of the welding unit, and determining the cause of the welding abnormality based on the obtained production data of the welding equipment during welding of the battery includes:

[0024] The sensor data is identified by a PHM management system to determine whether the operation of the pressing unit is abnormal.

[0025] The welding parameters of the welding unit are identified by a third identification model to determine whether the operation of the welding unit is abnormal, the third identification model being an anomaly detection model, and the anomaly detection model being obtained by combining a deep learning model and a correction model.

[0026] In some embodiments, the production data includes debugging data, a first set of debugging parameters before debugging of the welding equipment, and a second set of debugging parameters after debugging of the welding equipment, and determining the cause of the welding abnormality based on the obtained production data of the welding equipment during welding of the battery includes:

[0027] The debugging data is identified by an experience model to obtain an intermediate identification result, the intermediate identification result being used to indicate whether the debugging process is abnormal, and the experience model being obtained by training a deep learning model based on historical welding data and historical maintenance data in a historical welding process.

[0028] The first set of debugging parameters and the second set of debugging parameters are subjected to feature extraction processing to obtain first parameter features and second parameter features.

[0029] The first parameter features and the second parameter features are identified by a fourth identification model to obtain a third identification result, the third identification result being used to indicate whether the debugging parameters are abnormal.

[0030] The intermediate identification result and the third identification result are combined to determine whether there is a human parameter adjustment abnormality.

[0031] In a second aspect, the embodiments of the present application provide a welding detection device, comprising:

[0032] An acquisition module is configured to acquire a target data set, the target data set being welding data of a welding device on a battery, and the target data set comprising at least one of:

[0033] An extraction module is configured to perform feature extraction processing on each target data set to obtain a target feature set corresponding to each target data set.

[0034] An identification module is configured to determine a first identification model according to a data amount of the target data set, and perform identification processing on the target feature set corresponding to the target data set by using the first identification model to obtain a first identification result corresponding to the target data set, the first identification result comprising welding normality or welding abnormality, the first identification model being a statistical model or an abnormality detection model, and the abnormality detection model being obtained by combining a deep learning model and a correction model.

[0035] In a third aspect, the embodiments of the present application provide a computer device, comprising a memory and a processor, the memory storing a computer program capable of running on the processor, and the processor implements the method provided by the embodiments of the present application when executing the program.

[0036] In a fourth aspect, the embodiments of the present application provide a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the method provided by the embodiments of the present application.

[0037] The welding detection method, device, computer device and computer readable storage medium provided by the embodiments of the present application can acquire a target data set, the target data set being welding data of a welding device on a battery; perform feature extraction processing on each target data set to obtain a target feature set corresponding to each target data set; determine a first identification model according to a data amount of the target data set, and perform identification processing on the target feature set corresponding to the target data set by using the first identification model to obtain a first identification result corresponding to the target data set, the first identification result comprising welding normality or welding abnormality, the first identification model being a statistical model or an abnormality detection model, and the abnormality detection model being obtained by combining a deep learning model and a correction model. In this way, the welding quality of the battery can be effectively monitored, thereby improving the welding efficiency and solving the technical problems proposed in the background art. BRIEF DESCRIPTION OF DRAWINGS

[0038] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and serve to explain the technical solutions of the present application together with the specification.

[0039] Figure 1An application scenario of a welding detection method provided by the embodiment of the present application is shown in the figure.

[0040] Figure 2 An implementation flowchart of a welding detection method provided by the embodiment of the present application is shown in the figure.

[0041] Figure 3 An implementation flowchart of another welding detection method provided by the embodiment of the present application is shown in the figure.

[0042] Figure 4 An implementation flowchart of determining a cause of welding abnormality provided by the embodiment of the present application is shown in the figure.

[0043] Figure 5 An implementation flowchart of another determining a cause of welding abnormality provided by the embodiment of the present application is shown in the figure.

[0044] Figure 6 An implementation flowchart of still another determining a cause of welding abnormality provided by the embodiment of the present application is shown in the figure.

[0045] Figure 7 An implementation flowchart of cause analysis of welding result abnormality provided by the embodiment of the present application is shown in the figure.

[0046] Figure 8 A structure diagram of a welding detection device provided by the embodiment of the present application is shown in the figure.

[0047] Figure 9 A structure diagram of a computer device provided by the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the specific technical scheme of the present application will be further described in detail below with reference to the drawings in the embodiments of the present application. The following embodiments are used to illustrate the present application, but not to limit the scope of the present application.

[0049] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs. The terms used herein are only for the purpose of describing the embodiments of the present application, and are not intended to limit the present application.

[0050] In the following description, "some embodiments" are described, which describe a subset of all possible embodiments, but it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict.

[0051] It should be noted that the terms "first", "second", "third" in the embodiments of the present application are used to distinguish similar or different objects, and do not represent a specific order of the objects. Understandably, "first", "second", "third" can be interchanged in a specific order or sequence as allowed, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0052] In the middle stage of lithium battery production, regardless of the form of the battery, the related workstations of the tab will involve the process of ultrasonic welding. Although the current ultrasonic welding process is relatively mature, ultrasonic welding still has many shortcomings.

[0053] On the one hand, the ultrasonic welding process is affected by many factors, such as welding pressure, amplitude, time, welding surface state, etc. If these parameters are not accurately controlled, it may lead to unstable welding quality. Uneven welding or insufficient strength may cause safety hazards in the use of the battery.

[0054] On the other hand, ultrasonic welding involves many process parameters, such as welding time, amplitude, pressure, etc. These parameters need to be accurately adjusted according to the specific welding materials and requirements, and the process debugging and optimization process is relatively complex. After the material is replaced or the consumable is replaced, the on-site engineer needs to spend a lot of time to debug and test samples to ensure that the welding requirements meet the effect.

[0055] On the other hand, some small defects in the ultrasonic welding process, such as cracks, pores, and virtual welding, are difficult to detect by conventional detection methods. These hidden defects may cause potential safety risks during battery use, so ensuring the reliability of welding quality is an important challenge.

[0056] In addition, ultrasonic welding equipment is relatively expensive, and in addition, the welding equipment needs to be regularly maintained to ensure its normal operation, which increases the production cost and maintenance cost, and the complexity of the equipment requires the operator to have high professional knowledge and skills.

[0057] As can be seen, in the current various ultrasonic welding scenarios, there is no effective means for detecting welding quality, so the welding quality cannot be effectively monitored; the ultrasonic welding equipment can only be maintained regularly by manual, and there is no effective means to identify whether the equipment has problems that cause the welding to not meet the requirements; after the welding problem or equipment problem is found, a lot of manpower and material resources are needed to troubleshoot the problem, and after the problem is solved, the solution needs to be verified through multiple rounds of trial and error before it can be finally confirmed.

[0058] Therefore, the embodiments of the present application provide a welding detection method, which can effectively monitor the welding quality of the battery, thereby improving the welding efficiency.

[0059] Figure 1 An application scenario diagram of a welding detection method provided by an embodiment of the present application is given. As shown in the figure, Figure 1 the method is applied to a welding system. The welding system includes a welding device, a data acquisition module, a control module, a storage module, and an anomaly detection module.

[0060] The welding device can include an ultrasonic welding unit and a welding down unit. The ultrasonic welding unit can include a controller, a generator, a transducer, an amplitude horn, etc., and the welding unit is a set of device units for ultrasonic welding; the welding down unit can include a servo motor or a pneumatic cylinder, and the down welding head can be used to provide pressure during the welding process.

[0061] The data acquisition module can include data acquisition software and HMI. The data acquisition software is used to acquire various welding parameter values during the welding process; and the HMI is a unit for the user to set relevant parameters for the ultrasonic welding unit.

[0062] The control module can be a programmable logic controller (PLC), which is a controller unit for controlling the normal operation of the welding device, and can also provide relevant data during the welding process.

[0063] The storage module is used to store process data of various parameters collected from different data sources during the welding process.

[0064] The anomaly detection module is used to detect abnormal problems occurring during the welding process.

[0065] Figure 2 An implementation flow diagram of the welding detection method provided by the embodiment of the present application is given. As shown in the figure, Figure 2 the method can include the following steps 201 to 203:

[0066] Step 201, obtaining a target data set, the target data set being welding data of the welding device on the battery, and the target data set including at least one.

[0067] In the embodiment of the present application, the target data set can be welding data collected by the welding device during the welding process of the battery. In this way, the welding data can include a plurality of welding times, welding parameters corresponding to each welding time, etc.

[0068] In the embodiments of the present application, the number of target data sets obtained is not limited, for example, only one target data set can be obtained, and the welding quality is identified by using the target data set to obtain the welding result; or, two or more target data sets in the welding process can be obtained, and the final welding quality identification result is obtained by using the welding quality identification results of the plurality of target data sets.

[0069] In step 202, feature extraction processing is performed on each target data set to obtain a target feature set corresponding to each target data set.

[0070] In the embodiments of the present application, the implementation manner of the feature extraction processing is not limited. The feature extraction processing can be learning the representation of the features from the welding data of the target data set.

[0071] For example, the principal component analysis method can be used to map the welding data in the original target data set to a low-dimensional space through linear transformation while retaining the main features of the data as much as possible, thereby obtaining the target feature set corresponding to each target data set.

[0072] In some embodiments, before the feature extraction processing is performed on each target data set, each target data set can also be preprocessed first, and the preprocessing can include one or more of data cleaning, normalization, and standardization. In this way, the feature extraction processing can be performed based on the preprocessed target data set, so that the processing efficiency is higher.

[0073] In other embodiments, after obtaining the preprocessed target data set, the preprocessed target data set can also be subjected to feature selection to select the most useful features for target identification from all possible features, and then the selected features are subjected to feature extraction processing, so that the processing efficiency is higher.

[0074] In step 203, a first identification model is determined according to the data amount of the target data set, and the target feature set corresponding to the target data set is subjected to identification processing by using the first identification model to obtain a first identification result corresponding to the target data set, the first identification result including welding normal or welding abnormal, the first identification model being a statistical model or an anomaly detection model, and the anomaly detection model being obtained by combining a deep learning model and a correction model.

[0075] In the embodiments of the present application, different first identification models can be selected to identify the target data set according to the different data amounts of the target data set to obtain the corresponding first identification result.

[0076] In the embodiments of the present application, the specific structure of the statistical model is not limited. The statistical model is a model established based on probability theory and using mathematical statistical methods. For example, the statistical model can be a regression model, such as linear regression, multiple linear regression, time series regression, etc.; or the statistical model can also be a classification model, such as the classification model can include logistic regression, support vector machine (SVM), decision tree, K-nearest neighbor, etc., which is used to classify data into different categories; or the statistical model can also be a clustering model, such as the clustering model can be K-Means, Gaussian mixture model (GMM), hierarchical clustering, etc., which is used to group similar objects in the data set; or the statistical model can also be a dimensionality reduction model, such as the dimensionality reduction model can be a principal component analysis (PCA) model, etc.

[0077] Through the statistical model, the target data set with a small amount of data can be subjected to anomaly detection processing to obtain a first identification result corresponding to the target data set, and the first identification result includes welding normal or welding abnormal. The welding normal means that the welding quality of the tab of the battery is normal, i.e., no empty welding or false welding, etc. The welding abnormal means that the welding result of the tab of the battery has empty welding or false welding, etc.

[0078] In the embodiments of the present application, the specific structure of the anomaly detection model is not limited, which can be a detection model obtained by combining a deep learning model and a correction model.

[0079] Among them, the type of deep learning model is not limited, such as the deep learning model can be a support vector machine (SVM), which can be used for classification tasks, and the model can identify abnormal points by maximizing the interval between different categories. Or the deep learning model can also be an unsupervised learning algorithm, such as Isolation Forest, etc., which is suitable for data sets without labels, and judges the anomaly by the density or isolation degree of data points. Although the deep learning model has high efficiency and accuracy, there may still be false positives or false negatives in some complex or specific scenarios, so a correction algorithm can also be combined to further correct the identification result.

[0080] Among them, the type of correction algorithm is also not limited, such as the correction algorithm can be an empirical formula method, such as an empirical formula obtained by experimental data and experience is used to calculate the correction value, and the deep learning model is corrected based on the correction value. Or the correction algorithm can also be a computer simulation method, such as using computer software to simulate the measurement process, and calculating the correction value according to the simulation result, and correcting the deep learning model based on the correction value. Or the correction algorithm can also be a least squares method, which finds the best fitting curve by minimizing the sum of squared residuals between the predicted value and the actual value, and corrects the deep learning model through the best fitting curve.

[0081] The target data set with a large amount of data can be subjected to the abnormality detection processing by the abnormality detection model to obtain a first identification result corresponding to the target data set, and the first identification result includes welding normal or welding abnormal. The welding normal means that the welding quality of the tab of the battery is normal, that is, no empty welding or false welding occurs. The welding abnormal means that the welding result of the tab of the battery has empty welding or false welding.

[0082] As in an exemplary embodiment, the abnormality detection model can include a combined model of a support vector machine model and a correction model. In this way, after the target data set collected in the welding process of the battery is obtained, the target data set can be subjected to feature extraction processing to obtain a target feature set corresponding to the target data set; the target feature set is input into the support vector machine model to obtain a preliminary identification result; and the preliminary identification result is subjected to correction processing by the correction model to obtain a final first identification result.

[0083] It should be noted that the number of first identification results obtained is different based on the number of target data sets. For example, in the case of a single target data set, the first identification result obtained is the final identification result of the welding quality of the welding equipment; in the case of two or more target data sets, the final identification result of the welding quality of the welding equipment can be obtained by comprehensive analysis of the first identification result corresponding to each target data set. Here, the implementation manner of the comprehensive analysis is not limited, such as selecting the identification result with the highest occurrence rate from the plurality of first identification results as the final identification result, or performing weighted processing on each first identification result to obtain the final identification result, and the like.

[0084] In the embodiments of the present application, by obtaining the target data set, the target data set is the welding data of the battery obtained by the welding equipment; each target data set is subjected to feature extraction processing to obtain a target feature set corresponding to each target data set; a first identification model is determined according to the data amount of the target data set, and the target feature set corresponding to the target data set is subjected to identification processing by the first identification model to obtain a first identification result corresponding to the target data set, the first identification result includes welding normal or welding abnormal, the first identification model is a statistical model or an abnormality detection model, and the abnormality detection model is obtained by combining a deep learning model and a correction model. In this way, the welding quality of the battery can be effectively monitored to improve the welding efficiency.

[0085] Figure 3 An implementation flowchart of the welding detection method provided in the embodiments of the present application is shown in FIG. 3. Figure 3 As shown in FIG. 3, the method can include the following steps 301 to 306:

[0086] Step 301, obtaining a target data set, the target data set being welding data of a welding device on a battery obtained, the target data set including at least one.

[0087] Here, the manner of implementing step 301 is the same as that of implementing step 201 in the above embodiment, and will not be repeated here.

[0088] Step 302, performing feature extraction processing on each target data set to obtain a target feature set corresponding to each target data set.

[0089] Here, the manner of implementing step 302 is the same as that of implementing step 202 in the above embodiment, and will not be repeated here.

[0090] Step 303, in a case where a data quantity of the target data set is less than a data quantity threshold, performing identification processing on the target feature set by a statistical model to obtain a first identification result corresponding to the target data set; in a case where the data quantity of the target data set is greater than or equal to the data quantity threshold, performing identification processing on the target feature set by an anomaly detection model to obtain the first identification result corresponding to the target data set, the anomaly detection model being obtained by combining a deep learning model and a correction model.

[0091] In the embodiment of the present application, the type of the identification model selected is different based on the different data quantities of the target data set.

[0092] Among them, the statistical model usually has a lower calculation complexity and is suitable for fast running in a case where resources are limited (such as computing resources, time, etc.); and a small data set is prone to cause an overfitting problem, that is, a model performs well on training data but performs poorly on new data, and the statistical model can often avoid this problem due to its simplicity. Therefore, in a case where the data quantity of the target data set is less than the data quantity threshold, the statistical model can be selected as the identification model.

[0093] And the deep learning model can process complex nonlinear relationships and learn high-dimensional features in data through a multi-layer network structure, which is particularly important when processing large data sets, because large data sets often contain more information and more complex patterns; and when the data quantity is large enough, the deep learning model can capture the internal laws and patterns of data through learning a large number of samples, thereby performing better generalization on unseen data. Therefore, in a case where the data quantity of the target data set is greater than or equal to the data quantity threshold, the anomaly detection model obtained by combining the deep learning model and the correction model can be selected as the identification model.

[0094] In this way, different identification models are selected based on the different data quantities of the target data set, which is more targeted and has higher identification efficiency.

[0095] At step 304, data stability analysis is performed on the target data set to obtain a stability analysis result.

[0096] It can be understood that during the welding process, there can be a change in type and a change in material.

[0097] The change in type refers to a change in the model, fixture, tooling or type and specification of the welding object during the welding process. The change in material refers to a change in welding material, including welding rods, welding wires, welding fluxes, etc.

[0098] And based on the difference in the type of the welding model or the difference in the welding material, there is also a difference in identifying the welding quality. Therefore, it is also necessary to update the identification model according to whether there is a change in type or a change in material during welding, so that the updated identification model is adapted to the type of the changed welding model or the welding material.

[0099] The data stability can be an important indicator for detecting whether a change in type or a change in material occurs. When a change in type or a change in material occurs during the welding process, the process parameters are often adjusted and the equipment state is changed, and these changes will directly affect the stability of the welding data.

[0100] Data stability is an indicator for measuring data volatility and dispersion. The smaller the data volatility and the smaller the dispersion, the higher the stability. If the welding data maintains a high stability during the welding process, it means that the welding process parameters and the equipment state are under good control, and the welding process is stable and reliable. On the contrary, if the welding data shows obvious fluctuations or dispersion, it may mean that there are abnormal situations in the welding process, such as a change in type, a change in material, equipment failure or improper adjustment of process parameters, etc.

[0101] Therefore, data stability analysis can also be performed on the target data set to obtain a stability analysis result corresponding to the target data set, and the stability analysis result is used to preliminarily judge whether there is a change in type and / or a change in material in the welding process.

[0102] At step 305, the stability analysis result corresponding to each target data set and the first identification result are identified by a second identification model to obtain a second identification result, the second identification result including whether the welding equipment and / or the welding material is changed, and the second identification model is a deep learning model.

[0103] In the embodiments of the present application, the specific type of the second identification model is not limited.

[0104] The second recognition model can be a convolutional neural network (CNN), a recurrent neural network (RNN), or a deep belief network.

[0105] After obtaining the stability analysis result corresponding to each target data set, the at least one stability analysis result and the at least one first recognition result are input into a second recognition model, and whether the welding equipment or the welding material is replaced during the welding process is determined by the second recognition model.

[0106] In a case where the second recognition result indicates that the welding equipment and / or the welding material is replaced, the first recognition model is updated to obtain an updated first recognition model, and the next target data set is recognized by the updated first recognition model.

[0107] In the embodiments of the present application, if it is determined according to the second recognition result that the welding equipment and / or the welding material is replaced during the welding process, the first recognition model can be updated, that is, the recognition for identifying the welding quality is updated, so that the updated first recognition model is consistent with the welding after the welding equipment and / or the welding material is replaced.

[0108] In the embodiments of the present application, if it is determined according to the second recognition result that the welding equipment and / or the welding material is replaced during the welding process, the first recognition model can be updated, that is, the recognition for identifying the welding quality is updated, so that the updated first recognition model is consistent with the welding after the welding equipment and / or the welding material is replaced.

[0109] In the embodiments of the present application, by obtaining the target data set, the target data set is the welding data of the battery obtained by the welding equipment; the feature extraction processing is performed on each target data set to obtain a target feature set corresponding to each target data set; the first recognition model is determined according to the data amount of the target data set, and the target feature set corresponding to the target data set is recognized by the first recognition model to determine whether the welding result is normal, which can effectively monitor the welding quality of the battery, thereby improving the welding efficiency.

[0110] Further, the data stability analysis is performed on the target data set to obtain a stability analysis result; the second identification model is used to identify the stability analysis result and the first identification result corresponding to each target data set to determine whether the welding equipment and / or the welding material are replaced during the welding process, and the first identification model is updated to obtain an updated first identification model when the welding equipment and / or the welding material are replaced, and the next target data set is identified by using the updated first identification model. In this way, the quality monitoring demand of different welding equipment or welding material can be met, the monitoring range is wider, and the monitoring efficiency is higher.

[0111] It can be understood that the ultrasonic welding equipment can only be maintained periodically by manual operation at present, and there is no effective means to identify whether the equipment has problems leading to unsatisfactory welding. After the welding problem or equipment problem is found at present, a large amount of manpower and material resources are needed to troubleshoot the problem, and after the problem is solved, the solution needs to be verified through multiple rounds of trial and error before it can be finally confirmed.

[0112] Based on this, in some embodiments, in the case that the first identification result is welding abnormality obtained by implementing the above-mentioned embodiments, the production data of the battery in the welding process of the welding equipment can be used to determine the cause of the welding abnormality, so as to give an early warning of the problem of the ultrasonic welding equipment and perform maintenance on the equipment in advance, and locate the cause of the welding quality problem, and give reasonable suggestions according to the problem cause.

[0113] The cause of the welding abnormality includes one or more of welding equipment parameter abnormality, welding device abnormality, or human parameter adjustment abnormality.

[0114] Here, the process of determining the cause of the welding abnormality is different based on the different causes of the welding abnormality.

[0115] For example, in some embodiments, the welding equipment includes a welding machine, and the production data includes the setting parameters of the welding equipment and the welding parameters of the welding machine. In this way, when determining the cause of the welding abnormality, the following steps 401 to 403 can be performed:

[0116] Step 401: The first sub-identification model is used to identify the setting parameters to obtain a first sub-identification result, the first sub-identification result includes normal setting parameters of the welding equipment or abnormal setting parameters of the welding equipment, and the first sub-identification model is a statistical model.

[0117] In the embodiments of the present application, the setting parameters of the welding equipment are not limited, for example, the setting parameters can include one or more of welding current, welding voltage, welding speed, welding angle, welding time, and preheating time.

[0118] After the setting parameters of the welding equipment are acquired, the setting parameters are identified by the first sub-identification model to determine whether the setting parameters of the welding equipment are set normally.

[0119] In the embodiments of the present application, the specific structure of the first sub-identification model is not limited. For example, the first sub-identification model can be a regression model, such as linear regression, multiple linear regression, time series regression, etc.; or the first sub-identification model can also be a classification model, such as a classification model that can include logistic regression, support vector machine (SVM), decision tree, K-nearest neighbor, etc., for classifying data into different categories; or the first sub-identification model can also be a clustering model, such as a clustering model that can be K-Means, Gaussian Mixture Model (GMM), hierarchical clustering, etc., for grouping similar objects in the data set; or the first sub-identification model can also be a dimensionality reduction model, such as a principal component analysis (PCA) model, etc.

[0120] At step 402, the welding parameters of the welding machine are identified by the second sub-identification model to obtain a second sub-identification result, the second sub-identification result including whether the welding parameters of the welding machine are normal or abnormal, and the second sub-identification model is an anomaly detection model obtained by combining a deep learning model and a correction model.

[0121] Similarly, in the embodiments of the present application, the welding parameters of the welding machine are not limited, such as the welding parameters of the welding machine that can include one or more of power source types, such as alternating current power, direct current power, etc., rated power, load duration rate, control system, including manual control or automatic control, cooling system, etc.

[0122] After the welding parameters of the welding machine are acquired, the welding parameters of the welding machine are identified by the second sub-identification model to determine whether the welding parameters of the welding machine are set normally.

[0123] In the embodiments of the present application, the specific structure of the anomaly detection model is not limited. For example, the deep learning model in the anomaly detection model can be a support vector machine (SVM) or an isolation forest model, etc.

[0124] And the type of the correction algorithm is not limited, such as the correction algorithm can be any one of an empirical formula method, a computer simulation method, a least squares method, etc.

[0125] At step 403, the first sub-identification result and the second sub-identification result are combined to determine whether the welding equipment parameters are abnormal.

[0126] In the embodiment of the present application, after it is determined whether the setting parameters of the welding equipment are set abnormally and whether the welding parameters of the welding machine are set abnormally, whether the welding equipment parameters are abnormal can be determined according to the two identification results.

[0127] By implementing the embodiment, the cause of the welding quality problem can be located, and the welding problem can be quickly solved.

[0128] In other embodiments, the welding equipment includes a pressing unit and a welding unit, and the production data includes sensor data collected in the working process of the pressing unit and welding parameters of the welding unit. In this way, when determining the cause of the welding abnormality, the following steps 501 to 502 can be performed:

[0129] Step 501: The PHM management system identifies and processes the sensor data to determine whether the pressing unit works abnormally.

[0130] In the embodiment of the present application, the welding equipment can include a pressing unit and a welding unit. During the welding process, the pressing unit may work abnormally or the welding unit may work abnormally. Therefore, it is necessary to monitor whether the devices of the welding equipment are abnormal to discover the abnormal cause and abnormal position in time and help quickly troubleshoot the problem.

[0131] Here, the prognostics and health management (PHM) is a system capable of monitoring, predicting and managing the health status of equipment. It obtains sensor data such as vibration, temperature, pressure and other key parameters in the running process of the equipment in real time, analyzes and processes the sensor data to identify whether the pressing unit works abnormally, and further locates the fault source when the pressing unit works abnormally.

[0132] Step 502: The welding parameters of the welding unit are identified and processed by a third identification model to determine whether the welding unit works abnormally. The third identification model is an anomaly detection model, which is obtained by combining a deep learning model and a correction model.

[0133] Here, the welding parameters of the welding unit can be related setting parameters of the welding unit collected during the welding process. The anomaly detection model can be used to identify and process the welding parameters of the welding unit to determine whether the welding unit works abnormally.

[0134] In the embodiments of the present application, the specific structure of the anomaly detection model is not limited. For example, the deep learning model in the anomaly detection model can be a support vector machine (SVM), an isolation forest model, or the like.

[0135] Moreover, the type of the correction algorithm is not limited. For example, the correction algorithm can be any one of an empirical formula method, a computer simulation method, a least square method, or the like.

[0136] By implementing the embodiments, the cause of the welding quality problem can be located, and the welding problem can be quickly solved.

[0137] In some other embodiments, the production data includes debugging data, a first set of debugging parameters before the welding equipment is debugged, and a second set of debugging parameters after the welding equipment is debugged. In this way, when determining the cause of the welding anomaly, the following steps 601 to 604 can be performed:

[0138] Step 601: performing identification processing on the debugging data by using an empirical model to obtain an intermediate identification result, the intermediate identification result being used to indicate whether an anomaly occurs in the debugging process, and the empirical model being obtained by training a deep learning model according to historical welding data and historical maintenance data in a historical welding process.

[0139] It can be understood that the cause of the welding quality anomaly can also be caused by an error in the parameter adjustment of the welding equipment by the debugging personnel, that is, the welding quality anomaly can be caused by a human parameter adjustment anomaly. Based on this, in the embodiments of the present application, whether a human parameter adjustment anomaly exists in the welding process can also be detected to quickly locate the cause of the anomaly.

[0140] Here, the deep learning model can be trained according to the historical welding data and the historical maintenance data in the historical welding process in advance to obtain the empirical model. The historical maintenance data can be data obtained in the welding process and in the debugging stage, which is used as a feature input value in the deep learning model for training, so that the empirical model can be obtained.

[0141] In this way, the debugging data obtained in the current welding process can be input into the pre-trained empirical model to obtain the intermediate identification result used to indicate whether an anomaly occurs in the debugging process.

[0142] Step 602: performing feature extraction processing on the first set of debugging parameters and the second set of debugging parameters to obtain first parameter features and second parameter features.

[0143] Here, the content of the set of debugging parameters is not limited. For example, the set of debugging parameters can include welding current, welding voltage, welding speed, welding angle, laser power, spot size, focal length, and gas flow, or the like.

[0144] Thus, after obtaining the first set of debugging parameters before debugging and the second set of debugging parameters after debugging of the welding equipment, feature extraction processing can be performed on the first set of debugging parameters and the second set of debugging parameters to obtain the first parameter features and the second parameter features.

[0145] Step 603: The first parameter feature and the second parameter feature are identified and processed by the fourth identification model to obtain the third identification result. The third identification result is used to indicate whether the debugging parameters are abnormal. The fourth identification model is a deep learning model.

[0146] After obtaining the first feature and the second parameter feature, a deep learning model can be used to identify and process the first feature and the second parameter feature to determine whether the debugging parameters are abnormal.

[0147] Here, there are no restrictions on the type of deep learning model, such as Support Vector Machine (SVM), Isolation Forest model, etc.

[0148] Step 604: Combine the intermediate identification results and the third identification results to determine whether there is any abnormality in human parameter tuning.

[0149] In the embodiments of this application, the implementation method for determining whether there is a human-induced parameter tuning anomaly by combining the intermediate identification results and the third identification results is not limited. For example, if either the intermediate identification result or the third identification result is abnormal, it can be determined that there is a human-induced parameter tuning anomaly; or, if both the intermediate identification result and the third identification result are abnormal, then it can be determined that there is a human-induced parameter tuning anomaly.

[0150] Implementing this embodiment can pinpoint the cause of welding quality problems and help resolve welding issues quickly.

[0151] In some embodiments, after determining that the welding equipment parameters are abnormal, solutions can be provided, such as parameter adjustment suggestions, and the results can be displayed through electronic devices.

[0152] In this embodiment, problems with ultrasonic welding equipment can be warned in advance and the equipment can be maintained in advance to minimize losses; and under the premise of automatically identifying workpiece quality problems, the cause of welding quality problems can be located and reasonable opinions can be given based on the cause of the problem.

[0153] Figure 7 A flowchart illustrating the cause analysis of abnormal welding results is provided.

[0154] like Figure 7 As shown, if the welding result is determined to be abnormal, the cause of the welding abnormality can be further determined. The cause of the welding abnormality may include one or more of the following: abnormal welding equipment parameters, abnormal welding components, or abnormal manual parameter adjustment.

[0155] Among them, for whether the welding equipment parameter is abnormal, according to the setting parameters of the welding equipment and the key parameters of the welding equipment itself, the judgment of two identification models is carried out respectively, and finally the identification result of the welding equipment itself reason is output. Specifically, the first sub-identification model is to collect the data in the production process according to the device setting parameters as the input quantity, and the abnormal situation is identified by statistical method; the second sub-identification model is to collect some key parameters and their characteristic quantities generated in the welding process of the welding machine as the input quantity, and the second sub-identification model is an abnormal detection model, which is obtained by combining a deep learning model and a correction model. Finally, the identification results of the two models are combined to form the final identification result, and whether the welding equipment parameter is abnormal is determined.

[0156] For whether the welding device is abnormal, the PHM model can be used to identify the abnormality of the production process data collected by various sensors, and determine whether the down unit works abnormally; and the third identification model is to collect the parameters related to the welding unit, extract the characteristic quantities thereof as the input quantity of the model, and determine whether the welding unit works abnormally by means of the third identification model. The third identification model is an abnormal detection model, which is obtained by combining a deep learning model and a correction model.

[0157] For whether the human parameter adjustment is abnormal, an experience model can be trained according to the welding production process data and some maintenance process data maintained by artificial maintenance. Secondly, according to the fourth identification model, the parameter problems in the production process are identified from the welding production process data, and the final identification result is output according to the located parameter problems and the experience model mentioned before.

[0158] Among them, the experience model is obtained by deep learning training with the data features of on-site artificial debugging and debugging stage as input quantity; and the fourth identification model is to identify the key parameters and the change characteristics of the features before and after the parameter adjustment as input quantity.

[0159] In addition, after the welding abnormality reason is located as welding equipment parameter and welding device abnormality, the corresponding problem point maintenance suggestion can be given. When the human parameter adjustment is abnormal, the final parameter suggestion is also given. And the running effect and conclusion of various models can be displayed through electronic equipment, so that users can operate according to the effect and conclusion.

[0160] It should be understood that although each step in the above flowcharts is shown in sequence according to the direction of the arrow, these steps are not necessarily executed in the order indicated by the arrow. Unless otherwise explicitly stated herein, there is no strict order limitation for the execution of these steps, and these steps can be executed in other orders. Moreover, at least part of the steps in the above flowcharts can include multiple sub-steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily sequential, but can be alternately executed with at least part of other steps or sub-steps or stages of other steps.

[0161] Based on the foregoing embodiments, the embodiments of the present application provide a welding detection device, which comprises the modules included and the units included in the modules, and can be realized by a processor; of course, it can also be realized by a specific logic circuit; in the implementation process, the processor can be a central processing unit (CPU), a microprocessor (MPU), a digital signal processor (DSP), or a field programmable gate array (FPGA), etc.

[0162] Figure 8 The structure diagram of the welding detection device provided by the embodiments of the present application is shown in Figure 8 As shown in the figure, the device 800 comprises an acquisition module 801, an extraction module 802, and an identification module 803, wherein:

[0163] The acquisition module 801 is configured to acquire a target data set, wherein the target data set is welding data of a welding device on a battery, and the target data set comprises at least one of:

[0164] The extraction module 802 is configured to perform feature extraction processing on each target data set to obtain a target feature set corresponding to each target data set.

[0165] The identification module 803 is configured to determine a first identification model according to the data amount of the target data set, and perform identification processing on the target feature set corresponding to the target data set by using the first identification model to obtain a first identification result corresponding to the target data set, wherein the first identification result comprises welding normal or welding abnormal, the first identification model is a statistical model or an anomaly detection model, and the anomaly detection model is obtained by combining a deep learning model and a correction model.

[0166] In some embodiments, the identification module 803 is specifically configured to, in a case where the data amount of the target data set is less than a data amount threshold, perform identification processing on the target feature set by using the statistical model to obtain the first identification result corresponding to the target data set.

[0167] In a case where the data quantity of the target data set is greater than or equal to the data quantity threshold, the target feature set is subjected to identification processing by the anomaly detection model, to obtain a first identification result corresponding to the target data set.

[0168] In some embodiments, the apparatus further comprises an analysis module;

[0169] The analysis module is configured to perform data stability analysis on the target data set, to obtain a stability analysis result.

[0170] The identification module 803 is further configured to perform identification processing on the stability analysis result and the first identification result corresponding to each target data set by a second identification model, to obtain a second identification result, the second identification result including whether to replace the welding equipment and / or welding material, the second identification model being a deep learning model.

[0171] In some embodiments, the apparatus further comprises an update module;

[0172] The update module is configured to, in a case where the second identification result indicates that the welding equipment and / or welding material is to be replaced, update the first identification model to obtain an updated first identification model, and perform identification processing on a next target data set by the updated first identification model.

[0173] In some embodiments, the apparatus further comprises a determination module;

[0174] The determination module is configured to, in a case where the first identification result is a welding anomaly, determine a cause of the welding anomaly by the obtained production data of the welding process of the battery by the welding equipment, the cause of the welding anomaly including one or more of a welding equipment parameter anomaly, a welding device anomaly, or a human parameter adjustment anomaly.

[0175] In some embodiments, the welding equipment includes a welding machine, the production data includes a setting parameter of the welding equipment and a welding parameter of the welding machine, and the apparatus further comprises a comprehensive module;

[0176] The identification module 803 is further configured to perform identification processing on the setting parameter by a first sub-identification model, to obtain a first sub-identification result, the first sub-identification result including whether the setting parameter of the welding equipment is normal or abnormal, the first sub-identification model being a statistical model.

[0177] The second sub-identification model is used for identifying the welding parameters of the welding machine, to obtain a second sub-identification result, the second sub-identification result including that the welding parameters of the welding machine are normal or the welding parameters of the welding machine are abnormal, the second sub-identification model being an anomaly detection model, the anomaly detection model being obtained by combining a deep learning model and a correction model;

[0178] The comprehensive module is configured to comprehensively integrate the first sub-identification result and the second sub-identification result, to determine whether the welding equipment parameters are abnormal.

[0179] In some embodiments, the welding equipment includes a pressing unit and a welding unit, and the production data includes sensor data collected during the operation of the pressing unit and welding parameters of the welding unit.

[0180] The identification module 803 is further configured to identify the sensor data by using a PHM management system, to determine whether the operation of the pressing unit is abnormal.

[0181] The third identification model is used for identifying the welding parameters of the welding unit, to determine whether the operation of the welding unit is abnormal, the third identification model being an anomaly detection model, the anomaly detection model being obtained by combining a deep learning model and a correction model.

[0182] In some embodiments, the production data includes debugging data, a first set of debugging parameters before the welding equipment is debugged, and a second set of debugging parameters after the welding equipment is debugged.

[0183] The identification module 803 is further configured to identify the debugging data by using an experience model, to obtain an intermediate identification result, the intermediate identification result being used for indicating whether the debugging process is abnormal, the experience model being obtained by training a deep learning model according to historical welding data and historical maintenance data in a historical welding process.

[0184] The extraction module 802 is further configured to perform feature extraction processing on the first set of debugging parameters and the second set of debugging parameters, to obtain first parameter features and second parameter features.

[0185] The identification module 803 is further configured to identify the first parameter features and the second parameter features by using a fourth identification model, to obtain a third identification result, the third identification result being used for indicating whether the debugging parameters are abnormal, and to comprehensively integrate the intermediate identification result and the third identification result, to determine whether there is an artificial parameter adjustment abnormality.

[0186] The descriptions of the above device embodiments are similar to those of the above method embodiments, and have similar beneficial effects. For technical details not disclosed in the device embodiments of this application, please refer to the descriptions of the method embodiments of this application for understanding.

[0187] It should be noted that, in the embodiments of this application... Figure 8 The module division of the welding inspection device shown is illustrative and represents only one logical functional division; in actual implementation, other division methods may be used. Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, exist as separate physical units, or be integrated into one unit by two or more units. The integrated units can be implemented in hardware, as software functional units, or a combination of both.

[0188] It should be noted that, in the embodiments of this application, if the above-described methods are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, or the parts that contribute to related technologies, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an electronic device to execute all or part of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), magnetic disks, or optical disks. Thus, the embodiments of this application are not limited to any specific hardware and software combination.

[0189] This application provides a computer device, which may be a server, and its internal structure diagram may be as follows: Figure 9 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores data. The network interface communicates with external terminals via a network connection. When the computer program is executed by the processor, it implements the methods described above.

[0190] This application provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the method provided in the above embodiments.

[0191] The embodiment of the present application provides a computer program product containing instructions, which, when running on a computer, causes the computer to execute the steps in the method provided by the above method embodiment.

[0192] Those skilled in the art can understand that, Figure 9 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0193] In one embodiment, the welding detection device provided by the present application can be implemented in the form of a computer program, which can run on a computer device as shown in the figure. Figure 9 The memory of the computer device can store various program modules constituting the above device. The computer program constituted by the various program modules causes the processor to execute the steps in the method of each embodiment of the present application described in the specification.

[0194] It should be noted that the description of the above storage medium and device embodiments is similar to the description of the above method embodiments, and has similar beneficial effects to the method embodiments. For technical details not disclosed in the storage medium, storage medium and device embodiments of the present application, please refer to the description of the method embodiments of the present application.

[0195] It should be understood that the "one embodiment" or "an embodiment" or "some embodiments" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, "in one embodiment" or "in an embodiment" or "in some embodiments" appearing throughout the specification does not necessarily mean the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in various embodiments of the present application, the size of the serial number of the above processes does not mean the execution order, and the execution order of the processes should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application. The serial number of the above embodiments of the present application is only for description, not representing the advantages and disadvantages of the embodiments. The above description of each embodiment tends to emphasize the differences between each embodiment, and the same or similar parts can be referred to each other. For the sake of brevity, this paper will not repeat here.

[0196] The term "and / or" in this paper is only a description of the association relationship of the associated objects, which means that there can be three kinds of relationships, for example, object A and / or object B, which can represent three cases of existence of object A, existence of object A and object B, and existence of object B.

[0197] It should be noted that, in the present document, the terms "comprising", "containing" or any other similar term are intended to encompass non-exclusive inclusion, such that processes, methods, articles, or apparatuses that comprise a list of elements are not limited to those elements, but can also include other elements not expressly listed, or inherent to such processes, methods, articles, or apparatuses. Without further limitation, an element defined by the phrase "comprising a" does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0198] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The above-described embodiments are merely illustrative, for example, the division of the modules is only a logical function division, and actual implementation can have another division manner, such as: a plurality of modules or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed components can be indirect coupling or communication connection between some interfaces, devices or modules, which can be electrical, mechanical or other forms.

[0199] The above-described modules explained as separate components can or can not be physically separated, and the components displayed as modules can or can not be physical modules; they can be located in one place or distributed on multiple network units; and part or all of the modules can be selected according to actual needs to achieve the purpose of the present embodiment.

[0200] In addition, each functional module in each embodiment of the present application can be integrated in one processing unit, or each module can be a separate unit, or two or more modules can be integrated in one unit; the above integrated modules can be realized in the form of hardware or hardware plus software functional units.

[0201] Those of ordinary skill in the art can understand that all or part of the steps of the above method embodiments can be completed by program instruction-related hardware, and the foregoing program can be stored in a computer-readable storage medium, and the program executes the steps of the above method embodiments when executed; and the foregoing storage medium includes mobile storage devices, read-only memories (ROM), magnetic discs or optical discs, and various media that can store program codes.

[0202] Alternatively, the above-mentioned integrated units of the present application, if realized in the form of software function modules and sold or used as independent products, can also be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the embodiments of the present application can be embodied in the form of a software product, and the computer software product is stored in a storage medium, and includes several instructions for causing an electronic device to execute all or part of the methods described in the embodiments of the present application. The aforementioned storage medium includes: mobile storage devices, ROM, magnetic disks or optical disks, and various media capable of storing program codes.

[0203] The methods disclosed in the several method embodiments provided by the present application can be combined arbitrarily without conflict, to obtain new method embodiments.

[0204] The features disclosed in the several product embodiments provided by the present application can be combined arbitrarily without conflict, to obtain new product embodiments.

[0205] The features disclosed in the several method or device embodiments provided by the present application can be combined arbitrarily without conflict, to obtain new method or device embodiments.

[0206] The above is only an implementation of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method of detecting a weld, characterized by, The method comprises: obtaining a target data set, the target data set being welding data of a welding device on a battery, the target data set comprising at least one, the welding data comprising a plurality of welding times, welding parameters corresponding to each welding time; performing feature extraction processing on each target data set to obtain a target feature set corresponding to each target data set; determining a first identification model according to the data volume of the target data set, and performing identification processing on the target feature set corresponding to the target data set through the first identification model to obtain a first identification result corresponding to the target data set, the first identification result comprising welding normal or welding abnormal, the first identification model being a statistical model or an anomaly detection model, the anomaly detection model being obtained by combining a deep learning model and a correction model; in the case that the first identification result is welding abnormal, determining the cause of the welding abnormal by obtaining production data of the welding device during the welding process of the battery, the cause of the welding abnormal comprising one or more of welding device parameter abnormal, welding device abnormal or human parameter adjustment abnormal; wherein the welding device comprises a pressing unit and a welding unit, the production data comprising sensor data collected during the operation of the pressing unit and welding parameters of the welding unit, and the determination of the cause of the welding abnormal by obtaining the production data of the welding device during the welding process of the battery comprises: performing identification processing on the sensor data through a PHM management system to determine whether the pressing unit operation is abnormal, and performing identification processing on the welding parameters of the welding unit through a third identification model to determine whether the welding unit operation is abnormal, the third identification model being an anomaly detection model, and the anomaly detection model being obtained by combining a deep learning model and a correction model; the production data comprising debugging data, a first debugging parameter set before debugging of the welding device and a second debugging parameter set after debugging of the welding device, and the determination of the cause of the welding abnormal by obtaining the production data of the welding device during the welding process of the battery comprises: performing identification processing on the debugging data through an experience model to obtain an intermediate identification result, the intermediate identification result being used to indicate whether the debugging process is abnormal, the experience model being obtained by training a deep learning model according to historical welding data and historical maintenance data in a historical welding process; performing feature extraction processing on the first debugging parameter set and the second debugging parameter set to obtain first parameter features and second parameter features; performing identification processing on the first parameter features and the second parameter features through a fourth identification model to obtain a third identification result, the third identification result being used to indicate whether the debugging parameters are abnormal, the fourth identification model being a deep learning model; and comprehensively determining whether there is human parameter adjustment abnormal based on the intermediate identification result and the third identification result.

2. The method of claim 1, wherein, The identifying processing on the target feature set corresponding to the target data set by the first identification model comprises: In a case where the data quantity of the target data set is less than a data quantity threshold, identifying processing on the target feature set by the statistical model is performed to obtain the first identification result corresponding to the target data set; In a case where the data quantity of the target data set is greater than or equal to the data quantity threshold, identifying processing on the target feature set by the anomaly detection model is performed to obtain the first identification result corresponding to the target data set.

3. The method of claim 1, wherein, After the identification result corresponding to the target data set is obtained, the method further comprises: performing data stability analysis on the target data set to obtain a stability analysis result; performing identifying processing on the stability analysis result and the first identification result corresponding to each target data set by a second identification model to obtain a second identification result, the second identification result comprising whether to replace the welding equipment and / or welding material, the second identification model being a deep learning model.

4. The method of claim 3, wherein, After the second identification result is obtained, the method further comprises: in a case where the second identification result indicates that the welding equipment and / or welding material is to be replaced, updating the first identification model to obtain an updated first identification model, and performing identifying processing on a next target data set by the updated first identification model.

5. The method of claim 1, wherein, The welding equipment comprises a welding machine, the production data comprises setting parameters of the welding equipment and welding parameters of the welding machine, and determining the cause of the welding anomaly in the welding process of the battery by the obtained production data of the welding equipment comprises: performing identifying processing on the setting parameters of the welding equipment by a first sub-identification model to obtain a first sub-identification result, the first sub-identification result comprising whether the setting parameters of the welding equipment are normal or abnormal, the first sub-identification model being a statistical model; performing identifying processing on the welding parameters of the welding machine by a second sub-identification model to obtain a second sub-identification result, the second sub-identification result comprising whether the welding parameters of the welding machine are normal or abnormal, the second sub-identification model being an anomaly detection model, the anomaly detection model being obtained by combining a deep learning model and a correction model; comprehensively determining whether the welding equipment parameters are abnormal based on the first sub-identification result and the second sub-identification result.

6. A weld detection apparatus characterized by, comprise: an acquisition module configured to acquire target data sets, the target data sets being welding data of the welding equipment on the battery, the target data sets comprising at least one, and the welding data comprising a plurality of welding times and welding parameters corresponding to each welding time; an extraction module configured to perform feature extraction processing on each target data set to obtain a target feature set corresponding to each target data set; The identification module is configured to determine a first identification model according to an amount of data of the target data set, and perform identification processing on a target feature set corresponding to the target data set by using the first identification model to obtain a first identification result corresponding to the target data set, the first identification result including welding normality or welding abnormality, the first identification model being a statistical model or an abnormality detection model, and the abnormality detection model being obtained by combining a deep learning model and a correction model. The determination module is configured to, in a case where the first identification result is welding abnormality, determine a cause of the welding abnormality by using obtained production data of a welding process of the battery by the welding equipment, the cause of the welding abnormality including one or more of welding equipment parameter abnormality, welding device abnormality, or human parameter adjustment abnormality, the welding equipment including a pressing unit and a welding unit, and the production data including sensor data collected during working of the pressing unit and welding parameters of the welding unit. The identification module is further configured to perform identification processing on the sensor data by using a PHM management system to determine whether the pressing unit works abnormally, and perform identification processing on the welding parameters of the welding unit by using a third identification model to determine whether the welding unit works abnormally, the third identification model being an abnormality detection model, the abnormality detection model being obtained by combining a deep learning model and a correction model, and the production data including debugging data, a first debugging parameter set before debugging of the welding equipment, and a second debugging parameter set after debugging of the welding equipment. The identification module is further configured to perform identification processing on the debugging data by using an experience model to obtain an intermediate identification result, the intermediate identification result being used to indicate whether an abnormality occurs in a debugging process, and the experience model being obtained by training a deep learning model according to historical welding data and historical maintenance data in a historical welding process. The extraction module is further configured to perform feature extraction processing on the first debugging parameter set and the second debugging parameter set to obtain first parameter features and second parameter features. The identification module is further configured to perform identification processing on the first parameter features and the second parameter features by using a fourth identification model to obtain a third identification result, the third identification result being used to indicate whether an abnormality occurs in the debugging parameters, and the fourth identification model being a deep learning model; and determine whether the human parameter adjustment abnormality exists by comprehensively considering the intermediate identification result and the third identification result. 7.A computer device, comprising a memory and a processor, wherein the memory stores a computer program capable of running on the processor, and the computer device is characterized in that, The processor executes the program to implement the steps of the method in any one of claims 1 to 5.

8. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the method in any one of claims 1 to 5.

Images