An online fault diagnosis method and system for laser welding wire feeder

Abstract

The present invention relates to the technical field of online fault diagnosis of laser welding wire feeders, and discloses an online fault diagnosis method and system for laser welding wire feeders, wherein the method comprises: real-time acquisition of wire feeding state parameters of the laser welding wire feeder; calculation of welding wire fluctuation state score and welding wire offset, and judgment of abnormal bending fluctuation fault in the guiding mechanism of the welding wire feeder; outputting fault classification results of the laser welding wire feeder by using convolutional neural network and long short-term memory network in combination with wire feeder fault model; generating an online fault diagnosis report for the laser welding wire feeder; compared with the technical problem that it is difficult to cope with multi-dimensional changes in the operation process of the welding wire feeder in the prior art, resulting in failure to timely discover the fault of the welding wire feeder, the present invention realizes accurate diagnosis and efficient classification of wire feeder faults through multi-modal data fusion, calculation of welding wire fluctuation state score and welding wire offset, and application of machine algorithm, thereby improving the fault recognition rate.

Description

Technical Field

[0001] The invention belongs to the technical field of online fault diagnosis of a laser welding wire feeder, and in particular relates to an online fault diagnosis method and system for a laser welding wire feeder. Background Art

[0002] At present, laser welding wire feeders have been widely used in the field of industrial automated welding, especially in high-precision welding operations, where the stability of laser welding wire feeders directly affects the welding quality. However, the welding wire feeders in the prior art still have some shortcomings during operation. For example, the existing fault diagnosis methods often rely on a single fault detection method, such as judging faults by current fluctuations or load monitoring of the wire feeder motor, but these methods cannot fully meet the needs of efficient and real-time fault diagnosis in complex welding environments. In particular, under the conditions of temperature fluctuations, unstable welding current or environmental vibration, the existing technology cannot timely and accurately identify the potential faults of the welding wire feeder. Therefore, there is an urgent need for a solution that can fuse multimodal data (including physical sensor data and image data) and combine deep learning technology for fault diagnosis. This solution should be able to monitor the status of the wire feeder in real time under complex and changeable welding environments, timely identify deviations, jamming, straightening problems that may occur during wire feeding, and provide accurate fault classification and fault repair suggestions, thereby improving the stability of the laser welding wire feeder, reducing equipment downtime and improving welding quality. Summary of the invention

[0003] In view of the above-mentioned technical deficiencies, the purpose of the present invention is to propose an online fault diagnosis method for a laser welding wire feeder, aiming to solve the technical problem that the prior art is difficult to cope with the multi-dimensional changes during the operation of the welding wire feeder, resulting in the inability to timely detect the faults of the welding wire feeder.

[0004] In order to solve the above technical problems, the present invention adopts the following technical solutions: The present invention provides an online fault diagnosis method for a laser welding wire feeder.

[0005] The online fault diagnosis method of the laser welding wire feeder comprises:

[0006] Step S10: collecting wire feeding state parameters of the laser welding wire feeder in real time;

[0007] Step S20: collecting welding image data in real time through an industrial camera, and calculating a welding wire fluctuation state score and a welding wire offset according to the welding image data;

[0008] Step S30: Preset a threshold value of the welding wire fluctuation state score. If the welding wire fluctuation state score is greater than the threshold value of the welding wire fluctuation state score, it is determined that the guide mechanism of the welding wire feeder has an abnormal bending fluctuation fault; otherwise, proceed to the next step;

[0009] Step S40: A multimodal feature matrix is ​​obtained by fusing wire feeding state parameters, welding wire fluctuation state score and welding wire offset, a convolutional neural network is used to extract the multimodal feature matrix to obtain multidimensional spatial feature information, and the multimodal feature matrix is ​​processed by a long short-term memory network to obtain time series feature information, and the multidimensional spatial feature information and time series feature information are used as inputs of a preset wire feeder fault model, and the wire feeder fault model outputs a fault classification result of the laser welding wire feeder;

[0010] Step S50: Obtain historical fault classification results, calculate the probability of missed reports based on the historical fault classification results combined with anomaly detection algorithms, and generate an online fault diagnosis report for the laser welding wire feeder.

[0011] Preferably, in step S20, the step of obtaining the welding wire fluctuation state score according to the welding image data specifically includes:

[0012] Extract the welding wire pixels based on the welding image data combined with Canny edge detection;

[0013] The horizontal coordinate of the center line of the welding wire is calculated according to the pixel points of the welding wire. The calculation formula is:

[0014]

[0015] Among them, x w is the horizontal coordinate of the center line of the welding wire, x i is the horizontal coordinate of the welding wire pixel point i in the welding image data, and N is the total number of welding wire pixels.

[0016] Preferably, in step S20, the calculation formula of the welding wire fluctuation state score is:

[0017]

[0018] Among them, y w The horizontal axis is x w The vertical coordinate of the welding wire pixel point, C w is the welding wire fluctuation state score, x w+1 ,y w+1 and x w+2 ,y w+2 are the coordinates of the two pixels to the right of the horizontal coordinate of the center line of the welding wire, x w+2 -2x w+1 +x w and w+2 -2y w+1 +yw are the second-order discrete differences of the welding wire in the x and y directions, respectively.

[0019] Preferably, in step S10, the wire feeding state parameters include wire feeding speed, wire feeding acceleration, wire feeding driving torque, welding current and welding power.

[0020] Preferably, in step S20, the calculation formula of the welding wire offset is:

[0021]

[0022] Among them, x i is the horizontal coordinate of the welding wire pixel point i in the welding image data, Δθ w is the wire offset, y i is the ordinate of the welding wire pixel point i in the welding image data, is the horizontal coordinate of the welding wire pixel point i in the preset standard welding image data, is the ordinate of the welding wire pixel point i in the preset standard welding image data.

[0023] Preferably, in step S40, the calculation formula of the multimodal feature matrix is:

[0024] F fusion =α1·F sensor +α2·F image +α3·Δθ w

[0025] Among them, F fusion is the multimodal feature matrix, F sensor is the wire feeding state characteristic matrix constructed according to the wire feeding state parameters, F image is the welding wire fluctuation state score, Δθ w is the wire offset, α1, α2, α3 are the dynamic weighting coefficient matrices.

[0026] Preferably, in step S40, the preset wire feeder fault model is constructed based on a support vector machine algorithm, and uses the multi-dimensional spatial feature information and time series feature information of the preset historical fault data as input, and the preset standard fault cause template of the historical fault data as output to train the model; the wire feeder fault model outputs the fault classification results of the laser welding wire feeder, specifically including abnormalities in the wire feeding mechanism, blockage in the wire feeding tube, and abnormalities in the wire straightening mechanism.

[0027] The present invention also provides an online fault diagnosis system for a laser welding wire feeder, comprising:

[0028] Wire feeding state parameter acquisition module, used for real-time acquisition of wire feeding state parameters of laser welding wire feeder;

[0029] An image analysis module is used to collect welding image data in real time through an industrial camera and calculate the welding wire fluctuation state score and welding wire offset according to the welding image data;

[0030] The fluctuation fault judgment module is used to preset the welding wire fluctuation state score threshold. If the welding wire fluctuation state score is greater than the welding wire fluctuation state score threshold, it is judged that the guide mechanism of the welding wire feeder has an abnormal bending fluctuation fault; otherwise, it proceeds to the next step;

[0031] A fault diagnosis module is used to fuse wire feeding state parameters, welding wire fluctuation state scores and welding wire offsets to obtain a multimodal feature matrix, extract the multimodal feature matrix using a convolutional neural network to obtain multidimensional spatial feature information, and process the multimodal feature matrix through a long short-term memory network to obtain temporal feature information. The multidimensional spatial feature information and temporal feature information are used as inputs of a preset wire feeder fault model, and the wire feeder fault model outputs a fault classification result of the laser welding wire feeder;

[0032] The result verification module is used to obtain historical fault classification results, calculate the probability of missed reports based on the historical fault classification results combined with the anomaly detection algorithm, and generate an online fault diagnosis report for the laser welding wire feeder.

[0033] The present invention also provides a computer program product, including an online fault diagnosis program for a laser welding wire feeder, wherein the online fault diagnosis program for a laser welding wire feeder implements the online fault diagnosis method for a laser welding wire feeder when executed by a processor.

[0034] The beneficial effects of the present invention are as follows: compared with the technical problem that it is difficult to cope with the multi-dimensional changes in the operation process of the welding wire feeder in the prior art, resulting in the inability to timely discover the faults of the welding wire feeder, the present invention successfully realizes the accurate diagnosis and efficient classification of wire feeder faults through multimodal data fusion, combined with the calculation of the welding wire fluctuation state score and the welding wire offset, and the application of machine learning algorithms (such as convolutional neural network (CNN) and long short-term memory network (LSTM). Through this multi-angle analysis method, the working state of the wire feeder can be more comprehensively evaluated, so as to timely identify the types of faults that may occur in the welding wire feeding process at an early stage, such as welding wire offset, wire feeding instability, welding wire jamming, etc.;

[0035] Since the diagnosis method based on multimodal data can cope with complex and dynamic working environments, the present invention has higher stability and applicability in practical applications, and greatly improves the maintenance efficiency and production benefits of welding equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0037] Figure 1 It is a flow chart of a first embodiment of an online fault diagnosis method for a laser welding wire feeder according to the present invention.

[0038] Figure 2 It is a schematic diagram of equipment for an online fault diagnosis method of a laser welding wire feeder according to the present invention. DETAILED DESCRIPTION

[0039] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0040] Embodiment 1: Figure 1 , which is a flow chart of the first embodiment of the online fault diagnosis method for the laser welding wire feeder of the present invention, and provides the first embodiment of the online fault diagnosis method for the laser welding wire feeder of the present invention.

[0041] In a first embodiment, the online fault diagnosis method of the laser welding wire feeder includes:

[0042] Step S10: collecting wire feeding state parameters of the laser welding wire feeder in real time;

[0043] It should be noted that in step S10, the wire feeding status parameters include wire feeding speed, wire feeding acceleration, wire feeding drive torque, welding current and welding power. These parameters are used to monitor the status of the welding wire feeder in real time during operation, so as to help determine whether there are potential faults or abnormalities. The wire feeding speed indicates the feeding speed of the welding wire during the wire feeding process, and the unit is meter per second (m / s). This parameter is used to monitor the stability of the welding wire and ensure that the welding wire is fed into the welding area at a predetermined speed. The wire feeding acceleration indicates the acceleration of the welding wire during the wire feeding process, and the unit is meter per second (m / s2). 2), this parameter is used to evaluate whether the welding wire is stable during the start or stop process, so as to avoid affecting the welding quality due to excessive acceleration or deceleration; the wire feeding drive torque, the unit is Newton meter (Nm), reflects the load and driving capacity of the wire feeding motor. Monitoring the wire feeding drive torque helps to identify faults such as wire jamming or uneven wire feeding; the welding current unit is ampere (A), which is used to monitor the current changes during the welding process in real time. The stability of the welding current directly affects the quality of the welding pool, so it is necessary to monitor the current fluctuation in real time; the welding power unit is watt (W), which indicates the power used in the laser welding process, which is usually related to the welding current and voltage. The stability of the welding power is crucial to the welding quality.

[0044] It is understandable that the collection of wire feeding status parameters is the basis for ensuring the stability of the wire feeding process, and can help identify abnormal fluctuations in the wire feeding process in the early stages, such as unstable wire feeding, wire deviation, wire jamming and other problems.

[0045] It should be understood that in step S10, the data collected by the sensor may be interfered by various external factors, such as ambient temperature, humidity or electromagnetic interference, etc. In order to ensure the accuracy of the data, methods such as Kalman filtering are usually used to denoise and smooth the collected data to ensure the reliability of subsequent analysis results.

[0046] For example, during the operation of a welding wire feeder, the data monitored by the wire feeding speed suddenly fluctuates, which may indicate that there is a problem with the wire guiding mechanism or the wire straightening mechanism, causing the trajectory of the wire feeding into the welding point to be offset or blocked. If the wire feeding drive torque is too large and the welding current fluctuates at the same time, it may indicate that the wire is stuck or the wire feeding is uneven during the wire feeding process, and further inspection is needed to see if the wire feeding path is unobstructed.

[0047] Step S20: collecting welding image data in real time through an industrial camera, and calculating a welding wire fluctuation state score and a welding wire offset according to the welding image data;

[0048] It should be noted that the Canny edge detection algorithm is used to identify the edge of the welding wire in the image in order to accurately locate the position of the welding wire. Through edge detection, the outline of the welding wire in the welding image is obtained, and the horizontal coordinate of the center line of the welding wire is the main path of the welding wire in the image. By extracting the welding wire pixel points in the image, the horizontal coordinate of the center line of the welding wire can be calculated. . In step S20, the step of obtaining the welding wire fluctuation state score according to the welding image data specifically includes: extracting the welding wire pixel points according to the welding image data combined with the Canny edge detection; calculating the horizontal coordinate of the center line of the welding wire according to the welding wire pixel points, and the calculation formula is: Among them, x w is the horizontal coordinate of the center line of the welding wire, x iis the horizontal coordinate of the welding wire pixel point i in the welding image data, and N is the total number of welding wire pixels.

[0049] In step S20, the calculation formula of the welding wire fluctuation state score is:

[0050]

[0051] Among them, y w The horizontal axis is x w The vertical coordinate of the welding wire pixel point, C w is the welding wire fluctuation state score, x w+1 ,y w+1 and x w+2 ,y w+2 are the coordinates of the two pixels to the right of the horizontal coordinate of the center line of the welding wire, x w+2 -2x w+1 +x w and w+2 -2y w+1 +y w are the second-order discrete differences of the welding wire in the x and y directions, respectively.

[0052] In step S20, the calculation formula of the welding wire offset is:

[0053]

[0054] Among them, x i is the horizontal coordinate of the welding wire pixel point i in the welding image data, Δθ w is the wire offset, y i is the ordinate of the welding wire pixel point i in the welding image data, is the horizontal coordinate of the welding wire pixel point i in the preset standard welding image data, is the ordinate of the welding wire pixel point i in the preset standard welding image data.

[0055] It is understandable that the wire fluctuation state score is used to measure whether the wire has irregular fluctuations during the wire feeding process. If the fluctuation state score is too large, it indicates that there is unstable airflow or abnormal guidance mechanism during the wire feeding process, which affects the welding quality. The wire offset reflects the difference between the actual path of the wire and the preset path. A large offset may cause the wire to fail to be accurately delivered to the welding point, thus affecting the welding accuracy.

[0056] It should be understood that during the image processing process, the deviation and fluctuation of the welding wire are inevitable due to the shape and material of the welding wire itself. The calculation of the welding wire fluctuation state score and the offset is based on the results of real-time analysis of image data and is used to monitor abnormal conditions during the wire feeding process. The calculation of the welding wire offset not only depends on the lateral position of the welding wire, but also needs to consider its actual deviation relative to the preset trajectory to ensure that the welding wire is accurately fed into the welding area.

[0057] Step S30: Preset a threshold value of the welding wire fluctuation state score. If the welding wire fluctuation state score is greater than the threshold value of the welding wire fluctuation state score, it is determined that the guide mechanism of the welding wire feeder has an abnormal bending fluctuation fault; otherwise, proceed to the next step;

[0058] It should be noted that the wire fluctuation state score threshold is a key value determined through actual testing or standardization process. When the wire fluctuation state score is greater than the threshold, it can be considered that there is an abnormal fluctuation fault of the guide mechanism.

[0059] For example, by setting the threshold of the welding wire fluctuation state score, it is possible to effectively distinguish between normal fluctuations and abnormal fluctuations in the wire feeding process. If the fluctuation score exceeds the set threshold, it is determined that the wire feeder's guide mechanism has an abnormal bending fluctuation fault, which triggers a fault repair suggestion; otherwise, the next operation or fault diagnosis will be continued. This judgment mechanism helps to efficiently and real-time detect and locate potential faults in the welding wire feeder, thereby ensuring the stability of the welding process and the welding quality.

[0060] Step S40: A multimodal feature matrix is ​​obtained by fusing wire feeding state parameters, welding wire fluctuation state score and welding wire offset, a convolutional neural network is used to extract the multimodal feature matrix to obtain multidimensional spatial feature information, and the multimodal feature matrix is ​​processed by a long short-term memory network to obtain time series feature information, and the multidimensional spatial feature information and time series feature information are used as inputs of a preset wire feeder fault model, and the wire feeder fault model outputs a fault classification result of the laser welding wire feeder;

[0061] It should be noted that in step S40, the calculation formula of the multimodal feature matrix is:

[0062] F fusion =α1·F sensor +α2·F image +α3·Δθ w

[0063] Among them, F fusion is the multimodal feature matrix, F sensor is the wire feeding state characteristic matrix constructed according to the wire feeding state parameters, F image is the welding wire fluctuation state score, Δθ wis the wire offset, α1, α2, α3 are the dynamic weighting coefficient matrices.

[0064] In step S40, the preset wire feeder fault model is constructed based on the support vector machine algorithm, and uses the multi-dimensional spatial feature information and time series feature information of the preset historical fault data as input, and the preset standard fault cause template of the historical fault data as output to train the model; the wire feeder fault model outputs the fault classification results of the laser welding wire feeder, specifically including abnormal wire feeding mechanism, wire feeding tube blockage, and wire straightening mechanism abnormality.

[0065] It can be understood that by weighted fusion of features from different data sources, a comprehensive multimodal feature matrix can be obtained, which contains information on spatial features and temporal features. Through the weighted fusion of the multimodal feature matrix, multiple key factors in the wire feeding process are comprehensively considered, such as the wire fluctuation state score, wire offset, and wire feeding state parameters, which can more comprehensively reflect the working state of the wire feeder.

[0066] It should be understood that step S40 achieves accurate diagnosis of welding wire feeder faults by fusing multiple data sources (such as wire feeding state parameters, wire fluctuation state scores, and wire offsets) and combining deep learning models (CNN and LSTM). Through the application of multimodal feature matrix and SVM fault classification model, the wire feeder fault model can efficiently and accurately classify various fault types of welding wire feeders. This method improves the fault recognition rate during welding and significantly enhances the real-time and accuracy of fault detection.

[0067] For example, if the wire feeding current fluctuates greatly during the wire feeding process, resulting in abnormal wire offset, the model will identify the abnormality of the wire feeding mechanism through multimodal data fusion (including wire feeding current and wire offset) and output the diagnostic result of "wire feeding mechanism failure" through the SVM classifier.

[0068] Step S50: Obtain historical fault classification results, calculate the probability of missed reports based on the historical fault classification results combined with anomaly detection algorithms, and generate an online fault diagnosis report for the laser welding wire feeder.

[0069] It should be noted that by obtaining historical fault classification results and comparing them with the current real-time fault diagnosis results, anomaly detection algorithms (such as Isolation Forest, One-Class SVM, etc.) are used to calculate the probability of missing faults. In this way, the accuracy of the fault diagnosis model of the wire feeder can be verified and whether there are potential faults that have not been detected can be evaluated.

[0070] It should be understood that the historical fault classification results provide a reference benchmark for the wire feeder fault model, but the current diagnosis results may not be completely consistent with the historical data. By comparing with historical faults, the anomaly detection algorithm can find possible deficiencies and missed reports in the current model. The calculation of the missed report probability does not mean that there is a problem with the diagnosis result, but provides an evaluation standard through a quantitative method. If the missed report probability is high, the missed reports can be reduced by updating and training the fault classification model.

[0071] For example, the diagnostic results do not report a wire jam fault, but historical data shows that wire jam is one of the common faults. In this case, the probability of false negatives may be calculated to be 10%, and feedback may prompt the operator to perform further model parameter calibration to reduce the possibility of false negatives.

[0072] Embodiment 2: In addition, the online fault diagnosis system of a laser welding wire feeder provided by the present invention adopts the online fault diagnosis method of a laser welding wire feeder in the above embodiment, which can solve the technical problem of online fault diagnosis of a laser welding wire feeder. Compared with the prior art, the beneficial effects of the online fault diagnosis system of a laser welding wire feeder provided by the present invention are the same as the beneficial effects of the online fault diagnosis method of a laser welding wire feeder provided by the above embodiment, and the other technical features of the online fault diagnosis system of a laser welding wire feeder are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0073] Embodiment 3: The present invention provides an online fault diagnosis device for a laser welding wire feeder, please refer to Figure 2, an online fault diagnosis device for a laser welding wire feeder includes: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the online fault diagnosis method for a laser welding wire feeder in the above-mentioned embodiment 1. An online fault diagnosis device for a laser welding wire feeder in an embodiment of the present invention may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. An online fault diagnosis device for a laser welding wire feeder is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention. An online fault diagnosis device for a laser welding wire feeder may include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM) 1004. Various programs and data required for the operation of an online fault diagnosis device for a laser welding wire feeder are also stored in RAM 1004. The processing device 1001, ROM 1002, and RAM 1004 are connected to each other via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 1003 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 1009. The communication device 1009 can allow an online fault diagnosis device for a laser welding wire feeder to communicate wirelessly or wired with other devices to exchange data. Although the figure shows an online fault diagnosis device for a laser welding wire feeder with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems may be implemented or have instead.

[0074] Embodiment 4: The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the above-mentioned online fault diagnosis method for a laser welding wire feeder. The computer program product provided by the present invention can solve the technical problem of online fault diagnosis of a laser welding wire feeder. Compared with the prior art, the beneficial effects of the computer program product provided by the present invention are the same as the beneficial effects of the online fault diagnosis method for a laser welding wire feeder provided in the above-mentioned embodiment, which will not be repeated here.

[0075] In particular, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through a communication device, or installed from the storage device 1003, or installed from the ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present invention are executed.

[0076] It should be understood that the various parts disclosed in the present invention can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0077] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. An online fault diagnosis method for a laser welding wire feeder, characterized in that: Methods include: Step S10: collecting wire feeding state parameters of the laser welding wire feeder in real time; Step S20: collecting welding image data in real time through an industrial camera, and calculating a welding wire fluctuation state score and a welding wire offset according to the welding image data; Step S30: Preset a threshold value of the welding wire fluctuation state score. If the welding wire fluctuation state score is greater than the threshold value of the welding wire fluctuation state score, it is determined that the guide mechanism of the welding wire feeder has an abnormal bending fluctuation fault; otherwise, proceed to the next step; Step S40: A multimodal feature matrix is ​​obtained by fusing wire feeding state parameters, welding wire fluctuation state score and welding wire offset, a convolutional neural network is used to extract the multimodal feature matrix to obtain multidimensional spatial feature information, and the multimodal feature matrix is ​​processed by a long short-term memory network to obtain time series feature information, and the multidimensional spatial feature information and time series feature information are used as inputs of a preset wire feeder fault model, and the wire feeder fault model outputs a fault classification result of the laser welding wire feeder; Step S50: Obtain historical fault classification results, calculate the probability of missed reports based on the historical fault classification results combined with anomaly detection algorithms, and generate an online fault diagnosis report for the laser welding wire feeder.

2. The online fault diagnosis method of a laser welding wire feeder according to claim 1, characterized in that: In step S20, the step of obtaining the welding wire fluctuation state score according to the welding image data specifically includes: Extract the welding wire pixels based on the welding image data combined with Canny edge detection; The horizontal coordinate of the center line of the welding wire is calculated according to the pixel points of the welding wire. The calculation formula is: Among them, x w is the horizontal coordinate of the center line of the welding wire, x i is the horizontal coordinate of the welding wire pixel point i in the welding image data, and N is the total number of welding wire pixels; The welding wire fluctuation state score is calculated using the three-point difference method based on the welding wire centerline.

3. The online fault diagnosis method of a laser welding wire feeder according to claim 2, characterized in that: In step S20, the calculation formula of the welding wire fluctuation state score is: Among them, y w The horizontal axis is x w The vertical coordinate of the welding wire pixel point, C w is the welding wire fluctuation state score, x w+1 ,y w+1 and x w+2 ,y w+2 are the coordinates of the two pixels to the right of the horizontal coordinate of the center line of the welding wire, x w+2 -2x w+1 +x w and w+2 -2y w+1 +y w are the second-order discrete differences of the welding wire in the x and y directions, respectively.

4. The online fault diagnosis method of a laser welding wire feeder according to claim 1, characterized in that: In step S10, the wire feeding state parameters include wire feeding speed, wire feeding acceleration, wire feeding driving torque, welding current and welding power.

5. The online fault diagnosis method of a laser welding wire feeder according to claim 1, characterized in that: In step S20, the calculation formula of the welding wire offset is: Among them, x i is the horizontal coordinate of the welding wire pixel point i in the welding image data, Δθ w is the wire offset, y i is the ordinate of the welding wire pixel point i in the welding image data, is the horizontal coordinate of the welding wire pixel point i in the preset standard welding image data, is the ordinate of the welding wire pixel point i in the preset standard welding image data.

6. The online fault diagnosis method of a laser welding wire feeder according to claim 1, characterized in that: In step S40, the calculation formula of the multimodal feature matrix is: F fusion =α1·F sensor +α2·F image +α2·Δθ w Among them, F fusion is the multimodal feature matrix, F sensor is the wire feeding state characteristic matrix constructed according to the wire feeding state parameters, F image is the welding wire fluctuation state score, Δθ w is the wire offset, α1, α2, α3 are the dynamic weighting coefficient matrices.

7. The online fault diagnosis method of a laser welding wire feeder according to claim 1, characterized in that: In step S40, the preset wire feeder fault model is constructed based on the support vector machine algorithm, and uses the multi-dimensional spatial feature information and time series feature information of the preset historical fault data as input, and the preset standard fault cause template of the historical fault data as output to train the model; the wire feeder fault model outputs the fault classification results of the laser welding wire feeder, specifically including abnormal wire feeding mechanism, wire feeding tube blockage, and wire straightening mechanism abnormality.

8. An online fault diagnosis system for a laser welding wire feeder, characterized in that: The online fault diagnosis system of the laser welding wire feeder comprises: Wire feeding state parameter acquisition module, used for real-time acquisition of wire feeding state parameters of laser welding wire feeder; An image analysis module is used to collect welding image data in real time through an industrial camera and calculate the welding wire fluctuation state score and welding wire offset according to the welding image data; The fluctuation fault judgment module is used to preset the welding wire fluctuation state score threshold. If the welding wire fluctuation state score is greater than the welding wire fluctuation state score threshold, it is judged that the guide mechanism of the welding wire feeder has an abnormal bending fluctuation fault; otherwise, it proceeds to the next step; A fault diagnosis module is used to fuse wire feeding state parameters, welding wire fluctuation state scores and welding wire offsets to obtain a multimodal feature matrix, extract the multimodal feature matrix using a convolutional neural network to obtain multidimensional spatial feature information, and process the multimodal feature matrix through a long short-term memory network to obtain temporal feature information. The multidimensional spatial feature information and temporal feature information are used as inputs of a preset wire feeder fault model, and the wire feeder fault model outputs a fault classification result of the laser welding wire feeder; The result verification module is used to obtain historical fault classification results, calculate the probability of missed reports based on the historical fault classification results combined with the anomaly detection algorithm, and generate an online fault diagnosis report for the laser welding wire feeder.

9. An online fault diagnosis device for a laser welding wire feeder, characterized in that: The online fault diagnosis device of the laser welding wire feeder comprises: a memory, a processor and an online fault diagnosis program of the laser welding wire feeder stored in the memory and executable on the processor. When the online fault diagnosis program of the laser welding wire feeder is executed by the processor, the online fault diagnosis method of the laser welding wire feeder described in any one of claims 1 to 7 is implemented.

10. A computer program product, characterized in that The computer program product includes an online fault diagnosis program for a laser welding wire feeder, and when the online fault diagnosis program for a laser welding wire feeder is executed by a processor, the online fault diagnosis method for a laser welding wire feeder described in any one of claims 1 to 7 is implemented.

Images