Pipeline welding gun swing amplitude prediction method and device based on deep learning

Abstract

The invention discloses a pipeline welding gun swing amplitude prediction method and device based on deep learning. The method comprises the steps that a welding state data set is continuously obtained and preprocessed, and a preprocessed welding state data set is obtained; according to the preprocessed welding state data set, a first state data set and a second state data set are determined; the first state data set is input into a first prediction model, and the distance between the welding gun and the groove edge is obtained; the second state data set is input into a second prediction model, and the distance between the welding gun and the lower edge of the groove is obtained; and determining the swing amplitude of the welding gun according to the distance between the welding gun and the edge of the groove and the distance between the welding gun and the lower edge of the groove. According to the method, through two prediction models constructed based on deep learning, according to the actual groove width and the position of the welding gun, the swing amplitude of the welding gun can be automatically adjusted in real time, the requirements for irregular groove widths and various welding bead thickness changes are met, all-position welding of pipelines with any groove width is completed, manual intervention is not needed, and the cost and the system size are reduced.

Description

Technical Field

[0001] The present application relates to a method and device for predicting the swing amplitude of a pipeline welding torch based on deep learning. Background Art

[0002] Long-distance oil and gas pipelines have large diameters and wall thicknesses, so groove welding is required. Due to the deformation of the weld bead caused by the pipeline groove processing accuracy, rail assembly errors, and the heating and heat dissipation conditions of the workpiece during the welding process, it is impossible to form a constant weld bead thickness and the groove width cannot be consistent according to the welding process requirements. Therefore, when welding, the swing amplitude of the welding gun needs to be continuously adjusted. Pipeline welding is full-position welding, and the welding gun angle changes continuously from 0 to 180 degrees. The welding process parameters at each position may be different. Therefore, even if the groove width is the same, the swing amplitude may need to be adjusted according to the position.

[0003] In previous welding construction, the welding gun swing width was preset in advance. During the welding process, the welder observed the spatial position of the groove and the welding gun in real time and adjusted the preset swing amplitude in real time. However, the welder's work intensity was relatively high and the observation accuracy was limited. Summary of the invention

[0004] In order to better predict the swing amplitude of the welding torch during pipeline welding, an embodiment of the present application provides a method and device for predicting the swing amplitude of a pipeline welding welding torch based on deep learning.

[0005] In a first aspect, an embodiment of the present application provides a method for predicting the swing amplitude of a pipeline welding torch based on deep learning, the method comprising:

[0006] Continuously acquiring and preprocessing the welding state data group to obtain the preprocessed welding state data group;

[0007] Determining a first state data group and a second state data group according to the preprocessed welding state data group;

[0008] Inputting the first state data group into a first prediction model to obtain the distance between the welding gun and the groove edge;

[0009] Inputting the second state data group into a second prediction model to obtain a distance between the welding gun and the lower edge of the groove;

[0010] The swing amplitude of the welding gun is determined according to the distance between the welding gun and the edge of the groove and the distance between the welding gun and the lower edge of the groove.

[0011] In an optional implementation of the embodiment of the present application, the first prediction model is obtained by training in the following manner:

[0012] Collecting multiple sets of historical welding state data sets and preprocessing them to obtain corresponding preprocessed historical welding state data sets; the historical welding state data sets include welding speed, welding angle, dry extension length, distance between welding gun and lower edge of groove, swing speed, current, voltage, wire feeding speed, gas flow and welding layer;

[0013] Constructing a first sample set according to the plurality of pre-processed historical welding state data groups and the corresponding distances between the welding gun and the groove edge;

[0014] Dividing the first sample set into a first training set and a first test set;

[0015] Inputting the first training set into a pre-built initial prediction model for training to obtain a trained prediction model;

[0016] Inputting the first test set into the trained prediction model for testing to obtain a test result of the distance between the welding gun and the groove edge;

[0017] Compare the test result of the distance between the welding gun and the groove edge with the corresponding actual distance between the welding gun and the groove edge, and update the trained prediction model;

[0018] The above training process is repeated until the accuracy of the test result of the distance between the welding gun and the groove edge meets the preset conditions, and the first prediction model is obtained.

[0019] In an optional implementation of the embodiment of the present application, the second prediction model is obtained by training in the following manner:

[0020] Collecting multiple sets of historical welding state data sets and preprocessing them to obtain corresponding preprocessed historical welding state data sets; the historical welding state data sets include welding speed, welding angle, dry extension length, distance between welding gun and groove edge, swing speed, current, voltage, wire feeding speed, gas flow and welding layer;

[0021] Constructing a second sample set according to the plurality of pre-processed historical welding state data groups and the corresponding distances between the welding gun and the lower edge of the groove;

[0022] Dividing the second sample set into a second training set and a second test set;

[0023] Inputting the second training set into a pre-built initial prediction model for training to obtain a trained prediction model;

[0024] Inputting the second test set into the trained prediction model for testing to obtain a test result of the distance between the welding gun and the lower edge of the groove;

[0025] Compare the test result of the distance between the welding gun and the lower edge of the groove with the corresponding actual distance between the welding gun and the lower edge of the groove, and update the trained prediction model;

[0026] The above training process is repeated until the accuracy of the test result of the distance between the welding gun and the lower edge of the groove meets the preset conditions, and the second prediction model is obtained.

[0027] In an optional implementation of the embodiment of the present application, the initial prediction model is constructed in the following manner:

[0028] Based on the LSTM neural network, the optimal initial hyperparameters are set to build the initial prediction model.

[0029] In an optional implementation of the embodiment of the present application, the welding state data group includes welding speed, welding angle, dry extension length, distance between welding gun and groove edge, distance between welding gun and groove lower edge, swing speed, current, voltage, wire feeding speed, gas flow and welding layer; the pre-processed welding state data group is obtained by the following method:

[0030] The welding state data group is cleaned by a normal distribution algorithm to obtain a cleaned welding state data group;

[0031] The welding state data group after cleaning is subjected to frequency synchronization processing according to a preset frequency to obtain the welding state data group after preprocessing.

[0032] In an optional implementation of the embodiment of the present application, after determining the first state data group and the second state data group, the method further includes:

[0033] The first state data group and the second state data group are respectively subjected to data standardization processing.

[0034] In a second aspect, an embodiment of the present application provides a device for predicting the swing amplitude of a pipeline welding torch based on deep learning, the device comprising:

[0035] An acquisition module is used to continuously acquire a welding state data group and perform preprocessing to obtain a preprocessed welding state data group;

[0036] A first determination module, used for determining a first state data group and a second state data group according to the preprocessed welding state data group;

[0037] A first prediction module, used for inputting the first state data group into a first prediction model to obtain the distance between the welding gun and the groove edge;

[0038] A second prediction module, used for inputting the second state data group into a second prediction model to obtain the distance between the welding gun and the lower edge of the groove;

[0039] The second determination module is used to determine the swing amplitude of the welding gun according to the distance between the welding gun and the edge of the groove and the distance between the welding gun and the lower edge of the groove.

[0040] In a third aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned method for predicting the swing amplitude of a pipeline welding torch based on deep learning.

[0041] In a fourth aspect, an embodiment of the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for predicting the swing amplitude of a pipeline welding gun based on deep learning as described above is implemented.

[0042] In a fifth aspect, an embodiment of the present application provides a computer program product comprising instructions. When the computer program product is run on a computer device, the computer device executes the above-mentioned deep learning-based pipeline welding torch swing amplitude prediction method.

[0043] In a sixth aspect, an embodiment of the present application provides a chip, the chip including a processor and a communication interface, the communication interface and the processor are coupled, and the processor is used to run a computer program or instruction to implement the above-mentioned deep learning-based pipeline welding gun swing amplitude prediction method.

[0044] The beneficial effects of the above technical solution provided by the embodiment of the present application include at least:

[0045] The deep learning-based pipeline welding torch swing amplitude prediction method provided in the embodiment of the present application continuously acquires the welding state data group and performs preprocessing to obtain the first state data group and the second state data group, and then respectively inputs the trained first prediction model and the second prediction model to predict the distance between the welding torch and the groove edge and the distance between the welding torch and the lower edge of the groove, and finally predicts the swing amplitude of the welding torch. In the pipeline full-position welding, this method can automatically adjust the welding torch swing amplitude in real time according to the actual groove width and welding torch position through two prediction models constructed based on deep learning, and can adapt to the requirements of irregular groove width and various weld thickness changes, and complete the full-position welding of pipelines with any groove width without manual intervention, and no additional components are added to the welding trolley part, reducing costs and system volume.

[0046] Other features and advantages of the present application will be described in the following description, and partly become apparent from the description, or be understood by practicing the present application. The purpose and other advantages of the present application can be realized and obtained by the structures specifically pointed out in the written description, claims, and drawings.

[0047] The technical solution of the present application is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The accompanying drawings are used to provide a further understanding of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings:

[0049] Figure 1 A schematic diagram of a welding gun swinging to a certain position during a welding process provided in an embodiment of the present application;

[0050] Figure 2 A schematic diagram of the steps of a method for predicting the swing amplitude of a pipeline welding torch based on deep learning provided in an embodiment of the present application;

[0051] Figure 3 A schematic diagram of the structure of an LSTM network provided in an embodiment of the present application;

[0052] Figure 4 A schematic diagram of a process of training a first prediction model and a second prediction model and using the two prediction models to predict the swing amplitude provided in an embodiment of the present application;

[0053] Figure 5 A schematic diagram of the structure of a pipeline welding torch swing amplitude prediction device based on deep learning provided in an embodiment of the present application. DETAILED DESCRIPTION

[0054] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0055] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, wholes, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or combinations thereof.

[0056] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0057] As used in the specification and appended claims of this application, the term "if" can be interpreted as "when" or "uponce" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "uponce it is determined" or "in response to determining" or "uponce [described condition or event] is detected" or "in response to detecting [described condition or event]", depending on the context.

[0058] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0059] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0060] It should be understood that the size of the serial numbers of the steps in the following embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0061] In order to illustrate the technical solution of the present application, a specific embodiment is provided below for illustration.

[0062] The inventors found that in the prior art, during the welding process, the welder observes the spatial position of the groove and the welding gun in real time and adjusts the preset swing amplitude in real time, but this makes the welder's work intensity relatively high and the observation accuracy is also limited. There is also a weld tracking system based on laser vision, which can detect the weld width in real time and adjust the swing amplitude of the welding gun. However, the laser vision weld tracking system needs to add components such as laser light source and camera, which increases the cost and system volume. Moreover, during use, these components are also easily contaminated by dust, welding smoke and welding spatter, which does not meet the inventor's expectations. Based on this, the inventors have made this application after further research and development, providing a method and device for predicting the swing amplitude of a pipeline welding gun based on deep learning.

[0063] Embodiment 1

[0064] The present application embodiment provides a method for predicting the swing amplitude of a pipeline welding torch based on deep learning. In this method, the swing amplitude of the welding torch is determined by predicting the distance between the welding torch and the edge of the groove and the distance between the welding torch and the lower edge of the groove during pipeline welding. Figure 1 As shown, it is a schematic diagram of the welding gun swinging to a certain position during the welding process. The horizontal arrow in the figure indicates the distance between the welding gun and the left and right edges of the groove, which is referred to as the distance between the welding gun and the edge of the groove in this application; the vertical arrow in the figure indicates the distance between the welding gun and the lower edge of the groove.

[0065] The pipeline welding torch swing amplitude prediction method based on deep learning provided in the embodiment of the present application is referred to Figure 2 As shown, the method includes:

[0066] S101: continuously acquiring and preprocessing a welding state data group to obtain a preprocessed welding state data group.

[0067] In the embodiment of the present application, the welding state data group includes welding speed, welding angle, dry extension length, distance between welding gun and groove edge, distance between welding gun and groove lower edge, swing speed, current, voltage, wire feeding speed, gas flow and welding layer; the pre-processed welding state data group is obtained by the following method:

[0068] The welding state data group is cleaned by a normal distribution algorithm to obtain a cleaned welding state data group;

[0069] The welding state data group after cleaning is subjected to frequency synchronization processing according to a preset frequency to obtain the welding state data group after preprocessing.

[0070] In an embodiment of the present application, the status data provided by the sensors installed in each unit of the automatic welding during the swing welding process are continuously collected, such as welding speed, welding angle, dry extension length, swing speed, current, voltage, wire feeding speed, gas flow and weld layer data, and the distance between the welding gun and the edge of the groove and the distance between the welding gun and the lower edge of the groove are determined by calculation, and together with the previous status data, constitute a welding status data group.

[0071] After obtaining the welding state data group, each state data is preprocessed, including filtering, data outlier removal, frequency matching, data standardization, etc. Among them, the specific processing method for data outlier removal is as follows:

[0072] 1. Remove missing values ​​by deleting specific rows. For continuous value types in the welding status data group, use algorithms such as normal distribution and clustering filters to remove noise and abnormal data in the collected data;

[0073] 2. Specifically calculate the mathematical expectation and standard deviation of the normal population, select appropriate probability data to retain, and eliminate other values.

[0074] Since the acquisition frequencies of the sensors installed in each unit of the automatic welding are different, but all welding status data must be processed at the same frequency, the collected welding status data is processed at the same frequency through the preset frequency. The specific method of frequency processing is as follows:

[0075] For the welding status data group including welding speed, welding angle, dry extension length, distance between welding gun and groove edge, distance between welding gun and right edge of groove, distance between welding gun and lower edge of groove, swing speed, current and voltage, the Hilbert filter is used to select the preset frequency for frequency reduction of high-frequency signals; the resampling algorithm is used to perform frequency-homing processing on low-frequency signals. Among them, the preset frequency is set according to the actual situation, and a reasonable frequency is selected for frequency reduction.

[0076] After the welding state data group is subjected to data outlier elimination and frequency isochronous processing in the above manner, a preprocessed welding state data group is obtained.

[0077] S102: Determine a first state data group and a second state data group according to the preprocessed welding state data group.

[0078] In the embodiment of the present application, since the swing amplitude of the welding gun is predicted by the first prediction model and the second prediction model respectively, it is necessary to determine the welding state data groups input into the two models respectively.

[0079] The preprocessed welding state data group includes welding speed, welding angle, dry extension length, distance between welding gun and groove edge, distance between welding gun and groove bottom edge, swing speed, current, voltage, wire feeding speed, gas flow and welding layer. If the distance between welding gun and groove edge at the next moment is to be predicted by the welding state data acquired in real time, the first state data group input into the first prediction model should include welding speed, welding angle, dry extension length, distance between welding gun and groove bottom edge, swing speed, current, voltage, wire feeding speed, gas flow and welding layer; if the distance between welding gun and groove bottom edge at the next moment is to be predicted, the second state data group input into the second prediction model should include welding speed, welding angle, dry extension length, distance between welding gun and groove edge, swing speed, current, voltage, wire feeding speed, gas flow and welding layer.

[0080] In the embodiment of the present application, after determining the first state data group and the second state data group, the method further includes:

[0081] The first state data group and the second state data group are respectively subjected to data standardization processing.

[0082] In an embodiment of the present application, after determining the first state data group and the second state data group, each state data in the two groups can also be standardized. By adopting the standardization with a mean of 0 and a standard deviation of 1, each state data in the first state data group and the second state data group is under the same standard, which is biased towards subsequent model calculations.

[0083] S103: Input the first state data group into a first prediction model to obtain the distance between the welding gun and the groove edge.

[0084] In the embodiment of the present application, the first prediction model is obtained by training in the following manner:

[0085] Collecting multiple sets of historical welding state data sets and preprocessing them to obtain corresponding preprocessed historical welding state data sets; the historical welding state data sets include welding speed, welding angle, dry extension length, distance between welding gun and lower edge of groove, swing speed, current, voltage, wire feeding speed, gas flow and welding layer;

[0086] Constructing a first sample set according to the plurality of pre-processed historical welding state data groups and the corresponding distances between the welding gun and the groove edge;

[0087] Dividing the first sample set into a first training set and a first test set;

[0088] Inputting the first training set into a pre-built initial prediction model for training to obtain a trained prediction model;

[0089] Inputting the first test set into the trained prediction model for testing to obtain a test result of the distance between the welding gun and the groove edge;

[0090] Compare the test result of the distance between the welding gun and the groove edge with the corresponding actual distance between the welding gun and the groove edge, and update the trained prediction model;

[0091] The above training process is repeated until the accuracy of the test result of the distance between the welding gun and the groove edge meets the preset conditions, and the first prediction model is obtained.

[0092] In the embodiment of the present application, during the training process of the first prediction model, the first sample set is first constructed. Multiple groups of historical welding state data sets at different times are collected, including welding speed, welding angle, dry extension length, distance between welding gun and lower edge of groove, swing speed, current, voltage, wire feeding speed, gas flow and welding layer data at different historical times. Since data at different historical times are collected, multiple groups of historical welding state data sets can be regarded as time series data sets with time as a variable. Then, multiple groups of historical welding state data sets are preprocessed. The preprocessing method also includes filtering processing, data outlier removal, frequency isochronization processing, data standardization processing, etc. The specific implementation process can refer to the method of preprocessing the welding state data set in S101, which will not be repeated here.

[0093] A plurality of preprocessed historical welding state data groups are obtained, and the distance between the welding gun and the groove edge corresponding to each historical welding state data group is obtained. The distance between the welding gun and the groove edge is used as the objective function to construct the first sample set as shown in the following formula 1:

[0094] D1={(y1|x1,x2,x3,x4,…,x n )} Formula 1;

[0095] Where D1 is the first sample set; x = {x1, x2, x3, x4, ..., x n} is the historical welding state data set after preprocessing; y1 is the distance between the welding gun and the groove edge.

[0096] After the first sample set is constructed, it is divided into a first training set and a first test set according to a preset ratio, which are used to train the initial prediction model and determine the optimal first prediction model.

[0097] In the embodiment of the present application, the initial prediction model is constructed in the following manner:

[0098] Based on the LSTM neural network, the optimal initial hyperparameters are set to build the initial prediction model.

[0099] In the embodiment of the present application, since the acquired welding state data set has time series, the inventor considers using the recurrent neural network LSTM algorithm as the basic algorithm of the initial prediction model, such as Figure 3The figure shows a schematic diagram of the structure of an LSTM network. The root mean square error and mean absolute error are used to evaluate the prediction model results, and the optimal input layer, hidden layer, activation function and other initial hyperparameters are set to build the model and construct the initial prediction model. In the subsequent training process, the model can be optimized by adjusting the number of LSTM layers, input and output dimensions, number of neurons, batch size, number of epochs, optimizer and learning rate, dropout rate, and input gate, output gate, and forget gate of the initial prediction model.

[0100] The first training set is input into the initial prediction model for training to obtain the trained prediction model, and then the test set is used to test the effect of the trained prediction model in the current stage. According to the comparison between the test results and the actual results, that is, the comparison between the test results of the distance between the welding gun and the edge of the groove and the corresponding actual distance between the welding gun and the edge of the groove, the hyperparameters of the trained prediction model are optimized and updated, and then the optimized and updated model is continued to be trained and tested, and the above training process is repeated until the accuracy of the test results of the distance between the welding gun and the edge of the groove meets the preset conditions. At this time, the hyperparameters of the prediction model are the optimal hyperparameters, thereby obtaining the trained first prediction model.

[0101] The standardized first state data group obtained in S102 is input into the trained first prediction model to obtain a prediction result of the distance between the welding gun and the groove edge at the next moment.

[0102] S104: Input the second state data group into a second prediction model to obtain the distance between the welding gun and the lower edge of the groove.

[0103] In the embodiment of the present application, the second prediction model is obtained by training in the following manner:

[0104] Collecting multiple sets of historical welding state data sets and preprocessing them to obtain corresponding preprocessed historical welding state data sets; the historical welding state data sets include welding speed, welding angle, dry extension length, distance between welding gun and groove edge, swing speed, current, voltage, wire feeding speed, gas flow and welding layer;

[0105] Constructing a second sample set according to the plurality of pre-processed historical welding state data groups and the corresponding distances between the welding gun and the lower edge of the groove;

[0106] Dividing the second sample set into a second training set and a second test set;

[0107] Inputting the second training set into a pre-built initial prediction model for training to obtain a trained prediction model;

[0108] Inputting the second test set into the trained prediction model for testing to obtain a test result of the distance between the welding gun and the lower edge of the groove;

[0109] Compare the test result of the distance between the welding gun and the lower edge of the groove with the corresponding actual distance between the welding gun and the lower edge of the groove, and update the trained prediction model;

[0110] The above training process is repeated until the accuracy of the test result of the distance between the welding gun and the lower edge of the groove meets the preset conditions, and the second prediction model is obtained.

[0111] In the embodiment of the present application, similar to the process of constructing the first prediction model, in the training process of the second prediction model, the second sample set is first constructed. Collect multiple groups of historical welding state data at different times, including welding speed, welding angle, dry extension length, distance between welding gun and groove edge, swing speed, current, voltage, wire feeding speed, gas flow and welding layer data at different times. Since the data collected are data at different historical times, the multiple groups of historical welding state data can be regarded as a time series data set with time as a variable. Then, preprocess the multiple groups of historical welding state data. The preprocessing method also includes filtering processing, data outlier removal, frequency isochronization processing, data standardization processing, etc. The specific implementation process can refer to the method of preprocessing the welding state data group in S101, which will not be repeated here.

[0112] A plurality of preprocessed historical welding state data sets are obtained, and the distance between the welding gun and the lower edge of the groove corresponding to each historical welding state data set is obtained. The distance between the welding gun and the lower edge of the groove is used as the objective function to construct the second sample set as shown in the following formula 2:

[0113] D2={(y2|x1,x2,x3,x4,…,x n )} Formula 2;

[0114] Where D2 is the second sample set; x = {x1, x2, x3, x4, ..., x n} is the historical welding state data set after preprocessing; y2 is the distance between the welding gun and the lower edge of the groove.

[0115] After the second sample set is constructed, it is divided into a second training set and a second test set according to a preset ratio, which are used to train the initial prediction model and determine the optimal second prediction model.

[0116] The second training set is input into the initial prediction model with LSTM as the basic algorithm for training to obtain the trained prediction model. The test set is then used to test the effect of the trained prediction model in the current stage. According to the comparison between the test results and the actual results, that is, the comparison between the test results of the distance between the welding gun and the lower edge of the groove and the corresponding actual distance between the welding gun and the lower edge of the groove, the hyperparameters of the trained prediction model are optimized and updated, and then the optimized and updated model is continued to be trained and tested. The above training process is repeated until the accuracy of the test results of the distance between the welding gun and the lower edge of the groove meets the preset conditions. At this time, the hyperparameters of the prediction model are the optimal hyperparameters, thereby obtaining the trained second prediction model.

[0117] The standardized second state data group obtained in S102 is input into the trained second prediction model to obtain a prediction result of the distance between the welding gun and the lower edge of the groove at the next moment.

[0118] S105: Determine the swing amplitude of the welding gun according to the distance between the welding gun and the groove edge and the distance between the welding gun and the groove lower edge.

[0119] In the embodiment of the present application, the swing amplitude of the welding gun is determined by the distance between the welding gun and the edge of the groove and the distance between the welding gun and the lower edge of the groove. After predicting the distance between the welding gun and the edge of the groove and the distance between the welding gun and the lower edge of the groove at the next moment through the above steps S101 to S104, the swing amplitude of the welding gun at the next moment can be determined to guide the next swing of the welding gun.

[0120] In the present application embodiment, Figure 4As shown, it is a schematic diagram of the process of training the first prediction model and the second prediction model and using the two prediction models to predict the swing amplitude. In the model training process in the figure, the historical welding speed, welding angle, dry extension length, welding gun distance from groove edge, welding gun distance from groove lower edge, swing speed, current, voltage, wire feeding speed, gas flow and welding layer data are first collected. After completing the data collection, the collected multiple groups of historical welding state data groups are preprocessed, including data cleaning, frequency synchronization and standardization, and the first sample set and the second sample set are obtained respectively. The two sample sets are divided into training sets and test sets respectively, which are used to train the optimal first prediction model and the second prediction model, that is, the distance between the welding gun and the groove edge in the figure. A timing model of the distance between the welding gun and the lower edge of the groove and a timing model of the distance between the welding gun and the lower edge of the groove; in the swing amplitude prediction process in the figure, the welding state data group is obtained through the sensor parameters collected in the welding machine, including welding speed, welding angle, dry extension length, distance between the welding gun and the edge of the groove, distance between the welding gun and the lower edge of the groove, swing speed, current, voltage, wire feeding speed, gas flow and weld layer, and data processing, i.e. preprocessing, including data cleaning, frequency synchronization and standardization, is performed to obtain the first state data group and the second state data group respectively, and then input them into the first prediction model and the second prediction model respectively for prediction. After determining the distance between the welding gun and the edge of the groove and the distance between the welding gun and the lower edge of the groove at the next moment, the swing amplitude of the welding gun at the next moment can be determined.

[0121] The deep learning-based pipeline welding torch swing amplitude prediction method provided in the embodiment of the present application continuously acquires the welding state data group and performs preprocessing to obtain the first state data group and the second state data group, and then respectively inputs the trained first prediction model and the second prediction model to predict the distance between the welding torch and the groove edge and the distance between the welding torch and the lower edge of the groove, and finally predicts the swing amplitude of the welding torch. In the pipeline full-position welding, this method can automatically adjust the welding torch swing amplitude in real time according to the actual groove width and welding torch position through two prediction models constructed based on deep learning, and can adapt to the requirements of irregular groove width and various weld thickness changes, and complete the full-position welding of pipelines with any groove width without manual intervention, and no additional components are added to the welding trolley part, reducing costs and system volume.

[0122] Embodiment 2

[0123] Based on the same inventive concept, the embodiment of the present application also provides a pipeline welding torch swing amplitude prediction device based on deep learning, referring to Figure 5 As shown, the device comprises:

[0124] An acquisition module 101 is used to continuously acquire a welding state data group and perform preprocessing to obtain a preprocessed welding state data group;

[0125] A first determining module 102, configured to determine a first state data group and a second state data group according to the preprocessed welding state data group;

[0126] A first prediction module 103, used for inputting the first state data group into a first prediction model to obtain the distance between the welding gun and the groove edge;

[0127] A second prediction module 104, used for inputting the second state data group into a second prediction model to obtain the distance between the welding gun and the lower edge of the groove;

[0128] The second determination module 105 is used to determine the swing amplitude of the welding gun according to the distance between the welding gun and the edge of the groove and the distance between the welding gun and the lower edge of the groove.

[0129] Embodiment 3

[0130] Based on the same inventive concept, an embodiment of the present application also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the method for predicting the swing amplitude of a pipeline welding torch based on deep learning as described in the above embodiment 1 is implemented.

[0131] Embodiment 4

[0132] Based on the same inventive concept, an embodiment of the present application also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for predicting the swing amplitude of a pipeline welding gun based on deep learning as described in the above-mentioned embodiment 1 is implemented.

[0133] Embodiment 5

[0134] Based on the same inventive concept, an embodiment of the present application also provides a computer program product comprising instructions. When the computer program product is run on a computer device, the computer device executes the pipeline welding torch swing amplitude prediction method based on deep learning as described in the above embodiment 1.

[0135] Embodiment 6

[0136] Based on the same inventive concept, an embodiment of the present application also provides a chip, the chip includes a processor and a communication interface, the communication interface and the processor are coupled, and the processor is used to run a computer program or instruction to implement the pipeline welding gun swing amplitude prediction method based on deep learning as described in the above embodiment one.

[0137] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) that contain computer-usable program code.

[0138] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0139] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0140] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0141] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for predicting the swing amplitude of a pipeline welding torch based on deep learning, characterized in that: include: Continuously acquiring and preprocessing the welding state data group to obtain the preprocessed welding state data group; Determining a first state data group and a second state data group according to the preprocessed welding state data group; Inputting the first state data group into a first prediction model to obtain the distance between the welding gun and the groove edge; Inputting the second state data group into a second prediction model to obtain a distance between the welding gun and the lower edge of the groove; The swing amplitude of the welding gun is determined according to the distance between the welding gun and the edge of the groove and the distance between the welding gun and the lower edge of the groove.

2. The method according to claim 1, characterized in that The first prediction model is trained in the following way: Collecting multiple sets of historical welding state data sets and preprocessing them to obtain corresponding preprocessed historical welding state data sets; the historical welding state data sets include welding speed, welding angle, dry extension length, distance between welding gun and lower edge of groove, swing speed, current, voltage, wire feeding speed, gas flow and welding layer; Constructing a first sample set according to the plurality of pre-processed historical welding state data groups and the corresponding distances between the welding gun and the groove edge; Dividing the first sample set into a first training set and a first test set; Inputting the first training set into a pre-built initial prediction model for training to obtain a trained prediction model; Inputting the first test set into the trained prediction model for testing to obtain a test result of the distance between the welding gun and the groove edge; Compare the test result of the distance between the welding gun and the groove edge with the corresponding actual distance between the welding gun and the groove edge, and update the trained prediction model; The above training process is repeated until the accuracy of the test result of the distance between the welding gun and the groove edge meets the preset conditions, and the first prediction model is obtained.

3. The method according to claim 1, characterized in that The second prediction model is trained in the following way: Collecting multiple sets of historical welding state data sets and preprocessing them to obtain corresponding preprocessed historical welding state data sets; the historical welding state data sets include welding speed, welding angle, dry extension length, distance between welding gun and groove edge, swing speed, current, voltage, wire feeding speed, gas flow and welding layer; Constructing a second sample set according to the plurality of pre-processed historical welding state data groups and the corresponding distances between the welding gun and the lower edge of the groove; Dividing the second sample set into a second training set and a second test set; Inputting the second training set into a pre-built initial prediction model for training to obtain a trained prediction model; Inputting the second test set into the trained prediction model for testing to obtain a test result of the distance between the welding gun and the lower edge of the groove; Compare the test result of the distance between the welding gun and the lower edge of the groove with the corresponding actual distance between the welding gun and the lower edge of the groove, and update the trained prediction model; The above training process is repeated until the accuracy of the test result of the distance between the welding gun and the lower edge of the groove meets the preset conditions, and the second prediction model is obtained.

4. The method according to claim 2 or 3, characterized in that The initial prediction model is constructed as follows: Based on the LSTM neural network, the optimal initial hyperparameters are set to build the initial prediction model.

5. The method according to claim 1, characterized in that The welding state data group includes welding speed, welding angle, dry extension length, distance between welding gun and groove edge, distance between welding gun and groove lower edge, swing speed, current, voltage, wire feeding speed, gas flow and welding layer; the pre-processed welding state data group is obtained by the following method: The welding state data group is cleaned by a normal distribution algorithm to obtain a cleaned welding state data group; The welding state data group after cleaning is subjected to frequency synchronization processing according to a preset frequency to obtain the welding state data group after preprocessing.

6. The method according to claim 1, characterized in that After determining the first state data set and the second state data set, the method further includes: The first state data group and the second state data group are respectively subjected to data standardization processing.

7. A pipeline welding torch swing amplitude prediction device based on deep learning, characterized in that: include: An acquisition module is used to continuously acquire a welding state data group and perform preprocessing to obtain a preprocessed welding state data group; A first determination module, configured to determine a first state data group and a second state data group according to the preprocessed welding state data group; A first prediction module, used for inputting the first state data group into a first prediction model to obtain the distance between the welding gun and the groove edge; A second prediction module, used for inputting the second state data group into a second prediction model to obtain the distance between the welding gun and the lower edge of the groove; The second determination module is used to determine the swing amplitude of the welding gun according to the distance between the welding gun and the edge of the groove and the distance between the welding gun and the lower edge of the groove.

8. A computer-readable storage medium, in which a computer program is stored. When the program is executed by a processor, the processor executes the method for predicting the swing amplitude of a pipeline welding torch based on deep learning as described in any one of claims 1 to 6.

9. A computer device, characterized in that: It includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for predicting the swing amplitude of a pipeline welding torch based on deep learning as described in any one of claims 1 to 6 is implemented.

10. A computer program product comprising instructions, which, when executed on a computer device, enables the computer device to execute the method for predicting the swing amplitude of a pipeline welding torch based on deep learning as described in any one of claims 1 to 6.

11. A chip, comprising a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run a computer program or instruction to implement the deep learning-based pipeline welding torch swing amplitude prediction method as described in any one of claims 1 to 6.

Images