Self-learning method and device for quantitatively representing welding parameters, equipment and storage medium
By constructing a neural network model and a welding parameter expert system, welding parameters are quantitatively characterized, solving the problem that manual arc welding cannot be de-skilled, and realizing automated management of welding parameters and high-quality robotic welding.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING INSTITUTE OF PETROCHEMICAL TECHNOLOGY
- Filing Date
- 2022-10-12
- Publication Date
- 2026-04-21
AI Technical Summary
Manual arc welding cannot be decoupled from skill, welding parameters are difficult to use rationally, and welding experience and techniques are difficult to pass on, resulting in a limited number of experienced welders and a long training cycle.
By constructing a neural network model and a welding parameter expert system, welding process parameters are quantitatively characterized. The neural network model is used to train and optimize the parameters, which are then input into the welding parameter expert system for autonomous judgment and updating, thereby achieving automated management of welding parameters.
It has enabled the automation and intelligentization of manual arc welding, improved the accuracy and efficiency of welding quality prediction, broken through the bottleneck of welding experience and technology dissemination, and supported high-quality automated robotic welding.
Smart Images
Figure CN115815751B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of welding robot technology, and in particular to a self-learning method, apparatus, equipment and storage medium for the quantitative characterization of welding parameters. Background Technology
[0002] Manual arc welding is characterized by its high flexibility, simple equipment, strong operability, and high welding quality, and is widely used in various industries, and will continue to play a role in the foreseeable future. However, the welding quality of manual arc welding depends on the welder's skill level, and the number of experienced and skilled welders is limited, the training period is long, and due to various objective factors, their welding experience and skills are difficult to pass on to others.
[0003] In other words, existing manual arc welding technology cannot be de-skilled and cannot make reasonable use of the obtained welding parameters. This results in a limited number of experienced welders with good welding skills, a long training cycle, and difficulty in passing on welding experience and skills to others. Summary of the Invention
[0004] To overcome, to some extent, the problem in related technologies that manual arc welding cannot be de-skilled and that the obtained welding parameters can be used reasonably, this application provides a self-learning method, apparatus, equipment, and storage medium for the quantitative characterization of welding parameters.
[0005] The proposed solution is as follows:
[0006] In a first aspect, this application provides a self-learning method for the quantitative characterization of welding parameters, the method comprising:
[0007] Obtain relevant parameters for the welding process;
[0008] The welding-related parameters are input into a pre-built neural network model to obtain the optimal welding parameters.
[0009] Using the preferred parameters, input into a pre-built welding parameter expert system, we can obtain weld quality prediction schemes for different welding parameters.
[0010] Furthermore, the welding-related parameters include: welding parameters, spatial position parameters, and environmental parameters;
[0011] The welding parameters include: welding voltage, welding current, welding speed, wire extension length, and shielding gas flow rate;
[0012] The spatial position parameters include: welding torch end position parameters and welding torch attitude parameters;
[0013] The environmental parameters include: environmental pressure parameters, environmental humidity parameters, and environmental temperature parameters.
[0014] Furthermore, the step of using the welding-related parameters and inputting them into a pre-built neural network model to obtain optimal welding parameters includes:
[0015] S1. Construct a neural network model;
[0016] S2. Using the neural network model, train it to obtain the trained neural network model;
[0017] S3. Input the welding-related parameters into the trained neural network model to obtain the optimal welding parameters.
[0018] Further, the neural network model is constructed, including:
[0019] A neural network model was built using the TensoFlow framework, employing the cross-entropy cost function, mini-batch gradient descent algorithm, Sigmoid activation function, L2 regularization, and Adam optimizer. Each batch iterated 30 times, with a maximum of 1500 iterations.
[0020] Furthermore, the neural network model also includes:
[0021] The input layer includes: welding voltage, welding current, welding speed, shielding gas flow rate, shielding gas type, welding torch end position, weld spatial position, ambient pressure parameters, ambient humidity parameters, and ambient temperature parameters.
[0022] The hidden layer includes: welding arc, droplet transfer, temperature field, molten pool state, metal crystallization, secondary phase transformation, mass transfer, heat transfer, forces, and welding magnetic field;
[0023] The output layer includes: weld appearance quality, weld joint mechanical properties, and joint microstructure.
[0024] Furthermore, the neural network model is trained to obtain the trained neural network model, including:
[0025] S1. Evaluate and test the relevant parameters of the welding process to obtain the preferred parameters of the welding process. The evaluation and testing include: evaluating the appearance of the weld, testing its mechanical properties, and observing its microstructure.
[0026] S2. The preferred parameters of the welding process are used as input variables for two-layer loop training. By importing the library, downloading the parameter set, and standardizing the data, the prediction confidence of the trained neural network is not less than 0.95.
[0027] Furthermore, the process of inputting the preferred parameters into a pre-built welding parameter expert system to obtain an automated welding process solution includes:
[0028] Using the trained neural network model, the optimized parameters are input into the welding parameter expert system as the input layer to perform autonomous judgment and screening of welding parameters, realize the autonomous learning and updating of the welding parameter expert system, and obtain weld quality prediction schemes for different welding parameters.
[0029] Secondly, this application provides a self-learning device for quantitative characterization of welding parameters, the device comprising:
[0030] The parameter acquisition module is used to obtain relevant parameters for the welding process.
[0031] The screening module is used to input the welding-related parameters into a pre-built neural network model to obtain the preferred welding parameters.
[0032] The prediction module is used to input the preferred parameters into a pre-built welding parameter expert system to obtain a weld quality prediction scheme for different welding parameters.
[0033] Thirdly, this application provides a self-learning device for the quantitative characterization of welding parameters, the device comprising:
[0034] Memory, on which executable programs are stored;
[0035] A processor for executing the executable program in the memory to implement the steps of any of the methods described above.
[0036] Fourthly, this application provides a computer-readable storage medium storing computer instructions for causing a computer to perform the steps of any of the methods described above.
[0037] The technical solution provided in this application may include the following beneficial effects:
[0038] This application obtains welding process-related parameters; inputs these parameters into a pre-built neural network model to obtain optimal welding parameters; and inputs these optimal parameters into a pre-built welding parameter expert system to obtain weld quality prediction schemes for different welding parameters. This application contributes to the de-skilling of manual arc welding by obtaining detailed welding parameters and rationally applying these parameters to obtain weld quality prediction schemes for different welding parameters, providing a process solution for high-quality automated robotic welding under complex working conditions.
[0039] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description
[0040] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.
[0041] Figure 1 This is a flowchart of a self-learning method for quantitative characterization of welding parameters provided in one embodiment of this application;
[0042] Figure 2 This is a diagram of the self-learning device for quantitative characterization of welding parameters provided in another embodiment of this application;
[0043] Figure 3 This is a diagram of the self-learning device for quantitative characterization of welding parameters provided in another embodiment of this application;
[0044] Figure 4 This is a schematic diagram of the neural network model structure for the quantitative characterization of welding parameters provided in another embodiment of this application. Detailed Implementation
[0045] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.
[0046] Example 1
[0047] Please see Figure 1 , Figure 1 This is a flowchart of a self-learning method for quantitative characterization of welding parameters provided in one embodiment of this application. The method includes:
[0048] S1. Obtain relevant parameters for the welding process;
[0049] S2. Using the welding-related parameters, input them into a pre-built neural network model to obtain the optimal welding parameters;
[0050] S3. Using the preferred parameters, input them into a pre-built welding parameter expert system to obtain weld quality prediction schemes for different welding parameters.
[0051] The welding process parameters obtained according to step S1 include:
[0052] Multiple welding experts carried out welding operations on typical welding structures, collecting and recording parameters for each welding process, including welding parameters, spatial position parameters, and environmental parameters;
[0053] In specific implementation, the quantitative characterization methods for the welding parameters, spatial position parameters, and environmental parameters are as follows: welding voltage u(t) and welding current i(t) can be obtained directly from the welding machine; welding speed v(t) is obtained by recording the welding start / end time and combining it with the weld length; welding wire extension length e(t) is measured before welding begins; shielding gas flow rate g(t) is measured by a gas flow meter; welding torch end position parameters x(t), y(t), z(t) and attitude parameters α(t), β(t), γ(t) can be measured by an attitude sensor mounted on the welding torch; environmental pressure p(t) can be measured by an environmental barometer; environmental humidity h(t) can be measured by a hygrometer; and environmental temperature k(t) can be measured by a thermometer.
[0054] According to step S2, the welding-related parameters are input into a pre-built neural network model to obtain optimal welding parameters, including:
[0055] S21. Construct a neural network model;
[0056] In this embodiment, constructing the neural network model includes:
[0057] A neural network model was built using the TensoFlow framework, employing the cross-entropy cost function, mini-batch gradient descent algorithm, Sigmoid activation function, L2 regularization, and Adam optimizer. Each batch iterated 30 times, with a maximum of 1500 iterations.
[0058] Specifically, a neural network model is built based on the TensorFlow framework, using the cross-entropy cost function: Increase training speed, where x represents the number of samples and n represents the total number of samples; mini-batch gradient descent algorithm: To save computational resources, f(x; θ) represents a deep neural network, where θ represents the network parameters and the Lpgistic activation function. "Squeeze" the input from the real number field into (0, 1), using L2 regularization: To avoid overfitting and improve generalization ability, where λ is the regularization coefficient (set to 0.8), m is the number of samples, w is the number of samples, and Adam optimizer: m t =β1m t-1 +(1-β1)g t m0 is initialized to 0, β1 is the exponential decay rate, which controls the weight allocation and has a value of 0.9. Each batch iterates 30 times, and the maximum number of iterations is 1500.
[0059] In practical implementation, the neural network model further includes: an input layer, a hidden layer, and an output layer, as detailed below. Figure 4 As shown;
[0060] The input layer includes: welding voltage, welding current, welding speed, shielding gas flow rate, shielding gas type, welding torch end position, weld spatial position, ambient pressure parameters, ambient humidity parameters, and ambient temperature parameters.
[0061] The hidden layer includes: welding arc, droplet transfer, temperature field, molten pool state, metal crystallization, secondary phase transformation, mass transfer, heat transfer, force, and welding magnetic field;
[0062] The output layer includes: weld appearance quality, weld joint mechanical properties, and joint microstructure.
[0063] S22. Using the neural network model, train it to obtain the trained neural network model;
[0064] S22.1. Evaluate and test the relevant parameters of the welding process to obtain the preferred parameters of the welding process. The evaluation and testing include: evaluating the appearance of the weld, testing its mechanical properties, and observing its microstructure.
[0065] Specifically, multiple welding experts carry out welding operations on typical welding structures, collecting and recording parameters for each welding process, including welding parameters, spatial position parameters, and environmental parameters;
[0066] The evaluation and inspection of the welded workpiece mainly includes weld appearance evaluation, mechanical property testing and microstructure observation to obtain the comprehensive performance of the weld.
[0067] The collected welding process parameters are screened, and the welding parameters of each weld are evaluated based on the obtained comprehensive performance of the weld. The welding parameters of the weld with good comprehensive performance will be retained to obtain the optimal welding process parameters and build a welding parameter expert system.
[0068] It should be noted that in the comprehensive performance evaluation of the weld, the weld appearance evaluation includes the evaluation of weld formation and appearance defects, mainly to quantitatively evaluate the quality of weld formation and the number of appearance defects; the mechanical property testing includes the testing of tensile strength, hardness, bending strength, impact toughness, etc. of the weld, so as to obtain quantitative data; the microstructure observation is to observe the metallographic structure of the weld and the heat-affected zone, and to corroborate the metallographic structure of the weld and the heat-affected zone with the mechanical properties.
[0069] S22.2. The preferred parameters of the welding process are used as input variables for double-layer loop training. By importing the library, downloading the parameter set, and standardizing the data, the prediction confidence of the trained neural network is not less than 0.95.
[0070] Specifically, the neural network is trained using optimized parameters (more than n records) as input variables, ensuring that the prediction confidence of the trained neural network is not less than 0.95. The trained neural network is then used as a prediction and evaluation tool to autonomously judge and filter newly acquired welding parameters, enabling the welding expert system to learn and update autonomously.
[0071] The aforementioned multi-input multi-output neural network has the following input layers: welding voltage, welding current, welding speed, shielding gas flow rate, shielding gas type, welding torch end pose, weld spatial location, ambient pressure, temperature, and humidity; hidden layers: welding arc, droplet transfer, temperature field, molten pool state, metal crystallization, secondary phase transformation, mass transfer, heat transfer, force, and welding magnetic field; and output layers: weld appearance quality (weld formation, appearance defects), weld joint mechanical properties (tensile strength, hardness, flexural strength, impact toughness, weld zone microstructure, heat-affected zone microstructure), and joint microstructure (weld zone microstructure, heat-affected zone microstructure).
[0072] S23. Input the welding-related parameters into the trained neural network model to obtain the optimal welding parameters:
[0073] Specifically, a multi-input multi-output neural network is constructed, and the optimized parameters (n records) are used as input variables for training. The trained neural network is used as a means of predicting weld quality under different welding parameters, thereby enabling autonomous judgment and screening of newly acquired welding parameters.
[0074] According to step S3, the preferred parameters are input into a pre-built welding parameter expert system to obtain weld quality prediction schemes for different welding parameters, including:
[0075] Using the trained neural network model, the optimized parameters are input into the welding parameter expert system as the input layer to perform autonomous judgment and screening of welding parameters, realize the autonomous learning and updating of the welding parameter expert system, and obtain weld quality prediction schemes for different welding parameters.
[0076] In one embodiment, the expert system self-learning method utilizes the expert system's automatic learning and continuously expanding welding process parameter library to provide ideal welding process parameters for robotic welding.
[0077] Specifically, the advantages of this invention compared to existing technologies are as follows: it fills the gaps in existing technologies, proposes a quantitative characterization method for arc welding parameters and an expert system learning method, clarifies the quantitative design method and steps for manual arc welding experience parameters, improves the accuracy and intelligence of my country's use of manual arc welding to guide welding robots, breaks through the problem of the difficulty in recording and disseminating manual arc welding technical experience, and provides ideal welding process parameters for robot welding by utilizing the automatic learning of expert systems and the continuously enriched welding process parameter library.
[0078] Example 2
[0079] Please see Figure 2 , Figure 2 This is a diagram illustrating the composition of a self-learning device for quantifying and characterizing welding parameters according to another embodiment of this application. The device includes:
[0080] Parameter acquisition module 101 is used to obtain welding process related parameters;
[0081] The screening module 102 is used to input the welding-related parameters into a pre-built neural network model to obtain the preferred welding parameters.
[0082] The prediction module 103 is used to input the preferred parameters into a pre-built welding parameter expert system to obtain a weld quality prediction scheme for different welding parameters.
[0083] Example 3
[0084] Please see Figure 3 , Figure 3 This is a diagram illustrating the composition of a self-learning device for quantitative characterization of welding parameters, provided in another embodiment of this application. The device includes:
[0085] Memory 31, on which an executable program is stored;
[0086] Processor 32 is configured to execute the executable program in the memory 31 to implement the steps of any of the methods described above.
[0087] Furthermore, this application provides a computer-readable storage medium storing computer instructions for causing a computer to perform the steps of any of the methods described above. The storage medium may be a magnetic disk, optical disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk drive (HDD), or solid-state drive (SSD), etc.; the storage medium may also include combinations of the above types of memory.
[0088] It is understood that the same or similar parts in the above embodiments can be referred to each other, and the contents not described in detail in some embodiments can be referred to the same or similar contents in other embodiments.
[0089] It should be noted that in the description of this application, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, in the description of this application, unless otherwise stated, "a plurality of" means at least two.
[0090] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the function involved, as will be understood by those skilled in the art to which embodiments of this application pertain.
[0091] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.
[0092] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.
[0093] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.
[0094] The storage media mentioned above can be read-only memory, disk, or optical disk, etc.
[0095] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.
[0096] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.
Claims
1. A self-learning method of quantitatively characterizing welding parameters, characterized in that, The method includes: Obtain relevant parameters for the welding process; The welding process-related parameters are input into a pre-built neural network model to obtain the optimal welding parameters. Using the preferred parameters, input them into a pre-built welding parameter expert system to obtain weld quality prediction schemes for different welding parameters; The welding process-related parameters include: welding parameters, spatial position parameters, and environmental parameters; The welding parameters include: welding voltage, welding current, welding speed, wire extension length, and shielding gas flow rate; The spatial position parameters include: welding torch end position parameters and welding torch attitude parameters; The environmental parameters include: environmental pressure parameters, environmental humidity parameters, and environmental temperature parameters; The process of inputting welding process-related parameters into a pre-built neural network model to obtain optimal welding parameters includes: S1. Construct a neural network model; S2. Using the neural network model, train it to obtain the trained neural network model; S3. Input the welding process-related parameters into the trained neural network model to obtain the optimal welding parameters; The neural network model also includes: The input layer includes: welding voltage, welding current, welding speed, shielding gas flow rate, shielding gas type, welding torch end position, weld spatial position, ambient pressure parameters, ambient humidity parameters, and ambient temperature parameters. The hidden layer includes: welding arc, droplet transfer, temperature field, molten pool state, metal crystallization, secondary phase transformation, mass transfer, heat transfer, forces, and welding magnetic field; The output layer includes: weld appearance quality, weld joint mechanical properties, and joint microstructure; The neural network model is trained to obtain the trained neural network model, including: S1. Evaluate and test the relevant parameters of the welding process to obtain the preferred parameters of the welding process. The evaluation and testing include: evaluating the appearance of the weld, testing its mechanical properties, and observing its microstructure. S2. The preferred parameters of the welding process are used as input variables for two-layer loop training. By importing the library, downloading the parameter set, and standardizing the data, the prediction confidence of the trained neural network is not less than 0.
95. The process involves inputting the preferred parameters into a pre-built welding parameter expert system to obtain weld quality prediction schemes for different welding parameters, including: Using the trained neural network model, the optimized parameters are input into the welding parameter expert system as the input layer to perform autonomous judgment and screening of welding parameters, realize the autonomous learning and updating of the welding parameter expert system, and obtain weld quality prediction schemes for different welding parameters.
2. The method of claim 1, wherein, Constructing the neural network model includes: A neural network model was built using the TensoFlow framework, employing the cross-entropy cost function, mini-batch gradient descent algorithm, Sigmoid activation function, L2 regularization, and Adam optimizer. Each batch iterated 30 times, with a maximum of 1500 iterations.
3. The self-learning device for quantitatively characterizing welding parameters, applied to the self-learning method for quantitatively characterizing welding parameters according to any one of claims 1-2, characterized in that, The device includes: The parameter acquisition module is used to obtain relevant parameters for the welding process. The screening module is used for inputting the welding process related parameters into a pre-constructed neuron network model to obtain welding preferred parameters; The prediction module is used for inputting the preferred parameters into a pre-constructed welding parameter expert system to obtain different welding parameter weld quality prediction schemes.
4. Self-learning device for quantitatively characterizing welding parameters, characterized in that The device comprises: a memory having an executable program stored thereon; a processor configured to execute the executable program in the memory to implement the steps of the method of any one of claims 1-2.
5. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer instructions for causing a computer to execute the steps of the method of any one of claims 1-2.