A Smart Selection Method for Welding Materials to Match the Strength of Pipeline Circumferential Welds

Abstract

This invention provides an intelligent selection method for welding materials in pipeline circumferential weld strength matching. Applied to the field of welding material selection technology, the method includes: acquiring the physicochemical properties data of the base material, service environment conditions data, and proposed welding process data of the pipeline to be welded; performing vectorization processing to construct a feature vector of the matching conditions; retrieving a historical welding case database, extracting several sets of historical condition feature vectors, and obtaining the measured yield strength of the weld and the measured yield strength of the base material corresponding to each set of historical condition feature vectors; calculating the historical strength matching coefficient to determine qualified and unqualified samples; fusing the material property data of candidate welding materials into the feature vector of the matching conditions to generate candidate input vectors, and inputting the candidate input vectors into a trained intelligent matching evaluation model to output the predicted fit; when the predicted fit is greater than the recommended confidence threshold, the candidate welding material is determined as the recommended welding material. This improves the efficiency of welding material selection.

Description

Technical Field

[0001] This invention relates to the field of welding material selection technology, and in particular to an intelligent method for selecting welding materials to match the strength of pipe circumferential welds. Background Technology

[0002] Intelligent selection of welding materials for matching the strength of pipeline circumferential welds is crucial for high-quality construction of pipeline projects in industries such as oil and gas and chemicals. By integrating multi-dimensional data such as base material performance, operating parameters, and welding processes through algorithms, the strength and toughness indicators of welding materials are precisely matched. This ensures synergy between the mechanical properties of the weld and the base material, reducing the risk of cracking and failure. Simultaneously, intelligent selection improves material selection efficiency, reduces trial-and-error costs, shortens the construction cycle, and enables optimized configuration of welding materials, reducing resource consumption. Furthermore, this technology provides data support for the quality control of pipeline circumferential welds, helping projects meet safety standards and improving the reliability and stability of pipelines in long-term service.

[0003] Currently, the selection of welding materials for existing pipeline circumferential welds requires manual verification of the base material properties, operating parameters, and welding material models one by one. The influence of variables such as welding process and ambient temperature on the actual strength is not considered. The selection relies on experience and trial selection, and repeated verification is time-consuming, resulting in low efficiency in the selection of welding materials.

[0004] Therefore, there is an urgent need for an intelligent selection method for welding materials that can match the strength of pipe circumferential welds with high selection efficiency. Summary of the Invention

[0005] To address the problems existing in the prior art, the present invention provides a method for intelligent selection of welding materials for matching the strength of pipe circumferential welds.

[0006] According to a first aspect of the present invention, a method for intelligent selection of welding materials for strength matching of pipe circumferential welds is provided. The method includes: S1. Obtain the physical and chemical properties data of the base material of the pipeline to be welded, the service environment conditions data, and the proposed welding process data. Perform vectorization processing on the above data to construct the feature vector of the working conditions to be matched. S2, retrieve the historical welding case library, extract several sets of historical working condition feature vectors, and obtain the measured yield strength of the weld and the measured yield strength of the base material corresponding to each set of historical working condition feature vectors; S3. Based on the ratio of the measured yield strength of the weld to the measured yield strength of the base material, calculate the historical strength matching coefficient and determine whether the historical strength matching coefficient is within the preset preferred matching range. S4. If the historical intensity matching coefficient is within the preset preferred matching interval, the corresponding historical working condition feature vector is marked as a qualified sample; if the historical intensity matching coefficient is not within the preset preferred matching interval, it is marked as an unqualified sample. S5. Train an intelligent matching evaluation model based on the qualified samples and the unqualified samples. Calculate the model generalization error during the training process and determine whether the model generalization error is less than the convergence threshold in order to determine whether the model parameters are fixed. S6, the material property data of the candidate welding materials are fused into the feature vector of the working condition to be matched to generate a candidate input vector, and the candidate input vector is input into the trained intelligent matching evaluation model to output the predicted fit. S7. Determine whether the predicted fitness is greater than the recommended confidence threshold. When the predicted fitness is greater than the recommended confidence threshold, determine the candidate welding material as the recommended welding material.

[0007] Further, in S3, determining whether the historical intensity matching coefficient is within a preset preferred matching range includes: If the historical strength matching coefficient is less than the lower limit threshold of the preset preferred matching interval, the sample is determined to be an undermatched unqualified sample, and a first penalty weight is assigned to the sample. If the historical strength matching coefficient is greater than the upper limit threshold of the preset preferred matching interval, the sample is determined to be an over-matched unqualified sample, and a second penalty weight is assigned to the sample. Wherein, the first penalty weight is greater than the second penalty weight.

[0008] Furthermore, the method also includes: Calculate the standardized Z-score value for each data point in the historical welding case library; Determine whether the absolute value of the Z-score standardized value is greater than the anomaly detection threshold. If the absolute value of the Z-score standardized value is greater than the anomaly detection threshold, the data point is determined to be an outlier. Determine whether the feature type to which the outlier belongs is a key mechanical feature. If it is a key mechanical feature, remove the historical cases containing the outlier from the database to avoid misleading model training. If it is not a key mechanical feature, correct the outlier using the mean of the nearest samples.

[0009] Further, in S5, determining whether the model generalization error is less than the convergence threshold includes: If the model's generalization error is greater than or equal to the convergence threshold, the model is determined to be in an underfitting or overfitting state. The model state type is determined based on the difference between the training set error and the validation set error. When the model is determined to be underfitting, the maximum depth parameter of the decision tree is increased. When the model is determined to be overfitting, increase the regularization coefficient. The model is retrained using the adjusted parameters until its generalization error is less than the convergence threshold.

[0010] Furthermore, prior to generating the candidate input vector, the method further includes: Calculate the carbon equivalent of the parent material based on the chemical composition of the parent material in the feature vector of the working condition to be matched; Determine whether the carbon equivalent of the base material is greater than the cold cracking sensitivity threshold. If it is greater than the cold cracking sensitivity threshold, it is determined that there is a preheating requirement in the welding process. Check the diffusible hydrogen content of the candidate welding materials; If preheating is required and the diffusible hydrogen content index is higher than the low hydrogen standard value, the candidate welding material is determined to be incompatible and is directly eliminated without further compatibility prediction.

[0011] Furthermore, after outputting the predicted fitness in S6, the method further includes: Calculate the Euclidean distance between the candidate input vector and each sample vector in the historical welding case library; The smallest Euclidean distance is selected as the working condition variability. Determine whether the operating condition difference is greater than the extrapolation warning threshold. When the operating condition difference is greater than the extrapolation warning threshold, determine that the current prediction belongs to the data extrapolation scenario. The fitness probability of the original model output is weighted using a confidence decay coefficient to obtain the corrected prediction fitness.

[0012] Furthermore, after determining the candidate welding material as the recommended welding material in S7, the method further includes: Obtain the measured historical strength matching coefficient after actual welding tests using the recommended welding materials; Calculate the prediction deviation between the measured historical intensity matching coefficient and the expected historical intensity matching coefficient corresponding to the predicted fit in S6; Determine whether the prediction deviation exceeds the model update threshold; If the model update threshold is exceeded, the working condition data of this experiment and the measured results will be used to generate new training samples. The new training samples are injected into the historical welding case library, and the incremental learning process of the model is triggered to correct the model parameters.

[0013] Furthermore, after determining the candidate welding material as the recommended welding material in S7, the method further includes: Based on the characteristics of the recommended welding materials and the feature vector of the working condition to be matched, the process parameter database is searched; Determine whether there are any historical process records that have a similarity to the recommended welding material and the feature vector of the working condition to be matched that is higher than the process reuse threshold. If it exists, the historical process record will be retrieved directly as the suggested process procedure; If not, the theoretical welding current and voltage range are calculated based on the deposition characteristics of the recommended welding materials, and a new experimental process specification is generated.

[0014] According to a second aspect of the present invention, an electronic device is provided. The electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the program to implement the method.

[0015] According to a third aspect of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the method.

[0016] This invention achieves standardized characterization of welding requirements and improves the accuracy of welding material selection by vectorizing base material, working conditions, and process data and constructing feature vectors; it improves the matching reliability of intelligent models by retrieving historical cases and selecting qualified samples based on strength matching coefficients; and it achieves automated prediction of welding material suitability by training an intelligent matching evaluation model and inputting candidate vectors that integrate welding material attributes, thereby improving the efficiency of welding material selection.

[0017] It should be understood that the description in the Summary of the Invention is not intended to limit the key or essential features of the embodiments of the present invention, nor is it intended to restrict the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0018] The above and other features, advantages, and aspects of the various embodiments of the present invention will become more apparent from the accompanying drawings and the following detailed description. The drawings are provided for a better understanding of the invention and are not intended to limit the scope of the invention. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein: Figure 1 A flowchart is shown below illustrating a method for intelligent selection of welding materials for matching the strength of pipe circumferential welds according to an embodiment of the present invention. Figure 2 A block diagram of an exemplary electronic device capable of implementing embodiments of the present invention is shown. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0021] Figure 1 A flowchart is shown below illustrating a method for intelligent selection of welding materials for matching the strength of pipe circumferential welds according to an embodiment of the present invention. The method includes: S1. Obtain the physical and chemical properties data of the base material of the pipeline to be welded, the service environment conditions data, and the proposed welding process data. Perform vectorization processing on the above data to construct the feature vector of the working conditions to be matched. For example, the pipeline to be welded is an X70 grade natural gas transmission pipeline. The physical and chemical properties of the base material include tensile strength of 570 MPa, yield strength of 480 MPa, elongation of 22%, Brinell hardness of 180 HBW, carbon content of 0.06 wt%, and manganese content of 1.45 wt%. Service environment conditions include a design pressure of 10 MPa, a design temperature of -20℃, a medium of natural gas (98% methane by volume), an annual corrosion rate of 0.01 mm / a, and a pipeline burial depth of 1.5 m. The proposed welding process data includes a flux-cored wire gas shielded welding method, a welding current of 220 A, a welding voltage of 28 V, a welding speed of 15 cm / min, an interpass temperature of 150℃, and a preheating temperature of 80℃. The above quantitative data are assigned values ​​sequentially according to the order of base material properties - service environment - welding process. The feature vector of the matching conditions is [570, 480, 22, 180, 0.06, 1.45, 10, -20, 98, 0.01, 1.5, 220, 28, 15, 150, 80].

[0022] S2, retrieve the historical welding case library, extract several sets of historical working condition feature vectors, and obtain the measured yield strength of the weld and the measured yield strength of the base material corresponding to each set of historical working condition feature vectors; In some embodiments, the method further includes: calculating the Z-score standardized value of each data point in the historical welding case library; determining whether the absolute value of the Z-score standardized value is greater than an anomaly determination threshold; when the absolute value of the Z-score standardized value is greater than the anomaly determination threshold, determining that the data point is an outlier; determining whether the feature type to which the outlier belongs is a key mechanical feature; if it belongs to a key mechanical feature, removing the historical case containing the outlier from the library to avoid misleading model training; if it does not belong to a key mechanical feature, correcting the outlier using the mean of neighboring samples. According to embodiments of the present invention, Z-score standardization unifies the numerical feature dimensions, eliminating the interference of dimensional differences on anomaly detection and improving the accuracy of outlier identification; outlier identification is based on anomaly detection threshold quantification, avoiding subjective judgment and accurately locating abnormal data; outliers are processed differently according to feature importance: cases with outliers of key mechanical features are removed to eliminate core data misleading information and ensure model training quality; non-key feature outliers are corrected to retain the value of case data and reduce data loss; the optimized case library improves data quality, making model training more accurate and welding material matching recommendations more efficient, thereby improving the efficiency of welding material selection.

[0023] S3. Based on the ratio of the measured yield strength of the weld to the measured yield strength of the base material, calculate the historical strength matching coefficient and determine whether the historical strength matching coefficient is within the preset preferred matching range. In some embodiments, in S3, determining whether the historical strength matching coefficient is within a preset preferred matching interval includes: if the historical strength matching coefficient is less than the lower threshold of the preset preferred matching interval, then the sample is determined to be an under-matched unqualified sample, and a first penalty weight is assigned to the sample; if the historical strength matching coefficient is greater than the upper threshold of the preset preferred matching interval, then the sample is determined to be an over-matched unqualified sample, and a second penalty weight is assigned to the sample; wherein, the first penalty weight is greater than the second penalty weight. According to embodiments of the present invention, by differentially assigning penalty weights to under-matched and over-matched samples, the negative impact of different unqualified samples can be more accurately distinguished, reducing the interference of unqualified samples on the matching determination of welding materials, making the determination result more in line with the core requirements for strength matching in actual applications, thereby quickly screening out better welding materials and improving the selection efficiency of welding materials.

[0024] For example, the preset preferred matching interval is set to [0.95, 1.10], with a lower threshold of 0.95 and an upper threshold of 1.10. The first penalty weight is set to 1.8, and the second penalty weight is set to 0.6. Three sets of sample data are extracted from the historical welding case database. The measured yield strength of the weld in sample 1 is 450 MPa, the measured yield strength of the base metal is 480 MPa, and the historical strength matching coefficient is 450 ÷ 480 = 0.9375. Since 0.9375 < 0.95, it is judged as an under-matched and unqualified sample, and a penalty weight of 1.8 is assigned. The measured yield strength of the weld in sample 2 is 55 MPa. The measured yield strength of the base material is 480 MPa, and the historical strength matching coefficient is 550 ÷ 480 ≈ 1.1458. Since 1.1458 > 1.10, it is judged as an over-matched and unqualified sample, and a penalty weight of 0.6 is assigned. The measured yield strength of the weld of sample 3 is 500 MPa, and the measured yield strength of the base material is 480 MPa. The calculated historical strength matching coefficient is 500 ÷ 480 ≈ 1.0417. Since 0.95 ≤ 1.0417 ≤ 1.10, it is judged as a qualified sample, and no penalty weight is assigned.

[0025] S4. If the historical intensity matching coefficient is within the preset preferred matching interval, the corresponding historical working condition feature vector is marked as a qualified sample; if the historical intensity matching coefficient is not within the preset preferred matching interval, it is marked as an unqualified sample. S5. Train an intelligent matching evaluation model based on the qualified samples and the unqualified samples. Calculate the model generalization error during the training process and determine whether the model generalization error is less than the convergence threshold in order to determine whether the model parameters are fixed. In some embodiments, in S5, determining whether the model generalization error is less than the convergence threshold includes: if the model generalization error is greater than or equal to the convergence threshold, determining that the model is in an underfitting or overfitting state; determining the model state type based on the difference between the training set error and the validation set error; when determined to be in an underfitting state, increasing the maximum depth parameter of the model decision tree; when determined to be in an overfitting state, increasing the regularization coefficient of the model; and retraining using the adjusted parameters until the model generalization error is less than the convergence threshold. According to embodiments of the present invention, by real-time determination of the model's underfitting / overfitting state, and targeted adjustment of the maximum depth of the decision tree or the regularization coefficient, the model parameters are quickly optimized and retrained until the generalization error reaches the target, thereby improving the model training convergence speed and prediction accuracy, and thus improving the efficiency of welding material selection.

[0026] For example, the model convergence threshold is set to 0.03, the initial maximum depth parameter of the decision tree is 8, and the regularization coefficient is 0.01. 1000 sets of historical working condition feature vectors are selected as training data, and the training set (700 sets) and validation set (300 sets) are divided in a 7:3 ratio. After training the intelligent matching evaluation model, the generalization error is calculated to be 0.052. Since 0.052 ≥ 0.03, the model is determined not to have met the convergence requirement. The training set error is calculated to be 0.028, and the validation set error is 0.080. The difference between the two is 0.052 (the validation set error is significantly greater than the training set error), indicating that the model is overfitting. The model regularization coefficient is increased from 0.01 to 0.08, while keeping the maximum depth parameter of the decision tree unchanged. The model is retrained using the adjusted parameters. After retraining, the generalization error is calculated to be 0.027. Since 0.027 < 0.03, the convergence requirement is met, training is stopped, and the model parameters are fixed.

[0027] S6, the material property data of the candidate welding materials are fused into the feature vector of the working condition to be matched to generate a candidate input vector, and the candidate input vector is input into the trained intelligent matching evaluation model to output the predicted fit. In some embodiments, before generating the candidate input vector, the method further includes: calculating the carbon equivalent of the base material based on the base material chemical composition in the feature vector of the working condition to be matched; determining whether the carbon equivalent of the base material is greater than the cold crack sensitivity threshold; if it is greater than the cold crack sensitivity threshold, then determining that there is a preheating requirement in the welding process; checking the diffusible hydrogen content index of the candidate welding materials; if there is a preheating requirement and the diffusible hydrogen content index is higher than the low hydrogen standard value, then determining that the candidate welding material is incompatible, directly eliminating the candidate welding material, and not performing subsequent compatibility prediction. According to the embodiments of the present invention, the preheating requirement is determined by the carbon equivalent, locking the core screening dimension of the welding material and reducing invalid compatibility calculations; for working conditions with preheating requirements, incompatible materials are directly eliminated by using the diffusible hydrogen content as a hard index, without entering the subsequent complex compatibility prediction process, greatly reducing the amount of calculation and screening steps; candidate materials that obviously do not meet the process safety requirements are eliminated from the source, narrowing the subsequent matching range, thereby improving the selection efficiency of welding materials.

[0028] For example, the chemical composition data of the base material in the feature vector of the working condition to be matched are: carbon content 0.06wt%, manganese content 1.45wt%, silicon content 0.25wt%, phosphorus content 0.015wt%, sulfur content 0.008wt%, chromium content 0.10wt%, molybdenum content 0.08wt%, and vanadium content 0.05wt%. Using the IIW carbon equivalent calculation formula (CE(IIW)=C+Mn / 6+(Cr+Mo+V) / 5+(Ni+Cu) / 15), the carbon equivalent of the base material CE(IIW) is calculated as: CE(IIW)=0.06 + 1.45 / 6 + (0.10+0.08+0.05) / 5 + 0 / 15≈0.06+ 0.2417 + 0.046 + 0 = 0.3477wt%; the cold crack sensitivity threshold is set to 0.32wt%. Since 0.3477wt% > 0.32wt%, it is determined that preheating is required in the welding process. The low hydrogen standard value is set to 8mL / 100g. The diffusible hydrogen content index of candidate material A is 12mL / 100g. Since 12mL / 100g > 8mL / 100g, it is determined to be incompatible and directly eliminated. The diffusible hydrogen content index of candidate material B is 6mL / 100g. Since 6mL / 100g ≤ 8mL / 100g, it is determined to be compatible and retained for subsequent compatibility prediction. The diffusible hydrogen content index of candidate material C is 9mL / 100g. Since 9mL / 100g > 8mL / 100g, it is determined to be incompatible and directly eliminated.

[0029] In some embodiments, after outputting the prediction fit in S6, the method further includes: calculating the Euclidean distance between the candidate input vector and each sample vector in the historical welding case library; selecting the smallest Euclidean distance as the working condition difference; determining whether the working condition difference is greater than the extrapolation warning threshold; when the working condition difference is greater than the extrapolation warning threshold, determining that the current prediction belongs to the data extrapolation scenario; and using the confidence decay coefficient to weight the fit probability of the original output of the model to obtain the corrected prediction fit. According to embodiments of the present invention, by introducing Euclidean distance to calculate the difference in working conditions, the matching degree between the current welding working conditions and historical cases is quantified, data extrapolation scenarios are quickly identified, and unreliable results are avoided from the model outputting in unsuitable data scenarios, reducing invalid welding material recommendations. Extrapolation scenarios are determined by extrapolation warning thresholds, and the prediction fit is specifically corrected by weighting the confidence decay coefficient, improving the accuracy of prediction results and reducing repeated verification of material selection due to result distortion. The difference calculation, scenario determination, and result correction are completed automatically, replacing the process of manually identifying the working condition matching and adjusting the prediction results, shortening the decision time for welding material selection, and thus improving the efficiency of welding material selection.

[0030] For example, the extrapolation warning threshold is set to 8.5, and the confidence decay coefficient is 0.75; the feature vector of the working condition to be matched is [570,480,22,180,0.06,1.45,10,-20,98,0.01,1.5,220,28,15,150,80], and the material property data of the candidate welding material B are [tensile strength 580MPa, yield strength 490MPa, elongation 23%]. The Brinell hardness is 185 HBW, and the diffusible hydrogen content is 6 mL / 100g. These parameters are fused into the feature vector of the working condition to be matched, generating a candidate input vector of [570, 480, 22, 180, 0.06, 1.45, 10, -20, 98, 0.01, 1.5, 220, 28, 15, 150, 80, 580, 490, 23, 185, 6]. Three sets of sample vectors are extracted from the historical welding case database. Sample vector 1 is [560,470,21,175,0.05,1.40,9,-18,97,0.009,1.4,210,27,14,140,75,570,480,22,180,5], Sample vector 2 is [580,490,23,185,0.07,1.50,11,-22,99,0.011,1.6,230,29,16,160,85,590,500,24,190,7], Sample vector 3 is [550,460,20,170,0.04,1.35,8,-15,96,0.008,1.3,200,26,13,130,70,560,470,21,175,4]; The Euclidean distance between the candidate input vector and sample vector 1 is √[(570-560)²+(480-470)²+...+(6-5)²]≈18.3; the Euclidean distance between the candidate input vector and sample vector 2 is √[(570-580)²+(480-490)²+...+(6-7)²]≈19.1; and the Euclidean distance between the candidate input vector and sample vector 3 is √[(570-550)²+(480-460)²+...+(6-4)²]≈25.7. The minimum Euclidean distance of 18.3 was selected as the working condition difference. Since 18.3 > 8.5, the current prediction is determined to be a data extrapolation scenario. The original output of the intelligent matching evaluation model has a fit probability of 0.85. After weighting with a confidence decay coefficient of 0.75, the corrected prediction fit is 0.85 × 0.75 = 0.6375.

[0031] S7. Determine whether the predicted fitness is greater than the recommended confidence threshold. When the predicted fitness is greater than the recommended confidence threshold, determine the candidate welding material as the recommended welding material.

[0032] In some embodiments, after determining the candidate welding material as the recommended welding material in S7, the method further includes: obtaining the measured historical strength matching coefficient after conducting actual welding tests using the recommended welding material; calculating the prediction deviation between the measured historical strength matching coefficient and the expected historical strength matching coefficient corresponding to the predicted fit in S6; determining whether the prediction deviation exceeds the model update threshold; if it exceeds the model update threshold, generating new training samples from the working condition data and measured results of this test; injecting the new training samples into the historical welding case library and triggering the incremental learning program of the model to correct the model parameters. According to the embodiments of the present invention, the prediction model is corrected by feedback from measured data, reducing the prediction error of subsequent welding material fit and reducing the selection of invalid welding materials due to inaccurate prediction; incremental learning continuously optimizes the model parameters, improves the model's adaptability to different working conditions, and makes the recommendation of candidate welding materials more accurate; new training samples enrich the case library, allowing the model to learn more comprehensive welding working conditions and welding material matching rules, shortening the trial and error cycle of welding material selection, and thus improving the selection efficiency of welding materials.

[0033] In some embodiments, after determining the candidate welding material as the recommended welding material in S7, the method further includes: searching a process parameter database based on the characteristics of the recommended welding material and the feature vector of the working condition to be matched; determining whether there is a historical process record with a similarity higher than the process reuse threshold between the recommended welding material and the feature vector of the working condition to be matched; if there is, directly retrieving the historical process record as a suggested process procedure; if there is, calculating the theoretical welding current and voltage range based on the deposition characteristics of the recommended welding material, and generating a new experimental process procedure.

[0034] For example, setting the process reuse threshold to 90%, the recommended welding material is B, with the following characteristics: tensile strength 580MPa, yield strength 490MPa, elongation 23%, Brinell hardness 185HBW, and diffusible hydrogen content 6mL / 100g; the feature vector of the working condition to be matched is [570,480,22,180,0.06,1.45,10,-20,98,0.01,1.5,220,28,15,150,80]; searching the process parameter database, extracting the corresponding associated data for two sets of historical process records, historical record 1: associated welding material characteristics are tensile strength 575MPa, yield strength 485MPa, elongation 22.5%, Brinell hardness 182HBW, and diffusible hydrogen content 5.5mL / 100g; the associated working condition feature vector is... [568,478,21.8,178,0.058,1.43,9.8,-19,97.5,0.0095,1.48,218,27.5,14.8,148,78]; Calculations show that this record has a 92% similarity to the recommended welding materials and the feature vector of the working condition to be matched; Historical record 2: The associated welding material characteristics are tensile strength 590MPa, yield strength 500MPa, elongation 24%, Brinell hardness 188HBW, and diffusible hydrogen content 7mL / 100g; The associated working condition feature vector is [585,495,2...]. [3.5,183,0.065,1.48,10.5,-21,98.5,0.0105,1.55,225,28.5,15.2,153,82]; After calculation, the similarity between this record and the feature vector of the recommended welding materials and the working conditions to be matched is 85%; Since the similarity of historical record 1 is 92%>90%, its corresponding historical process record is directly retrieved as the suggested process procedure, which includes welding current 215A, welding voltage 27V, welding speed 14.5cm / min, interpass temperature 145℃, and preheating temperature 78℃; If there are no historical process records with a similarity higher than 90% in the process parameter database, and the known deposition characteristics of recommended welding material B are 85% deposition efficiency and 7.8 g / cm³ deposition metal density, based on the parameters of the pipe wall thickness of 12 mm and the weld joint gap of 2 mm in the working condition to be matched, the theoretical welding current range is calculated to be 210A-230A and the theoretical welding voltage range is 26V-29V. A new experimental process specification is generated, including welding current of 220A, welding voltage of 27.5V, welding speed of 15cm / min, interpass temperature of 150℃, and preheating temperature of 80℃.

[0035] According to embodiments of the present invention, the present invention also provides an electronic device and a readable storage medium.

[0036] Figure 2 A schematic block diagram of an electronic device that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0037] The electronic device includes a computing unit 201, which can perform various appropriate actions and processes based on a computer program stored in ROM 202 or a computer program loaded into RAM 203 from storage unit 208. RAM 203 can also store various programs and data required for the operation of the electronic device. The computing unit 201, ROM 202, and RAM 203 are interconnected via bus 204. I / O interface 205 is also connected to bus 204.

[0038] Multiple components in the electronic device are connected to the I / O interface 205, including: an input unit 206, such as a keyboard, mouse, etc.; an output unit 207, such as various types of displays, speakers, etc.; a storage unit 208, such as a disk, optical disk, etc.; and a communication unit 209, such as a network card, modem, wireless transceiver, etc. The communication unit 209 allows the electronic device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0039] The computing unit 201 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 201 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 201 performs the various methods and processes described above, such as the intelligent selection method for welding materials for matching the strength of pipe circumferential welds. For example, in some embodiments, the intelligent selection method for welding materials for matching the strength of pipe circumferential welds can be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as storage unit 208. In some embodiments, part or all of the computer program can be loaded and / or installed on an electronic device via ROM 202 and / or communication unit 209. When the computer program is loaded into RAM 203 and executed by the computing unit 201, one or more steps of the intelligent selection method for welding materials for matching the strength of pipe circumferential welds described above can be performed. Alternatively, in other embodiments, the computing unit 201 may be configured by any other suitable means (e.g., by means of firmware) to perform a method for intelligent selection of welding materials for matching the strength of pipe circumferential welds.

[0040] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0041] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0042] In the context of this invention, a readable storage medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A readable storage medium can be a machine-readable signal medium or a machine-readable storage medium. A readable storage medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0043] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including voice input, speech input, or tactile input).

[0044] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0045] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0046] It should be understood that the various processes described above can be used to rearrange, add, or delete steps. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0047] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for intelligent selection of welding material for pipe girth weld strength matching, characterized by, include: S1. Obtain the physical and chemical properties data of the base material of the pipeline to be welded, the service environment conditions data, and the proposed welding process data. Perform vectorization processing on the above data to construct the feature vector of the working conditions to be matched. S2, retrieve the historical welding case library, extract several sets of historical working condition feature vectors, and obtain the measured yield strength of the weld and the measured yield strength of the base material corresponding to each set of historical working condition feature vectors; S3. Based on the ratio of the measured yield strength of the weld to the measured yield strength of the base material, calculate the historical strength matching coefficient and determine whether the historical strength matching coefficient is within the preset preferred matching range. S4. If the historical intensity matching coefficient is within the preset preferred matching interval, then the corresponding historical working condition feature vector is marked as a qualified sample. If the historical intensity matching coefficient is not within the preset preferred matching range, it is marked as a non-qualified sample; S5. Train an intelligent matching evaluation model based on the qualified samples and the unqualified samples. Calculate the model generalization error during the training process and determine whether the model generalization error is less than the convergence threshold in order to determine whether the model parameters are fixed. S6, the material property data of the candidate welding materials are fused into the feature vector of the working condition to be matched to generate a candidate input vector, and the candidate input vector is input into the trained intelligent matching evaluation model to output the predicted fit. S7. Determine whether the predicted fitness is greater than the recommended confidence threshold. When the predicted fitness is greater than the recommended confidence threshold, determine the candidate welding material as the recommended welding material.

2. The pipe girth weld strength- matching welding material intelligent selection method according to claim 1, characterized in that, In S3, determining whether the historical intensity matching coefficient is within a preset preferred matching range includes: If the historical strength matching coefficient is less than the lower limit threshold of the preset preferred matching interval, the sample is determined to be an undermatched unqualified sample, and a first penalty weight is assigned to the sample. If the historical strength matching coefficient is greater than the upper limit threshold of the preset preferred matching interval, the sample is determined to be an over-matched unqualified sample, and a second penalty weight is assigned to the sample. Wherein, the first penalty weight is greater than the second penalty weight.

3. The pipe girth weld strength- matched welding material intelligent selection method according to claim 2, characterized by, The method further includes: Calculate the standardized Z-score value for each data point in the historical welding case library; Determine whether the absolute value of the Z-score standardized value is greater than the anomaly detection threshold. If the absolute value of the Z-score standardized value is greater than the anomaly detection threshold, the data point is determined to be an outlier. Determine whether the feature type to which the outlier belongs is a key mechanical feature. If it is a key mechanical feature, remove the historical cases containing the outlier from the database to avoid misleading model training. If it is not a key mechanical feature, correct the outlier using the mean of the nearest samples.

4. The pipe girth weld strength- matching welding material intelligent selection method according to claim 3, characterized in that, In S5, determining whether the model generalization error is less than the convergence threshold includes: If the model's generalization error is greater than or equal to the convergence threshold, the model is determined to be in an underfitting or overfitting state. The model state type is determined based on the difference between the training set error and the validation set error. When the model is determined to be underfitting, the maximum depth parameter of the decision tree is increased. When the model is determined to be overfitting, increase the regularization coefficient. The model is retrained using the adjusted parameters until its generalization error is less than the convergence threshold.

5. The pipe girth weld strength- matching welding material intelligent selection method according to claim 4, characterized in that, Prior to generating the candidate input vector, the method further includes: Calculate the carbon equivalent of the parent material based on the chemical composition of the parent material in the feature vector of the working condition to be matched; Determine whether the carbon equivalent of the base material is greater than the cold cracking sensitivity threshold. If it is greater than the cold cracking sensitivity threshold, it is determined that there is a preheating requirement in the welding process. Check the diffusible hydrogen content of the candidate welding materials; If preheating is required and the diffusible hydrogen content index is higher than the low hydrogen standard value, the candidate welding material is determined to be incompatible and is directly eliminated without further compatibility prediction.

6. The pipe girth weld strength- matched welding material intelligent selection method according to claim 5, characterized by, After outputting the predicted fitness in S6, the method further includes: Calculate the Euclidean distance between the candidate input vector and each sample vector in the historical welding case library; The smallest Euclidean distance is selected as the working condition variability. Determine whether the operating condition difference is greater than the extrapolation warning threshold. When the operating condition difference is greater than the extrapolation warning threshold, determine that the current prediction belongs to the data extrapolation scenario. The fitness probability of the original model output is weighted using a confidence decay coefficient to obtain the corrected prediction fitness.

7. The pipe girth weld strength- matched welding material intelligent selection method according to claim 6, characterized by, After determining the candidate welding material as the recommended welding material in S7, the method further includes: Obtain the measured historical strength matching coefficient after actual welding tests using the recommended welding materials; Calculate the prediction deviation between the measured historical intensity matching coefficient and the expected historical intensity matching coefficient corresponding to the predicted fit in S6; Determine whether the prediction deviation exceeds the model update threshold; If the model update threshold is exceeded, the working condition data of this experiment and the measured results will be used to generate new training samples. The new training samples are injected into the historical welding case library, and the incremental learning process of the model is triggered to correct the model parameters.

8. The pipe girth weld strength- matching welding material intelligent selection method according to claim 6, characterized in that, After determining the candidate welding material as the recommended welding material in S7, the method further includes: Based on the characteristics of the recommended welding materials and the feature vector of the working condition to be matched, the process parameter database is searched; Determine whether there are any historical process records that have a similarity to the recommended welding material and the feature vector of the working condition to be matched that is higher than the process reuse threshold. If it exists, the historical process record will be retrieved directly as the suggested process procedure; If not, the theoretical welding current and voltage range are calculated based on the deposition characteristics of the recommended welding materials, and a new experimental process specification is generated.

9. An electronic device, comprising: include: At least one processor; A memory that is communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the method described in any one of claims 1-8.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-8.

Images