Weld joint quality prediction and process optimization method based on welding cloud platform

Abstract

The invention provides a welding seam quality prediction and process optimization method based on a welding cloud platform. The initial optimization parameters of the weld joint process parameters are determined on the basis of the weld joint quality data and in combination with the preset target requirements, it is ensured that the optimization direction meets the actual production requirements, a scientific basis is provided for real-time adjustment of welding equipment, real-time optimization is conducted through a welding equipment controller according to the initial optimization parameters, and the welding efficiency is improved. The current welding seam quality data in the optimization process are collected in real time and fed back to the welding cloud platform to update the welding seam quality prediction model, quality fluctuation in the welding process is quickly responded, defect expansion is avoided, the welding quality stability is improved, and the welding quality prediction efficiency is improved. The current welding seam quality data in the optimization process is fed back to the cloud platform, and the model is updated, so that the model can continuously adapt to the change of the welding scene, the prediction and optimization accuracy is gradually improved, and the iterative upgrading of the technical scheme is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of welding intelligent manufacturing, in particular to a welding seam quality prediction and process optimization method based on a welding cloud platform. BACKGROUND

[0002] In modern equipment manufacturing industries such as steel structure manufacturing, the welding quality directly determines the load-carrying capacity and service life of the components. In particular, in large-scale engineering construction projects, welding defects may cause serious safety hazards. Traditional welding quality control mainly relies on the experience judgment of operators and post-detection, which has problems such as response lag, low precision, and large blindness in process adjustment, and is difficult to meet the needs of large-scale and intelligent production.

[0003] With the development of cloud computing and artificial intelligence technology, welding quality control technology based on cloud platform has gradually become a research hotspot. In the prior art, there are related patents that have proposed application schemes of welding cloud platform, for example, the Chinese invention patent with the publication number CN113960114B, "Welding process quality online analysis system and method based on cloud server", which realizes the processing and analysis of electrical signals such as welding current and voltage through a cloud server, and provides characteristic index parameters for quality analysis. However, this technology only focuses on online analysis of data, and does not realize advanced prediction of welding seam quality and active optimization of process parameters, lacking a closed-loop control mechanism. For another example, the Chinese invention patent with the publication number CN118875566A, "Steel structure welding process quality evaluation method based on big data processing", which adopts multi-modal data fusion to construct a quality evaluation model. However, this method does not fully utilize the computing power advantage of the cloud platform, has low model training efficiency, and the process optimization strategy is single, which is difficult to adapt to the individual needs of different welding scenarios. SUMMARY

[0004] The present application provides a welding seam quality prediction and process optimization method based on a welding cloud platform to solve the problems raised in the background art.

[0005] A welding seam quality prediction and process optimization method based on a welding cloud platform, comprising: S1: Collecting welding process parameters, environmental parameters, and welding seam feature data based on pre-deployed sensors, and obtaining multi-source data after data preprocessing of the welding process parameters, environmental parameters, and welding seam feature data, and uploading the multi-source data to a welding cloud platform; S2: Inputting the multi-source data into a welding seam quality prediction model pre-trained based on the welding cloud platform to obtain welding seam quality data; S3: Determining initial optimization parameters for welding seam process parameters based on the welding seam quality data and combining with preset target requirements; S4: The welding equipment controller performs real-time optimization according to the initial optimization parameters, and collects current welding quality data in the optimization process in real time, and feeds back the current welding quality data to the welding cloud platform to update the welding quality prediction model.

[0006] Preferably, in S1, after data preprocessing of the welding process parameters, environmental parameters and weld feature data, multi-source data is obtained, including: The welding process parameters, environmental parameters and weld feature data are subjected to abnormal data rejection, missing value supplementation and repeated data deduplication processing to obtain processed data. The processed data is subjected to data standardization processing to obtain multi-source data.

[0007] Preferably, in S1, the multi-source data is uploaded to the welding cloud platform, including: The multi-source data is encapsulated and format-unified by deploying an edge computing node near the welding station, and the multi-source data is encapsulated into a data frame; Based on the communication link in the edge computing node, the data frame is sent to the dedicated data receiving port of the welding cloud platform.

[0008] Preferably, in S2, the multi-source data is input into the welding quality prediction model pre-trained based on the welding cloud platform to obtain the welding quality data, including: According to a preset timing rule, the multi-source data is grouped and integrated to obtain an input data set matching the input dimension of the welding quality prediction model; The input data set is input into the welding quality prediction model, and the input data set is automatically subjected to feature mining and operation analysis to identify the association rule between the multi-source data and the welding quality, and the welding quality data is output.

[0009] Preferably, in S3, based on the welding quality data, the initial optimization parameters of the welding process parameters are determined in combination with the preset target requirements, including: The first target is obtained by first disassembling the preset target requirements, and the quality standard target, the production efficiency target and the energy consumption cost target are obtained, and the first target is secondly disassembled to obtain quantifiable sub-targets. The first judgment matrix and the second judgment matrix are generated based on the importance between any two first targets and the importance between any two quantifiable sub-targets, and the first maximum eigenvalue and the second maximum eigenvalue of the first judgment matrix and the second judgment matrix are obtained respectively. If the consistency check passes, the eigenvectors corresponding to the first judgment matrix and the second judgment matrix are calculated based on the eigenvalue method, the eigenvectors are normalized to obtain initial main weights of the first target and initial sub-weights of the quantifiable sub-target, and an initial weight scheme is obtained based on the initial main weights and the initial sub-weights; otherwise, the scale value is adjusted, and the latest judgment matrix is regenerated until the consistency check passes; Based on the historical weight scheme and the corresponding historical effect data, the theoretical effect data under the initial weight scheme is predicted, the deviation value of the theoretical effect data and the historical effect data is obtained, the deviation value is fitted and analyzed through a linear regression algorithm, a correction coefficient corresponding to the weight is obtained, the initial main weight and the initial sub-weight are corrected based on the correction coefficient, and a target weight is obtained, and an optimized target system is established based on the first target, the quantifiable sub-target and the target weight; Based on the defect type and the defect correlation parameter in the weld quality data, the built-in process parameter-defect correlation rule library of the welding cloud platform is called to reversely map the process parameter set corresponding to the defect type and the defect correlation parameter, and a parameter constraint boundary is constructed based on the steel structure welding material characteristics and the welding equipment rated parameters; Based on the optimized target system, the parameter constraint boundary is used as a constraint to iteratively calculate the process parameter set to obtain an optimal process parameter scheme, and the initial optimization parameter of the weld process parameter is determined based on the optimal process parameter scheme.

[0010] Preferably, the consistency check based on the first maximum eigenvalue and the second maximum eigenvalue comprises: A first difference value between the first maximum eigenvalue or the second maximum eigenvalue and the order of the corresponding judgment matrix, and a second difference value between the order of the corresponding judgment matrix and 1 are obtained; The ratio of the first difference value and the second difference value is used as a consistency index value, if the consistency index value is less than a preset value, it is determined that the consistency check passes, otherwise, it is determined that the consistency check fails.

[0011] Preferably, the iterative calculation of the process parameter set to obtain the optimal process parameter scheme comprises: The process parameter set is iteratively calculated, and a plurality of groups of candidate process parameter schemes are generated according to the iteration results; The plurality of groups of candidate process parameter schemes are virtually welded and simulated based on the digital twin model, and the optimal process parameter scheme is selected according to the simulation results.

[0012] Preferably, in S4, the current weld quality data is fed back to the welding cloud platform to update the weld quality prediction model, comprising: The current weld quality data is extracted according to quality characteristics, parameter characteristics, scene characteristics and operation characteristics, and four-dimensional traceability samples of quality-parameter-scene-operation are formed according to the extraction results; The four-dimensional traceability samples are input into the pre-established welding Bayesian network, the root node layer is used to analyze the process parameters and environmental parameters, the intermediate hidden node layer is used to analyze the indirect parameters, and the leaf node layer is used to analyze the defect type and defect level, according to the analysis results, the leaf node layer is used as the reason to perform backward reasoning to the intermediate hidden node layer and the root node layer, the posteriori causal probability that the current parameter abnormality causes the defect to occur is obtained, the causal contribution degree that the current parameter abnormality causes the defect to occur is obtained based on the sum of the posteriori causal probabilities that all parameter abnormalities cause the same defect to occur, and the core parameter causing the defect to occur is determined based on the causal contribution degree; The causal relationship between the core parameter and the defect is divided into an invariable relationship and a variable relationship according to whether it is affected by the scene, the general critical interval of the continuous process parameter in the invariable relationship is determined, the common characteristics of the discrete process parameters in the variable relationship are extracted, and the scene-free general rule is established based on the general critical interval and the common characteristics; The key position corresponding to the core parameter is marked to obtain an updated local module, the updated local module is updated first, and the scene-free general rule is added to the data-scarce model scene layer for second updating, the welding quality prediction model is parameter-optimized based on the results of the first updating and the second updating, and the latest prediction model is obtained.

[0013] Preferably, the leaf node layer is used as the reason to perform backward reasoning to the intermediate hidden node layer and the root node layer, and the posteriori causal probability that the current parameter abnormality causes the defect to occur is obtained, and the posteriori causal probability that the current parameter abnormality causes the defect to occur is obtained. ; Wherein, represents the posteriori causal probability that the current parameter abnormality causes the defect to occur, represents the probability that the defect occurs when the current parameter is abnormal, represents the probability that the current parameter is abnormal, represents the probability that the defect occurs.

[0014] Preferably, the general critical interval of the continuous process parameter in the invariable relationship is determined, and the parameter deviation degree is obtained by taking the ratio of the specific value of the continuous process parameter in all scenes to the critical reference state value in the scene. The maximum value and the minimum value are obtained from all the values of the parameter deviation degree, and the general critical interval is determined.

[0015] ​Compared with the prior art, the present application has the following beneficial effects: By collecting welding process parameters, environmental parameters and weld characteristic data through pre-deployed sensors, after data preprocessing of the welding process parameters, environmental parameters and weld characteristic data, multi-source data is obtained, the multi-source data is uploaded to the welding cloud platform, the standardization and precision of the multi-source data are realized, the centralized retrieval, sharing and subsequent reuse of the multi-source data are realized, sufficient data support is provided for model training and parameter optimization, by inputting the multi-source data into the weld quality prediction model pre-trained based on the welding cloud platform, weld quality data is obtained, the objectivity and accuracy of the weld quality evaluation are improved, it is suitable for large-scale and standardized production scene, by determining the initial optimization parameters of the weld process parameters based on the weld quality data and combining the preset target demand, it is ensured that the optimization direction is consistent with the actual production demand, a scientific basis is provided for the real-time adjustment of the welding equipment, the welding equipment controller is optimized in real time according to the initial optimization parameters, and the current weld quality data in the optimization process is collected in real time, the current weld quality data is fed back to the welding cloud platform to update the weld quality prediction model, the quality fluctuation in the welding process is quickly responded, the defect expansion is avoided, and the welding quality stability is improved, by feeding back the current weld quality data in the optimization process to the cloud platform and updating the model, the model can continuously adapt to the changes of the welding scene, the precision of prediction and optimization is gradually improved, and iterative upgrading of the technical scheme is realized.

[0016] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the present application. The objects and other advantages of the present application can be realized and obtained by the structure particularly pointed out in the application.

[0017] The technical solutions of the present application will be further described in detail below with the help of the drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0018] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, together with the embodiments of the present application, to explain the present application, and do not constitute a limitation on the present application. In the drawings: Figure 1 A flowchart of the weld quality prediction and process optimization method based on the welding cloud platform in the embodiments of the present application is shown in the figure. Figure 2 A flowchart of obtaining weld quality data in the embodiments of the present application is shown in the figure. DETAILED DESCRIPTION

[0019] The preferred embodiments of the present application are described below in combination with the drawings, and it should be understood that the preferred embodiments described herein are only used to illustrate and explain the present application, and do not limit the present application.

[0020] Embodiment 1: The embodiment of the present application provides a welding seam quality prediction and process optimization method based on a welding cloud platform, as shown in the figure, comprising: Figure 1 S1: based on the pre-deployed sensor, the welding process parameters, the environmental parameters and the welding seam characteristic data are collected, the welding process parameters, the environmental parameters and the welding seam characteristic data are pre-processed to obtain multi-source data, and the multi-source data is uploaded to the welding cloud platform; S2: inputting the multi-source data into the welding seam quality prediction model pre-trained based on the welding cloud platform to obtain welding seam quality data; S3: based on the welding seam quality data, the initial optimization parameters of the welding seam process parameters are determined in combination with the preset target requirements; S4: the welding equipment controller performs real-time optimization according to the initial optimization parameters, and real-time collection of the current welding seam quality data in the optimization process is performed, and the current welding seam quality data is fed back to the welding cloud platform to update the welding seam quality prediction model. In this embodiment, the welding process parameters refer to the core technical parameters that can be controlled in the welding process, including welding current, voltage, welding speed, protective gas flow, welding wire feeding speed, etc.

[0021] In this embodiment, the environmental parameters refer to the environmental conditions of the welding site, including environmental temperature, humidity, wind speed, etc.

[0022] In this embodiment, the welding seam characteristic data refers to the parameters describing the appearance and internal state of the welding seam, including welding seam appearance size (such as fusion width, fusion depth, excess height), internal defect information (such as the position and size of pores, cracks, incomplete penetration and other defects), etc.

[0023] In this embodiment, the welding seam quality prediction model is an intelligent analysis model (such as a neural network, a machine learning model, etc.) pre-trained by historical welding data, which can input multi-source welding data (process, environment, welding seam characteristics) and output welding seam quality related results (such as defect type, qualified grade, risk level, etc.).

[0024] In this embodiment, the initial optimization parameters are determined based on the welding seam quality prediction results and the preset target (such as quality, efficiency, cost target), and are used to guide the initial value of the process parameters adjusted by the welding equipment.

[0025] In this embodiment, the welding equipment controller refers to the core component for controlling the operation of the welding equipment, which can receive the optimization parameters issued by the cloud platform and adjust the operation state of the welding equipment in real time.

[0026] In this embodiment, the pre-deployed sensors include voltage, current, temperature, speed, air pressure, laser, ultrasonic sensor, etc.

[0027] In this embodiment, the pre-deployed sensors include voltage, current, temperature, speed, air pressure, laser, ultrasonic sensor, etc.

[0028] In this embodiment, the preset target requirement is, for example, a requirement for quality compliance, efficiency improvement, cost reduction, etc.

[0029] The beneficial effects of the above design scheme are: the welding process parameters, the environmental parameters and the weld characteristic data are collected by the pre-deployed sensors, the welding process parameters, the environmental parameters and the weld characteristic data are pre-processed to obtain multi-source data, the multi-source data is uploaded to the welding cloud platform, the standardization and precision of the multi-source data are realized, the centralized retrieval, sharing and subsequent reuse of the multi-source data are realized, sufficient data support is provided for model training and parameter optimization, the multi-source data is input into the weld quality prediction model pre-trained based on the welding cloud platform, the weld quality data is obtained, the objectivity and accuracy of the weld quality evaluation are improved, it is suitable for large-scale and standardized production scenarios, the initial optimization parameters of the weld process parameters are determined based on the weld quality data and the preset target requirement, the optimization direction is ensured to meet the actual production requirements, a scientific basis is provided for real-time adjustment of the welding equipment, the welding equipment controller performs real-time optimization according to the initial optimization parameters, and the current weld quality data in the optimization process is collected in real time, the current weld quality data is fed back to the welding cloud platform to update the weld quality prediction model, the quality fluctuation in the welding process is quickly responded, the defect expansion is avoided, and the welding quality stability is improved, the current weld quality data in the optimization process is fed back to the cloud platform and the model is updated, so that the model can continuously adapt to the changes of the welding scene, the accuracy of prediction and optimization is gradually improved, and the iteration and upgrading of the technical scheme are realized.

[0030] In embodiment 2, based on the basis of embodiment 1, a weld quality prediction and process optimization method based on a welding cloud platform is provided, and in S1, after the welding process parameters, the environmental parameters and the weld characteristic data are pre-processed, multi-source data is obtained, including: The welding process parameters, the environmental parameters and the weld characteristic data are subjected to abnormal data elimination, missing value supplement and repeated data deduplication processing to obtain processed data. The processed data is subjected to data standardization processing to obtain multi-source data.

[0031] In this embodiment, the abnormal data elimination is realized based on the process threshold range and statistical law of the process parameters, the environmental parameters and the weld characteristic data of the steel structure welding.

[0032] In this embodiment, based on the characteristics that the welding process is a time-continuous process, the linear interpolation method or the adjacent value filling method with strong adaptability in the industrial scene is adopted to realize the missing value supplement.

[0033] In this embodiment, the repeated data deduplication is realized based on the dual judgment criteria of time stamp and core feature.

[0034] In this embodiment, the data standardization processing of the processing data is, for example, that the reasonable range of the welding current is 200-300 A, and a certain collected value is 250 A, and then the standardized value is (250-200) / (300-200)=0.5.

[0035] The beneficial effects of the above design scheme are that: by performing abnormal data elimination, missing value supplement, and repeated data deduplication processing on the welding process parameters, environmental parameters, and weld feature data, the processing data is obtained, the inherent defects of the welding site original data are solved, the data authenticity and integrity are ensured, by performing data standardization processing on the processing data, the multi-source data is obtained, the dimensional difference of the multi-source data is eliminated, the fairness and effectiveness of the model operation are ensured, the preprocessed data can truly and accurately reflect the actual influence of each parameter on the weld quality, the weld quality prediction result obtained by the cloud platform based on the data is more in line with the actual situation, and the subsequent process optimization parameters determined in combination with the prediction result are also more targeted.

[0036] Embodiment 3: Based on the basis of embodiment 1, the present embodiment provides a weld quality prediction and process optimization method based on a welding cloud platform, in S1, the multi-source data is uploaded to the welding cloud platform, including: By deploying the edge computing node near the welding station, the multi-source data is encapsulated and format unified, and the multi-source data is encapsulated into a data frame; Based on the communication link in the edge computing node, the data frame is sent to the exclusive data receiving port of the welding cloud platform.

[0037] In this embodiment, the format of the data frame is, for example: welding station number + sensor type + millisecond timestamp + standardized value.

[0038] In this embodiment, the communication link is an industrial communication link, adopts a dual-link redundancy architecture of wired transmission as the main and wireless transmission as the auxiliary, and is suitable for complex working conditions of a steel structure welding site.

[0039] In this embodiment, encapsulating the multi-source data into a data frame can ensure that a single data frame corresponds to a complete set of welding process multi-source data, and realize data uniqueness and traceability.

[0040] The beneficial effects of the above design scheme are: through the edge computing node deployed near the welding station, the multi-source data is encapsulated and format unified, the multi-source data is encapsulated into a data frame, the encapsulation and format unification of the multi-source data can be completed in the edge computing node, without the need to transmit the scattered data to the remote device for processing, greatly shortening the link time consumption of data preprocessing and encapsulation, adapting to the real-time uploading demand of multi-source data in the welding process, and format unified encapsulation into a data frame, guaranteeing the integrity and traceability of data transmission, through the communication link in the edge computing node, the data frame is sent to the exclusive data receiving port of the welding cloud platform, which can reduce the risk of loose link interface, interference invasion and other risks, ensure that the data frame is stably sent to the exclusive receiving port of the cloud, avoid data transmission interruption caused by link failure, guarantee continuous uploading of multi-source data, and provide uninterrupted data input for subsequent weld quality prediction model, the welding cloud platform sets an exclusive data receiving port, which only interfaces the data frame sent by each station edge computing node, can effectively isolate the interference and illegal access of external irrelevant data, reduce the risk of industrial data leakage and tampering, and guarantee the security of welding core data.

[0041] In the embodiment 4, based on the embodiment 1, a weld quality prediction and process optimization method based on a welding cloud platform is provided, as shown in the following figure. Figure 2 In S2, the multi-source data is input into the weld quality prediction model pre-trained based on the welding cloud platform, and weld quality data is obtained, including: According to the preset time sequence rule, the multi-source data is grouped and integrated to obtain an input data set matched with the input dimension of the weld quality prediction model; The input data set is input into the weld quality prediction model, and feature mining and operation analysis are automatically performed on the input data set to identify the association rule between the multi-source data and the weld quality, and the weld quality data is output.

[0042] In this embodiment, the input data set is associated with the welding station number and the millisecond level timestamp.

[0043] In this embodiment, the weld quality prediction model is deployed on the welding cloud platform server cluster, and is trained and generated based on historical welding data and corresponding weld quality detection results, and has multi-source data feature extraction and quality prediction capability.

[0044] In this embodiment, the preset time rule refers to the data grouping rule pre-set by the welding cloud platform to adapt to the welding operation characteristics and model input requirements, and the core purpose is to classify and integrate the time sequence multi-source data uploaded to the structured database according to the welding operation period, process stage or time window, form a complete data sequence conforming to the model input dimension, and at the same time, retain the time sequence association and traceability of the data.

[0045] The beneficial effects of the above design scheme are: by grouping and integrating the multi-source data according to the preset timing rule, the input data set matching the input dimension of the weld quality prediction model is obtained, the scattered standardized data is associated as a complete data sequence corresponding to a single welding operation or a specific process stage, and the mixed data of different time periods is avoided, by inputting the input data set into the weld quality prediction model, the input data set is automatically mined and analyzed, the association rule between the multi-source data and the weld quality is identified, and the weld quality data is output, the grouped and integrated data has matched the model input dimension in advance, avoiding secondary grouping or dimension correction after the model receives the data, reducing the invalid computing power consumption of the welding cloud platform, and improving the prediction efficiency and standardization level.

[0046] In the embodiment 5 based on the embodiment 1, a weld quality prediction and process optimization method based on a welding cloud platform is provided, and in S3, based on the weld quality data, the initial optimization parameters of the weld process parameters are determined in combination with the preset target demand, including: The preset target demand is first decomposed to obtain a quality standard target, a production efficiency target and an energy consumption cost target, and the first target is second decomposed to obtain quantifiable sub-targets. Based on the scale method, the importance between any two first targets and the importance between any two quantifiable sub-targets are generated to generate a first judgment matrix and a second judgment matrix, and the first maximum eigenvalue and the second maximum eigenvalue of the first judgment matrix and the second judgment matrix are obtained respectively. Based on the first maximum eigenvalue and the second maximum eigenvalue, consistency check is performed, if passed, the characteristic vectors corresponding to the first judgment matrix and the second judgment matrix are calculated based on the eigenvalue method, and the characteristic vectors are normalized and processed as the initial main weight of the first target and the initial sub-weight of the quantifiable sub-targets, and based on the initial main weight and the initial sub-weight, the initial weight scheme is obtained, otherwise, the scale value is adjusted, the latest judgment matrix is regenerated until the consistency check is passed. Based on the historical weight scheme and the corresponding historical effect data, the theoretical effect data under the initial weight scheme is predicted, the deviation value of the theoretical effect data and the historical effect data is obtained, the deviation value is fitted and analyzed by a linear regression algorithm, the correction coefficient corresponding to the weight is obtained, the initial main weight and the initial sub-weight are corrected based on the correction coefficient, the target weight is obtained, and the optimization target system is established based on the first target, the quantifiable sub-target and the target weight. Based on the defect type and the defect association parameter in the weld quality data, the process parameter-defect association rule library built in the welding cloud platform is called, and the process parameter set corresponding to the defect type and the defect association parameter is reversely mapped, and the parameter constraint boundary is constructed based on the steel structure welding material characteristics and the welding equipment rated parameters. Based on the optimization target system as the basis, with parameter constraint boundary as the constraint, the process parameter set is iteratively calculated to obtain the optimal process parameter scheme, and the initial optimization parameter of the weld process parameter is determined based on the optimal process parameter scheme.

[0047] In this embodiment, the quantifiable sub-targets are, for example, the quality standard target is decomposed into defect rate control sub-target, weld grade standard sub-target, the production efficiency target is decomposed into weld speed improvement sub-target, operation cycle stability sub-target, the energy consumption cost target is decomposed into electric energy consumption reduction sub-target, consumable loss control sub-target, etc.

[0048] In this embodiment, the scale method is initially set to use the 1-9 scale method, and the comparison standard is defined in combination with the steel structure welding scene, 1 represents the same importance of two targets, 3 represents that one target is slightly more important than the other target, 5 represents that one target is obviously more important than the other target, 7 represents that one target is strongly more important than the other target, and 9 represents that one target is extremely more important than the other target, 2, 4, 6, and 8 are intermediate values of the corresponding scale, and the inverse of 1-9 represents the opposite importance of the two targets, for example, 1 / 3 represents that one target is slightly less important than the other target; the welding cloud platform automatically pops up a pairwise comparison interface, the operator inputs the scale value based on the welding scene demand, and the system automatically generates an upper triangular or lower triangular judgment matrix.

[0049] In this embodiment, the welding cloud platform automatically retrieves the historical optimization data matched with the current welding scene (workpiece type, welding material, equipment model, and consistent environmental conditions) in the past three years, and filters out the weight scheme and the corresponding actual application effect data.

[0050] In this embodiment, the core characteristic parameters corresponding to the first and second judgment matrices are the key indicators for measuring the consistency of the judgment matrix, which are solved through matrix operation. If the judgment matrix is completely consistent, the maximum eigenvalue is equal to the matrix order, and the larger the deviation, the larger the difference between the maximum eigenvalue and the matrix order.

[0051] In this embodiment, the linear regression algorithm is used for fitting analysis of the historical weight scheme, effect data, and current deviation value to generate a weight correction coefficient algorithm. The core is to construct a linear correlation model of the deviation value and the weight to realize accurate correction of the deviation.

[0052] In this embodiment, the process parameter-defect association rule base is a core database built in the welding cloud platform, which integrates industry standards, historical data, and defect mechanisms, and stores rules according to defect type-core influence parameter-parameter adjustment trend, which is used for reverse positioning of process parameters affecting defects; for example, the rule base stores the core parameters of gas hole defect as protective gas flow and welding speed, and the adjustment trend is to increase the protective gas flow (15→22 L / min) and reduce the welding speed (5→3.8 mm / s), which directly maps the corresponding parameters for the predicted gas hole defect.

[0053] In this embodiment, the parameter constraint boundary is a process parameter value range constructed based on hardware, process, compliance requirements, used to limit the boundary of iterative calculation.

[0054] The beneficial effects of the above design scheme are: through first disassembly of the preset target demand, the first target of quality standard target, production efficiency target and energy consumption cost target is obtained, the first target is secondly disassembled to obtain quantifiable sub-target, the fuzzy preset target demand is converted into clear hierarchical target system, the three core demands of quality, efficiency and cost are considered, the actual production demand of steel structure welding multi-target balance is adapted, the quantification design of sub-target provides clear index for subsequent weight calculation and effect verification, avoids the deviation of optimization direction caused by fuzzy target, the importance between any two first targets and the importance between any two quantifiable sub-targets are generated based on scale method to generate first judgment matrix and second judgment matrix, the first maximum eigenvalue and the second maximum eigenvalue of the first judgment matrix and the second judgment matrix are obtained respectively, the consistency check is carried out based on the first maximum eigenvalue and the second maximum eigenvalue, if it passes, the characteristic vectors corresponding to the first judgment matrix and the second judgment matrix are calculated based on the eigenvalue method, and the characteristic vectors are normalized and processed as the initial main weight of the first target and the initial sub weight of the quantifiable sub-target, based on the initial main weight and the initial sub weight, the initial weight scheme is obtained, otherwise, the scale value is adjusted, the latest judgment matrix is regenerated until the consistency check passes, the deviation of subjective judgment is forced to be corrected, the rationality of the initial main weight and the initial sub weight is ensured, a reliable foundation is laid for subsequent optimization target system construction, the process parameter optimization failure caused by weight distortion is avoided, the theoretical effect data under the initial weight scheme is predicted based on the historical weight scheme and the corresponding historical effect data, the deviation value of the theoretical effect data and the historical effect data is obtained, the deviation value is fitted and analyzed by linear regression algorithm, the correction coefficient corresponding to the weight is obtained, the initial main weight and the initial sub weight are corrected based on the correction coefficient to obtain the target weight, the optimized target system is established based on the first target, the quantifiable sub-target and the target weight, the process parameter-defect association rule library built in the welding cloud platform is called based on the defect type and the defect association parameter in the weld quality data, the process parameter set corresponding to the defect type and the defect association parameter is obtained through reverse mapping, and the parameter constraint boundary is constructed based on the material characteristics of steel structure welding and the rated parameters of welding equipment, the defect type in the weld quality data is accurately optimized, blind traversal of all parameters is avoided, and the iteration calculation efficiency is greatly improved.Meanwhile, three constraint boundaries are constructed in combination with material properties, equipment rated parameters and industry specifications, so as to limit the parameter range from three dimensions of hardware, process and compliance, to eliminate the problems of exceeding the bearing capacity of the equipment, violating the material properties or industry standards, to ensure that the initial optimization parameters can be directly applied, to reduce the trial and error cost, to take the optimization target system as the basis, to take the parameter constraint boundary as the constraint, to perform iterative calculation on the process parameter set, to obtain the optimal process parameter scheme, to determine the initial optimization parameters of the weld process parameters based on the optimal process parameter scheme, and to generate the initial optimization parameters which are consistent with the theoretical optimum and adapt to the actual conditions of the on-site welding equipment, materials and process, thereby providing a high-quality benchmark for subsequent real-time optimization of the welding equipment.

[0055] Embodiment 6: Based on the basis of embodiment 5, the embodiment of the application provides a weld quality prediction and process optimization method based on a welding cloud platform, and the consistency check is performed based on the first maximum eigenvalue and the second maximum eigenvalue, including: obtaining a first difference value between the first maximum eigenvalue or the second maximum eigenvalue and the corresponding judgment matrix order, and a second difference value between the corresponding judgment matrix order and 1; taking the ratio of the first difference value and the second difference value as a consistency index value, if the consistency index value is less than a preset value, determining that the consistency check is passed, otherwise, determining that the consistency check is not passed.

[0056] The beneficial effects of the above design scheme are that the table lookup and matching process of the random consistency index is not introduced, the algorithm logic is greatly simplified, the consistency check is pre-processed, a reliable foundation is built for subsequent weight correction, optimization target system construction and process parameter iterative calculation, and the adaptability of the initial optimization parameters is improved from the source.

[0057] Embodiment 7: Based on the basis of embodiment 5, the embodiment of the application provides a weld quality prediction and process optimization method based on a welding cloud platform, and the process parameter set is iteratively calculated to obtain an optimal process parameter scheme, including: iteratively calculating the process parameter set, and generating multiple groups of candidate process parameter schemes according to the iteration result; performing virtual welding simulation on the multiple groups of candidate process parameter schemes based on the digital twin model, and selecting the optimal process parameter scheme according to the simulation result.

[0058] The beneficial effects of the above design scheme are: multiple candidate schemes are generated through iterative calculation, global optimization and parameter constraint boundary are considered, local optimal solution is avoided, digital twin virtual simulation does not need physical welding test, can accurately simulate welding process, predict weld quality and defects, shorten optimization cycle, combined with simulation results screening, can accurately match optimization target system, ensure that the optimal scheme meets the quality, efficiency and cost requirements, and is suitable for on-site equipment and material characteristics, and improves the reliability of initial optimization parameters.

[0059] In the embodiment 8 based on the embodiment 1, a welding cloud platform based weld quality prediction and process optimization method is provided, in S4, the current weld quality data is fed back to the welding cloud platform to update the weld quality prediction model, including: The current weld quality data is extracted according to quality characteristics, parameter characteristics, scene characteristics and operation characteristics, and four-dimensional traceability samples of quality-parameter-scene-operation are formed according to the extraction results; The four-dimensional traceability samples are input into the pre-established welding Bayesian network, the process parameters and environmental parameters are analyzed by the root node layer, the indirect parameters are analyzed based on the intermediate hidden node layer, the defect types and defect levels are analyzed based on the leaf node layer, according to the analysis results, the leaf node layer is taken as the reason, the intermediate hidden node layer and the root node layer are reversely inferred, the posteriori causal probability that the current parameter abnormality causes the defect to occur is obtained, the causal contribution degree of the current parameter abnormality causing the defect to occur is obtained based on the sum of the posteriori causal probabilities that all parameter abnormalities cause the same defect to occur, and the core parameter causing the defect to occur is determined based on the causal contribution degree; The causal relationship between the core parameter and the defect is obtained according to whether it is affected by the scene to obtain an immutable relationship and a variable relationship, the general critical interval of the continuous process parameter in the immutable relationship is determined, the common characteristics of the discrete process parameters in the variable relationship are extracted, and the scene-free general rule is established based on the general critical interval and the common characteristics; The key positions corresponding to the core parameters are marked to obtain a local module of this update, the local module of this update is updated first, and the scene-free general rule is added to the data scarce model scene layer for second update, the parameter optimization of the weld quality prediction model is performed based on the results of the first update and the second update, and the latest prediction model is obtained.

[0060] In this embodiment, the quality features are, for example, weld defect types (porosity / cracks), defect levels, defect locations, etc., the parameter features are, for example, process parameter time curves during the welding of the weld, environmental parameter fluctuation data, equipment operating state data (such as welding machine current stability), the scene features are, for example, welding stations, steel grades, welding methods (such as submerged arc welding / gas shielded welding), etc., and the operation features are, for example, welder operation habit labels (such as changes in welding gun angles, fluctuations in walking speeds).

[0061] In this embodiment, the welding Bayesian network is a welding process causally exclusive architecture, the network nodes are arranged in a hierarchical manner according to a causal conduction logic, there are no messy correlations, all directed edges only represent the welding physical logic of cause and effect, rather than meaningless correlation connections, and the node levels are fixed at three levels.

[0062] In this embodiment, the immutable relationships are objective and permanent physical relationships in welding metallurgy, fluid mechanics, and materials science, and the variable relationships are relationships that are strongly bound to specific welding scenes and change in values, thresholds, states, conditions, etc. with changes in scenes.

[0063] In this embodiment, the causal contribution degree of the current parameter abnormality to the occurrence of a defect is the ratio of the posteriori causal probability of the current parameter abnormality to the occurrence of the defect to the sum of the posteriori causal probabilities of all parameter abnormalities to the occurrence of the same defect.

[0064] In this embodiment, the indirect parameter is, for example, a sudden drop in the protective gas flow, and the probability of the poor protection effect of the molten pool is 92%.

[0065] The beneficial effects of the above design scheme are: four-dimensional features of quality, parameters, scenes, and operations are extracted and correlated into traceable samples to provide real and traceable input data for subsequent causal reasoning, to eliminate analysis bias caused by missing data dimensions, to quantitatively distinguish core parameters by reverse deducing the posteriori causal probability of parameter abnormalities from defect results, to break through the limitations of conventional correlation analysis, to accurately locate the dominant causes of defects, to avoid optimizing all parameters without distinction, to greatly improve the targeting of model updates, to reduce the waste of computing power, to determine the continuous parameter general critical interval by splitting the variable and immutable relationships of the causal relationship, to extract the common features of discrete parameters, to refine the general rules that are independent of specific scenes, to solve the model optimization problems in data-scarce scenes such as special steels and unpopular welding processes, to realize cross-scene knowledge reuse, to mark the corresponding modules of core parameters for local updates, to avoid knowledge forgetting caused by full training, and to improve the update efficiency; the general rules are embedded into the data-scarce scene layer to fill the data gap, and the model parameters are optimized by double updates, which not only strengthens the defect prediction accuracy of core parameters, but also enhances the generalization ability of the model in unpopular scenes, forming a closed-loop optimization.

[0066] Embodiment 9: Based on the basis of embodiment 8, the embodiment of the application provides a welding seam quality prediction and process optimization method based on a welding cloud platform, and the leaf node layer is taken as a reason to perform reverse reasoning on the intermediate hidden node layer and the root node layer to obtain a posteriori causal probability that a current parameter anomaly causes a defect to occur, and the specific process is as follows: ; wherein, represents the posteriori causal probability that the current parameter anomaly causes the defect to occur, represents a probability that the defect occurs when the current parameter anomaly occurs, represents a probability that the current parameter anomaly occurs, represents a probability that the defect occurs.

[0067] The beneficial effect of the above design scheme is that the posteriori causal probability that the parameter anomaly is reversely deduced from the defect result, and a basis is provided for the causal contribution degree calculation.

[0068] Embodiment 10: Based on the basis of embodiment 8, the embodiment of the application provides a welding seam quality prediction and process optimization method based on a welding cloud platform, and a general critical interval of continuous process parameters in an immutable relationship is determined, including: a ratio of a specific value of the continuous process parameter in all scenes to a critical reference state value in the scene is taken as a parameter deviation degree; when the causal contribution degree of the continuous process parameter is greater than a preset contribution degree, all values of the parameter deviation degree are obtained, the maximum value and the minimum value are obtained from all the values, and the general critical interval is determined.

[0069] In this embodiment, when the deviation degree of a certain parameter is greater than the preset contribution degree, it is indicated that the parameter is a leading cause of the defect.

[0070] The beneficial effect of the above design scheme is that the scene attribute is stripped through the dimensionless deviation degree, the deviation degree range is screened through the causal contribution degree threshold, the general critical interval of the core parameter is accurately locked, the rule cross-scene adaptability is improved, a reliable basis is provided for the model targeted update, and the accuracy and the universality are considered.

[0071] Obviously, those skilled in the art can make various modifications and variations to the application without departing from the spirit and scope of the application. Thus, if these modifications and variations of the application belong to the scope of the present application and its equivalent technologies, the application also intends to include these modifications and variations.

Claims

1. A method for weld quality prediction and process optimization based on a welding cloud platform, characterized in that, include: S1: Based on pre-deployed sensors, welding process parameters, environmental parameters and weld feature data are collected. After data preprocessing of the welding process parameters, environmental parameters and weld feature data, multi-source data is obtained and uploaded to the welding cloud platform. S2: Input the multi-source data into the weld quality prediction model pre-trained based on the welding cloud platform to obtain weld quality data; S3: Based on weld quality data and combined with preset target requirements, determine the initial optimization parameters for weld process parameters; S4: The welding equipment controller performs real-time optimization according to the initial optimization parameters and collects the current weld quality data in real time during the optimization process. The current weld quality data is then fed back to the welding cloud platform to update the weld quality prediction model.

2. The method for weld quality prediction and process optimization based on a welding cloud platform according to claim 1, characterized in that, In step S1, after preprocessing the welding process parameters, environmental parameters, and weld characteristic data, multi-source data is obtained, including: The welding process parameters, environmental parameters, and weld characteristic data are processed by removing abnormal data, supplementing missing values, and deduplicating duplicate data to obtain the processed data. Data standardization is performed on the processed data to obtain multi-source data.

3. The method for weld quality prediction and process optimization based on a welding cloud platform according to claim 1, characterized in that, In step S1, uploading multi-source data to the welding cloud platform includes: By deploying edge computing nodes near the welding station, the multi-source data is encapsulated and formatted into data frames. Based on the communication link in the edge computing node, the data frame is sent to the dedicated data receiving port of the welding cloud platform.

4. The method for weld quality prediction and process optimization based on a welding cloud platform according to claim 1, characterized in that, In step S2, multi-source data is input into a weld quality prediction model pre-trained based on a welding cloud platform to obtain weld quality data, including: The multi-source data are grouped and integrated according to the preset time series rules to obtain an input dataset that matches the input dimension of the weld quality prediction model. The input dataset is fed into the weld quality prediction model, which automatically performs feature mining and computational analysis on the input dataset, identifies the correlation between multi-source data and weld quality, and outputs weld quality data.

5. The method for weld quality prediction and process optimization based on a welding cloud platform according to claim 1, characterized in that, In step S3, based on weld quality data and in conjunction with preset target requirements, initial optimization parameters for the weld process parameters are determined, including: The preset target requirements are first decomposed to obtain the first target, which includes the quality compliance target, the production efficiency target, and the energy consumption cost target. The first target is then second decomposed to obtain quantifiable sub-targets. Based on the scaling method, a first judgment matrix and a second judgment matrix are generated to determine the importance between any two first objectives and the importance between any two quantifiable sub-objectives, and the first maximum eigenvalue and the second maximum eigenvalue of the first judgment matrix and the second judgment matrix are obtained respectively. Consistency verification is performed based on the first and second largest eigenvalues. If it passes, the eigenvectors corresponding to the first and second judgment matrices are calculated based on the eigenvalue method. The eigenvectors are normalized and used as the initial sovereign weight of the first target and the initial sub-weights of the quantifiable sub-targets. Based on the initial sovereign weights and the initial sub-weights, the initial weight scheme is obtained. Otherwise, the scale value is adjusted and the latest judgment matrix is ​​regenerated until the consistency verification passes. Based on historical weighting schemes and their corresponding historical performance data, the theoretical performance data under the initial weighting scheme is predicted, the deviation between the theoretical performance data and the historical performance data is obtained, and the deviation is fitted and analyzed by linear regression algorithm to obtain the correction coefficient of the corresponding weight. Based on the correction coefficient, the initial main weight and initial sub-weight are corrected to obtain the target weight. Based on the first target, quantifiable sub-targets and target weight, the optimized target system is established. Based on the defect types and defect association parameters in the weld quality data, the process parameter-defect association rule library built into the welding cloud platform is retrieved, and the process parameter set corresponding to the defect type and defect association parameter is obtained by reverse mapping. At the same time, parameter constraint boundaries are constructed based on the characteristics of steel structure welding materials and the rated parameters of welding equipment. Based on the optimization target system and the parameter constraint boundary, the process parameter set is iteratively calculated to obtain the optimal process parameter scheme. Based on the optimal process parameter scheme, the initial optimization parameters for the weld process parameters are determined.

6. The method for weld quality prediction and process optimization based on a welding cloud platform according to claim 5, characterized in that, Consistency verification is performed based on the first and second largest eigenvalues, including: Obtain the first difference between the first or second largest eigenvalue and the order of the corresponding judgment matrix, and the second difference between the order of the corresponding judgment matrix and 1; The consistency index is determined based on the ratio of the first difference and the second difference. If the consistency index is less than the preset value, the consistency verification is determined to be successful; otherwise, the consistency verification is determined to be unsuccessful.

7. The method for weld quality prediction and process optimization based on a welding cloud platform according to claim 5, characterized in that, The optimal process parameter scheme is obtained by iterative calculation of the set of process parameters, including: The process parameter set is iteratively calculated, and multiple sets of candidate process parameter schemes are generated based on the iteration results. Virtual welding simulations were performed on multiple candidate process parameter schemes based on digital twin models, and the optimal process parameter scheme was selected based on the simulation results.

8. The method for weld quality prediction and process optimization based on a welding cloud platform according to claim 1, characterized in that, In step S4, the current weld quality data is fed back to the welding cloud platform to update the weld quality prediction model, including: Feature extraction is performed on the current weld quality data according to quality characteristics, parameter characteristics, scenario characteristics and operation characteristics, and a four-dimensional traceability sample of quality-parameter-scenario-operation is formed based on the extraction results; The four-dimensional traceability sample is input into a pre-established welding Bayesian network. The root node layer is used to analyze process parameters and environmental parameters, the intermediate hidden node layer is used to analyze indirect parameters, and the leaf node layer is used to analyze defect type and defect level. Based on the analysis results, the leaf node layer is used as the cause to reason backward to the intermediate hidden node layer and the root node layer to obtain the posterior causal probability that the current parameter abnormality causes the defect to occur. Based on the sum of the posterior causal probabilities that all parameter abnormalities cause the same defect to occur, the causal contribution of the current parameter abnormality to the defect is obtained. Based on the causal contribution, the core parameters for the occurrence of the defect are determined. The causal relationship between core parameters and defects is divided into immutable and variable relationships according to whether it is affected by the scenario. A general critical interval is determined for the continuous process parameters in the immutable relationship, and common features are extracted for the discrete process parameters in the variable relationship. Based on the general critical interval and common features, a scenario-free general rule is established. The key positions corresponding to the core parameters are obtained from the weld quality prediction model and marked to obtain the local module for this update. The local module for this update is updated for the first time, and the general rule without scenarios is added to the data-scarce model scenario layer for the second update. Based on the results of the first and second updates, the parameters of the weld quality prediction model are optimized to obtain the latest prediction model.

9. The method for weld quality prediction and process optimization based on a welding cloud platform according to claim 8, characterized in that, Starting from the leaf node layer as the cause, we reason backwards towards the intermediate hidden node layer and the root node layer to obtain the posterior causal probability that the current parameter anomaly causes the defect to occur, specifically: ； in, This represents the posterior causal probability that an abnormality in the current parameters leads to the occurrence of a defect. This indicates the probability of a defect occurring when the current parameter is abnormal. This indicates the probability that the current parameter is abnormal. This indicates the probability of a defect occurring.

10. The method for weld quality prediction and process optimization based on a welding cloud platform according to claim 8, characterized in that, Determining the general critical interval for continuous process parameters in invariant relationships includes: The ratio of the specific values ​​of continuous process parameters in all scenarios to the critical reference state values ​​in the scenarios is used as the parameter deviation. When the causal contribution of a continuous process parameter is greater than a preset contribution, obtain all values ​​of the parameter deviation, and then obtain the maximum and minimum values ​​from all values ​​to determine the general critical interval.

Images