A Leakage Monitoring Method and System Associated with Welding Defects of Brazed Parts

Abstract

The present application provides a leakage monitoring method and system associated with welding defects of brazed parts, relating to the field of brazing technology. It locates the coordinate set of the same type of welding defects under the defect type; under the combined action of the characteristics of the base material and the filler metal, a first-layer defect association topology structure is established; under the combined action of the brazing process characteristics and the interface characteristics, a second-layer defect association topology structure is established; the first and second-layer defect association topology structures are fused to establish a defect association multi-domain fusion model; the pressure change data during the welding process is collected, and anomaly recognition is performed according to the model, potential leakage points and leakage risks are marked, and a leakage discrimination mechanism is established for real-time leakage monitoring. The present application solves the technical problem in the prior art that due to the lack of real-time monitoring means for brazing leakage, it is difficult to timely detect and warn of leakage risks, affecting the brazing quality, improves the real-time and forward-looking of brazing leakage detection, and thus improves the brazing quality.

Description

Technical Field

[0001] The present application relates to the field of brazing technology, and in particular to a method and system for monitoring leakage associated with welding defects of brazed parts. Background Art

[0002] Brazing is a widely used connection technology in metal processing and assembly. It achieves connection through the flow and solidification of brazing material on the surface of the base material. Its advantage is that it can achieve efficient and precise joining, especially for high-performance components that need to be sealed. However, the internal leakage problem that may occur during brazing seriously affects the quality and reliability of the product.

[0003] At present, the monitoring of brazing internal leakage mainly relies on some traditional detection methods, such as the bubble method, pressure decay method, helium detection method, etc. These methods determine whether there is a problem by detecting the leakage after the welding is completed. For example, the bubble method detects leakage by immersing the weldment in liquid and applying pressure to observe the generation of bubbles; the helium detection method detects the concentration of leaked helium by filling helium into the weldment and using a helium mass spectrometer. Although these methods can detect the existence of leakage to a certain extent, most of them can only be detected after the welding is completed, and it is impossible to monitor the leakage in real time during the welding process. Therefore, it is difficult to warn of potential leakage risks in advance, so that the internal leakage problem can only be discovered after the fact, and it is impossible to intervene in time, which greatly affects the brazing quality and increases the welding cost and time consumption. Summary of the invention

[0004] The present application provides a leakage monitoring method and system associated with welding defects of brazed parts, which solves the technical problem that the prior art lacks real-time monitoring means for leakage during the brazing process, resulting in difficulty in timely detection and warning of leakage risks, thereby affecting the brazing quality. The application achieves the technical effect of improving the real-time and foresight of brazing leakage detection, thereby improving the brazing quality.

[0005] In view of the above problems, on the one hand, the present application provides a leakage monitoring method under the association of welding defects of brazed parts, the method comprising: based on a closed brazing assembly, locating a similar welding defect coordinate set under a defect type, the closed brazing assembly comprises a plurality of brazing parts, and the defect types include crack type, pore type, and slag type; introducing basic information of the base material and basic information of the brazing material, and under the parallel combination of the base material characteristics and the brazing material characteristics, establishing a first-layer defect association topological architecture associated with the similar welding defect coordinate set; introducing brazing process parameters and interface gap, and under the parallel combination of the brazing process characteristics and the interface characteristics, establishing a first-layer defect association topological architecture associated with the similar welding defect coordinate set. A second-layer defect association topological architecture is provided; the first-layer defect association topological architecture and the second-layer defect association topological architecture are integrated, the defect type is used for boundary definition, and a defect association multi-domain fusion model is established, and the defect association multi-domain fusion model is mapped one-to-one with the compressive stress distribution between multiple brazing parts; pressure change data is collected during the welding process of the closed brazing assembly, anomaly identification is performed according to the defect association multi-domain fusion model, and potential leakage points and leakage risks are marked; according to the potential leakage points and leakage risks, a leakage discrimination mechanism is established, real-time leakage monitoring is performed on multiple brazing parts of the closed brazing assembly, and leakage warning information is output.

[0006] On the other hand, the present application also provides a leakage monitoring system associated with welding defects of brazed parts, the system comprising: a defect coordinate positioning module, used to locate similar welding defect coordinate sets under defect types based on a closed brazing assembly, the closed brazing assembly comprising multiple brazing parts, and the defect types comprising cracks, pores, and slag inclusions; a first defect association topology module, used to introduce basic information of a base material and basic information of a brazing filler metal, and under the parallel combination of the base material characteristics and the brazing filler metal characteristics, establish a first layer of defect association topology architecture associated with similar welding defect coordinate sets; a second defect association topology module, used to introduce brazing process parameters and interface gaps, and under the parallel combination of the brazing process characteristics and the interface characteristics, establish a first layer of defect association topology architecture associated with similar welding defect coordinate sets. a second-layer defect association topological architecture connected to the first layer; a defect association multi-domain fusion module, used to fuse the first-layer defect association topological architecture and the second-layer defect association topological architecture, use the defect type to perform boundary definition, establish a defect association multi-domain fusion model, and the defect association multi-domain fusion model is mapped one-to-one with the compressive stress distribution between multiple brazing parts; a welding anomaly recognition module, used to collect pressure change data during the welding process of the closed brazing assembly, perform anomaly recognition according to the defect association multi-domain fusion model, and mark potential leakage points and leakage risks; a real-time leakage monitoring module, used to establish a leakage discrimination mechanism according to the potential leakage points and leakage risks, perform real-time leakage monitoring on multiple brazing parts of the closed brazing assembly, and output leakage warning information.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] First, by locating the type and coordinates of welding defects, basic data is collected to provide support for subsequent analysis. Next, the basic information of the base material and the brazing filler metal is introduced to establish the first-layer defect association model, which explains the formation of welding defects from the perspective of material properties. Then, the brazing process parameters and interface gap are introduced to establish the second-layer defect association model, which further explains the formation of welding defects from the perspective of process and structure. By fusing these two models and using defect types for boundary definition, a defect association multi-domain fusion model is established to achieve comprehensive analysis of multi-dimensional information. During the welding process, pressure change data is collected in real time, and anomalies are identified based on the multi-domain fusion model to promptly discover potential leakage points and risks. Finally, by establishing a leakage discrimination mechanism, real-time leakage monitoring of brazed parts is carried out, and leakage reminder information is output, realizing closed-loop management from monitoring to early warning.

[0009] In summary, this application realizes real-time monitoring and early warning of potential leakage risks in the brazing process through the construction of a multi-level defect association topological architecture and a multi-domain fusion model. By combining the multi-dimensional analysis of the base material, brazing material and process parameters, welding defects can be identified comprehensively and accurately, and reliable technical support is provided for brazing quality control. By collecting pressure change data and performing abnormality identification during the welding process, potential leakage points can be predicted and marked in advance, effectively avoiding the limitation of traditional methods that leakage problems are discovered only after the fact. Overall, this solution improves the accuracy and reliability of welding defect detection, and can also monitor leakage in real time during the welding process and provide early warning, effectively avoiding product performance degradation and safety hazards caused by leakage, and significantly improving the brazing quality and reliability of welding products.

[0010] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 A schematic flow chart of a leakage monitoring method associated with welding defects of brazed parts provided in an embodiment of the present application.

[0012] Figure 2 A schematic diagram of a flow chart of obtaining a parent material characteristic queue and a brazing filler metal characteristic queue in a leakage monitoring method associated with welding defects of a brazed part provided in an embodiment of the present application.

[0013] Figure 3A schematic flow chart of association rule mapping in a leakage monitoring method associated with welding defects of brazed parts provided in an embodiment of the present application.

[0014] Figure 4 A structural schematic diagram of a leakage monitoring system associated with welding defects of brazed parts provided in an embodiment of the present application.

[0015] Explanation of the accompanying drawings: defect coordinate positioning module 10, first defect association topology module 20, second defect association topology module 30, defect association multi-domain fusion module 40, welding anomaly identification module 50, real-time leakage monitoring module 60. DETAILED DESCRIPTION

[0016] The embodiments of the present application provide a leakage monitoring method and system associated with welding defects of brazed parts, thereby solving the technical problem that the prior art lacks real-time monitoring means for leakage during the brazing process, resulting in difficulty in timely detection and warning of leakage risks, thereby affecting the brazing quality. The embodiment of the present application achieves the technical effect of improving the real-time and foresight of brazing leakage detection, thereby improving the brazing quality.

[0017] Embodiment 1, as Figure 1 As shown, the embodiment of the present application provides a leakage monitoring method associated with welding defects of brazed parts, the method comprising:

[0018] Step S1: locating a similar welding defect coordinate set under a defect type based on a closed brazing assembly, wherein the closed brazing assembly includes a plurality of brazing parts, and the defect types include cracks, pores, and slag inclusions.

[0019] Specifically, a closed brazed assembly refers to an assembly composed of multiple brazed parts that forms a closed structure during the brazing process. For example, a closed box formed by multiple metal plates connected by brazing. Defect types refer to the types of defects that may occur during the welding process, including cracks, pores, slag inclusions, etc. The same type of welding defect coordinate set is a collection of the location coordinates of each defect in historical welding assemblies with similar structures to the closed brazed assembly, for welding defects of different types (such as cracks).

[0020] According to the structural characteristics of the closed brazed assembly, the welding data of the brazed assembly with the same structure is searched from historical records or data, including the brazing defect type and the corresponding brazing position. The welding data of the brazed assembly with the same structure can be obtained by scanning the closed brazed assembly through non-destructive testing technology (such as ultrasonic testing, X-ray testing, etc.) to collect the data of welding defects. For example, the closed brazed assembly is viewed using X-ray detection equipment, and the location of the welding defect is determined by analyzing the X-ray image. Then, according to the type of defect (crack type, pore type, slag type), its position coordinates are recorded respectively to form a set of coordinates of the same type of welding defects.

[0021] Step S2: introducing the basic information of the base material and the basic information of the solder, and establishing the first-level defect association topological framework associated with the same type of welding defect coordinate set under the parallel combination of the base material characteristics and the solder characteristics.

[0022] Specifically, the basic information of the base material refers to the basic characteristic information of the base material to be brazed, including the chemical composition, mechanical properties (such as strength, hardness, etc.), physical properties (such as thermal conductivity, thermal expansion coefficient, etc.) of the material. The basic information of the solder refers to the relevant basic characteristics of the filler material used for brazing, such as the chemical composition, melting point, wettability, etc. of the solder. The first-level defect association topological architecture is a structural model that associates the same welding defect coordinate set with the characteristics of the base material and solder, showing the connection between the defect location and the characteristics of the base material and solder.

[0023] Obtain basic information such as the chemical composition and mechanical properties of the parent material through material manuals or experimental measurements. At the same time, obtain basic information such as the chemical composition, melting point and fluidity of the solder through data provided by the solder supplier or experimental measurements. Associate similar welding defect coordinate sets with parent material and solder characteristics by establishing mathematical models or relationship charts. For example, the collected information such as thermal conductivity and hardness of the copper parent material and the melting point and wettability of the silver-based solder are linked to the previously determined welding defect coordinate set to construct the first-level defect association topology architecture.

[0024] This step establishes a preliminary correlation analysis of welding defects from the perspective of material properties, which helps to deeply understand the relationship between defect generation and material properties, and provides an important dimension for comprehensive analysis of welding defects.

[0025] Step S3: introducing brazing process parameters and interface gap, and establishing a second-level defect association topology framework associated with the same type of welding defect coordinate set under the parallel combination of brazing process characteristics and interface characteristics.

[0026] Specifically, brazing process parameters refer to the parameters that control the welding quality during the brazing process, including brazing temperature, brazing time, brazing pressure, etc. The interface gap refers to the gap size between the brazed parts, which will affect the flow of the brazing material and the welding quality. The second-level defect association topological framework is a structural model that associates the same welding defect coordinate set with the brazing process parameters and interface characteristics. It is similar to the topological framework in step S2, but focuses on the process and interface factors.

[0027] Through process tests or production records, brazing process parameters such as welding temperature, welding time, and welding pressure are obtained. The gap between brazed parts is measured by measuring tools (such as micrometers, vernier calipers, etc.). These brazing process parameters and interface gap data are recorded and stored, and then these data are linked to similar welding defect coordinate sets through data analysis software to build a second-level defect association topology architecture.

[0028] This step supplements the correlation analysis of welding defects from the perspective of process and interface, further improves the comprehensive understanding of the causes of welding defects, provides a more comprehensive analysis framework for welding defects, and improves the accuracy of defect prediction.

[0029] Step S4: integrating the first layer defect association topology architecture and the second layer defect association topology architecture, using the defect type for boundary definition, and establishing a defect association multi-domain fusion model, wherein the defect association multi-domain fusion model is mapped one-to-one with the compressive stress distribution between multiple brazing parts.

[0030] Specifically, the defect correlation multi-domain fusion model is a comprehensive mathematical model that integrates the characteristics of multiple fields, including the characteristics of the base material and the brazing filler metal, the brazing process parameters, the interface gap, etc. Through multi-level fusion, it can comprehensively analyze and predict the occurrence and development of defects in the brazing process. Compressive stress distribution refers to the pressure distribution generated inside the brazed parts due to the thermal expansion, contraction and structural constraints of the material during the brazing process.

[0031] The first layer defect association topology established in step S2 and the second layer defect association topology established in step S3 are fused. The fused data is bounded according to the defect type to ensure the accuracy and relevance of the data. Based on the fused data, a machine learning algorithm (such as support vector machine, neural network, etc.) is used to train the defect association multi-domain fusion model. During the model construction process, the compressive stress distribution of the brazed parts is simulated by finite element analysis software, and then the simulated compressive stress distribution data is associated with the corresponding parameters in the model, so that the model can reflect the influence of compressive stress distribution on welding defects.

[0032] This step creates a comprehensive, multi-factor correlation model and establishes a mapping relationship with the compressive stress distribution, providing a comprehensive theoretical basis for subsequent anomaly identification and leakage monitoring.

[0033] Step S5: collecting pressure change data during the welding process of the closed brazing assembly, performing abnormality identification according to the defect association multi-domain fusion model, and marking potential leakage points and leakage risks.

[0034] Specifically, during the welding process of the closed brazing assembly, pressure change data is collected using pressure sensors and other equipment. The collected data is then input into the defect-associated multi-domain fusion model established in step S4 for analysis, abnormal pressure points are identified, and the abnormal positions of the corresponding closed brazing assemblies are marked as potential leakage points, and the leakage risks that may exist at the potential leakage points are predicted. By combining the actual collected pressure change data with the theoretical model, abnormal conditions during the welding process can be accurately identified, potential leakage points and leakage risks can be marked, and key information can be provided for preventing leakage problems in advance.

[0035] Step S6: establishing a leakage identification mechanism according to the potential leakage points and leakage risks, performing real-time leakage monitoring on the multiple brazing parts of the closed brazing assembly, and outputting leakage warning information.

[0036] Specifically, the leakage discrimination mechanism refers to the rules and methods for determining whether a leakage has occurred based on potential leakage points and risks. Leakage reminder information refers to the reminder information output when it is determined that a potential leakage may occur, which is used to notify the operator to take measures. According to the potential leakage points and leakage risks marked in step S5, a leakage discrimination mechanism is established by writing a special algorithm or program. This mechanism can be run in a computer control system to monitor multiple brazed parts in a closed brazing assembly in real time. For example, on the brazing production line of the heat sink of an electronic device, a PLC (programmable logic controller) is used in combination with a written discrimination algorithm to monitor each brazing part in real time. Once the leakage risk is detected to reach the set threshold, a leakage reminder message is output.

[0037] This step realizes real-time leakage monitoring of brazed parts and timely outputs leakage warning information, which can effectively avoid quality problems caused by leakage and improve the quality and reliability of brazing.

[0038] Further, step S2 includes:

[0039] Step S21: Perform data mining based on the parent material basic information and the solder basic information to establish a substrate information database.

[0040] Step S22: performing a feature sorting operation in the substrate information library to obtain a parent material characteristic queue and a solder characteristic queue.

[0041] Step S23: performing association rule mapping through the parent material characteristic queue and the solder characteristic queue.

[0042] Specifically, the base material information database is a database that stores data related to the basic information of base materials and solders. Collect basic information of various types of base materials and solders, which can come from data sheets provided by material suppliers, previous brazing experiment records, etc. Use data mining tools (such as Python's Pandas library and MATLAB's data analysis toolbox) to process and analyze the collected data and extract useful information. Use Python's Pandas library for data cleaning, screening and statistical analysis. For example, use the Pandas library to clean and screen the chemical composition data of the base material to extract key information. Use data visualization tools (such as Python's Matplotlib library and Seaborn library) to visualize the data to help understand the data distribution and characteristics. For example, use the Matplotlib library to draw a distribution map of the tensile strength of the base material. After organizing the mined data information, establish a base material information database. Through data mining, the basic information of the base material and solder is integrated, and a comprehensive base material information database is established, which provides data support for subsequent feature sorting and association rule mapping.

[0043] Use feature selection algorithms (such as information gain, similarity analysis, etc.) to perform feature sorting on the base material basic information and solder basic information data in the base material information library, and generate base material characteristic queues and solder characteristic queues according to the results of the feature sorting operation. Feature sorting can make the data more orderly, highlight the relative importance of various characteristics, and help to map association rules more targetedly in the future.

[0044] The obtained parent material characteristic queue and solder characteristic queue are mapped with association rules. The parent material characteristic queue and solder characteristic queue are analyzed by using association rule algorithms in data mining, such as Apriori algorithm or FP-Growth algorithm, to determine the interaction and relationship between the parent material and solder characteristics. For example, certain parent material characteristics and solder characteristics may cause specific defects (such as pores, cracks, etc.) in the welding process. For example, in the association analysis of titanium alloy parent material and titanium-based solder, the FP-Growth algorithm is used to analyze the association between the parent material's hardness, toughness and other characteristics and the solder's wettability, solidification rate and other characteristics, and establish association rule mapping. The association rule mapping provides a more detailed relationship basis for establishing the first-level defect association topology architecture associated with the same welding defect coordinate set, which helps to more accurately analyze the relationship between welding defects and material characteristics.

[0045] Further, such as Figure 2 As shown, step S22 includes:

[0046] Step S221: extracting a first set of key characteristics from the base material information library using the base material basic information.

[0047] Step S222: taking a first key characteristic element randomly selected from the first key characteristic set as a reference characteristic, and using a distance metric to propose a similarity ranking.

[0048] Step S223: Taking the first key characteristic element as a benchmark characteristic, using information gain to propose a contribution ranking.

[0049] Step S224: Based on the similarity ranking and the contribution ranking, a self-balancing process is performed by a feature ranking operation.

[0050] Specifically, the first key characteristic set is a set of characteristics that have an important impact on the brazing process or welding defects, extracted from the basic information of the parent material in the base material information library. In the base material information library, the key characteristics in the basic information of the parent material are screened out based on the prior knowledge of the brazing process and welding defects. This prior knowledge may come from a large number of brazing experimental results or theoretical studies. For example, in the brazing of copper materials, if it is known that the purity, thermal conductivity and other characteristics of the copper material have a great influence on the welding quality, these characteristics are extracted from the basic information of the parent material in the base material information library to form the first key characteristic set.

[0051] A characteristic is randomly selected from the first key characteristic set and recorded as the first key characteristic element (such as the hardness of the base material). For the first key characteristic element, a data record is randomly selected from the base material information library and assigned a value (such as the specific hardness value of a certain base material) as the reference characteristic. Then, for other base material basic information in the base material information library, the distance between the first key characteristic element value and the reference characteristic in these data is calculated using a distance measurement method (such as Euclidean distance, Manhattan distance, etc.). The similarity is determined based on the distance, and finally the base material basic information is sorted according to the similarity to obtain a similarity sort.

[0052] In information theory, information gain refers to the degree of increase in the amount of information that an attribute can bring to classification. In the embodiment of the present application, information gain is used to measure the contribution of a certain characteristic to the welding process or welding defect analysis. Contribution ranking is the result of arranging the characteristics from large to small (or small to large) according to the size of information gain, reflecting the contribution of the characteristics to the classification related to welding defects.

[0053] Similarly, the first key characteristic element extracted previously (such as the hardness of the parent material) is used as the benchmark characteristic, and the information gain of the first key characteristic element value in other parent material basic information relative to the benchmark characteristic is calculated. For example, when analyzing the relationship between welding defects and parent material hardness, the contribution of different parent material basic information data to data richness is determined by calculating information gain, such as parent material basic information data with different carbon content, yield strength and other characteristics under similar hardness conditions. Contribution ranking is performed according to the calculation results. When calculating information gain, algorithm formulas related to information theory can be used, and calculations can be implemented with the help of data analysis software (such as related functions in the Scikit-learn library in Python). By using information gain to propose contribution ranking, the contribution of each characteristic to welding defects can be determined, which helps to identify key characteristics that have a greater impact on welding defects and provides an important basis for subsequent comprehensive ranking.

[0054] Based on the similarity sorting and contribution sorting results obtained above, the self-balancing processing of the feature sorting operation is performed. The self-balancing binary tree algorithm can be used for processing, and the sorting results obtained according to the similarity sorting and contribution sorting results are used as nodes. The first data in a sorting result (such as similarity sorting) can be selected as the root node, and for subsequent data, the insertion operation is performed according to the comprehensive order of the data in the two sortings. If a data (represented as data node B), according to the comparison rules of the self-balancing binary tree (such as the size of the comprehensive sorting value of data node B and root node A), if data node B is smaller than the root node feature A, then data node B is inserted into the left subtree of the root node; if feature B is greater than the root node feature A, then it is inserted into the right subtree of the root node. During the insertion process, after each node is inserted, check whether the binary tree is balanced, that is, the absolute value of the height difference between the left subtree and the right subtree of each node does not exceed 1. If it is found that the absolute value of the height difference between the left subtree and the right subtree of a node is greater than 1, an adjustment operation is required. By repeatedly inserting data and adjusting operations, a self-balancing binary tree is constructed. The node data is output according to the in-order traversal sequence of the binary tree (left-root-right), and the feature sorting result after self-balancing processing is obtained.

[0055] The feature sorting after self-balancing processing comprehensively considers the similarity between the characteristics and the contribution to the welding process or defect analysis, making the sorting result more scientific and reasonable, and providing a more accurate basis for subsequent operations such as obtaining the parent material characteristic queue.

[0056] Furthermore, step S224 also includes:

[0057] Step S224 - 1 : using the first key characteristic element as a reference characteristic, and using a feature sorting operation to determine a first parent material characteristic queue.

[0058] Step S224 - 2 : According to a preset ratio, randomly select the first key characteristic element, the second key characteristic element, up to the Nth key characteristic element, and determine the first parent material characteristic queue, the second parent material characteristic queue, up to the Nth parent material characteristic queue.

[0059] Step S224 - 3 : based on the first key characteristic element and the first parent material characteristic queue, the second key characteristic element and the second parent material characteristic queue, and so on to the Nth key characteristic element and the Nth parent material characteristic queue, a parent material characteristic queue is obtained.

[0060] Specifically, the preset ratio scale is a preset extraction ratio used to control the number of extracted key characteristic elements. The Nth key characteristic element is the Nth characteristic element randomly extracted from the first key characteristic set according to the preset ratio scale. The Nth parent material characteristic queue is a parent material characteristic queue determined based on the Nth key characteristic element.

[0061] The first key characteristic element is taken as the reference characteristic, and self-balancing processing is performed by feature sorting operation according to similarity sorting and contribution sorting to determine the first parent material characteristic queue.

[0062] According to a preset scale, a first key characteristic element, a second key characteristic element, and finally an Nth key characteristic element are randomly extracted from the first key characteristic set, wherein N is a positive integer representing the total number of key features to be extracted from the first key characteristic set according to the preset scale.

[0063] Referring to the generation method of the first parent material characteristic queue, the extracted second key characteristic element to the Nth key characteristic element are used as the benchmark to perform feature sorting operations to determine the corresponding second parent material characteristic queue to the Nth parent material characteristic queue. The first key characteristic element is combined with the first parent material characteristic queue, the second key characteristic element is combined with the second parent material characteristic queue to the Nth key characteristic element and the Nth parent material characteristic queue to determine the final parent material characteristic queue. The weighted average method can be used to set the weight of each characteristic queue according to the importance of each key characteristic element to the accuracy of welding defect identification, and then the sorting results of multiple characteristic queues are weighted and combined to determine the comprehensive score of each group of wood characteristic data, and then the final feature sorting operation is performed based on the comprehensive score.

[0064] By integrating multiple key characteristic elements and corresponding characteristic queues, the comprehensiveness and representativeness of the final parent material characteristic queue are ensured, providing an ordered feature list for subsequent association rule mapping and defect association topology architecture establishment.

[0065] Further, step S224-2 includes:

[0066] Perform preliminary classification according to the first key characteristic set to set characteristic clusters; and flexibly adjust the preset proportion scale through the characteristic clusters.

[0067] Specifically, each characteristic in the first key characteristic set is analyzed and preliminarily classified according to the type of characteristic (such as physical characteristics, chemical characteristics, etc.), the similarity of the influence on the welding results (such as the characteristics that affect the welding strength are classified into one category), or the value range, etc. For example, for the parent material, if the characteristics related to hardness (such as the crystal structure of the material, the content of elements that affect the hardness in the alloy composition, etc.) are classified into one characteristic cluster, the characteristics related to heat conduction (such as thermal conductivity, specific heat capacity, etc.) are classified into another characteristic cluster.

[0068] Determine the distribution of characteristic clusters, including the size of the clusters (i.e., the number of characteristics in each cluster) and the importance of the characteristics within the clusters (which can be determined by expert experience). For example, if a characteristic cluster contains key characteristics that have a significant impact on welding defects and the cluster is large, then it may be necessary to increase the preset scale associated with this cluster. This adjustment process can be achieved by writing an algorithm program, which sets different adjustment coefficients based on the importance and size according to the input cluster information (such as cluster size, importance weight, etc.), and calculates the adjusted scale.

[0069] Flexible adjustment of the preset scale can make the sampling operation more flexible and reasonable, better adapt to the distribution and importance of characteristics, improve the accuracy and effectiveness of the operation, and further optimize the entire welding defect analysis process.

[0070] Further, such as Figure 3 As shown, step S23 includes:

[0071] Step S231: extracting a second set of key characteristics from the substrate information database using the solder basic information.

[0072] Step S232: cross-combining the first key characteristic set and the second key characteristic set to form a solder-base material characteristic association pair.

[0073] Step S233: constructing an association rule mapping table through the solder-base material characteristic association pairs.

[0074] Specifically, in the base material information library, the basic information of the solder is analyzed, and the characteristics that have a greater impact on the welding quality are extracted therefrom to form a new characteristic set, which is recorded as the second key characteristic set. In the same way as the parent material characteristic queue is determined from the first key characteristic set, a similar feature sorting operation is performed on the second key characteristic set to generate a solder characteristic queue.

[0075] Traverse the parent material characteristic queue corresponding to the first key characteristic set, and traverse the brazing material characteristic queue corresponding to the second key characteristic set at the same time, enumerate the elements in the two characteristic queues in pairs, and obtain multiple brazing material-parent material characteristic association pairs. After determining multiple brazing material-parent material characteristic association pairs, it is also necessary to verify the brazing feasibility of the brazing material-parent material characteristic association pairs to ensure the feasibility of these brazing material-parent material characteristic association pairs in the actual welding process. Exemplarily, basic verification can be performed based on the brazing principle. For example, the melting point of the brazing material must be lower than the melting point of the parent material, which is the basic condition for brazing. Therefore, the comparison between the melting point of the brazing material and the melting point of the parent material is used as an important indicator. Welding sample data sets related to the brazing material-parent material characteristic association pairs can also be collected from welding experiment records, operation data of actual welding projects, and experience data of welding process experts. The welding sample data set includes data on parent material characteristics, brazing material characteristics, and welding results (such as welding quality, whether welding defects occur, etc.) under different welding conditions. The solder-base material characteristic association pairs in the welding sample data set are subjected to concentration value analysis. The solder-base material characteristic association pairs with a higher occurrence frequency are screened using a preset threshold. The screened results are matched with the multiple solder-base material characteristic association pairs combined above, and the solder-base material characteristic association pairs that do not appear in the screened results are eliminated.

[0076] Based on the verified solder-base material characteristic association pairs, an association rule mapping table is constructed to systematically organize and store the relationship between the solder-base material characteristic association pairs and other factors in the welding process, providing an important reference for subsequent welding process optimization, welding defect prediction, etc.

[0077] Further, step S4 includes:

[0078] Step S41: integrating the first layer defect association topology architecture and the second layer defect association topology architecture, using the defect type to define the boundary, and setting a multi-domain feature space.

[0079] Step S42: configuring a multi-domain feature vector based on the multi-domain feature space.

[0080] Step S43: Using the multi-domain feature vector, construct an input layer and an output layer of the defect association multi-domain fusion model, wherein the input layer is connected to the multi-domain feature space, and the output layer is used to distinguish between a normal welding state and a leakage risk state during the welding process.

[0081] Specifically, following the steps similar to the first-layer defect association topology architecture, the brazing process parameters and interface gap are introduced, and the second-layer defect association topology architecture associated with the same type of welding defect coordinate set is constructed under the parallel combination of the brazing process characteristics and the interface characteristics. The constructed first-layer defect association topology architecture and the second-layer defect association topology architecture are merged. The data related to material properties (from the first layer) and the data related to the process and interface (from the second layer) are merged into one data set according to the associated similar welding defect coordinate set. For example, the base material hardness data, the solder melting point data, the brazing temperature data, and the interface roughness data are integrated together. The association relationships in the integrated data set are sorted out, and the association relationships between the various characteristics are re-analyzed. For example, the joint influence of the base material hardness and the brazing temperature on a certain defect type, or the interaction relationship between the solder melting point and the interface roughness when a specific defect is generated, is explored.

[0082] For each defect type, analyze the range of characteristic values ​​associated with it in the fusion architecture dataset. Taking pore defects as an example, analyze the range of values ​​of characteristics such as the oxygen content of the base material, the fluidity of the brazing material, the shielding gas flow rate during brazing, and the cleanliness of the interface when pore defects occur. Establish clear boundary conditions based on the analysis of the characteristic range of each defect type. For example, for crack defects, set the upper and lower limits of characteristics such as the stress concentration factor of the base material, the solidification rate of the brazing material, and the cooling rate after brazing as boundary conditions. Set the multi-domain characteristic space with these boundary conditions as constraints.

[0083] Based on the multi-domain feature space that has been set up, each characteristic factor is quantified and a multi-domain feature vector is configured. This multi-domain feature vector contains multiple characteristic information in the multi-domain feature space and can be used to describe various states in the welding process. For example, the multi-domain feature space contains characteristics such as the hardness of the base material, the melting point of the solder, and the brazing temperature. These characteristics are organized in a certain order to form a vector, which is the multi-domain feature vector. The configuration of the multi-domain feature vector can represent the complex characteristic information in the multi-domain feature space in a concise and orderly form, which is convenient for the subsequent construction of the input layer and output layer of the defect association multi-domain fusion model, and provides an effective data representation for the establishment of the model.

[0084] The configured multi-domain feature vector is used to construct the input layer and output layer of the defect-associated multi-domain fusion model. The multi-domain feature vector is used as the input content of the input layer, and the connection between the input layer and the multi-domain feature space is established, so that the defect-associated multi-domain fusion model can obtain various characteristic information in the multi-domain feature space as input. For the output layer, it is designed to output two states, namely, normal welding state and leakage risk state. For example, a deep learning framework (such as TensorFlow or PyTorch) can be used to construct a defect-associated multi-domain fusion model using a neural network model, and the number of neurons in the input layer is determined according to the dimension of the multi-domain feature vector, and the number of neurons in the output layer is determined according to the two states to be output (such as one neuron can be used, and the output 0 indicates a normal welding state, and the output 1 indicates a leakage risk state). The model is trained using labeled welding data (labeled as normal or defective state, etc.). During the training process, the weight and bias of the model are adjusted according to the set loss function (such as the mean square error function, etc.) to make the output of the model as close to the real welding state label as possible. The model is optimized through methods such as cross-validation to improve its generalization ability, ensure that the model can accurately predict defect associations on unseen data, and finally establish a defect association multi-domain fusion model.

[0085] In summary, the leakage monitoring method for brazing parts associated with welding defects provided in the embodiments of the present application has the following technical effects:

[0086] By locating the type and coordinates of welding defects, basic data is collected to provide support for subsequent analysis. Next, the basic information of the base material and the brazing filler metal is introduced. Through data mining and feature sorting, key characteristics are extracted from the basic information of the base material and the brazing filler metal, and a base material information library is established. By cross-combining the key characteristics of the base material and the brazing filler metal, characteristic association pairs are formed, and an association rule mapping table is constructed to establish the association relationship between the base material characteristics and the brazing filler metal characteristics, forming the first layer of defect association model, which explains the formation of welding defects from the perspective of material characteristics. Then, the brazing process parameters and interface gap are introduced to establish the second layer of defect association model, which further explains the formation of welding defects from the perspective of process and structure. By fusing these two layers of models and using defect types for boundary definition, a defect association multi-domain fusion model is established, which realizes the comprehensive analysis of multi-dimensional information. During the welding process, the pressure change data is collected in real time, and anomalies are identified according to the multi-domain fusion model to timely discover potential leakage points and risks. By establishing a leakage discrimination mechanism, real-time leakage monitoring of brazed parts is carried out, and leakage reminder information is output, realizing closed-loop management from monitoring to early warning.

[0087] Overall, the embodiments of the present application achieve real-time monitoring and early warning of potential leakage risks during the brazing process through the construction of a multi-level defect association topology architecture and a multi-domain fusion model. By combining the multi-dimensional analysis of the base material, brazing material and process parameters, welding defects can be identified comprehensively and accurately, and reliable technical support is provided for brazing quality control. By collecting pressure change data and performing abnormality identification during the welding process, potential leakage points can be predicted and marked in advance, effectively avoiding the limitations of traditional methods that only discover leakage problems after the fact, improving the accuracy and reliability of welding defect detection, and being able to monitor leakage in real time during the welding process and provide early warning, effectively avoiding product performance degradation and safety hazards caused by leakage, and significantly improving the brazing quality and reliability of welding products.

[0088] Embodiment 2, as Figure 4 As shown, the embodiment of the present application provides a leakage monitoring system associated with welding defects of brazed parts, the system comprising:

[0089] The defect coordinate positioning module 10 is used to locate the same type of welding defect coordinate set under the defect type based on the closed brazing assembly, the closed brazing assembly includes a plurality of brazing parts, and the defect types include crack type, pore type, and slag inclusion type.

[0090] The first defect association topology module 20 is used to introduce the basic information of the base material and the basic information of the solder, and establish the first layer of defect association topology architecture associated with the same type of welding defect coordinate set under the parallel combination of the base material characteristics and the solder characteristics.

[0091] The second defect association topology module 30 is used to introduce brazing process parameters and interface gaps, and establish a second layer of defect association topology architecture associated with the same type of welding defect coordinate set under the parallel combination of brazing process characteristics and interface characteristics.

[0092] The defect-related multi-domain fusion module 40 is used to fuse the first-layer defect-related topological architecture and the second-layer defect-related topological architecture, use the defect type to define the boundary, and establish a defect-related multi-domain fusion model. The defect-related multi-domain fusion model is mapped one-to-one with the compressive stress distribution between multiple brazing parts.

[0093] The welding anomaly identification module 50 is used to collect pressure change data during the welding process of the closed brazing assembly, perform anomaly identification according to the defect association multi-domain fusion model, and mark potential leakage points and leakage risks.

[0094] The real-time leakage monitoring module 60 is used to establish a leakage identification mechanism according to the potential leakage points and leakage risks, perform real-time leakage monitoring on the multiple brazing parts of the closed brazing assembly, and output leakage warning information.

[0095] Furthermore, the first defect-associated topology module 20 in the embodiment of the present application is further configured to perform the following steps:

[0096] Data mining is performed based on the basic information of the base material and the basic information of the solder to establish a base material information library; feature sorting operations are performed in the base material information library to obtain a base material characteristic queue and a solder characteristic queue; and association rule mapping is performed through the base material characteristic queue and the solder characteristic queue.

[0097] Furthermore, the first defect-associated topology module 20 in the embodiment of the present application is further configured to perform the following steps:

[0098] In the substrate information library, a first key characteristic set is extracted through the substrate basic information; a first key characteristic element randomly selected from the first key characteristic set is used as a reference characteristic, and a similarity sorting is proposed using a distance metric; a contribution sorting is proposed using the first key characteristic element as a reference characteristic using information gain; based on the similarity sorting and the contribution sorting, a self-balancing process is performed using a feature sorting operation.

[0099] Furthermore, the first defect-associated topology module 20 in the embodiment of the present application is further configured to perform the following steps:

[0100] Taking the first key characteristic element as the reference characteristic, a feature sorting operation is used to determine the first parent material characteristic queue; according to a preset proportion scale, the first key characteristic element, the second key characteristic element, up to the Nth key characteristic element are randomly selected to determine the first parent material characteristic queue, the second parent material characteristic queue, up to the Nth parent material characteristic queue; based on the first key characteristic element and the first parent material characteristic queue, the second key characteristic element and the second parent material characteristic queue, up to the Nth key characteristic element and the Nth parent material characteristic queue, the parent material characteristic queue is obtained.

[0101] Furthermore, the first defect-associated topology module 20 in the embodiment of the present application is further configured to perform the following steps:

[0102] Perform preliminary classification according to the first key characteristic set to set characteristic clusters; and flexibly adjust the preset proportion scale through the characteristic clusters.

[0103] Furthermore, the first defect-associated topology module 20 in the embodiment of the present application is further configured to perform the following steps:

[0104] In the substrate information library, a second key characteristic set is extracted through the solder basic information; the first key characteristic set and the second key characteristic set are cross-combined to form solder-parent material characteristic association pairs; and an association rule mapping table is constructed through the solder-parent material characteristic association pairs.

[0105] Furthermore, the defect association multi-domain fusion module 40 in the embodiment of the present application is also used to perform the following steps:

[0106] The first layer defect association topology architecture and the second layer defect association topology architecture are integrated, the defect type is used for boundary definition, and a multi-domain feature space is set; based on the multi-domain feature space, a multi-domain feature vector is configured; using the multi-domain feature vector, an input layer and an output layer of the defect association multi-domain fusion model are constructed, wherein the input layer is connected to the multi-domain feature space, and the output layer is used to distinguish between a normal welding state and a leakage risk state during the welding process.

[0107] Through the above-mentioned detailed description of a leakage monitoring method associated with a welding defect of a brazed part in this specification, those skilled in the art can clearly understand that a leakage monitoring system associated with a welding defect of a brazed part in this embodiment is a system. For the system disclosed in the second embodiment, since it corresponds to the method disclosed in the first embodiment and has corresponding functional modules and beneficial effects, the relevant parts can be referred to the description of the method part.

[0108] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for monitoring leakage under the association of welding defects of brazed parts, characterized in that: The method comprises: Based on a closed brazing assembly, locating a similar welding defect coordinate set under a defect type, the closed brazing assembly includes a plurality of brazing parts, and the defect types include crack type, porosity type, and slag inclusion type; The basic information of the base material and the solder are introduced. Under the parallel combination of the base material characteristics and the solder characteristics, the first-level defect association topology architecture associated with the same welding defect coordinate set is established, including: Perform data mining based on the basic information of the parent material and the basic information of the solder to establish a base material information database; Perform feature sorting operations in the substrate information database to obtain a parent material characteristic queue and a solder characteristic queue; Performing association rule mapping through the parent material characteristic queue and the solder characteristic queue; The brazing process parameters and interface gap are introduced, and under the parallel combination of brazing process characteristics and interface characteristics, a second-level defect association topological framework associated with the same welding defect coordinate set is established; The first layer defect association topology architecture and the second layer defect association topology architecture are integrated, the defect type is used for boundary definition, and a defect association multi-domain fusion model is established, wherein the defect association multi-domain fusion model is mapped one by one to the compressive stress distribution between multiple brazing parts; Collect pressure change data during the welding process of the closed brazing assembly, identify anomalies based on the defect correlation multi-domain fusion model, and mark potential leakage points and leakage risks; According to the potential leakage points and leakage risks, a leakage identification mechanism is established to perform real-time leakage monitoring on multiple brazing parts of the closed brazing assembly and output leakage warning information.

2. A method for monitoring leakage in association with welding defects of brazed parts according to claim 1, characterized in that: Performing a feature sorting operation in the substrate information library to obtain a parent material characteristic queue and a solder characteristic queue, the method comprising: In the substrate information library, extracting a first key characteristic set through the parent material basic information; Taking the first key characteristic element randomly selected from the first key characteristic set as the benchmark characteristic, and using the distance metric to propose a similarity ranking; Taking the first key characteristic element as the benchmark characteristic, information gain is used to propose contribution ranking; Based on the similarity ranking and the contribution ranking, a self-balancing process is performed by a feature ranking operation.

3. A method for monitoring leakage in association with welding defects of brazed parts as claimed in claim 2, characterized in that: Based on the similarity ranking and the contribution ranking, a self-balancing process is performed by a feature ranking operation, and the method further includes: Taking the first key characteristic element as a reference characteristic, a feature sorting operation is used to determine a first parent material characteristic queue; According to a preset proportion scale, randomly select the first key characteristic element, the second key characteristic element, and the Nth key characteristic element to determine the first parent material characteristic queue, the second parent material characteristic queue, and the Nth parent material characteristic queue; A parent material characteristic array is obtained based on the first key characteristic element and the first parent material characteristic array, the second key characteristic element and the second parent material characteristic array, and finally the Nth key characteristic element and the Nth parent material characteristic array.

4. A method for monitoring leakage in association with welding defects of brazed parts as claimed in claim 3, characterized in that: According to a preset scale, randomly extracting a first key characteristic element, a second key characteristic element, and finally an Nth key characteristic element, the method comprising: Performing preliminary classification according to the first key characteristic set and setting characteristic clusters; The preset scale is flexibly adjusted through the characteristic clusters.

5. A method for monitoring leakage in association with welding defects of brazed parts as claimed in claim 2, characterized in that: By using the parent material characteristic queue and the solder characteristic queue, association rule mapping is performed, and the method includes: In the substrate information database, extracting a second key characteristic set through the solder basic information; By cross-combining the first key characteristic set and the second key characteristic set, a brazing filler metal-base material characteristic correlation pair is formed; An association rule mapping table is constructed through the solder-base material characteristic association pairs.

6. A method for monitoring leakage in association with welding defects of brazed parts as claimed in claim 1, characterized in that: The first layer defect association topology architecture and the second layer defect association topology architecture are integrated, the defect type is used for boundary definition, and a defect association multi-domain fusion model is established. The method includes: The first layer defect association topology architecture and the second layer defect association topology architecture are integrated, the defect type is used for boundary definition, and a multi-domain feature space is set; Based on the multi-domain feature space, configuring a multi-domain feature vector; The multi-domain feature vector is used to construct the input layer and the output layer of the defect-associated multi-domain fusion model, wherein the input layer is connected to the multi-domain feature space, and the output layer is used to distinguish between a normal welding state and a leakage risk state during the welding process.

7. A leakage monitoring system associated with welding defects of brazed parts, characterized in that: The system is used to execute a leakage monitoring method associated with welding defects of a brazed part according to any one of claims 1 to 6, comprising: A defect coordinate positioning module is used to locate a set of similar welding defect coordinates under defect types based on a closed brazing assembly, wherein the closed brazing assembly includes a plurality of brazing parts, and the defect types include cracks, pores, and slag inclusions; The first defect association topology module is used to introduce the basic information of the base material and the basic information of the solder, and establish the first layer of defect association topology architecture associated with the same welding defect coordinate set under the parallel combination of the base material characteristics and the solder characteristics; The second defect association topology module is used to introduce brazing process parameters and interface gaps, and establish a second-level defect association topology architecture associated with the same welding defect coordinate set under the parallel combination of brazing process characteristics and interface characteristics; A defect-related multi-domain fusion module is used to fuse the first-layer defect-related topological architecture and the second-layer defect-related topological architecture, use the defect type to define the boundary, and establish a defect-related multi-domain fusion model, wherein the defect-related multi-domain fusion model is mapped one-to-one with the compressive stress distribution between multiple brazing parts; A welding anomaly identification module is used to collect pressure change data during the welding process of the closed brazing assembly, identify anomalies based on the defect association multi-domain fusion model, and mark potential leakage points and leakage risks; The real-time leakage monitoring module is used to establish a leakage identification mechanism according to the potential leakage points and leakage risks, perform real-time leakage monitoring on the multiple brazing parts of the closed brazing assembly, and output leakage warning information.

Images