Gradient welding strength control method and system for steel pipe

Abstract

The application provides a gradient welding strength control method and system for a steel pipe, and relates to the technical field of welding control. The method comprises the following steps: collecting welding process data and welding performance data of the target steel pipe with respect to a plurality of welding history samples, extracting a plurality of stage performance deviation sequences and stage control deviation sequences of each welding history sample, extracting a plurality of candidate deviation observation points and compensation end points of each welding history sample, identifying a plurality of deviation compensation event intervals and constructing a plurality of deviation compensation event units of each welding history sample, grouping the plurality of deviation compensation event units to construct a plurality of compensation event sets, performing compensation effect inversion processing on each compensation event set to generate a necessity event set, extracting a target welding compensation core corresponding to each necessity event set, and performing online welding control optimization on the target steel pipe based on the plurality of target welding compensation cores. The application realizes the improvement of the stability of the steel pipe welding process.

Description

Technical Field

[0001] This invention relates to the field of welding control technology, and in particular to a method and system for controlling the gradient welding strength of steel pipes. Background Technology

[0002] Steel pipes are widely used in oil and gas transportation, pressure pipelines, engineering structures, and equipment manufacturing. The quality of their welded joints directly affects the strength and safety of the overall structure. Steel pipe arc welding is typically not a single-pass process but rather a multi-pass, continuous welding process. Influenced by factors such as welding heat input, assembly errors, changes in welding position, and fluctuations in material properties, the welded joint will exhibit different microstructures and mechanical property distributions in different areas. If this distribution is not coordinated, localized areas may experience abrupt changes in properties, uneven hardness, or inconsistent quality between sections, thus affecting the overall quality of the joint.

[0003] Some steel pipe welding control methods focus on setting process parameters such as current, voltage, and welding speed, or adjusting for obvious anomalies at a certain moment during welding. These methods can handle some sudden deviations, but they lack effective solutions for the more common situation in actual production: the continuous accumulation of small fluctuations across multiple welding stages, gradually affecting the formation and performance distribution of different areas of the joint. In many cases, welding problems are not directly caused by a single instability in a particular stage, but rather gradually form through the interaction of preceding and subsequent stages, ultimately manifesting as uneven joint quality distribution, local performance imbalances, and a decline in overall consistency.

[0004] Therefore, how to more effectively control the quality distribution changes of welded joints in different regions and stages for the continuous production process of arc welding of steel pipes, and reduce the amplification of local anomalies, sudden changes in performance and decline in overall consistency caused by the imbalance of distribution changes and the accumulation of fluctuations, is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, this invention proposes a gradient welding strength control method and system for steel pipes, reducing the impact of local abrupt changes and subsequent cumulative instability on the welding quality of steel pipes.

[0006] To achieve the above objectives, the first aspect of the present invention provides a method for controlling the gradient welding strength of steel pipes, comprising: Welding process data and welding performance data of the target steel pipe based on multiple historical welding samples were collected to determine multiple welding stages and welding benchmark conditions of the target steel pipe. Based on welding benchmark data, multiple stage performance deviation sequences and stage control deviation sequences are extracted for each welding history sample. Performance deviation features of each stage performance deviation sequence are constructed, and multiple candidate deviation observation points and compensation termination points are extracted for each welding history sample. Multiple deviation compensation event intervals are identified based on multiple compensation end points and multiple candidate deviation observation points. Based on the control deviation characteristics of multiple welding stages in the deviation compensation event intervals, multiple deviation compensation event units are constructed for each welding history sample. Identify the deviation compensation matching status between any two deviation compensation event units, group multiple deviation compensation event units based on the deviation compensation matching status, and construct multiple compensation event sets corresponding to multiple deviation compensation event units; Each set of compensation events is processed by inversion to generate a set of necessary events. The target welding compensation core corresponding to each set of necessary events is extracted. Based on multiple target welding compensation cores, the online welding control of the target steel pipe is optimized.

[0007] Preferably, constructing performance deviation features for each stage's performance deviation sequence and extracting multiple candidate deviation observation points and compensation termination points for each welding history sample includes: For the stage control deviation sequence, the welding reference condition data includes the stage performance reference sequence and the stage control reference sequence for each welding stage. The sample stage performance sequence of each welding history sample in each welding stage is extracted from the welding process data. Based on the stage performance reference sequence, the performance deviation of the sample stage performance sequence of the welding history sample is detected, the stage performance deviation value of each welding performance parameter is extracted, and the stage performance deviation sequence of each welding history sample in different welding stages is generated. Multiple stage performance deviation values ​​in the fusion stage performance deviation sequence are used to generate stage performance deviation parameters, which serve as performance deviation features of the stage performance deviation sequence. Welding stages with performance deviation features not less than the deviation observation threshold are recorded as candidate deviation observation points, and welding stages with performance deviation features not greater than the stable observation threshold are recorded as candidate compensation end points. Multiple candidate compensation end points are continuously scanned to identify multiple continuous end segments composed of multiple candidate compensation end points. The first candidate compensation end point of each continuous end segment is recorded as the compensation end point.

[0008] Preferably, multiple deviation compensation event intervals are identified based on multiple compensation end points and multiple candidate deviation observation points. Based on the control deviation characteristics of multiple welding stages within each deviation compensation event interval, multiple deviation compensation event units are constructed for each welding history sample, including: Based on multiple compensation end points, multiple welding stages are divided into multiple candidate intervals. Candidate intervals containing at least one candidate deviation observation point are selected. The first candidate deviation observation point in the selected candidate interval is taken as the deviation observation start point. Welding stages before the deviation observation start point in each candidate interval are removed. The deviation observation start point and the compensation end point of the candidate interval are taken to form a deviation compensation event interval. The candidate deviation observation point with the largest control deviation characteristic in the deviation compensation event interval is identified and recorded as the compensation observation anchor point. The deviation observation start point, compensation observation anchor point and compensation end point belonging to the same deviation compensation event interval constitute one deviation compensation event unit of the welding history sample.

[0009] Preferably, performing compensation effect inversion processing on each compensation event set to generate a necessity event set includes: Determine the compensation window for each deviation compensation event unit in the compensation event set, where the upper and lower boundaries of the compensation window are the next welding stage and the compensation end point of the compensation observation anchor point. Align the compensation windows of multiple deviation compensation event units in the compensation event set and construct the standard control deviation matrix for each deviation compensation event unit, which includes multiple standard control deviation vectors. Each standard control deviation vector in the standard control deviation matrix is ​​marked as a compensation segment. The compensation effect inversion detection is performed on multiple compensation segments of each deviation compensation event unit. This includes constructing a control masking matrix of the deviation compensation event unit for each compensation segment, identifying the control events of the deviation compensation event unit under different compensation segments in the compensation event set based on the control masking matrix, and calculating the compensation loss parameter of each compensation segment based on the control events. Based on the compensation loss parameters, multiple target compensation segments are determined for each deviation compensation event unit, and the target compensation segments of multiple deviation compensation event units are combined to generate a necessary event set.

[0010] Preferably, extracting the target welding compensation core corresponding to each set of necessary events includes: Determine the multiple welding stages contained in each set of necessary events, perform necessity analysis on each welding stage based on the multiple target compensation segments of each deviation compensation event unit, and identify multiple welding compensation stages in the set of necessary events. For each welding compensation stage, the control direction synchronization status of multiple welding control parameters is identified to determine the control synchronization parameters of each welding control parameter in the welding compensation stage. Based on the control synchronization parameters, multiple compensation control parameters for each welding compensation stage are identified. Based on the multiple standard control deviation vectors corresponding to the compensation control parameters, the welding compensation information of each compensation control parameter is determined, including the compensation control direction and compensation control intensity of the compensation control parameter. The welding stage preceding the first welding compensation stage is marked as the online observation stage of compensation. The stage performance deviation sequence of each deviation compensation event unit in the online observation stage of compensation is extracted and fused to obtain the target compensation monitoring status sequence. Based on the welding compensation information of multiple compensation control parameters included in multiple welding compensation stages, and the target compensation monitoring state sequence of the online compensation observation stage, a target welding compensation kernel with a set of necessary events is constructed.

[0011] Preferably, the online welding control optimization based on multiple target welding compensation checks of the target steel pipe includes: During the welding process of the target steel pipe, after the completion of each welding stage, the online welding process data of the current welding stage is extracted, and the online stage performance deviation sequence corresponding to the online welding process data is extracted based on the welding reference working condition data. The online stage performance deviation parameter is calculated based on the online stage performance deviation sequence. If the online stage performance deviation parameter is not less than the deviation observation threshold, the target welding compensation core matching process is executed. This includes determining multiple target compensation monitoring state sequences corresponding to the welding stage to which the online stage performance deviation sequence belongs, calculating the state distance between the online stage performance deviation sequence and the multiple target compensation monitoring state sequences respectively, determining the minimum state distance and comparing it with the preset compensation trigger distance threshold. If the minimum state distance is not greater than the compensation trigger distance threshold, welding control compensation optimization is performed based on the target welding compensation core corresponding to the minimum state distance. Welding control compensation optimization includes performing phased compensation control on the subsequent welding stages of the target steel pipe based on multiple welding compensation stages corresponding to the target welding compensation core, as well as the compensation control direction and compensation control intensity of the compensation control parameters in each welding compensation stage. During the execution of compensation control, the online stage performance deviation parameters of the target steel pipe in the current welding stage are continuously updated. When the online stage performance deviation parameters fall below the stable observation threshold, the welding control compensation optimization process for the target welding compensation core ends.

[0012] Preferably, grouping multiple deviation compensation event units based on the deviation compensation matching state to construct multiple compensation event sets corresponding to multiple deviation compensation event units includes: Identify the compensation matching state of any two deviation compensation event units, including calculating the deviation state distance and compensation end state distance between the two deviation compensation event units based on the stage control deviation sequence of the compensation observation anchor point and the compensation end point, calculating the compensation window length distance between the two deviation compensation event units based on the compensation window of the deviation compensation event unit, and determining the compensation matching state between the two deviation compensation event units based on the deviation state distance, compensation end state distance and compensation window length distance. The compensation matching state includes the same compensation structure and different compensation structures. Based on the compensation matching status of any two deviation compensation event units, multiple deviation compensation event units are progressively merged and grouped to generate multiple compensation event sets corresponding to multiple deviation compensation event units.

[0013] A second aspect of the present invention provides a gradient welding strength control system for steel pipes, for implementing the above-described gradient welding strength control method for steel pipes, comprising: The steel pipe welding data acquisition module is used to collect welding process data and welding performance data of the target steel pipe based on multiple historical welding samples, and to determine multiple welding stages and welding benchmark working condition data of the target steel pipe. The welding deviation feature extraction module is used to extract multiple stage performance deviation sequences and stage control deviation sequences for each welding history sample based on welding benchmark working condition data, construct the performance deviation features of each stage performance deviation sequence, and extract multiple candidate deviation observation points and compensation termination points for each welding history sample. The compensation event unit identification module is used to identify multiple deviation compensation event intervals based on multiple compensation end points and multiple candidate deviation observation points. Based on the control deviation characteristics of multiple welding stages in the deviation compensation event interval, multiple deviation compensation event units are constructed for each welding history sample. The compensation event unit grouping module is used to identify the deviation compensation matching status between any two deviation compensation event units, group multiple deviation compensation event units based on the deviation compensation matching status, and construct multiple compensation event sets corresponding to multiple deviation compensation event units. The online welding control optimization module is used to perform compensation effect inversion processing on each compensation event set to generate a necessary event set, extract the target welding compensation core corresponding to each necessary event set, and perform online welding control optimization on the target steel pipe based on multiple target welding compensation cores.

[0014] The present invention has the following beneficial effects: This invention addresses the problem of joint quality distribution imbalance caused by multi-stage fluctuation accumulation during steel pipe arc welding. Based on historical welding sample data of the target steel pipe, the deviation entry, deviation accumulation, and recovery processes during welding are extracted into multiple deviation compensation event units. Event units with similar compensation structures are grouped to form compensation event sets. Further inversion analysis of the compensation effect is conducted to identify the target compensation segments that truly play a role in the compensation process, eliminate accompanying control changes, and extract target welding compensation cores corresponding to different deviation state types. During online control, the deviation state of the current welding stage is matched with the target compensation monitoring state of each target welding compensation core. When trigger conditions are met, the corresponding compensation core is invoked to implement staged compensation control for subsequent welding stages. This reduces the continuous accumulation and amplification of local deviations, improves the uncoordinated changes in joint quality distribution, and enhances the stability of the steel pipe welding process and the consistency of welded joint quality. Attached Figure Description

[0015] Figure 1 This is a flowchart of a gradient welding strength control method for steel pipes provided in an embodiment of the present invention.

[0016] Figure 2 This is a structural diagram of a gradient welding strength control system for steel pipes provided in an embodiment of the present invention. Detailed Implementation

[0017] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention.

[0018] Please see Figure 1 One aspect of this invention provides a gradient welding strength control method for steel pipes. Addressing the continuous influence relationships between different welding stages during arc welding of steel pipes, this method no longer focuses solely on anomalies at a single moment or for a single parameter. Instead, it analyzes the generation, recovery, and compensation behaviors of deviations in historical welding samples to extract stable compensating control patterns during the welding process. These extracted patterns are further transformed into online-callable welding compensation kernels for gradient control optimization of the target steel pipe welding process. This allows for more targeted compensation of local deviations during welding in subsequent stages, thereby improving the overall quality distribution coordination of the welded joint.

[0019] Specifically, the gradient welding strength control method for steel pipes includes the following steps: Step S1: Collect welding process data and welding performance data of the target steel pipe from multiple historical welding samples to determine multiple welding stages and welding reference conditions of the target steel pipe.

[0020] Specifically, welding process data refers to the process control information that can be directly adjusted and executed during the welding process. This includes specific control data for various common welding control parameters in arc welding, such as welding current, welding voltage, welding speed, welding torch oscillation amplitude, welding torch oscillation frequency, and the torch's travel angle and working angle. Welding performance data characterizes the staged welding state that has been achieved at a certain point in the welding process, such as weld width, penetration depth, and weld center offset.

[0021] Welding process and performance data from multiple collected historical welding samples were analyzed, and the welding process was divided into stages based on the target steel pipe's welding scenario. These welding stages can be rationally set according to the actual arc welding process of the steel pipe. For example, they can be determined based on weld bead levels and the internal process of a single pass, specifically including weld bead stages such as root pass, hot pass, fill pass, and cap pass, as well as internal stages of a single pass such as the starting stage, steady-state stage, lap stage, and arc termination stage. Simultaneously, the corresponding welding reference conditions for each welding stage were determined.

[0022] The welding reference condition data includes at least a stage performance reference sequence and a stage control reference sequence corresponding to each welding stage. The stage performance reference sequence characterizes the standard performance state of the target steel pipe at each welding stage under the reference condition, i.e., the benchmark reference values ​​for different welding performance parameters. The stage control reference sequence characterizes the standard control state of the target steel pipe at each welding stage under the reference condition, i.e., the benchmark reference values ​​for different welding control parameters. The welding reference condition data can be obtained by directly determining the specific values ​​of different welding control parameters and welding performance parameters for each welding stage based on expert experience. Alternatively, experts can determine multiple representative template samples based on actual production quality requirements, statistically analyze the welding performance parameters and welding control parameters according to the welding stage, and calculate the statistical values ​​(e.g., mean, median) of different welding performance parameters and welding control parameters at different welding stages in multiple template samples to generate the stage performance reference sequence and stage control reference sequence corresponding to each welding stage.

[0023] It is worth noting that the template samples that may be involved in the construction of welding benchmark data can specifically be historical samples selected by experts, characterized by small fluctuations in each welding stage, relatively stable overall process, and satisfactory post-weld quality results. These samples are used to generate stage performance reference sequences and stage control reference sequences for each welding stage. The acquired historical welding samples not only include the aforementioned template samples but also preferably include historical samples that exhibited certain stage deviations during the welding process but recovered to a stable state through compensation in subsequent welding stages, and whose post-weld quality results still meet the requirements. For example, among a large number of samples with qualified welding quality, if the deviations of different welding performance parameters and welding control parameters from their corresponding benchmark reference values ​​at different welding stages are all no greater than a preset deviation threshold, these samples are recorded as template samples, and the remaining samples are recorded as welding historical samples. This embodiment does not limit the specific deviation states of the template samples.

[0024] Step S2: Based on the welding benchmark working condition data, extract multiple stage performance deviation sequences and stage control deviation sequences for each welding history sample, construct the performance deviation features of each stage performance deviation sequence, and extract multiple candidate deviation observation points and compensation termination points for each welding history sample.

[0025] The stage performance deviation sequence is primarily used to characterize the overall deviation of welding performance parameters at each stage from the baseline condition, reflecting the stage-specific quality status of each welding stage. The stage control deviation sequence characterizes the control differences in welding control parameters at each stage relative to the baseline condition, reflecting the actual deviations from the baseline control behaviors adopted by the welding history samples during the welding process. Based on the stage performance deviation sequence, performance deviation characteristics for each stage can be further constructed, thereby identifying stages that have entered a deviation-concern state and stages that have returned to a stable state. The former serves as a candidate deviation observation point, characterizing that the welding state at a certain stage has shown deviations requiring subsequent attention; the latter serves as a candidate compensation termination point, characterizing that the welding quality at a certain stage has returned to the stable range.

[0026] For the stage performance deviation sequence and stage control deviation sequence, sample stage performance sequences and sample stage control sequences are extracted from the welding quality data for each welding history sample at each welding stage. The sample stage performance sequence specifically includes the characteristic values ​​of each welding performance parameter at the corresponding welding stage, and the sample stage control sequence specifically includes the characteristic values ​​of each welding control parameter at the corresponding welding stage. Based on the stage performance reference sequences and stage control reference sequences for different welding stages, the deviation between each characteristic value and the corresponding benchmark reference value is calculated. For example, the difference between the characteristic value and the benchmark reference value is calculated to obtain the stage performance deviation value for different welding performance parameters and the stage control deviation value for different welding control parameters. The sign of the difference indicates the direction of the deviation, and the numerical value indicates the magnitude of the deviation. Multiple stage performance deviation sequences and stage control deviation sequences are constructed for each welding history sample in this way.

[0027] The process of constructing the performance deviation features of each stage's performance deviation sequence and extracting multiple candidate deviation observation points and compensation termination points for each welding history sample specifically includes: Multiple stage performance deviation values ​​in the fusion stage performance deviation sequence are used to generate stage performance deviation parameters, which serve as performance deviation features of the stage performance deviation sequence. Specifically, fusion can be achieved by calculating the modulus of the stage performance deviation sequence to obtain the stage performance deviation parameters.

[0028] Based on this, welding stages with performance deviation characteristics not less than the deviation observation threshold are designated as candidate deviation observation points. These points characterize the current welding stage as having reached a deviation state requiring subsequent close monitoring, and this stage may subsequently enter a deviation accumulation or compensation recovery process. Simultaneously, welding stages with performance deviation characteristics not greater than the stability observation threshold are designated as candidate compensation termination points. These points characterize the overall deviation state of the current welding stage as having returned to a stable range, meeting the candidate conditions for the termination boundary of the deviation compensation process. Those skilled in the art can set the deviation observation threshold and the stability observation threshold according to actual needs. In this embodiment, it is preferable that the deviation observation threshold is greater than the stability observation threshold to ensure that the determination of entering and exiting the deviation monitoring state of the welding process is distinguishable.

[0029] Furthermore, a continuous scan is performed on multiple candidate compensation termination points to identify multiple consecutive termination segments consisting of one or more consecutive candidate compensation termination points. In practice, all candidate compensation termination points can be traversed sequentially according to the welding stages, and candidate compensation termination points that are consecutively distributed in the stage sequence can be merged into the same consecutive termination segment. For each consecutive termination segment, the first candidate compensation termination point in that segment is taken as the compensation termination point. The purpose of this setup is to use the welding stage where the welding process first recovers to the stable range as the actual compensation termination boundary, and to consider subsequent consecutive welding stages within the same stable recovery process as a continuation of the stable phase after that compensation termination boundary, thereby avoiding the repeated extraction of multiple compensation termination points within the same stable recovery process.

[0030] Through the above steps, multiple candidate deviation observation points and multiple compensation termination points can be obtained for each welding history sample. The candidate deviation observation points are used to characterize the starting position of deviation concerns during the welding process, and the compensation termination points are used to characterize the stable recovery boundary position during the welding process.

[0031] Step S3: Identify multiple deviation compensation event intervals based on multiple compensation end points and multiple candidate deviation observation points. Based on the control deviation characteristics of multiple welding stages in the deviation compensation event interval, construct multiple deviation compensation event units for each welding history sample.

[0032] It is worth noting that this invention does not process the deviation information of each welding stage separately. Instead, it identifies the process of deviation initiation, deviation accumulation, and eventual stabilization during welding and defines it as a deviation compensation event. The entire welding process is divided into multiple candidate intervals using nodes formed by multiple compensation endpoints. From each candidate interval, a deviation compensation event interval representing the process from the deviation initiation / exit position to the peak deviation, and the gradual compensation and correction to return to normal welding, is extracted. The control deviation characteristics of different welding stages within this interval are then integrated to form a complete deviation compensation event unit. This process organizes historical welding-related data distributed according to welding stages into multiple complete event samples representing how deviations are compensated.

[0033] The process of identifying multiple deviation compensation event intervals and constructing multiple deviation compensation event units for each welding history sample specifically includes: Multiple welding stages are divided into multiple candidate intervals based on multiple compensation end points. Specifically, the compensation end points are arranged in the order of the welding stages, and the welding stage interval between any two adjacent compensation end points is used as a candidate interval. For welding stages located before the first compensation end point, the welding stage interval from the welding start position to the first compensation end point can be used as the first candidate interval. This divides the entire welding process of the same welding history sample into multiple consecutively arranged candidate intervals, with each candidate interval ending at its corresponding compensation end point.

[0034] Based on this, each candidate interval is screened, retaining only those intervals containing at least one candidate deviation observation point. Specifically, if a candidate interval contains no candidate deviation observation points, it indicates that although the interval eventually reached a stable state, no welding stage meeting the deviation observation conditions occurred during the corresponding welding stage, and therefore, this candidate interval is not considered for subsequent deviation compensation event identification. Conversely, if a candidate interval contains one or more candidate deviation observation points, it indicates that a candidate deviation evolution process exists within the interval, from a deviation state to a stable state, and should be further used to extract deviation compensation event intervals.

[0035] For each selected candidate interval, the earliest candidate deviation observation point appearing in the welding stage sequence within that interval is taken as the deviation observation starting point. This is used to characterize the position where the deviation state in the current candidate interval first enters the stage requiring close attention. Welding stages before the deviation observation starting point in each candidate interval are removed, retaining only the welding stage interval from the deviation observation starting point to the corresponding compensation end point of that candidate interval. Thus, the welding stage interval between the deviation observation starting point and the compensation end point is defined as a deviation compensation event interval, representing a complete process of the welding process from entering the deviation concern state to returning to a stable state.

[0036] The candidate deviation observation point with the largest control deviation characteristic in the deviation compensation event interval is identified and recorded as the compensation observation anchor point. This anchor point is used to characterize the stage position where the deviation state reaches the representativeness or the compensation demand is most concentrated within the deviation compensation event interval. The deviation observation start point, compensation observation anchor point and compensation end point belonging to the same deviation compensation event interval constitute one deviation compensation event unit of the welding history sample. In this way, the deviation evolution process and compensation recovery process that were originally continuously distributed in multiple welding stages can be compressed into a deviation compensation event unit with clear boundaries, clear representative state and clear recovery end point.

[0037] It is worth noting that for the same welding history sample, since there may be multiple independent deviation entry and recovery processes during the welding process, multiple deviation compensation event units can be identified and constructed from the same welding history sample in the manner described above. Each deviation compensation event unit independently corresponds to the deviation compensation process in different candidate intervals. More preferably, when constructing deviation compensation event units, the number of welding stages contained in the deviation compensation event interval can also be constrained. When the number of welding stages contained in a certain deviation compensation event interval is less than the preset minimum number of stages, the deviation compensation event interval can be eliminated. This is because although a deviation compensation event interval that is too short may also represent the rapid recovery process after the deviation occurs, it contains fewer welding stages, which is not conducive to extracting stable compensation patterns from the control deviation characteristics.

[0038] Step S4: Identify the deviation compensation matching status between any two deviation compensation event units, group multiple deviation compensation event units based on the deviation compensation matching status, and construct multiple compensation event sets corresponding to multiple deviation compensation event units.

[0039] Although different deviation compensation events originate from different welding history samples, some of these events have similar deviation states, similar recovery results, and similar compensation stage structures. Therefore, they can be regarded as the same type of compensation problem. By identifying the deviation compensation matching state between deviation compensation event units, deviation compensation event units that can be compared and referenced with each other can be grouped into the same compensation event set. This allows for the extraction of stable and reproducible compensation patterns from multiple similar events, thereby avoiding the influence of the randomness of a single sample.

[0040] The process of constructing multiple compensation event sets as described above specifically includes: Identify the compensation matching state of any two deviation compensation event units, where the compensation matching state includes the same compensation structure and different compensation structures.

[0041] In this process, based on the stage control deviation sequences of the compensation observation anchor points and compensation end points, the deviation state distance and compensation end state distance between two deviation compensation event units are calculated. Specifically, the distance between the stage control deviation sequences corresponding to the compensation observation anchor points of the two deviation compensation event units is calculated as the deviation state distance, used to characterize the closeness of the two deviation compensation event units in representative deviation states. Similarly, the distance between the stage control deviation sequences corresponding to the compensation end points of the two deviation compensation event units is calculated as the compensation end state distance, used to characterize the closeness of the two deviation compensation event units in their end states when compensation recovery is completed. The deviation state distance and compensation end state distance can be characterized using Euclidean distance. Furthermore, based on the compensation windows corresponding to each of the two deviation compensation event units, the compensation window length distance between them is calculated. By comparing the compensation window lengths of the two deviation compensation events, the difference between the compensation window lengths, i.e., the difference in the number of welding stages included in the compensation window, is calculated as the compensation window length distance between the two deviation compensation event units, used to characterize whether the stage span experienced by the two deviation compensation event units in completing compensation recovery is similar.

[0042] Finally, the compensation matching state between two deviation compensation event units is determined based on the distance between the deviation state, the distance between the compensation end state, and the distance between the compensation window lengths. Specifically, if the distance between the deviation state and the compensation end state are not greater than a preset deviation state threshold, the distance between the compensation end state and the compensation window length is not greater than a preset length threshold, then the compensation matching state between the two deviation compensation event units is determined to be that their compensation structures are the same. Conversely, if at least one of the above three distances exceeds the corresponding preset threshold, then the compensation matching state between the two deviation compensation event units is determined to be that their compensation structures are different. This achieves constraints from three dimensions: representing the deviation state, the recovery end state, and the span of the compensation process. Those skilled in the art can reasonably set the values ​​of the preset deviation state threshold, the preset end state threshold, and the preset length threshold according to actual accuracy requirements; no specific limitations are imposed here.

[0043] After identifying the compensation matching state of any two deviation compensation event units, they are progressively merged and grouped according to the compensation matching state between multiple deviation compensation event units. Specifically, each deviation compensation event unit can be initially considered as an independent initial unit, and then deviation compensation event units with the same compensation structure are merged according to the identification result of the compensation matching state. For example, if the compensation matching state of the first deviation compensation event unit and the second deviation compensation event unit is the same compensation structure, then the first deviation compensation event unit and the second deviation compensation event unit are grouped into the same compensation event set; if the compensation matching state of the second deviation compensation event unit and the third deviation compensation event unit is also the same compensation structure, then the third deviation compensation event unit is further merged into the same compensation event set containing the first deviation compensation event unit and the second deviation compensation event unit. Through the above progressive merging method, multiple deviation compensation event units with the same or similar relationship in compensation structure can be gradually merged to form multiple compensation event sets. The deviation compensation event units in the compensation event set have high similarity in terms of representing deviation state, recovery end state, and compensation process span.

[0044] Step S5: Perform compensation effect inversion processing on each compensation event set to generate a necessary event set, extract the target welding compensation core corresponding to each necessary event set, and optimize the online welding control of the target steel pipe based on multiple target welding compensation cores.

[0045] The purpose of the compensation effect inversion process is to fragment each deviation compensation event unit and analyze which compensation segments are necessary for completing the compensation recovery and which segments are merely accompanying non-critical changes, in order to obtain a set of necessary events that retain only the necessary compensation segments. Based on this, the common necessary control structure that stably appears in the set of necessary events is identified, thereby extracting the target welding compensation kernels corresponding to different types of compensation events. The target welding compensation kernel can be understood as, for a certain type of welding deviation state, which welding control parameters need to be adjusted, and what the adjustment direction and intensity are in several subsequent welding stages. In online application, when the target steel pipe exhibits a welding state matching the deviation state corresponding to a certain compensation kernel after a certain welding stage, the corresponding target welding compensation kernel can be invoked. Within its subsequent compensation stage range, staged compensation control is implemented according to the control parameters, control direction, and control intensity information contained in the compensation kernel, thereby achieving online welding control optimization of the target steel pipe.

[0046] The process of performing compensation effect inversion processing on each compensation event set to generate a necessary event set specifically includes: Determine the compensation window for each deviation compensation event unit in the compensation event set. Specifically, for any deviation compensation event unit, it includes three key nodes: the deviation observation start point, the compensation observation anchor point, and the compensation end point. Since the actual compensation process mainly occurs after the representative deviation state, the upper and lower boundaries of the compensation window are taken as the next welding stage of the compensation observation anchor point and the compensation end point.

[0047] The compensation windows of multiple deviation compensation event units in the compensation event set are aligned. Since different deviation compensation event units belong to the same compensation structure, the number of welding stages included in their compensation windows may differ. To ensure comparability of the compensation effects between different deviation compensation event units, multiple compensation windows need to be uniformly converted into an expression with the same number of welding stages. Specifically, the compensation window with the largest number of welding stages is taken as the welding stage number template. The multiple welding stages included in the welding stage number template are determined. For compensation windows with fewer welding stages than the welding stage number template, the missing welding stages are identified, and the stage performance deviation sequence and stage control deviation sequence of the missing welding stages are used as filler data to ensure that each deviation compensation event unit contains the same number of stage control deviation sequences. These stage control deviation sequences are denoted as the standard control deviation vector, thereby constructing the standard control deviation matrix for each deviation compensation event unit.

[0048] Each standard control deviation vector in the standard control deviation matrix is ​​labeled as a compensation segment. In other words, in this embodiment, the compensation segment is not an additionally divided complex structure, but rather the standard control deviation vectors arranged in the compensation window order in the standard control deviation matrix are directly used as the basic compensation unit. This characterizes the net control deviation of multiple welding control parameters at this location relative to the reference condition.

[0049] Based on this, compensation effect inversion detection is performed on multiple compensation segments of each deviation compensation event unit. Specifically, for any deviation compensation event unit, a control masking matrix for each compensation segment is constructed. The control masking matrix for a compensation segment is a structural state in which the standard control deviation vector corresponding to the compensation segment is removed, while the remaining standard control deviation vectors remain unchanged. In practice, the standard control deviation vector at the corresponding position of the compensation segment in the standard control deviation matrix can be set to a zero-deviation state, that is, multiple elements in the standard control deviation matrix are all set to 0, thus constructing the control masking matrix for each compensation segment of the deviation compensation event unit. Based on the control masking matrix, the control events of the deviation compensation event unit under different compensation segments are identified in the compensation event set. Specifically, the control deviation distance between the control masking matrix and the standard control deviation matrix of the other deviation compensation event units in the compensation event set is calculated. For example, the Euclidean distance is calculated after expanding the matrix into a one-dimensional vector for characterization, and the deviation compensation event unit with the smallest control deviation distance is selected as the control event of the deviation compensation event unit under one of the compensation segments.

[0050] As a preferred implementation process, if the minimum value of the multiple control deviation distances between the deviation compensation event unit and other deviation compensation event units is not less than the preset control deviation threshold, then the deviation compensation event unit is considered to have no control event under the compensation segment.

[0051] After identifying the control events for different compensation segments within the deviation compensation event unit using the above method, the compensation loss parameter corresponding to each compensation segment is calculated based on the control events. This parameter characterizes the degree of loss in the overall compensation recovery effect of the deviation compensation event unit when the tested compensation segment is missing or weakened. Specifically, the difference in deviation recovery effect before and after compensation between the current deviation compensation event unit and its control event can be compared to obtain the compensation loss parameter corresponding to the tested compensation segment. For example, the difference sequence between the stage performance deviation sequences of the current deviation compensation event unit with respect to the compensation observation anchor point and the compensation end point can be calculated, as well as the difference sequence between the stage performance deviation sequences of the control event with respect to the compensation observation anchor point and the compensation end point can be calculated. The magnitude of the two difference sequences is calculated separately, and the difference between the magnitudes of the difference sequences between the deviation compensation event unit and the control event is used as the compensation loss parameter. The larger the compensation loss parameter, the more significant the decrease in compensation recovery ability after the missing compensation segment, i.e., the decrease in the total compensation amount after the missing compensation segment. This indicates that the compensation segment plays a more important role in completing the current compensation process. Conversely, the smaller the parameter, the weaker the impact of the compensation segment on the overall compensation recovery, and it may be an accompanying or substitutable compensation segment.

[0052] Based on the compensation loss parameter, multiple target compensation segments are determined for each deviation compensation event unit. For example, compensation segments with compensation loss parameters not less than a preset loss threshold are identified as target compensation segments. This allows for the selection of multiple target compensation segments that play a crucial role in compensation recovery from the original multiple compensation segments of each deviation compensation event unit. These target compensation segments from multiple deviation compensation event units are then combined to generate a necessary event set. In this way, redundant control actions, weakly correlated control actions, or accompanying control changes in the original compensation window can be eliminated, retaining only the compensation segments that truly play a major role in compensation recovery.

[0053] The process of extracting the target welding compensation core corresponding to each set of necessary events specifically includes: Each necessary event set is identified, comprising multiple welding stages. Necessity analysis is performed on each welding stage based on multiple target compensation segments of each deviation compensation event unit, identifying multiple welding compensation stages within the necessary event set. Specifically, since each deviation compensation event unit corresponds to one or more target compensation segments, stage mapping can be performed on the target compensation segments involved in all deviation compensation event units within the necessary event set to determine which welding stages these target compensation segments belong to. Subsequently, the occurrence of target compensation segments in each welding stage within the necessary event set is statistically analyzed and compared. If a certain welding stage repeatedly exhibits target compensation segments within the necessary event set—for example, if a target compensation segment exists in a welding stage exceeding a preset percentage of deviation compensation event units—then this welding stage can be considered to play a stable compensation role in the corresponding type of deviation recovery process, and is identified as a welding compensation stage. This process further refines the key welding stages that truly and continuously contribute to similar compensation processes from the necessary event set, while eliminating welding stages that only appear occasionally or lack stable compensation significance.

[0054] For each welding compensation stage, the control direction synchronization state of multiple welding control parameters is identified to determine the control synchronization parameter for each welding control parameter in the welding compensation stage. Specifically, for a certain welding compensation stage, multiple standard control deviation vectors corresponding to all deviation compensation event units in the necessary event set within that welding compensation stage can be extracted, and the direction of change of each welding control parameter in these standard control deviation vectors can be examined. The proportion of deviation compensation event units corresponding to each direction of change is determined, and the maximum value is recorded as the control synchronization parameter of the welding control parameter. If the control synchronization parameter is greater than a preset proportion, that is, in deviation compensation event units exceeding the preset proportion, the welding control parameter shows deviation in the same direction, then the welding control parameter is used as the compensation control parameter, thereby identifying multiple compensation control parameters for each welding compensation stage. This retains those welding control parameters that exhibit strong synchronization and stability in the necessary event set. The control effect in each welding compensation stage is further compressed into a small set of compensation control parameters with common compensation significance.

[0055] Based on multiple standard control deviation vectors corresponding to the compensation control parameters, welding compensation information for each compensation control parameter is determined, including the compensation control direction and compensation control intensity. Specifically, the compensation control direction refers to the offset direction of these compensation control parameters relative to the reference condition, characterizing the direction in which the compensation control parameter should be adjusted relative to the reference condition in the current welding compensation stage, such as compensation in an increasing or decreasing direction. The compensation control intensity is specifically the statistical characteristic value of the control deviation values ​​of these compensation control parameters across multiple stages, such as the mean or median. In this embodiment, the mean of the control deviation values ​​across multiple stages is used as the compensation control intensity of the compensation control parameter. Through the above method, multiple compensation control parameters, along with corresponding compensation control directions and compensation control intensities, can be formed for each welding compensation stage.

[0056] Furthermore, the welding stage preceding the first welding compensation stage is designated as the online observation stage for compensation. Specifically, since the compensation core should not be triggered only after compensation has begun when applied online, but rather the current state should be identified before the arrival of the first welding compensation stage, it is necessary to determine the corresponding online observation stage for each type of compensation core. Preferably, the welding stage preceding the first welding compensation stage is taken as the online observation stage for compensation. Based on this, for each deviation compensation event unit in the necessary event set, the stage performance deviation sequence of each deviation compensation event unit in the online observation stage for compensation is extracted. Since these stage performance deviation sequences correspond to the monitoring states of the same type of compensation event before actually entering the compensation process, they can be fused. For example, the mean of multiple stage performance deviation sequences can be calculated as the target compensation monitoring state sequence characterizing the triggering condition of this type of compensation core. The target compensation monitoring state sequence can be understood as follows: before entering the first welding compensation stage, if the current welding state matches the target compensation monitoring state sequence, it indicates that the corresponding target welding compensation core should be called to implement compensation control.

[0057] Finally, based on the welding compensation information of multiple compensation control parameters included in multiple welding compensation stages, and the target compensation monitoring state sequence in the online observation stage, a target welding compensation kernel with a necessary event set is constructed. Each target welding compensation kernel preferably includes at least two parts: first, a monitoring triggering part, which characterizes in which online observation stage and under what target compensation monitoring state the compensation kernel is triggered; second, a stage control part, which characterizes which compensation control parameters should be compensated in multiple welding compensation stages, as well as the corresponding compensation control direction and intensity. The target welding compensation kernel is not a single control quantity, but a staged compensation control rule structure extracted for the same type of deviation compensation process. By combining the target compensation monitoring state sequence in the online observation stage with the welding compensation information in subsequent welding compensation stages, the target welding compensation kernel can simultaneously possess both the information of when to trigger and how to compensate, thus enabling direct use for online control optimization in the target steel pipe welding process.

[0058] Specifically, for the online control optimization process of applying multiple target welding compensation cores to the target steel pipe welding process, firstly, during the welding process of the target steel pipe, the online welding process data of the current welding stage is extracted after the end of each welding stage.

[0059] During the welding process of the target steel pipe, after the completion of each welding stage, the online welding process data of the current welding stage is extracted. Based on the aforementioned welding benchmark data, the online stage performance deviation sequence corresponding to the online welding process data is extracted, and the online stage performance deviation parameters corresponding to the online stage performance deviation sequence are calculated. The online stage performance deviation parameters are then compared with a preset deviation observation threshold. If the online stage performance deviation parameter is less than the deviation observation threshold, it indicates that although the target steel pipe may have some fluctuations in the current welding stage, it has not yet reached the level of deviation concern requiring key intervention. In this case, the target welding compensation matching process is not initiated, and normal welding control for the next welding stage continues. If the online stage performance deviation parameter is not less than the deviation observation threshold, it indicates that the current welding stage has entered the deviation concern state, and the target welding compensation matching process is executed.

[0060] For the target welding compensation core matching process, multiple target compensation monitoring state sequences corresponding to the welding stage to which the online stage performance deviation sequence belongs are determined. The state distance between the online stage performance deviation sequence and each of the multiple target compensation monitoring state sequences is calculated to characterize the proximity between the current online welding state and the trigger states of each target welding compensation core. The Euclidean distance between the two sequences can be used as the state distance. After obtaining multiple state distances, the minimum state distance is determined and compared with a preset compensation trigger distance threshold. If the minimum state distance is not greater than the compensation trigger distance threshold, welding control compensation optimization is performed based on the target welding compensation core corresponding to the minimum state distance. The purpose of this setting is to avoid prematurely activating a compensation core simply because the current online welding state is relatively closest to one of the multiple target welding compensation cores.

[0061] Specifically, if the minimum state distance is greater than the compensation trigger distance threshold, it means that although the current state is closest to a certain target welding compensation core, it is still not close enough to the target compensation monitoring state sequence. In this case, no compensation core is activated, but monitoring continues in subsequent welding stages, and the online stage performance deviation sequence and online stage performance deviation parameters are recalculated after the next welding stage ends. If the minimum state distance is not greater than the compensation trigger distance threshold, it means that the current online welding state is close enough to the trigger state corresponding to a certain target welding compensation core. In this case, welding control compensation optimization is performed based on the target welding compensation core corresponding to the minimum state distance.

[0062] The aforementioned welding control compensation optimization specifically includes performing phased compensation control on subsequent welding stages of the target steel pipe based on multiple welding compensation stages corresponding to the target welding compensation core, and the compensation control direction and intensity of the compensation control parameters in each welding compensation stage. That is, according to the predefined sequence of compensation stages defined by the compensation core, the compensation control parameters corresponding to each subsequent welding stage are invoked, and the welding control parameters for the current stage are adjusted based on the corresponding compensation control direction and intensity. This approach makes the online compensation process more consistent with the actual characteristics of phased advancement and segmented compensation in steel pipe welding.

[0063] During the execution of compensation control, the online stage performance deviation parameters of the target steel pipe at the current welding stage are continuously updated. Specifically, after each subsequent welding stage, the online welding process data for that stage is repeatedly extracted, and the corresponding online stage performance deviation sequence and parameters are recalculated based on the welding baseline data. When the online stage performance deviation parameters fall below the stable observation threshold, it indicates that the welding state of the target steel pipe has re-entered the stable range, at which point the welding control compensation optimization process for the target welding compensation core ends. Through the above online control optimization process, the target steel pipe can be welded not only continuously using fixed process parameters, but also, after detecting a stage deviation and entering a state of interest, further matched with multiple target welding compensation cores. When the trigger condition is met, the most suitable target welding compensation core is called to perform staged compensation control for subsequent welding stages.

[0064] This invention mines the intrinsic relationship between deviation occurrence, deviation recovery, and compensation behavior from multiple historical welding samples, extracts a welding compensation core that can be stably reproduced, and uses it to guide subsequent online phased control. This can better handle the problem of uneven quality distribution caused by the accumulation of small fluctuations in multiple stages during the arc welding of steel pipes, reduce the continuous accumulation and amplification of local deviations, suppress the uncoordinated changes in the quality distribution of different areas of the joint, so as to achieve gradient adjustment of the welding strength distribution of steel pipes, improve the stability of the steel pipe welding process, and enhance the consistency of weld joint quality.

[0065] Please see Figure 2 Another aspect of the present invention provides a gradient welding strength control system for steel pipes, specifically for implementing the above-described gradient welding strength control method for steel pipes, including: The steel pipe welding data acquisition module is used to collect welding process data and welding performance data of the target steel pipe based on multiple historical welding samples, and to determine multiple welding stages and welding benchmark working condition data of the target steel pipe. The welding deviation feature extraction module is used to extract multiple stage performance deviation sequences and stage control deviation sequences for each welding history sample based on welding benchmark working condition data, construct the performance deviation features of each stage performance deviation sequence, and extract multiple candidate deviation observation points and compensation termination points for each welding history sample. The compensation event unit identification module is used to identify multiple deviation compensation event intervals based on multiple compensation end points and multiple candidate deviation observation points. Based on the control deviation characteristics of multiple welding stages in the deviation compensation event interval, multiple deviation compensation event units are constructed for each welding history sample. The compensation event unit grouping module is used to identify the deviation compensation matching status between any two deviation compensation event units, group multiple deviation compensation event units based on the deviation compensation matching status, and construct multiple compensation event sets corresponding to multiple deviation compensation event units. The online welding control optimization module is used to perform compensation effect inversion processing on each compensation event set to generate a necessary event set, extract the target welding compensation core corresponding to each necessary event set, and perform online welding control optimization on the target steel pipe based on multiple target welding compensation cores.

[0066] The above are merely specific embodiments of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art. Parts not described in detail in this specification are prior art known to those skilled in the art.

Claims

1. A method for controlling the gradient welding strength of steel pipes, characterized in that, include: Welding process data and welding performance data of the target steel pipe based on multiple historical welding samples were collected to determine multiple welding stages and welding benchmark conditions of the target steel pipe. Based on welding benchmark data, multiple stage performance deviation sequences and stage control deviation sequences are extracted for each welding history sample. Performance deviation features of each stage performance deviation sequence are constructed, and multiple candidate deviation observation points and compensation termination points are extracted for each welding history sample. Multiple deviation compensation event intervals are identified based on multiple compensation end points and multiple candidate deviation observation points. Based on the control deviation characteristics of multiple welding stages in the deviation compensation event intervals, multiple deviation compensation event units are constructed for each welding history sample. Identify the deviation compensation matching status between any two deviation compensation event units, group multiple deviation compensation event units based on the deviation compensation matching status, and construct multiple compensation event sets corresponding to multiple deviation compensation event units; Each set of compensation events is processed by inversion to generate a set of necessary events. The target welding compensation core corresponding to each set of necessary events is extracted. Based on multiple target welding compensation cores, the online welding control of the target steel pipe is optimized.

2. The method for controlling gradient welding strength of steel pipes according to claim 1, characterized in that, The performance deviation features of each stage's performance deviation sequence are constructed, and multiple candidate deviation observation points and compensation termination points are extracted for each welding history sample, including: For the stage control deviation sequence, the welding reference condition data includes the stage performance reference sequence and the stage control reference sequence for each welding stage. The sample stage performance sequence of each welding history sample in each welding stage is extracted from the welding process data. Based on the stage performance reference sequence, the performance deviation of the sample stage performance sequence of the welding history sample is detected, the stage performance deviation value of each welding performance parameter is extracted, and the stage performance deviation sequence of each welding history sample in different welding stages is generated. Multiple stage performance deviation values ​​in the fusion stage performance deviation sequence are used to generate stage performance deviation parameters, which serve as performance deviation features of the stage performance deviation sequence. Welding stages with performance deviation features not less than the deviation observation threshold are recorded as candidate deviation observation points, and welding stages with performance deviation features not greater than the stable observation threshold are recorded as candidate compensation end points. Multiple candidate compensation end points are continuously scanned to identify multiple continuous end segments composed of multiple candidate compensation end points. The first candidate compensation end point of each continuous end segment is recorded as the compensation end point.

3. The method for controlling gradient welding strength of steel pipes according to claim 2, characterized in that, Multiple deviation compensation event intervals are identified based on multiple compensation end points and multiple candidate deviation observation points. Based on the control deviation characteristics of multiple welding stages within each deviation compensation event interval, multiple deviation compensation event units are constructed for each welding history sample, including: Based on multiple compensation end points, multiple welding stages are divided into multiple candidate intervals. Candidate intervals containing at least one candidate deviation observation point are selected. The first candidate deviation observation point in the selected candidate interval is taken as the deviation observation start point. Welding stages before the deviation observation start point in each candidate interval are removed. The deviation observation start point and the compensation end point of the candidate interval are taken to form a deviation compensation event interval. The candidate deviation observation point with the largest control deviation characteristic in the deviation compensation event interval is identified and recorded as the compensation observation anchor point. The deviation observation start point, compensation observation anchor point and compensation end point belonging to the same deviation compensation event interval constitute one deviation compensation event unit of the welding history sample.

4. The method for controlling gradient welding strength of steel pipes according to claim 3, characterized in that, For each set of compensation events, a compensation effect inversion process is performed to generate a set of necessary events, including: Determine the compensation window for each deviation compensation event unit in the compensation event set, where the upper and lower boundaries of the compensation window are the next welding stage and the compensation end point of the compensation observation anchor point. Align the compensation windows of multiple deviation compensation event units in the compensation event set and construct the standard control deviation matrix for each deviation compensation event unit, which includes multiple standard control deviation vectors. Each standard control deviation vector in the standard control deviation matrix is ​​marked as a compensation segment. The compensation effect inversion detection is performed on multiple compensation segments of each deviation compensation event unit. This includes constructing a control masking matrix of the deviation compensation event unit for each compensation segment, identifying the control events of the deviation compensation event unit under different compensation segments in the compensation event set based on the control masking matrix, and calculating the compensation loss parameter of each compensation segment based on the control events. Based on the compensation loss parameters, multiple target compensation segments are determined for each deviation compensation event unit, and the target compensation segments of multiple deviation compensation event units are combined to generate a necessary event set.

5. The method for controlling gradient welding strength of steel pipes according to claim 4, characterized in that, Extracting the target welding compensation core corresponding to each set of necessary events includes: Determine the multiple welding stages contained in each set of necessary events, perform necessity analysis on each welding stage based on the multiple target compensation segments of each deviation compensation event unit, and identify multiple welding compensation stages in the set of necessary events. For each welding compensation stage, the control direction synchronization status of multiple welding control parameters is identified to determine the control synchronization parameters of each welding control parameter in the welding compensation stage. Based on the control synchronization parameters, multiple compensation control parameters for each welding compensation stage are identified. Based on the multiple standard control deviation vectors corresponding to the compensation control parameters, the welding compensation information of each compensation control parameter is determined, including the compensation control direction and compensation control intensity of the compensation control parameter. The welding stage preceding the first welding compensation stage is marked as the online observation stage of compensation. The stage performance deviation sequence of each deviation compensation event unit in the online observation stage of compensation is extracted and fused to obtain the target compensation monitoring status sequence. Based on the welding compensation information of multiple compensation control parameters included in multiple welding compensation stages, and the target compensation monitoring state sequence of the online compensation observation stage, a target welding compensation kernel with a set of necessary events is constructed.

6. The method for controlling gradient welding strength of steel pipes according to claim 5, characterized in that, Online welding control optimization based on multiple target welding compensation checks of the target steel pipe includes: During the welding process of the target steel pipe, after the completion of each welding stage, the online welding process data of the current welding stage is extracted, and the online stage performance deviation sequence corresponding to the online welding process data is extracted based on the welding reference working condition data. The online stage performance deviation parameter is calculated based on the online stage performance deviation sequence. If the online stage performance deviation parameter is not less than the deviation observation threshold, the target welding compensation core matching process is executed. This includes determining multiple target compensation monitoring state sequences corresponding to the welding stage to which the online stage performance deviation sequence belongs, calculating the state distance between the online stage performance deviation sequence and the multiple target compensation monitoring state sequences respectively, determining the minimum state distance and comparing it with the preset compensation trigger distance threshold. If the minimum state distance is not greater than the compensation trigger distance threshold, welding control compensation optimization is performed based on the target welding compensation core corresponding to the minimum state distance. Welding control compensation optimization includes performing phased compensation control on the subsequent welding stages of the target steel pipe based on multiple welding compensation stages corresponding to the target welding compensation core, as well as the compensation control direction and compensation control intensity of the compensation control parameters in each welding compensation stage. During the execution of compensation control, the online stage performance deviation parameters of the target steel pipe in the current welding stage are continuously updated. When the online stage performance deviation parameters fall below the stable observation threshold, the welding control compensation optimization process for the target welding compensation core ends.

7. The method for controlling gradient welding strength of steel pipes according to claim 4, characterized in that, Based on the deviation compensation matching state, multiple deviation compensation event units are grouped to construct multiple compensation event sets corresponding to multiple deviation compensation event units, including: Identify the compensation matching state of any two deviation compensation event units, including calculating the deviation state distance and compensation end state distance between the two deviation compensation event units based on the stage control deviation sequence of the compensation observation anchor point and the compensation end point, calculating the compensation window length distance between the two deviation compensation event units based on the compensation window of the deviation compensation event unit, and determining the compensation matching state between the two deviation compensation event units based on the deviation state distance, compensation end state distance and compensation window length distance. The compensation matching state includes the same compensation structure and different compensation structures. Based on the compensation matching status of any two deviation compensation event units, multiple deviation compensation event units are progressively merged and grouped to generate multiple compensation event sets corresponding to multiple deviation compensation event units.

8. A gradient welding strength control system for steel pipes, characterized in that, The system is used to implement the gradient welding strength control method for steel pipes according to any one of claims 1-7, comprising: The steel pipe welding data acquisition module is used to collect welding process data and welding performance data of the target steel pipe based on multiple historical welding samples, and to determine multiple welding stages and welding reference working condition data of the target steel pipe. The welding deviation feature extraction module is used to extract multiple stage performance deviation sequences and stage control deviation sequences for each welding history sample based on welding benchmark working condition data, construct the performance deviation features of each stage performance deviation sequence, and extract multiple candidate deviation observation points and compensation termination points for each welding history sample. The compensation event unit identification module is used to identify multiple deviation compensation event intervals based on multiple compensation end points and multiple candidate deviation observation points. Based on the control deviation characteristics of multiple welding stages in the deviation compensation event interval, multiple deviation compensation event units are constructed for each welding history sample. The compensation event unit grouping module is used to identify the deviation compensation matching status between any two deviation compensation event units, group multiple deviation compensation event units based on the deviation compensation matching status, and construct multiple compensation event sets corresponding to multiple deviation compensation event units. The online welding control optimization module is used to perform compensation effect inversion processing on each compensation event set to generate a necessary event set, extract the target welding compensation core corresponding to each necessary event set, and perform online welding control optimization on the target steel pipe based on multiple target welding compensation cores.

Images