A pass correct method for preventing wrinkle defect of high-strength steel welded pipe cold bending forming

Abstract

The application discloses a hole type correction method for preventing wrinkles of high-strength steel welded pipe cold bending forming, and relates to the technical field of cold bending forming, and aims to solve the technical problem that the R angle cracking source cannot be accurately and timely positioned in the process of the welded pipe cold bending forming and the hole type is corrected according to the positioning in the prior art, and the method comprises the following steps: arranging an acoustic emission sensor array in an R angle forming area of a cold bending forming unit, and pre-calibrating three-dimensional space coordinates of each acoustic emission sensor; in the process of the continuous cold bending forming of the high-strength steel welded pipe, the acoustic emission sensor array is used to collect acoustic emission signals of the R angle area in real time, and the acoustic emission signals are pre-processed; when the acoustic emission event is detected, the time when the signals of the acoustic emission event received by all the acoustic emission sensors arrive is recorded; the application realizes the technical leap from the "post-detection and experience trial and error" to the "online monitoring and quantitative correction", and effectively avoids the further expansion of the defects and the generation of waste products.

Description

Technical Field

[0001] This invention relates to the field of cold bending forming technology, and more specifically, to a method for correcting the hole shape to prevent wrinkling defects in the cold bending forming of high-strength steel welded pipes. Background Technology

[0002] High-strength steel welded pipes are increasingly widely used in engineering machinery, building structures, and oil and gas transportation due to their high strength, lightweight, and excellent load-bearing capacity. Cold bending is the core process in the production of high-strength steel square and rectangular tubes. Its basic principle is to pass a continuous strip of steel through multiple forming rolls, gradually bending and deforming it to the target cross-sectional shape at room temperature. During the cold bending process, the strip experiences the most severe bending deformation and compressive stress in the radius (R-angle) region. The plastic flow behavior of the material in this region directly determines the final forming quality of the welded pipe.

[0003] However, with the continuous improvement of the strength grade of high-strength steel, its plastic deformation range narrows and its work hardening tendency intensifies, making it extremely prone to wrinkling defects in the R-corner area during cold bending. The formation mechanism of this defect is as follows: when the compressive deformation in the R-corner area exceeds the critical wrinkling strain of the material, periodic wavy wrinkles appear on the surface of the strip steel, which can develop into micro or even macro cracks in severe cases, significantly reducing the load-bearing capacity and corrosion resistance of the welded pipe. Currently, the control of R-corner wrinkling defects mainly relies on two methods: one is experience-based die design adjustment, that is, after the defect is discovered, the process personnel modify the die parameters based on experience, and production can only be restarted after repeated trial molding verification. This method has a slow response, relies on personnel experience, and has high trial and error costs; the other is offline sampling inspection after the finished product is off the production line. When defects are found through cross-sectional measurement or non-destructive testing, the defective pipe has already been generated, and can only be downgraded or scrapped, making it impossible to intervene at the first time the defect occurs.

[0004] The technical root of the above problems lies in the fact that the cold bending of high-strength steel welded pipes is a continuous and high-speed dynamic process. Wrinkling and cracking in the radius (R-corner) region originate from plastic instability within the material, and their location is random. Furthermore, the time window from microscopic damage initiation to macroscopic defect formation is extremely short. Existing detection methods are all offline or quasi-static, lacking a technical means to monitor the spatial location of crack sources in the radius (R-corner) region in real time during the forming process and implement online correction accordingly. To address the shortcomings of existing technologies in accurately locating the radius (R-corner) crack source and implementing targeted die correction during the cold bending of welded pipes, this invention provides a die correction method to prevent wrinkling defects in the cold bending of high-strength steel welded pipes. This method uses an acoustic emission sensor array to capture cracking signals in the radius (R-corner) region in real time, and combines this with multi-sensor time-difference positioning technology to accurately solve for the three-dimensional spatial coordinates of the crack source. These coordinates are then mapped to specific passes, rolls, and correction parameters, achieving a rapid closed-loop response from crack source location to targeted die correction, thus enabling precise intervention in the early stages of wrinkling defect formation. In view of this, we propose a hole shape correction method to prevent wrinkling defects in cold bending of high-strength steel welded pipes. Summary of the Invention

[0005] The purpose of this invention is to provide a method for correcting the hole shape to prevent wrinkling defects in the cold bending of high-strength steel welded pipes, so as to solve the technical problem in the prior art that it is impossible to accurately locate the R-angle crack source in real time during the cold bending process of welded pipes and to perform hole shape correction accordingly.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a method for correcting the hole shape to prevent wrinkling defects in cold bending of high-strength steel welded pipes, comprising: An acoustic emission sensor array is arranged in the R-angle forming area of ​​the cold bending forming unit. The acoustic emission sensor array includes multiple acoustic emission sensors, and the three-dimensional spatial coordinates of each acoustic emission sensor are pre-calibrated. During the continuous cold bending process of high-strength steel welded pipe, the acoustic emission signal in the R-angle region is collected in real time by the acoustic emission sensor array, and the acoustic emission signal is preprocessed. When an acoustic emission event is detected, the arrival time of the signal received by all acoustic emission sensors is recorded, and the signal arrival time difference between each non-triggering sensor and the triggering sensor is calculated. The triggering sensor is the sensor that first detects the acoustic emission event. Based on the signal arrival time difference and the propagation speed of sound waves in high-strength steel, a three-dimensional spatial positioning equation system is established, and the three-dimensional spatial coordinates of the crack source are obtained by solving it. The three-dimensional spatial coordinates of the cracking source are matched with the preset mapping relationship of the die shape parameters to determine the forming pass number, roll number and die shape correction parameters corresponding to the cracking source. According to the pass shape correction parameters, a correction command is sent to the roll adjustment actuator of the corresponding pass to perform fixed-point correction of the pass shape of the corresponding pass.

[0007] This invention utilizes an array of acoustic emission sensors arranged in the R-corner forming area of ​​a cold bending forming unit. The sensors are distributed three-dimensionally along the forming direction and circumferential direction of the tube, forming a three-dimensional spatial coverage network. This allows for real-time acquisition of acoustic emission event signals generated in the R-corner area. By combining the time difference of arrival of multiple sensor signals and a three-dimensional spatial positioning equation system, the three-dimensional spatial coordinates of the crack source are accurately solved. These coordinates are then mapped to the forming pass number, roll number, and die correction parameters corresponding to the crack source. Based on this, correction commands are sent to the roll adjustment actuator of the corresponding pass, achieving complete online closed-loop control of the R-corner crack source from real-time capture and precise positioning to point-to-point correction. This technical solution solves the technical problem in existing technologies where R-corner wrinkling and cracking sources cannot be located online in real-time during cold bending forming, and defects can only be detected offline after the finished product is off the line, requiring delayed processing. It achieves a technological leap from "post-processing detection and trial-and-error" to "online monitoring and quantitative correction," enabling R-corner cracking defects to be detected and addressed in their early stages, effectively preventing further expansion of defects and the generation of defective products.

[0008] Preferably, the acoustic emission sensor array includes at least 6 acoustic emission sensors, and each acoustic emission sensor is distributed in at least two groups along the pipe forming direction, with at least 2 sensors distributed in each group along the circumferential direction of the pipe cross-section, forming a three-dimensional spatial coverage network.

[0009] Preferably, the calculation of the signal arrival time difference between each non-trigger sensor and the trigger sensor specifically includes: extracting the signal waveform within a preset time window before and after the triggering time of the acoustic emission event, calculating the similarity peak value between the signals of the trigger sensor and each non-trigger sensor through a cross-correlation algorithm, and obtaining the signal arrival time difference based on the time offset corresponding to the similarity peak value.

[0010] Preferably, the propagation speed of the sound wave in the high-strength steel is the sound velocity value after stress dynamic correction, which is calculated based on the actual stress value in the R-angle region and the reference sound velocity under stress-free conditions.

[0011] Preferably, the process of obtaining the three-dimensional spatial coordinates of the crack source specifically includes: selecting multiple different sensor combinations and substituting them into the three-dimensional spatial positioning equations to obtain multiple preliminary coordinates; determining weighting coefficients based on the signal-to-noise ratio of each sensor combination; and performing weighted fusion of the multiple preliminary coordinates to obtain the final three-dimensional spatial coordinates of the crack source.

[0012] Preferably, the mapping relationship of the hole shape parameters is as follows: the coordinates of the crack source along the tube forming direction are mapped to the forming pass number, the coordinates along the circumferential direction of the tube cross section are mapped to the roll number, and the coordinates along the tube wall thickness direction are mapped to the hole shape correction parameter type.

[0013] Preferably, the step of performing fixed-point correction of the corresponding pass profile specifically includes at least one of adjusting the roll gap of the corresponding roll, adjusting the radius of the arc of the corresponding roll, and adjusting the feed amount of the corresponding pass.

[0014] Preferably, after performing the fixed-point correction of the corresponding pass aperture, the method further includes: continuously monitoring the acoustic emission signal energy value at the location corresponding to the crack source within a preset verification time window, comparing the energy value with the reference energy value in the crack-free state, and determining that the correction is effective if the energy value is lower than a preset threshold; otherwise, increasing the correction amount or triggering an offline aperture redesign process.

[0015] Preferably, the method further includes: structuring and storing the three-dimensional coordinates of the crack source, the die correction parameters, and the correction effect data corresponding to each acoustic emission event into an experience database; when the correction records for the same forming pass and roll position accumulate to a preset number, extracting the statistically optimal correction parameters as the standard die parameter update value for that position, and using it for the die design of subsequent products of the same specification.

[0016] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a hole profile correction method for preventing wrinkling defects in cold bending of high-strength steel welded pipes.

[0017] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention arranges an array of acoustic emission sensors in the R-corner forming area of ​​a cold bending forming unit. The sensors are distributed three-dimensionally along the forming direction and circumferential direction of the tube, forming a three-dimensional spatial coverage network. This allows for real-time acquisition of acoustic emission event signals generated in the R-corner area. By combining the arrival time difference of multiple sensor signals and a three-dimensional spatial positioning equation set, the three-dimensional spatial coordinates of the crack source are accurately solved. These coordinates are then mapped to the forming pass number, roll number, and die correction parameters corresponding to the crack source. Based on this, correction commands are sent to the roll adjustment actuator of the corresponding pass, achieving complete online closed-loop control of the R-corner crack source from real-time capture and precise positioning to point-to-point correction. This technical solution solves the technical problem in existing technologies where R-corner wrinkling and cracking sources cannot be located online in real-time during cold bending forming, and defects can only be detected offline after the finished product is off the line, requiring delayed processing. It achieves a technological leap from "post-processing detection and trial-and-error" to "online monitoring and quantitative correction," enabling R-corner cracking defects to be detected and addressed in a timely manner at their initial stage, effectively preventing further expansion of defects and the generation of defective products.

[0018] 2. This invention further achieves precise three-dimensional spatial positioning of the crack source in the tube forming direction, cross-sectional circumferential direction, and wall thickness direction by employing a three-dimensional spatially distributed acoustic emission sensor array combined with a three-dimensional hyperbolic positioning equation system based on the time difference of arrival of multi-sensor signals. Based on this, the three-dimensional spatial coordinates of the crack source are mapped to the forming pass number, roll number, and die correction parameter type, establishing a direct quantitative correspondence between the crack source positioning result and the production process parameters. Compared to the limitations of existing technologies that cannot distinguish whether the crack source occurs on the tube surface or inside, or determine which pass and roll group the crack occurs in, this invention can accurately determine "which pass the crack occurs in, which roll group it corresponds to, and what parameters need to be corrected." This transforms die correction from blind global adjustment to precise point intervention, significantly improving correction efficiency and avoiding repeated adjustments and material waste caused by inaccurate positioning.

[0019] 3. This invention also achieves real-time verification of the correction effect of each round by continuously monitoring the acoustic emission signal energy value at the location of the crack source after the hole pattern is fixedly corrected and comparing it with the reference energy value in the crack-free state. Production can only continue after the correction is verified to be effective; if the correction is deemed ineffective, the correction amount is automatically increased or the offline hole pattern redesign process is triggered. Simultaneously, this invention stores the complete process data of each crack event from positioning, correction to verification in a structured form in an experience database. When a sufficient number of correction records are accumulated at the same location, the statistically optimal value is extracted through statistical analysis as the standard correction parameter update value for subsequent hole pattern design of products of the same specification. This "correction-verification-self-learning" mechanism ensures that the hole pattern correction accuracy continuously improves with the continuous accumulation of production data, avoiding the recurrence of defects caused by incomplete single corrections. While solving the problem of "only adjusting without verifying, only modifying without learning" in existing hole pattern correction technologies, it achieves a technological upgrade from single emergency response to continuous process optimization in hole pattern correction. Attached Figure Description

[0020] Figure 1 This is an overall flowchart of the method of the present invention; Figure 2 This is a flowchart of the signal processing and time difference extraction process of the present invention; Figure 3 This is a flowchart of the three-dimensional spatial positioning and coordinate mapping process of the present invention; Figure 4 This is a flowchart illustrating the hole shape positioning correction and execution process of the present invention; Figure 5 This is a flowchart illustrating the experimental verification and self-learning process of the present invention. Detailed Implementation

[0021] like Figures 1 to 4As shown, the present invention relates to a method for correcting the die shape to prevent wrinkling defects in the cold bending forming of high-strength steel welded pipes, comprising the following steps: Step S1: Arrange an acoustic emission sensor array in the R-angle forming area of ​​the cold bending forming unit. The sensor array contains multiple acoustic emission sensors. Each sensor is distributed in three dimensions along the forming direction of the pipe and the circumferential direction of the cross section to form a three-dimensional spatial coverage network. The spatial coordinates of each sensor are pre-calibrated and stored in the system. Acoustic emission refers to the physical phenomenon where strain energy is rapidly released in the form of elastic waves when a localized area within a material undergoes plastic deformation or crack propagation due to stress concentration. In the embodiments of this application, acoustic emission signals of a specific frequency band are generated before and at the moment of cracking at the radius (R-angle). Among them, an acoustic emission sensor refers to a transducer that can convert the surface micro-displacement caused by acoustic emission elastic waves into an electrical signal. In the embodiments of this application, multiple acoustic emission sensors constitute a sensor array. Step S2: During the continuous cold bending process of high-strength steel welded pipe, the acoustic emission signal of the R-angle region is collected in real time by the sensor array, and the signals of each channel are amplified and filtered. Step S3: When any sensor channel acquires an acoustic emission event signal, record the arrival time of the signal of the same acoustic emission event received by all sensor channels, and calculate the signal arrival time difference between each non-triggered sensor and the triggered sensor; Among them, the trigger sensor refers to the acoustic emission sensor in the sensor array whose received signal amplitude first exceeds the preset trigger threshold, and the triggering time of the sensor is used as the time reference. Step S4: Based on the arrival time difference of the signal and the pre-calibrated propagation speed of the sound wave in the high-strength steel, establish a three-dimensional spatial positioning equation set and solve for the three-dimensional spatial coordinates of the crack source. The crack source is the material damage location in the R-angle region where the acoustic emission event occurs. Step S5: Compare the three-dimensional spatial coordinates of the crack source with the preset mapping relationship of the die shape parameters to determine the forming pass number, roll number and die shape correction parameters corresponding to the crack source; Among them, the hole shape parameter mapping relationship refers to the pre-established mapping rules or functions that correspond the geometric coordinate position of the crack source in three-dimensional space to the cold bending forming process parameters (such as forming pass, roll number, correction amount type). Among them, the forming pass refers to the processing station or stand where each pair of rolls is located in the cold bending forming unit, and the tube is gradually bent and formed to the target cross-sectional shape through each pass in sequence. The pass shape correction parameters refer to the types and values ​​of parameters used to guide the corresponding rolls in the corresponding pass to perform pass shape correction at fixed points, including but not limited to roll gap adjustment amount, arc radius correction amount, and feed adjustment value.

[0022] Step S6: Based on the pass shape correction parameters, send a correction command to the roll adjustment actuator of the corresponding pass to perform fixed-point correction of the pass shape parameters.

[0023] In this embodiment, the above steps work collaboratively. Step S1 constructs a three-dimensional spatial coverage network composed of multiple acoustic emission sensors in the R-angle forming area, providing a hardware foundation for capturing and locating the acoustic emission signals of the crack source in three-dimensional space. Step S2 operates in real time during the dynamic process of continuous production of high-strength steel welded pipes, continuously acquiring and preprocessing the multi-channel acoustic emission signals. When the material cracks or wrinkles due to stress concentration, the released acoustic emission elastic waves are captured by the sensor array. Step S3 obtains the signal arrival time difference required for location by recording the difference in response time of each sensor to the acoustic emission event. Step S4 uses these time differences and the stress-corrected sound velocity value to convert the time difference information of the physical world into the precise coordinates of the crack source in three-dimensional space. Steps S5 and S6 map the invisible spatial coordinate information into operable process parameters and specific adjustment commands, driving the actuator to complete the fixed-point correction. The entire process solves the technical problem that existing technologies cannot monitor, accurately locate, and precisely correct the R-corner wrinkling defects in the cold bending process in real time, achieving a technological leap from "post-event inspection and trial and error based on experience" to "online monitoring and quantitative correction".

[0024] Taking a specific application scenario as an example, on a continuous cold bending forming production line for high-strength steel welded pipes, the pass spacing in the R-angle forming area is 1.5 meters. First, in step S1, three sets of sensor groups are arranged at intervals along the pipe's travel direction (denoted as the X-axis) on the side walls or frame of each pass in the R-angle forming area. Each group has two acoustic emission sensors installed vertically and horizontally along the circumferential direction of the pipe cross-section (denoted as the Y-axis), and sensors are also arranged at different heights along the pipe wall thickness direction (denoted as the Z-axis), forming a three-dimensional spatial coverage network containing more than six sensors. The three-dimensional installation coordinates of each sensor are... The laser tracker is used for precise calibration and data entry into the control module. After production begins, step S2 continues, with each sensor acquiring vibration signals in the R-angle region in real time. These signals are amplified by a preamplifier and then processed by a bandpass filter. At a certain moment, a microcrack occurs inside the R-angle of the 5th pass due to plastic flow instability, releasing an acoustic emission signal. This signal is first detected by a sensor (number 0, coordinates: ...). If a sensor captures a signal whose amplitude exceeds a preset trigger threshold, step S3 is activated, and the sensor becomes a trigger sensor. The system synchronously latches this trigger moment. All other sensors (numbered) within 500 microseconds before and after ) signal waveform and The precise signal arrival time difference between each non-triggered sensor and the triggered sensor is calculated using a cross-correlation algorithm. Then, in step S4, the stress distribution data of the R-angle region, fed back online by the stress sensors installed on the unit, is used to calculate the sound velocity under the current stress state using the stress-sound velocity dynamic correction formula. The spatial coordinates of each sensor, and Substituting the equations of the three-dimensional hyperbolic positioning system, and through weighted fusion calculation of multiple sensor combinations, the precise three-dimensional coordinates of the crack source are obtained. In step S5, by Coordinate differences and pass spacing map the passes where cracking occurs. ;pass and The coordinates map the crack location to the upper roll responsible for forming the inner side of the radius (R-angle) in the 5th pass; and because The coordinates indicate that the crack originates from the inner layer of the pipe wall, determining that the feed rate of the roll needs to be adjusted. In step S6, the control system determines the crack based on the geometric deviation. Calculate the feed rate adjustment value The system automatically sends a command to the adjustment actuator of the fifth rolling mill to increase the rolling feed, thereby optimizing the material flow at that point and inhibiting the continued propagation of cracks. This completes the closed-loop control from monitoring to decision-making to execution.

[0025] In one embodiment, the sensor array mentioned in step S1 above comprises at least six acoustic emission sensors. Each sensor is distributed in at least two groups along the axial direction of the pipe forming direction, and each group comprises at least two sensors distributed circumferentially along the pipe cross-section, creating a three-dimensional spatial coverage between the sensors. This arrangement ensures a sufficiently long positioning baseline along the pipe forming direction and forms a surrounding coverage along the circumferential direction of the cross-section, avoiding positioning blind spots. When a crack occurs at any position of the R-angle, the generated elastic wave can be effectively received by at least three non-coaxial sensors, constituting the spatial geometric conditions for solving the three-dimensional hyperbolic positioning equations. Compared to single-sided or planar sensor arrays, this three-dimensional distribution has the unique advantage of providing sensitivity to depth in the wall thickness direction (Z-axis), solving the problem that traditional planar or linear positioning networks cannot distinguish whether the crack origin occurs on the pipe surface or inside, and providing crucial Z-axis coordinate information for subsequent selection of wall thickness-related correction parameters.

[0026] In one embodiment, the filtering process for each channel signal mentioned in step S2 above specifically employs a bandpass filter to remove low-frequency vibration noise and high-frequency electrical noise from the molding equipment, while retaining the signal components in the frequency band where the acoustic emission event signal is located. The transfer function of the bandpass filter is expressed as follows: In the formula: The transfer function of the bandpass filter represents the ratio of the filter's output signal to its input signal in the complex frequency domain, and is used to describe the filter's response characteristics to different frequency components. Let be the Laplace operator, representing the frequency variable in the complex frequency domain. ,in Angular frequency; The center frequency of the bandpass filter, i.e., the frequency value corresponding to the maximum point of the passband amplitude response of the filter, is pre-selected based on the main frequency distribution range of the acoustic emission event signal; The quality factor is a dimensionless parameter characterizing the passband selectivity of a bandpass filter. A larger value indicates a narrower passband and better selectivity. The gain coefficient represents the amplification factor of the bandpass filter at the center frequency. By selecting a frequency band that matches the acoustic emission event signal. and The value effectively suppresses low-frequency vibration noise and high-frequency electrical noise in the forming equipment. On a cold-bending forming production line, the operation of the mechanical structure, the rotation of the motor, and the pulsation of the hydraulic system generate low-frequency strong vibration noise concentrated in the 0-50kHz range, while electromagnetic interference and electrical sparks generate high-frequency electrical noise above several hundred kHz. The acoustic emission signal generated by wrinkling or cracking of the R-angle of high-strength steel is mainly concentrated in a specific mid-frequency band. Using the aforementioned bandpass filter, the center frequency is reduced... Set it to this frequency band and configure an appropriate quality factor. To limit the passband width, so that the transfer function This filtering method maintains high gain within this frequency band while drastically attenuating low-frequency and high-frequency components on both sides of the passband. It effectively suppresses low-frequency vibration noise and high-frequency electrical noise from the molding equipment, while ensuring the preservation of signal components in the frequency band of acoustic emission events caused by R-corner cracking. This filtering process provides a signal-to-noise ratio-friendly foundation for accurate extraction of subsequent signal arrival times, effectively avoiding missed and false detections of acoustic emission event signals under strong vibration and noise environments.

[0027] In one embodiment, regarding the calculation of the signal arrival time difference between each non-triggered sensor and the triggered sensor mentioned in step S3 above, the extraction of the signal arrival time adopts a threshold triggering method. When the signal amplitude of any channel exceeds a preset trigger threshold, it is determined to be an acoustic emission event trigger. The signal waveforms of all channels within a predetermined time window before and after the trigger moment are recorded synchronously, and the signal arrival time difference between each channel is accurately calculated through a cross-correlation algorithm. For the trigger sensor and the... The cross-correlation function between the non-triggered sensors is calculated using the following formula: In the formula: Indicates the trigger sensor and the first The cross-correlation function value between two non-triggered sensors represents the degree of similarity between the acoustic emission signal waveforms received by the two sensors under different time delay conditions. To trigger the acoustic emission signal waveform received by the sensor, which is a waveform that varies with time. The amplitude function of a changing voltage signal; Indicates the first The waveform of the delayed signal received by the non-triggered sensor indicates the time delay signal received by the first sensor. The raw signal waveforms of each sensor are shifted along the time axis. The waveform function after that; The acoustic emission event trigger moment is the moment when the signal amplitude of the trigger sensor channel first exceeds the preset trigger threshold. The signal integration window duration is used to limit the time domain range of the signal waveform used for cross-correlation calculations; As a time delay variable, it represents the time offset of shifting the waveform of the non-triggered sensor signal, and is used as an independent variable in the cross-correlation function for traversal searching; No. The signal arrival time difference between a non-triggered sensor and a triggered sensor is determined by the following formula: In the formula: Indicates the first The signal arrival time difference between the non-triggered sensors and the triggered sensors, i.e., the time difference in the propagation of the acoustic emission signal from the crack source to the trigger sensor. The difference between the time required for propagation from a non-triggered sensor to a triggered sensor; The time delay reference value corresponding to the autocorrelation peak of the sensor is used to trigger the correlation operation between the signal and itself, and the peak value appears at [time delay reference value]. place, As a time correction reference; First, the amplitude threshold method is used to quickly pinpoint the occurrence of acoustic emission events, and based on the trigger time... A segment of the signal waveform that fully captures the characteristics of the acoustic emission event was extracted. Subsequently, the cross-correlation function was... A global search is performed on the entire waveform within a set time window, and the time offset of the two sensor signal waveforms is measured by integration. The similarity of the waveform is considered. When the signal waveform is distorted due to differences in stress paths at its location, this matching method based on the similarity of the complete waveform has a fundamental advantage in anti-interference compared to single-point time determination based solely on the first peak crossing the threshold. Its unique advantage is that it can overcome the limitation of traditional threshold methods that rely solely on signal amplitude judgment, and achieve microsecond-level precision time difference extraction by using similarity matching of complete waveform information. This calculation method can maintain a high accuracy rate in time difference extraction even under strong background noise conditions, providing a precise distance difference measurement basis for subsequent three-dimensional spatial localization of crack sources, and significantly improving the accuracy and reliability of crack source localization.

[0028] In one embodiment, regarding the sound wave propagation speed in high-strength steel mentioned in step S4 above, the pre-calibration of the sound wave propagation speed in high-strength steel considers the influence of the stress state at different parts of the high-strength steel during cold bending forming on the sound velocity, and dynamically corrects the sound velocity value according to the stress distribution state at different parts of the pipe. During the cold bending forming process of high-strength steel, the sound wave propagation speed at the location of the crack source is calculated using the following stress-sound velocity dynamic correction formula: Among them, the actual stress value at the location of the crack initiation point. The values ​​are obtained through finite element mechanical model calculations of the cold bending forming process, or through online detection using stress sensors placed on the forming frame. Where: The sound wave propagation speed is the stress-corrected value, which takes into account the influence of the stress state of high-strength steel during cold bending, and is used for the calculation of the subsequent three-dimensional positioning equations. The reference sound velocity is the sound velocity under stress-free conditions, that is, the reference sound velocity value of high-strength steel material obtained by sound velocity measurement experiment on standard sample; The sound velocity-stress sensitivity coefficient characterizes the sensitivity of the sound wave propagation velocity in high-strength steel materials to stress changes, and is determined by the acoustoelasticity of the material. represents the elastic modulus of high-strength steel, while represents the inherent mechanical property parameter of the material. The actual stress value at the location of the crack initiation is obtained by calculation using a finite element mechanical model of the cold bending forming process, or by online detection using stress sensors placed on the forming frame. The reference stress value is usually taken as the stress value under stress-free conditions. ) as the calibration reference point; In continuous cold bending, the tensile and compressive surfaces of the strip in the radius (R-angle) region exhibit significantly different stress distributions. This difference in stress state leads to acoustoelastic effects, meaning the propagation speed of sound waves in the material varies with the type (tension / compression) and magnitude of the stress. If no correction is made and the sound velocity along all paths is directly treated as a constant, a systematic deviation will occur in the final calculated three-dimensional coordinates. This formula uses the reference sound velocity under stress-free conditions. Based on this, through the sound speed-stress sensitivity coefficient With elastic modulus The ratio of stress deviation The linear mapping is the relative change in sound velocity. Using this method, the sound velocity value used in the three-dimensional positioning equations can be matched with the actual physical state of the crack source location, avoiding the systematic positioning deviation caused by the fixed sound velocity assumption, and significantly improving the positioning accuracy of the crack source in the wall thickness direction (Z direction) to a level acceptable for industrial applications.

[0029] In one embodiment, regarding the solution of the three-dimensional spatial coordinates of the crack source mentioned in step S4 above, multiple sets of sensor combinations are used to perform positioning calculations separately, and the positioning results of each set are fused and optimized to eliminate positioning deviations caused by abnormal signals from individual sensors. For any effective sensor combination, the three-dimensional spatial coordinates of the crack source are solved using the following set of three-dimensional hyperbolic positioning equations: In the formula: Let be the three-dimensional spatial coordinates of the crack source to be solved, where These are the position coordinates along the pipe forming direction. These are the position coordinates along the circumferential direction of the pipe cross-section. These are the position coordinates along the pipe wall thickness direction; The spatial coordinates of the trigger sensor are the installation coordinates of the sensor in three-dimensional space that first acquires the acoustic emission event signal in the sensor array. For the first The spatial coordinates of a non-triggering sensor, that is, the three-dimensional spatial installation position coordinates of the other sensors in the sensor array besides the triggering sensor; This represents the corrected speed of sound, i.e. The possible values ​​of ; Represents the calculated first... The signal arrival time difference between a non-triggered sensor and a triggered sensor; For at least three different sensor combinations, multiple sets of three-dimensional spatial localization results of crack sources are obtained by solving the problem. The final localization result after fusion is calculated using the following weighted fusion formula: In the formula: The final three-dimensional spatial coordinates of the crack source after weighted fusion are the final output positioning result; The total number of sensor combinations participating in the fusion, i.e., the number of different effective sensor combinations selected from the sensor array; This is a number assigned to the sensor combination, used to distinguish the positioning results and weighting coefficients of different combinations; Indicates the first The positioning result obtained by the combined solution of the sensor group is the crack origin coordinates calculated by the sensor group through the three-dimensional hyperbolic positioning equations. Indicates the first The weighting coefficients of a group of sensors represent the reliability of the localization results of that group in the fusion process; No. Weighting coefficients of sensor group combinations The signal-to-noise ratio of each channel in this combination is determined as follows: In the formula: Indicates the first The average signal-to-noise ratio of the signals from each sensor channel involved in the positioning process in a sensor group combination. The larger this value is, the better the signal quality of the sensor group and the more reliable the positioning result. This represents the maximum mean signal-to-noise ratio among all sensor combinations involved in the fusion, and is used as the baseline value for normalization. Due to the complex working conditions at the molding site, individual sensors may experience severe signal waveform distortion due to poor coupling or local interference, resulting in significant deviations in the positioning results based on those sensors. Using only a single sensor combination for a single positioning calculation carries a high risk of failure. By selecting multiple sensor combinations (e.g., selecting four sensors from a total of six to form a combination), multiple positioning results are independently calculated. Then, normalized weights are calculated using the signal-to-noise ratio (SNR) of each combination's channel signals, and weighted fusion is performed. Combinations with higher SNR receive the highest weights. Large, for the final fusion result The greater the contribution, the better. This operational logic effectively eliminates abnormal positioning deviations caused by signal interference or poor coupling of individual sensors, significantly improving the positioning accuracy of the crack source in both the pipe forming direction and the circumferential direction of the cross-section, and providing accurate spatial location basis for subsequent hole-shaped point correction.

[0030] In one embodiment, regarding the pass parameter mapping relationship mentioned in step S5 above, this pass parameter mapping relationship is the correspondence between the three-dimensional spatial coordinates of the cracking source and the forming pass number, roll number, and pass correction parameters; wherein, the coordinates of the cracking source along the tube forming direction are used to determine the forming pass number where the cracking occurred, the coordinates of the cracking source along the circumferential direction of the tube cross-section are used to determine the roll number corresponding to the cracking position in that pass, and the coordinates of the cracking source along the tube wall thickness direction are used to determine the type of pass correction parameters. With forming pass number The following mapping relationship is used to determine the relationship between them: In the formula: The forming pass number corresponding to the crack source is used to determine which forming pass of the cold bending forming unit the crack occurred in. The three-dimensional coordinates of the crack source along the pipe forming direction are the final output three-dimensional coordinates. In Quantity; The coordinates of the starting position of the cold bending forming are the coordinates of the preset forming starting point in the tube forming direction; The spacing between adjacent forming passes is determined by the pass layout of the cold bending forming unit; This is the floor operator, which takes the largest integer not greater than the value inside the parentheses. Coordinates of the crack initiation along the circumferential direction of the pipe cross-section Coordinates in the wall thickness direction With roll number The following mapping relationship is used to determine the relationship between them: In the formula: The roll number corresponding to the cracking source is used to determine which group or side of the forming rolls the cracking location specifically corresponds to. The coordinate-roll mapping function is established based on the geometric model of the tube cross-section, according to the circumferential position of the crack initiation point in the cross-section. and position in the wall thickness direction Output the corresponding roll number; Crack source coordinates and hole shape correction parameter types The following mapping relationship is used to determine the relationship between them: In the formula: This is the type of hole profile correction parameter, used to determine the type of hole profile parameter that should be adjusted. This is a parameter type mapping function based on the position in the wall thickness direction. When the crack source is located in the surface layer of the wall thickness, it outputs the roll gap adjustment type, and when the crack source is located in the inner layer of the wall thickness, it outputs the feed adjustment type. In step S5, the actual contour data of the pipe cross-section is acquired by a laser contour scanner / machine vision measuring device arranged on the outlet side of the cold bending forming unit or after the corresponding pass, and the actual R-angle radius value of the crack source location is extracted from it. and actual forming angle value The actual geometric deviation of the radius (R-angle) at the crack origin. Calculated using the following formula: ; ; In the formula: is the actual geometric deviation of the R-angle at the crack initiation location, and is the comprehensive deviation vector including the R-angle radius deviation and the forming angle deviation; This represents the deviation between the actual radius of the right angle and the target radius of the right angle. This represents the deviation between the actual forming angle and the target forming angle. The actual radius of the R-angle at the crack initiation location reflects the actual R-angle geometry after cold bending at that location; The target radius of the R-angle is the design target value for the R-angle of this specification of high-strength steel welded pipe; This is the actual forming angle value at the crack initiation location, reflecting the actual angle after cold bending at that location; The target forming angle value is the design target angle of the R-angle of this high-strength steel welded pipe. The above series of mapping and calculation processes will output an abstract pure geometric coordinate from step S4. This is broken down layer by layer into specific instructions that process engineers understand, such as "which pass," "which set of rolls," "which parameters to adjust," and "how much to adjust." For example, using the forming direction coordinates... Compared with track number benchmark The difference, divided by the spacing Rounding down gives the physical pass number of the cold bending forming machine. Using wall thickness coordinates... Call the decision function This method determines the type of correction and resolves the intrinsic differences in correction strategies between surface wrinkling and internal microcracks. Through the above calculations, a quantitative correspondence is established between the spatial positioning results and the forming process parameters, enabling rapid and accurate judgment of "which pass the crack occurred in, which set of rolls it corresponds to, and which parameters to correct," thus avoiding correction decision errors caused by experience differences in the traditional manual judgment mode.

[0031] In one embodiment, the fixed-point correction of the pass profile parameter mentioned in step S6 above includes the following methods: adjusting the roll gap of the corresponding roll in the corresponding pass, adjusting the radius of curvature of the corresponding roll in the corresponding pass, adjusting the feed amount of the corresponding pass, or any combination of the above methods. The adjustment amount of the roll gap... Calculated using the following formula: In the formula: This is the roll gap adjustment amount. A positive value indicates increasing the roll gap, and a negative value indicates decreasing the roll gap. It is used to adjust the gap size between the upper and lower rolls of the corresponding pass. is the roll gap-radius correction coefficient, which is a proportionality coefficient calibrated through experiments, characterizing the roll gap adjustment amount corresponding to a unit R-angle radius deviation; This represents the deviation between the determined actual radius of the radius (R) and the target radius of the radius (R). This is the roll gap compensation constant, a constant term used for zero-point correction or offset compensation of roll gap adjustment. Correction amount for arc radius Calculated using the following formula: In the formula: This is the radius correction amount; a positive value indicates an increase in the radius of the arc, and a negative value indicates a decrease in the radius of the arc. It is used to adjust the radius of the R-angle forming arc of the corresponding roll in the corresponding pass. The radius-radius correction factor is a proportionality coefficient calibrated through experiments, which characterizes the amount of roll arc radius correction corresponding to a unit R-angle radius deviation; This is a radius compensation constant, a constant term used for zero-point correction or offset compensation of the radius correction of the arc. Feed rate adjustment value Calculated using the following formula: In the formula: This is the feed rate adjustment value. A positive value indicates an increase in the feed rate, and a negative value indicates a decrease in the feed rate. It is used to adjust the pipe feed speed or advance rate for the corresponding pass. This represents the deviation between the determined actual forming angle and the target forming angle; , is the feed-angle correction coefficient, which is a proportional coefficient calibrated through experiments, representing the feed adjustment value corresponding to a unit forming angle deviation; This is the feed compensation constant, a constant term used for zero-point correction or offset compensation of the feed rate adjustment value; The above calculation formulas will *abstract geometric deviations* and This is linearly converted into a physical quantity that the actuator can directly respond to. Among these, the coefficients... , , It is a sensitivity coefficient established through offline experiments, reflecting the physical correction required per unit geometric deviation, and a compensation constant. , , These formulas are used to eliminate systematic errors such as mechanical backlash in the actuator. Using these formulas, the precise adjustment direction and amplitude can be automatically calculated based on the deviation characteristics of the actual crack source, achieving differentiated and precise corrections for different cracking modes and deviation types, and significantly improving the success rate of hole correction in one attempt.

[0032] In one embodiment, after completing step S6 above, the method further includes a correction effect verification step: after the aperture correction is completed, the acoustic emission signal at the location of the crack source is continuously monitored. If no acoustic emission event is triggered again within a predetermined time period, the correction is deemed effective; if an acoustic emission event is triggered again, the correction amount is increased or an offline aperture redesign process is triggered. The determination of the correction effectiveness is made in the following way: after the aperture correction is completed, the acoustic emission signal energy value at the location of the crack source is collected. And compared with the reference energy value in the crack-free state. Compare: In the formula: This represents the energy value of the acoustic emission signal collected after correction, characterizing the total energy of the acoustic emission signal at this location within the verification time window, and is used to quantitatively assess whether crack damage signals still exist at this location after correction. The correction completion time is the point in time when the hole pattern positioning correction is completed. To verify the time window, and to continuously monitor the acoustic emission signal for a period of time after correction, the energy of the acoustic emission signal is cumulatively calculated within this time window; Let be the instantaneous amplitude of the acoustic emission signal, and be a time-varying amplitude function of the voltage signal, the square of which is proportional to the instantaneous power of the signal; when In the verification time window The maximum value within the range remains consistently lower than the baseline energy value under the crack-free state. When the predetermined multiple is reached, the correction is deemed valid; Among them, the reference energy value in the crack-free state The baseline energy value is obtained in advance by collecting the acoustic emission signal energy value at the location under normal operating conditions, under the condition that the cold bending forming unit is operating normally and there are no crack defects in the R-angle area, and taking the statistical average value as the baseline energy value. The corrected signal energy is continuously monitored. It is the integral of the square of the instantaneous amplitude over time, reflecting the total strength of the signal within the statistical period. This is then compared with the reference energy. By comparing the amplitude with, rather than simply exceeding the threshold instantaneously, short-term high-frequency interference caused by accidental particle impacts or structural micro-movements during the molding process can be eliminated as a result of misjudgment. When the corrective measures effectively eliminate the crack source, the acoustic emission activity at that location should return to the normal background noise level. It will remain at a low level. This "correction-verification" closed-loop evaluation mechanism ensures that each round of correction achieves the expected technical effect, effectively avoiding repeated cracking caused by incomplete correction.

[0033] In one embodiment, the offline hole redesign process includes: packaging the three-dimensional spatial coordinates of the current cracking event, hole correction parameters, invalidation judgment results, and related process data to generate an anomaly report; pushing the anomaly report to the process engineer's terminal; and having the process engineer perform offline redesign and verification of the hole parameters based on the anomaly report.

[0034] In one embodiment, the method further includes a data recording and self-learning step: storing the three-dimensional spatial coordinates of the crack source, the corresponding aperture correction parameters, and the correction effect of each acoustic emission event into an empirical database for optimization reference in the design of apertures for subsequent products of the same specification or batch. In this data recording and self-learning step, the data is processed in the empirical database... A correction record, whose data items are represented as follows: In the formula: Indicates the first The entire set of data items in each correction record is stored in the empirical database in the form of structured data, representing the entire process of a complete cracking event from location to correction to verification; Indicates the first The three-dimensional spatial coordinates of the secondary crack source are the final output positioning result, recording the precise spatial location where the crack occurred; Indicates the first The forming pass number corresponding to the second crack is the pass mapping result; Indicates the first The roll number corresponding to the secondary crack is the roll mapping result; Indicates the first The roll gap adjustment amount used in this correction is the actual adjustment amount. Indicates the first The radius correction amount used in this correction is the actual adjustment amount. Indicates the first The feed rate adjustment value used in this correction is the actual adjustment value. Indicates the first The result of the validity determination of the second correction is the output judgment conclusion, which is usually expressed in Boolean form as correction valid or correction invalid; When the same forming pass number and the same roll number When the accumulated correction records reach a predetermined number, the correction parameters corresponding to that location in the experience database are statistically analyzed, and the statistically optimal value is taken as the updated value of the standard pass correction parameters for that forming pass and that roll position, for use in the pass design of subsequent products of the same specification. By structurally recording the complete process data from positioning, decision-making to correction and verification, a continuously growing process knowledge base is formed. When enough correction records are accumulated for the same specific location (same pass, same roll), the system can use statistical analysis to extract the most stable and effective combination of correction parameters for that location from historical data, replacing the correction amount that relied on the simplified model during the initial calculation. This self-learning mechanism enables the pass correction accuracy to continuously improve with the continuous accumulation of production data, realizing a technological upgrade from single correction to continuous optimization.

[0035] It should be noted that, in some optional implementations, the statistically optimal value can be determined by one of the following methods: taking the average, weighted average, or median of the correction parameters of all valid correction records at that position as the statistically optimal value.

[0036] It should be noted that, in some optional implementations, the acoustic emission sensor can be a piezoelectric ceramic broadband sensor with a response frequency range covering 100kHz to 1MHz. The sensor is mounted using magnetic clamps or threaded fasteners to the polished surface of the roll mill frame, with high-temperature vacuum silicone grease applied to the coupling surface to ensure acoustic coupling performance. The sensor signal line uses a shielded low-noise coaxial cable, which, after being amplified by 40dB by a preamplifier, is sent to a multi-channel synchronous data acquisition card. Each channel of the acquisition card has a sampling rate of no less than 10MS / s, and each channel has strict synchronous sampling and holding capabilities. Filtering can be achieved using a cascaded analog and digital filter. The amplified signal first passes through an analog anti-aliasing low-pass filter with an adjustable cutoff frequency, and then undergoes fine filtering by an FIR bandpass filter in the digital signal processor. The passband range can be set from 200kHz to 500kHz to suit the material characteristics of high-strength steel.

[0037] In other alternative implementations, the finite element mechanical model of the cold bending process can be established based on commercial finite element software to simulate the continuous elastoplastic deformation process of the strip steel through each pass in the die, obtaining the stress distribution cloud map of the strip steel at each node along the thickness direction in the R-angle region, thereby providing the stress-sound velocity dynamic correction formula. The system provides real-time data queries. For online stress sensor detection solutions, force sensors can be placed at the roll bearing housings or lead screws in key forming passes to indirectly obtain load information in the radius (R-angle) region. This information, combined with analytical models or pre-established force-stress conversion relationships, allows for the calculation of stress values ​​at crack initiation sites. Furthermore, the coordinate-roll mapping function... Specifically, this can be implemented as an offline lookup table, which, based on the cross-sectional geometric design parameters of the pipe specification, determines the circumferential position of the cross-section. and wall thickness location The gridded coordinates are mapped to the specific roll number responsible for forming that grid area. Parameter type mapping function Specifically, it can be implemented as follows: when When the coordinates are within 20% of the depth of the surface on both or one side of the wall thickness, it is determined to be a surface wrinkling mode, and the system is switched to roll gap or arc radius correction; when When the coordinates are located in the middle region of the wall thickness, it is determined to be an internal material flow instability mode, and the feed rate correction is switched. When the correction is ineffective and triggers the offline hole redesign process, all related data in this round can be automatically packaged to generate an anomaly report and pushed to the process engineer's terminal to facilitate root cause analysis.

[0038] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the hole profile correction method for preventing wrinkling defects in cold bending of high-strength steel welded pipes as described in any of the above embodiments. By executing program instructions including the aforementioned acoustic emission signal processing, spatial positioning, coordinate mapping, and correction calculation, this computer-readable storage medium can provide software support for the measurement and control system on the cold bending production line.

[0039] like Figure 5 As shown in the embodiments of this application, an experimental embodiment is also provided. To verify the effectiveness of the hole shape correction method for preventing wrinkling defects in cold bending of high-strength steel welded pipes proposed in this invention, a systematic experimental verification was conducted on a high-strength steel welded pipe production line. The experiment used Q355B grade high-strength steel strip with a width of 750mm and a thickness of 7.75mm. The target product was a 400mm×400mm square / rectangular tube with a radius of R20 for the R-angle design. The forming pass spacing was 1.5m, and a total of 12 forming rolls were set. Eight piezoelectric ceramic broadband sensors were used as acoustic emission sensors, arranged in four groups along the pipe forming direction, with two sensors arranged circumferentially around the pipe cross-section in each group, forming a three-dimensional spatial coverage network. The three-dimensional spatial coordinates of each sensor were calibrated using a laser tracker, with a calibration accuracy better than ±0.1mm. The data acquisition card sampling rate was set to 10MS / s, the bandpass filter center frequency was set to 350kHz, and the quality factor was set to... Set to 5. Preset verification threshold coefficient. The value was set to 0.70. A total of 120 batches were conducted, with each batch producing 1000 meters of pipe, for a total experimental pipe length of 120,000 meters. The control and experimental groups were conducted on the same production line, under the same operating conditions and with the same operators. The two groups were experimented on separate batches to avoid interference from sequence effects or learning effects on the experimental conclusions. The control group was a conventional production line of the same specifications and batches without the acoustic emission monitoring and point-of-care correction system, while the experimental group was a production line using the technical solution of this invention.

[0040] Filtering effect verification experiment: To verify the effect of the bandpass filter on improving the signal-to-noise ratio of acoustic emission signals, acoustic emission signal data before and after filtering were collected under the same molding conditions. The signal before filtering is the original signal collected by the sensor, and the signal after filtering is the signal processed by the bandpass filter. 100 signal samples were collected for each data point, and the average signal-to-noise ratio was calculated.

[0041] Table 1 Comparison of signal-to-noise ratio of acoustic emission signals before and after filtering.

[0042] After filtering, the signal-to-noise ratio (SNR) of the acoustic emission signals in the R-angle region of each pass improved from 9.6 dB–13.1 dB before filtering to 26.8 dB–30.2 dB after filtering. The average SNR increased from 11.4 dB to 28.4 dB, an average improvement of 17.0 dB. This significant improvement in SNR indicates that the bandpass filter effectively suppresses low-frequency vibration noise and high-frequency electrical noise from the molding equipment, providing a good signal foundation for accurate extraction of subsequent signal arrival times.

[0043] Cross-correlation time difference extraction accuracy verification experiment: To verify the accuracy of the cross-correlation algorithm in extracting the signal arrival time difference, a signal generator was used to simulate an acoustic emission source. Under the condition of known sensor spatial coordinates, the calculated time difference value was compared with the theoretical time difference value. The theoretical time difference value was calculated by dividing the spatial distance between the sound source coordinates and the sensor coordinates by the speed of sound. A total of 50 simulated acoustic emission events were conducted, and the response signals of 8 sensors were recorded simultaneously for each event. The table below selects 3 representative sensors to show their time difference data; the measurement error range of the remaining 5 sensors is 0.1μs to 0.4μs. The magnitude of the data is roughly the same as that in the table.

[0044] Table 2. Verification of the accuracy of cross-correlation algorithm for time difference extraction

[0045] The mean absolute error between the time difference calculated by the cross-correlation algorithm and the theoretical time difference is 0.25. The maximum absolute error does not exceed 0.4. The time difference extraction error at this level of accuracy corresponds to a spatial positioning error of approximately 0.08 mm to 0.15 mm, which is much smaller than the geometric scale of wrinkling defects in the R-corner region, indicating that the cross-correlation algorithm can achieve time difference extraction with microsecond-level accuracy.

[0046] Stress-sound velocity dynamic correction effect verification experiment: To verify the effect of the stress-sound velocity dynamic correction formula on improving the accuracy of three-dimensional positioning, under the same molding conditions, a fixed sound velocity value (based on the reference sound velocity in the stress-free state) was used. The crack source location was calculated using a fixed value and a sound velocity value after stress-sound velocity dynamic correction. The location accuracy was characterized by the deviation between the location result and the actual crack source location, which was obtained through post-experimental cross-sectional measurement. Location deviation data from 60 crack events were collected.

[0047] Table 3 Comparison of positioning accuracy before and after stress-velocity dynamic correction

[0048] When using a fixed sound velocity value for positioning, the average absolute deviations in the X, Y, and Z directions were 8.5 mm, 6.8 mm, and 15.2 mm, respectively, with a comprehensive three-dimensional spatial deviation of 18.7 mm. When using a sound velocity value after stress-sound velocity dynamic correction for positioning, the average absolute deviations in the X, Y, and Z directions decreased to 3.2 mm, 2.7 mm, and 2.8 mm, respectively, with a comprehensive three-dimensional spatial deviation reduced to 5.0 mm. The most significant improvement in positioning accuracy was observed in the wall thickness direction (Z direction), where the deviation decreased from 15.2 mm to 2.8 mm, an improvement of 12.4 mm. This result demonstrates that stress-sound velocity dynamic correction can effectively eliminate the systematic positioning deviation caused by the fixed sound velocity assumption, particularly playing a decisive role in improving positioning accuracy in the wall thickness direction (Z direction).

[0049] Verification Experiment of Multi-Sensor Combined Weighted Fusion Positioning Effect: To verify the improvement effect of the multi-sensor combined weighted fusion algorithm on positioning accuracy, comparative experiments were conducted using a single-sensor combination (4 sensors) positioning method and a multi-sensor combination (selecting multiple 4-sensor combinations from 8 sensors) weighted fusion positioning method. Positioning deviation data from 80 cracking events were collected.

[0050] Table 4 Comparison of single-group positioning and weighted fusion positioning accuracy

[0051] Compared to single-sensor combination positioning, the weighted fusion positioning method using multiple sensors reduced the average absolute deviations in the X, Y, and Z directions from 6.7 mm, 5.2 mm, and 6.3 mm to 2.8 mm, 2.3 mm, and 2.5 mm, respectively, and the overall 3D spatial deviation from 10.5 mm to 4.4 mm. The maximum single positioning deviation decreased from 18.3 mm to 7.2 mm, and the standard deviation of the positioning deviation decreased from 3.8 mm to 1.6 mm. These results demonstrate that the weighted fusion algorithm can effectively eliminate abnormal positioning deviations caused by signal interference or poor coupling of individual sensors, significantly improving the stability and consistency of the positioning results.

[0052] Verification Experiment of Hole Pattern Fixed-Point Correction Effect: To verify the effect of the hole pattern fixed-point correction method of the present invention on preventing R-corner wrinkling defects, under the same molding conditions, a control group without the correction method of the present invention and an experimental group using the correction method of the present invention were set up respectively. The R-corner wrinkling and cracking incidence and molding accuracy index of the two groups during continuous production were statistically analyzed. The control group was a conventional production line of the same specification and batch without the installation of acoustic emission monitoring and fixed-point correction system, and the experimental group was a production line using the technical solution of the present invention. The total length of pipe in each group was 120,000 meters. The cracking rate of 0.48% in the control group is basically consistent with the R-corner cracking rate of 0.5% recorded in the prior art for the same specification Q355B production line.

[0053] Table 5 Comparison of R-angle forming quality before and after hole shape correction.

[0054] The average single-correction response time of 0.42 seconds is broken down as follows: signal processing and positioning calculation approximately 0.18 seconds, coordinate mapping and correction calculation approximately 0.12 seconds, and actuator action approximately 0.12 seconds. This response time matches the 10MS / s sampling rate of the data acquisition card. Compared to the control group without this invention, the experimental group using the hole-shaped fixed-point correction method of this invention showed a reduction in the R-corner cracking rate from 0.48% to 0.09%, a decrease of 0.39 percentage points; the R-corner radius deviation decreased from ±1.8mm to ±0.8mm; the consistency of the four corners increased from 84.5% to 96.2%; the unevenness of the four planes decreased from 0.72mm to 0.38mm; and the R-corner wall thickness reduction rate decreased from 9.6% to 7.2%. The average single-correction response time of the experimental group was 0.42 seconds, and the success rate of hole-shaped correction on the first attempt was 87.5%. These results indicate that the method of this invention can effectively prevent the occurrence of R-corner wrinkling defects in the cold bending forming of high-strength steel welded pipes and significantly improve the forming accuracy of welded pipes.

[0055] Correction Effect Verification Experiment: To verify the accuracy of acoustic emission signal energy monitoring in determining the correction effect, 126 correction events occurred during the entire 120,000-meter production process of the experimental group. For each correction event, the acoustic emission signal energy values ​​before correction, after correction, and within the verification time window were recorded and compared with the baseline energy values. A preset verification threshold coefficient was used. .

[0056] Table 6. Statistical data on the verification of the correction effect

[0057] Of the 126 correction events, 109 were deemed effective after a single correction (86.5%), 14 required a second correction (11.1%), and 3 required triggering an offline aperture redesign process (2.4%). In effective correction events, the ratio of the corrected acoustic emission signal energy value to the reference energy value ranged from 0.12 to 0.58, all below the predetermined threshold of 0.70. In ineffective correction events, this ratio ranged from 0.72 to 1.42, all above the predetermined threshold. A clear numerical gap of 0.58–0.72 was formed between the energy ratios of effective and ineffective corrections, with the predetermined threshold of 0.70 falling within this gap. These results indicate that using the comparison of acoustic emission signal energy value with the reference energy value for correction effectiveness determination has high accuracy and reliability, and can effectively distinguish between effective and ineffective correction states.

[0058] Independent verification experiment on coordinate mapping accuracy: To verify the accuracy of mapping the three-dimensional spatial coordinates of the crack source to the forming pass number and roll number, acoustic emission simulation sources with known spatial locations were pre-positioned in the radius (R) region of each pass in the forming mill. The method of this invention was used for positioning and coordinate mapping, and the mapping results were compared with the actual pass and roll location of the sound source. Simulated sound sources were set at different R-angle positions in 12 passes, with each position tested 10 times.

[0059] Table 7. Verification Results of Coordinate Mapping Accuracy

[0060] The accuracy rates of the forming pass number mapping and the correction parameter type mapping were both 100%, the roll number mapping accuracy was 98.3%, and the overall mapping accuracy was 98.3%. These results indicate that the coordinate-process parameter mapping relationship has high accuracy, and the spatial location results of the crack source can be accurately converted into specific passes, rolls, and correction parameter types, providing a reliable decision-making basis for subsequent point-to-point correction.

[0061] Self-learning effect verification experiment: To verify the effect of data recording and self-learning steps on the optimization of die correction parameters, the inner roll position of the R angle in the 5th pass was selected as the analysis object, and the correction record data accumulated at this position during continuous production were statistically analyzed. The first 60 correction events were used as the training dataset, and the last 66 correction events were used as the verification dataset to compare the accuracy of the correction parameters before and after self-learning.

[0062] Table 8 Comparison of Parameter Optimization Before and After Self-Learning

[0063] After statistically analyzing the accumulated correction records for the inner roll position of the R-angle in the 5th pass and taking the statistically optimal value as the updated standard correction parameter value, the success rate of correction at this position increased from 78.3% to 92.4%, the average deviation rate of single correction decreased from 18.6% to 5.2%, the proportion requiring secondary adjustment after invalid correction decreased from 21.7% to 7.6%, and the final R-angle deviation at this position decreased from ±1.2mm to ±0.6mm. These results demonstrate that the data recording and self-learning mechanism can continuously optimize the accuracy of correction parameters by accumulating historical correction data, allowing the die correction accuracy to gradually improve with the continuous accumulation of production data.

[0064] Based on the above experimental data, the solution of this invention is compared with the existing technical solutions, and the results are summarized as follows: Table 9. Comprehensive Comparison of the Invention Solution and Existing Technical Solutions

[0065] The experimental data above demonstrate that the hole shape correction method for preventing wrinkling defects in cold bending of high-strength steel welded pipes provided by this invention has achieved significant technical effects in terms of improving the acoustic emission signal-to-noise ratio, time difference extraction accuracy, three-dimensional spatial positioning accuracy, especially wall thickness direction positioning accuracy, coordinate mapping accuracy, control of R-corner cracking defect incidence, improvement of forming accuracy, and improvement of correction efficiency. The experimental data verify the effectiveness of the complete methodological process of this invention, from acoustic emission signal monitoring, three-dimensional spatial positioning, coordinate mapping, hole shape fixed-point correction, correction effect verification to data self-learning.

[0066] The embodiments disclosed in this invention are preferred embodiments, but are not limited thereto. Those skilled in the art can easily understand the spirit of this invention based on the above embodiments and make different extensions and variations, but as long as they do not depart from the spirit of this invention, they are all within the protection scope of this invention.

Claims

1. A method for correcting the die shape to prevent wrinkling defects in cold bending of high-strength steel welded pipes, characterized in that, include: An acoustic emission sensor array is arranged in the R-angle forming area of ​​the cold bending forming unit. The acoustic emission sensor array includes multiple acoustic emission sensors, and the three-dimensional spatial coordinates of each acoustic emission sensor are pre-calibrated. During the continuous cold bending process of high-strength steel welded pipe, the acoustic emission signal in the R-angle region is collected in real time by the acoustic emission sensor array, and the acoustic emission signal is preprocessed. When an acoustic emission event is detected, the arrival time of the signal received by all acoustic emission sensors is recorded, and the signal arrival time difference between each non-triggering sensor and the triggering sensor is calculated. The triggering sensor is the sensor that first detects the acoustic emission event. Based on the signal arrival time difference and the propagation speed of sound waves in high-strength steel, a three-dimensional spatial positioning equation system is established, and the three-dimensional spatial coordinates of the crack source are obtained by solving it. The three-dimensional spatial coordinates of the cracking source are matched with the preset mapping relationship of the die shape parameters to determine the forming pass number, roll number and die shape correction parameters corresponding to the cracking source. According to the pass shape correction parameters, a correction command is sent to the roll adjustment actuator of the corresponding pass to perform fixed-point correction of the pass shape of the corresponding pass.

2. The method for correcting the die shape to prevent wrinkling defects in cold bending of high-strength steel welded pipes according to claim 1, characterized in that, The acoustic emission sensor array contains at least 6 acoustic emission sensors, and each acoustic emission sensor is distributed in at least two groups along the pipe forming direction. Each group has at least 2 sensors distributed along the circumferential direction of the pipe cross-section, forming a three-dimensional spatial coverage network.

3. The method for correcting the hole shape to prevent wrinkling defects in cold bending of high-strength steel welded pipes according to claim 1, characterized in that, The calculation of the signal arrival time difference between each non-triggering sensor and the triggering sensor specifically includes: extracting the signal waveform within a preset time window before and after the triggering time of the acoustic emission event, calculating the similarity peak value between the signals of the triggering sensor and each non-triggering sensor through a cross-correlation algorithm, and obtaining the signal arrival time difference based on the time offset corresponding to the similarity peak value.

4. The method for correcting the hole shape to prevent wrinkling defects in cold bending of high-strength steel welded pipes according to claim 1, characterized in that, The propagation speed of the sound wave in the high-strength steel is the sound velocity value after stress dynamic correction. The sound velocity value is calculated based on the actual stress value in the R-angle region and the reference sound velocity under stress-free conditions.

5. The method for correcting the hole shape to prevent wrinkling defects in cold bending of high-strength steel welded pipes according to claim 1, characterized in that, The process of obtaining the three-dimensional spatial coordinates of the crack source specifically includes: selecting multiple different sensor combinations and substituting them into the three-dimensional spatial positioning equations to obtain multiple preliminary coordinates; determining the weighting coefficients based on the signal-to-noise ratio of each sensor combination; and performing weighted fusion of the multiple preliminary coordinates to obtain the final three-dimensional spatial coordinates of the crack source.

6. The method for correcting the hole shape to prevent wrinkling defects in cold bending of high-strength steel welded pipes according to claim 1, characterized in that, The specific mapping relationship of the hole shape parameters is as follows: the coordinates of the crack source along the tube forming direction are mapped to the forming pass number, the coordinates along the circumferential direction of the tube cross section are mapped to the roll number, and the coordinates along the tube wall thickness direction are mapped to the hole shape correction parameter type.

7. The method for correcting the die shape to prevent wrinkling defects in cold bending of high-strength steel welded pipes according to claim 1, characterized in that, The specific steps of performing fixed-point correction for the corresponding pass profile include adjusting at least one of the following: adjusting the roll gap of the corresponding roll, adjusting the radius of the arc of the corresponding roll, and adjusting the feed amount of the corresponding pass.

8. The method for correcting the hole shape to prevent wrinkling defects in cold bending of high-strength steel welded pipes according to claim 1, characterized in that, After performing the fixed-point correction of the corresponding pass pattern, the method further includes: continuously monitoring the acoustic emission signal energy value at the location corresponding to the crack source within a preset verification time window, comparing the energy value with the reference energy value in the crack-free state, and determining that the correction is effective if the energy value is lower than a preset threshold; otherwise, increasing the correction amount or triggering the offline pass pattern redesign process.

9. The method for correcting the die shape to prevent wrinkling defects in cold bending of high-strength steel welded pipes according to claim 1, characterized in that, Also includes: The three-dimensional coordinates of the crack source, the die correction parameters, and the correction effect data corresponding to each acoustic emission event are structured and stored in the experience database. When the correction records of the same forming pass and roll position accumulate to a preset number, the optimal correction parameters are extracted and statistically analyzed as the standard die parameter update value for that position, which is then used for the die design of subsequent products of the same specification.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the hole profile correction method for preventing wrinkling defects in cold bending of high-strength steel welded pipes as described in any one of claims 1 to 9.

Images