A metal plate seam folding and uniform speed welding system

Abstract

This invention discloses a uniform-speed welding system for metal plate folds, relating to the field of uniform-speed welding technology. The system involves scanning the metal plate to be welded to construct three-dimensional data of its folds and analyzing and extracting candidate hotspot regions for preliminary welding tests. The optimal test section is selected from these candidate regions as the initial test section. Uniform-speed welding is then performed on this test section, and the welding feedback results are used to determine whether the test section length needs to be extended. If extension is necessary, the test section is adjusted, and uniform-speed welding continues. The feedback is then used to evaluate whether the remaining folds are suitable for continued uniform-speed welding. This method uses three-dimensional data to assist in selecting the test section, combined with dynamic adjustment based on feedback, balancing accurate prediction of welding quality with cost control. It avoids situations where the test section is too short, affecting judgment, or too long, leading to wasted materials and time, thus achieving a balance between welding quality assurance and economic benefits.

Description

Technical Field

[0001] This invention relates to the field of uniform speed welding technology, and more specifically to a uniform speed welding system for metal plate folds. Background Technology

[0002] In the field of light pole manufacturing, the folding and welding of metal plates is a crucial step in achieving structural sealing and strength integrity. Since most light poles have an axisymmetric structure, their folding paths often exhibit geometric continuity and regularity. Especially under standardized production conditions, parameters such as folding gaps, plate butt joint accuracy, and slope variations are relatively stable, making "uniform speed welding" a feasible and efficient welding method. The advantages of uniform speed welding are: constant heat input, which helps control the welding heat cycle and deformation accumulation; stable weld formation, which facilitates subsequent machining and surface treatment; and ease of robot path control and cycle time management, significantly improving automated production efficiency. This method has been widely used in the main seam welding of conical, hexagonal, and transitional light poles.

[0003] While uniform-speed welding is theoretically applicable to most standard folds, its feasibility still requires verification through a "trial welding" phase in practical applications. Currently, industrial sites commonly use manual methods to select a section of the fold for trial welding to assess whether a uniform-speed mode can be used for the overall welding. However, this manual selection of test sections lacks a systematic approach, often relying on experience or approximate equal division. This may result in a test section with low local complexity, which does not represent the possibility of uniform-speed welding of the entire fold, leading to misjudgments. On the one hand, if the selected section is too short, it may fail to expose potential structural fluctuations and operational disturbances, affecting the reliability of weld quality predictions. On the other hand, if the test section is too long, it will cause greater material and time losses when adaptation errors occur. Furthermore, existing methods lack an adaptive adjustment mechanism for the length of the test section, making it impossible to dynamically sense the actual welding response during the welding process and adjust subsequent strategies. If uniform-speed welding is still blindly adopted, it can lead to problems such as uneven weld penetration, hot cracking, and forming defects, reducing welding stability and yield. Summary of the Invention

[0004] The purpose of this invention is to solve the problems mentioned above and provide a uniform speed welding system for metal plate folds.

[0005] This invention proposes a uniform speed welding system for metal plate folds, the system comprising:

[0006] Preliminary selection module: Scans the metal plate to be welded, constructs three-dimensional data of the weld joint, analyzes the three-dimensional data of the weld joint, and extracts the preliminary selection area of ​​the weld joint that can be used for welding tests as a hot spot candidate area;

[0007] Preliminary test module: Screens hotspot candidate regions and selects the optimal test segment from the hotspot candidate regions as the preliminary test segment;

[0008] Judgment module: Perform uniform welding on the preliminary test section and determine whether to extend the length of the preliminary test section based on the welding feedback effect;

[0009] Uniform speed welding module: If it is necessary to extend the length of the preliminary test section, then continue uniform speed welding according to the extended preliminary test section, and judge whether the remaining folds to be welded can continue to be welded using uniform speed welding based on the feedback of the welding results.

[0010] Optionally, the preliminary selection module includes:

[0011] Path segmentation module: Divides the three-dimensional path of the weld seam into multiple continuous path segments at equal intervals;

[0012] Feature vector module: For each path segment, extract its descriptive geometric features and construct a feature vector. The feature vector includes the mean local curvature, the mean normal change angle, and the standard deviation of the fold width of the path segment.

[0013] Feature change rate module: Calculates the first-order difference between the feature vector of each path segment and the feature vector of the previous path segment, as the feature change rate of the path segment;

[0014] The variable acceleration module calculates the second-order difference based on the characteristic rate of change of the path segment, which is used as the variable acceleration of the path segment.

[0015] Disturbance intensity scoring module: The characteristic change rate and change acceleration of each path segment are weighted and summed to obtain the disturbance intensity score of the corresponding path segment;

[0016] Hotspot Candidate Region Module: Based on the disturbance intensity score of each path segment, the preliminary selection area for welding tests of the fold joint is extracted as the hotspot candidate region.

[0017] Optionally, the hotspot candidate region includes:

[0018] Candidate segment module: Compare the disturbance intensity score of each path segment with a preset threshold. If the disturbance intensity score is not less than the preset threshold, the corresponding path segment is recorded as a candidate segment.

[0019] Filtering module: The fold area with a number of consecutive candidate segments not less than a preset threshold is taken as a hot spot candidate area, and several hot spot candidate areas are obtained.

[0020] Optionally, the preliminary test module includes:

[0021] Test Segment Module: For each hotspot candidate region, n consecutive candidate segments are generated as test segments using a sliding window method. The total number of candidate segments contained in the test segment is no greater than the total number of candidate segments in the minimum hotspot candidate region.

[0022] First probability distribution module: For each test segment in the hotspot candidate region, the disturbance intensity of all path segments in the hotspot candidate region is normalized to obtain the proportion of the disturbance intensity of each path segment in the hotspot candidate region, which is used as the probability distribution of the disturbance intensity of the hotspot candidate region.

[0023] The second probability distribution module: The disturbance intensity of the path segment within each test segment is also normalized to form a probability distribution of the disturbance intensity within the test segment;

[0024] Information gain module: Calculates the Jensen-Shannon divergence between the probability distribution of disturbance intensity in the test segment and the probability distribution of disturbance intensity in the hotspot candidate region, and uses it as the information gain for each test segment;

[0025] Preliminary test segment module: The preset cost scores of all candidate segments included in each test segment are added together to obtain the test cost of the test segment. The value corresponding to the information gain of each test segment is divided by the value corresponding to the test cost to obtain the cost-effectiveness score of each test segment. The test segment with the highest cost-effectiveness score is recorded as the optimal test segment and is used as the preliminary test segment.

[0026] Optionally, the determination module includes:

[0027] The extension index module: The welding feedback results include the arc stability fluctuation index and the heat accumulation and heat fluctuation anomaly index. The arc stability fluctuation index and the heat accumulation and heat fluctuation anomaly index are normalized, and the normalized arc stability fluctuation index and the heat accumulation and heat fluctuation anomaly index are weighted and summed to obtain the extension index.

[0028] Final judgment module: Determines whether to extend the length of the initial test section based on the extension index.

[0029] Optionally, the extension index module includes:

[0030] Arc current time series module: acquires the arc current data at each time point during the uniform welding process in the initial test section, and obtains the arc current time series;

[0031] Current variation sequence module: Calculates the first-order difference sequence of adjacent sampling points of the arc current time series to obtain the current variation sequence:

[0032] Fluctuation amplitude sequence module: Takes the absolute value of the current change sequence to obtain the fluctuation amplitude sequence, and calculates the mean of the fluctuation amplitude sequence;

[0033] The autocorrelation coefficient module calculates the autocorrelation coefficient based on the previous and next values ​​of the fluctuation amplitude sequence, as well as the mean of the fluctuation amplitude sequence. This coefficient measures the continuity of fluctuations. The formula is as follows: In the formula, The autocorrelation coefficient is... This represents the total number of sampling points in the arc current time series. This represents the mean of the fluctuation range sequence. The first value in the fluctuation amplitude sequence A number, The first value in the fluctuation amplitude sequence A number.

[0034] Optionally, the extension index module further includes:

[0035] Interval probability distribution module: Divides the fluctuation amplitude sequence into Divide the sampling points in each interval into equal-width intervals, count the number of fluctuations in each interval, and divide the number of sampling points in each interval by the total number of fluctuations in the fluctuation amplitude sequence to obtain the probability distribution of each interval.

[0036] Entropy module: Calculates entropy based on the probability distribution of each interval, used to measure the complexity of fluctuations. The formula for calculating entropy is: In the formula, The entropy value. Indicates the first The probability distribution of each interval;

[0037] Arc stability fluctuation index module: The arc stability fluctuation index is calculated based on the autocorrelation coefficient and entropy value. The calculation formula is as follows: In the formula, This is the arc stability fluctuation index.

[0038] Optionally, the extension index module further includes:

[0039] Thermal image sequence module: During the uniform welding process of the initial test section of the metal plate fold, infrared thermography is used to continuously acquire thermal images of the welding area, resulting in a thermal image sequence consisting of several frames, each frame of which is a thermal image containing a two-dimensional temperature distribution.

[0040] Temperature range sequence module: For each frame of thermal image, extract the difference between the highest and lowest temperatures in the corresponding frame image to obtain the temperature range of each frame, and combine the temperature ranges of all frames into a temperature range sequence.

[0041] Thermocentroid amplitude sequence module: For each frame of thermal image, the coordinates of the thermal centroid are calculated. The coordinates of the thermal centroid are the center position of the image after temperature weighting. The Euclidean distance between the thermal centroids of the current frame and the previous frame is calculated as the amplitude of the thermal centroid, and a sequence of thermal centroid amplitude is formed.

[0042] Inter-frame temperature perturbation value sequence module: For each frame of thermal image, extract the temperature value sequence along the central cross section of the image, calculate the absolute difference between the temperature value of each pixel on the cross section of each frame and the previous frame, and take the average value as the inter-frame temperature perturbation value, thus forming the inter-frame temperature perturbation value sequence.

[0043] Abnormal mutation frequency module: For all inter-frame temperature disturbance values ​​in the inter-frame temperature disturbance value sequence, the inter-frame temperature disturbance values ​​that exceed the preset temperature change threshold are recorded as mutation values, and the total number of mutation values ​​is divided by the total number of inter-frame temperature disturbance values ​​to obtain the abnormal mutation frequency.

[0044] Optionally, the extension index module further includes:

[0045] Normalization module: Calculates the mean of the temperature range sequence, the mean of the amplitude sequence of the thermocentric beat, and the mean of the inter-frame temperature perturbation value sequence. Then, it normalizes the mean of the temperature range sequence, the mean of the amplitude sequence of the thermocentric beat, the mean of the inter-frame temperature perturbation value sequence, and the frequency of abnormal mutations, so that their values ​​are uniformly mapped to a closed interval of zero to one.

[0046] The module for thermal accumulation and thermal fluctuation anomaly index calculates the average values ​​of the normalized temperature range sequence, the amplitude sequence of thermocentric beats, and the inter-frame temperature perturbation value sequence to obtain the thermal accumulation and thermal fluctuation anomaly index.

[0047] Optionally, the final determination module includes:

[0048] The first judgment module compares the extension index with the preset extension index threshold. If the extension index is not less than the preset extension index threshold, it means that the length of the preliminary test section needs to be extended to continue transportation and welding. Welding continues at a constant speed based on the extended preliminary test section. Based on the welding results, it is judged whether the remaining folds to be welded can continue to be welded at a constant speed.

[0049] The second judgment module: If the extension index is less than the preset extension index threshold, it means that it is not necessary to extend the length of the preliminary test section to continue transportation and welding. Based on the welding results of the preliminary test section, it is judged whether the remaining folds to be welded can continue to be welded at a constant speed.

[0050] The beneficial effects of this invention are:

[0051] This invention proposes a uniform-speed welding system for metal plate folds. The system involves scanning the metal plate to be welded to construct three-dimensional data of the fold, analyzing this data to extract preliminary selectable areas for welding tests as hotspot candidate areas. These candidate areas are then screened to select the optimal test section as the preliminary test section. The preliminary test section is then welded at a uniform speed, and the length of the preliminary test section is determined based on the welding feedback.

[0052] If the length of the initial test section needs to be extended, welding continues at a constant speed based on the extended initial test section. The welding results are then used to determine whether the remaining weld joints can continue to be welded at a constant speed. Through this method, the weld joints of the metal plate are scanned to obtain their three-dimensional data. An initial welding test section of the weld joint is selected, and the welding results are used to determine whether to extend the welding section. This achieves a balance between accurately judging the quality of constant-speed welding and reducing costs, ensuring the reliability of welding quality predictions without causing greater material and time losses. Attached Figure Description

[0053] The invention will now be further described with reference to the accompanying drawings.

[0054] Figure 1 This is a frame diagram of a uniform speed welding system for metal plate folds. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] This invention provides a uniform-speed welding system for metal plate folded seams. See also... Figure 1 , Figure 1 A frame diagram of a uniform-speed welding system for metal plate folds provided in an embodiment of the present invention. The system includes the following steps:

[0057] Preliminary selection module: Scans the metal plate to be welded, constructs three-dimensional data of the weld joint, analyzes the three-dimensional data of the weld joint, and extracts the preliminary selection area of ​​the weld joint that can be used for welding tests as a hot spot candidate area;

[0058] Preliminary test module: Screens hotspot candidate areas, selects the optimal test segment from the hotspot candidate areas, and uses it as the preliminary test segment;

[0059] Judgment module: Perform uniform welding on the preliminary test section and determine whether to extend the length of the preliminary test section based on the welding feedback effect;

[0060] Uniform speed welding module: If it is necessary to extend the length of the preliminary test section, then continue uniform speed welding according to the extended preliminary test section, and judge whether the remaining folds to be welded can continue to be welded using uniform speed welding based on the feedback of the welding results.

[0061] Based on the embodiment of the present invention, a uniform speed welding system for metal plate folds is provided. By scanning the weld seam of the metal plate in the above manner, its three-dimensional data is obtained, and a preliminary welding test section of the fold is selected. Based on the actual welding results, it is determined whether to extend the welding section. This achieves a balance between accurately judging the quality of uniform speed welding and reducing costs. It ensures the reliability of welding quality prediction without causing greater material and time losses.

[0062] In one embodiment, the preliminary selection module scans the metal plate to be welded, constructs three-dimensional data of the weld joint, analyzes the three-dimensional data of the weld joint, and extracts the preliminary selection area of ​​the weld joint that can be used for welding tests as a hot spot candidate area.

[0063] It should be noted that scanning the metal plate to be welded to construct 3D data of the fold is typically achieved using industrial-grade 3D vision systems or laser measurement devices. Specifically, structured light scanners, laser contour sensors, or multi-axis laser scanning equipment can be used to acquire high-density point cloud data of the fold area. During the scanning process, these devices capture the 3D geometric topography information of the metal surface, including microscopic features such as the curvature of the fold, gap width, weld height difference, and fold angle. The raw point cloud data after scanning is preprocessed (e.g., denoising, filtering, registration, and reconstruction) to generate a continuous 3D model or curve representation of the fold. The final 3D fold data will serve as the basic input for subsequent welding test section extraction and risk analysis, providing geometric basis for the feasibility assessment of the entire uniform welding process.

[0064] In one implementation, the three-dimensional data of the weld fold to be welded is analyzed to extract a preliminary selectable area for welding tests, which is then used as a hot spot candidate area.

[0065] Specifically, the initial selection of modules includes:

[0066] Path segmentation module: Divides the 3D path of the weld joint to be welded into multiple continuous path segments at equal intervals; assuming the total length of the weld joint path is... Select the division step size as Division Each path segment; This corresponds to a small segment of three-dimensional space within the fold path;

[0067] Feature vector module: for each path segment Extract its descriptive geometric features and construct feature vectors. eigenvectors Including path segments Local curvature mean Mean value of normal variation angle Standard deviation of fold width , ;

[0068] Feature change rate module: calculates each path segment eigenvectors and the previous path segment eigenvectors The first difference, as a path segment characteristic rate of change ;

[0069] Variable acceleration module: based on path segment The characteristic rate of change (first-order difference) is used to calculate the second-order difference, which is then used as the path segment. Change in acceleration ;

[0070] Disturbance intensity scoring module: This module scores each path segment... characteristic rate of change and change acceleration The corresponding path segment is obtained by weighted summation. Disturbance intensity score , and when When it is 1, the corresponding and When it is 0, When it is 2, the corresponding =0;

[0071] Hotspot Candidate Region Module: Based on the disturbance intensity score of each path segment, the preliminary selection area for welding tests of the fold joint is extracted as the hotspot candidate region.

[0072] It should be noted that the 3D path is divided into multiple continuous path segments at equal intervals. The step size for this division is related to the overall length of the folds in the metal plate used to manufacture the lamp post, and is not specifically limited. For example, when the total fold length is long and the geometric changes are gentle, a larger step size can be used to improve processing efficiency; while in areas with short folds or drastic local changes, a smaller step size can be used to more accurately capture path disturbance features. Common step sizes such as 1cm and 2cm are all within an acceptable range, and can be selected according to production needs and equipment precision in practical applications. In addition, the mean local curvature, mean normal change angle, and standard deviation of fold width of the path segments can be obtained through geometric analysis of the fold point cloud data after 3D scanning. Specifically, the mean local curvature can be calculated based on the average value of the curvature of each point within the path segment. Common methods include fitting local curves or using discrete differential geometry algorithms. The mean normal variation angle is calculated by averaging the angle between the normal vectors of adjacent points. The standard deviation of the fold width can be calculated based on the distance between the left and right edge points on each cross section to obtain the width sequence, and then the standard deviation is calculated to reflect the stability and uniformity of the width.

[0073] It should be noted that dividing the weld seam into multiple continuous path segments aims to achieve localized and refined analysis of the complex geometric features of the seam. Since seams often exhibit deformation, abrupt curvature changes, or uneven width during actual welding, overall assessment may obscure local difficulties. Segmentation allows for more accurate identification of areas with concentrated welding challenges. Furthermore, by calculating the characteristic change rate (i.e., first-order difference) and acceleration (second-order difference) of each path segment, the geometric perturbation trend of that segment relative to adjacent segments can be quantified. The perturbation intensity score reflects whether a segment exhibits significant geometric changes. For example, a sudden increase in curvature or a drastic change in normal direction in a segment may indicate difficulty in adjusting the welding torch posture or reduced welding stability. This method effectively locates "hotspot candidate segments," providing more targeted decision support for subsequent test segment selection and welding strategy formulation. For instance, in the seam of a metal lamp post, if a segment suddenly becomes more curved or wider, the perturbation score can identify that segment, rather than it being obscured by average features.

[0074] In one embodiment, the preliminary selection area for welding tests of the fold is extracted based on the disturbance intensity score of each path segment, as a hot spot candidate area;

[0075] Hotspot candidate areas include:

[0076] Candidate segment module: Compare the disturbance intensity score of each path segment with a preset threshold. If the disturbance intensity score is not less than the preset threshold, the corresponding path segment is recorded as a candidate segment.

[0077] Filtering module: The fold area with a number of consecutive candidate segments not less than a preset threshold is taken as a hot spot candidate area, and several hot spot candidate areas are obtained.

[0078] It should be noted that the reason for comparing the disturbance intensity score of the path segment with a preset threshold to screen out hotspot candidate regions composed of continuous candidate segments is fundamentally to achieve precise positioning and structural extraction of welding difficulty areas. While folded seam paths are generally smooth macroscopically, there may be areas with significant geometrical disturbances at the microscopic level, such as sharp increases in curvature, discontinuities in the normal direction, and abrupt changes in width. These changes often directly affect the adjustment frequency of the welding torch posture, the uniformity of welding energy distribution, and the stability of welding quality. By setting a disturbance intensity threshold, stable path segments with minimal impact on welding can be effectively eliminated, retaining only "suspicious" segments with obvious changes. Further requiring these candidate segments to exhibit spatial continuity and reach a certain minimum length (preset quantity threshold) is to avoid misjudging accidental geometrical disturbances or noise as key welding difficulties, thereby ensuring that the extracted hotspot regions have sufficient representativeness and research value in actual welding operations. In other words, this screening mechanism can "automatically identify" typical areas that may pose a high challenge to the welding process within the overall folded seam path, improving analysis efficiency and providing a scientific basis for subsequent experimental segment selection.

[0079] In one implementation, the preliminary test module filters hotspot candidate regions and selects the optimal test segment from the hotspot candidate regions as the preliminary test segment;

[0080] The preliminary test module includes:

[0081] Test Segment Module: For each hotspot candidate region, n consecutive candidate segments are generated using a sliding window method as test segments. The total number of candidate segments in the test segment is no greater than the total number of candidate segments in the minimum hotspot candidate region.

[0082] The first probability distribution module: For each test segment's hotspot candidate region, the disturbance intensity of all path segments within the hotspot candidate region is normalized to obtain the proportion of disturbance intensity for each path segment in the hotspot candidate region, which serves as the probability distribution of the disturbance intensity in the hotspot candidate region; the calculation formula is: , , The first in the hotspot candidate region The probability distribution of the disturbance intensity of each path segment; A set of path segments within a hotspot candidate region;

[0083] The second probability distribution module: The disturbance intensity of each path segment within each test segment is also normalized to form a probability distribution of the disturbance intensity within the test segment; the calculation formula is: , , For the first test section The probability distribution of the disturbance intensity of each path segment; The set of path segments within the test section;

[0084] Information Gain Module: Calculates the Jensen-Shannon divergence between the probability distribution of disturbance intensity in the test segment and the probability distribution of disturbance intensity in the hotspot candidate region, as the information gain for each test segment; the calculation formula is: In the formula, For information gain, , representing the probability distribution of the disturbance intensity in the hotspot candidate region. , representing the probability distribution of disturbance intensity within the test section. ,express and average distribution This represents the Kullback-Leibler divergence of Q relative to M. This represents the Kullback-Leibler divergence of P relative to M;

[0085] Preliminary test segment module: The preset cost scores of all candidate segments included in each test segment are added together to obtain the test cost of the test segment. The value corresponding to the information gain of each test segment is divided by the value corresponding to the test cost to obtain the cost-effectiveness score of each test segment. The test segment with the highest cost-effectiveness score is recorded as the optimal test segment and is used as the preliminary test segment.

[0086] It should be noted that using n consecutive candidate segments as test segments, and then selecting test segments after extracting hotspot candidate regions, this hierarchical processing method has significant engineering implications and practical advantages. First, by generating a fixed number of consecutive candidate segments (i.e., test segments) within the hotspot candidate region using a sliding window, it is possible to comprehensively cover potentially representative local feature variation segments within that region, thereby enhancing the diversity and robustness of test segment selection and avoiding the omission of potential critical areas due to manual designation. Simultaneously, selecting a continuous path segment as a test segment better aligns with the requirements of actual welding processes for continuity, path integrity, and process stability. For example, the welding process requires continuous operation, posture maintenance, and uniform heating, making it more physically feasible and engineering valuable than discrete path segment selection. Second, extracting hotspot candidate regions first and then selecting test segments from them aims to focus on critical areas with significant welding feature changes from the global fold path, minimizing resource waste on redundant or unrepresentative areas. This not only improves the efficiency of test segment selection but also significantly enhances the "information content" of the selected test segments, i.e., their ability to characterize welding behavior and difficulty features.

[0087] It should be noted that the preset cost score is usually closely related to factors such as the physical location, processing complexity, welding accessibility, and assembly dependencies of the candidate segment corresponding to the test segment (i.e., the metal plate folding path segment) in the entire lamp post manufacturing process. For example, when the path segment is located at the upper part of the lamp post or near the end, the limited operating space, difficulty in adjusting the welding posture, and high difficulty in robot operation may lead to higher welding accuracy requirements and longer debugging time, resulting in a higher cost score. Similarly, if the path segment is located in the transition area or area of ​​drastic curvature change in the lamp post structure, its folding shape is complex, making it more prone to instability during welding, requiring higher control strategies and process requirements, thus increasing costs accordingly. Furthermore, if the path segment is adjacent to structurally sensitive areas such as reinforcing ribs, connecting flanges, or process holes, the risk of welding failure is high, and the cost of rework is significant, also leading to increased test costs. At the same time, if the segment is at the beginning or end of a critical welding process, it plays a crucial role in the entire welding process, potentially directly affecting the accuracy of subsequent processes and assembly progress; therefore, its test cost is higher and needs to be reflected in the cost score. The specific preset cost score is set according to the actual situation and is not limited.

[0088] It should be noted that the information gain of a test segment refers to the extent to which the distribution of disturbance intensity in that test segment can represent the characteristic differences and complexity of the overall disturbance intensity distribution in the candidate hotspot region. In simple terms, it measures how much "useful information" or "representativeness of structural features" the test segment contains. If the information gain of a test segment is high, it means that the distribution of its disturbance intensity is numerically close to or has high uncertainty in the entire hotspot region, thus covering a richer variety of welding feature types in that region, and the test results are more representative. On the other hand, the test cost measures the resource input required for the welding implementation of that segment, such as time, manpower, energy consumption, and attitude adjustment complexity. Therefore, the larger the ratio of information gain to test cost, the more feature information is obtained per unit cost, i.e., the higher the cost-effectiveness. If such a segment is selected as the initial test segment for uniform welding verification, the performance of typical welding conditions in the target hotspot region can be observed and evaluated more fully without increasing costs, thereby making more accurate predictions of the trends and impacts of complex welding behavior in the overall fold path.

[0089] It should be noted that a higher information gain to test cost ratio indicates more structural disturbance feature information can be obtained per unit test cost. Therefore, the higher the "information efficiency" of this test segment in the overall welding experiment plan, the more likely it is to become the optimal initial test segment. During the welding process, the disturbance intensity reflects the complexity of the path segment's geometric changes (such as abrupt curvature changes, rapid normal rotation, and crease width fluctuations). This complexity is a key factor leading to increased welding difficulty, unstable molten pool, and fluctuating welding quality. If a test segment has a high information gain, it means that its feature distribution can well cover the key disturbance types and trends in the hotspot area. In other words, the test data of welding this segment can highly represent the characteristics of other path segments in the area and has strong generalization ability. Conversely, a lower test cost means that more comprehensive disturbance response information can be obtained with less investment of resources (such as man-hours, materials, and adjustment costs). Therefore, test segments with a high information gain to cost ratio not only have stronger representativeness but also achieve more efficient and accurate parameter evaluation and prediction model training with limited resources. For example, in a certain hotspot area, if test segment A covers multiple feature points ranging from gradual change to rapid disturbance, its information gain is high. At the same time, since this segment is located in an area with relatively stable welding posture and good operability, the test cost is low. Therefore, this segment has a high cost-performance ratio and can be used as a preliminary test segment for welding. This not only captures diverse welding response features but also reduces unnecessary test expenses, thereby improving the accuracy and economy of overall folded weld performance prediction.

[0090] In one embodiment, the judgment module performs uniform welding on the preliminary test section and determines whether to extend the length of the preliminary test section based on the welding feedback effect.

[0091] The judgment module includes:

[0092] The extension index module: The welding feedback results include the arc stability fluctuation index and the heat accumulation and heat fluctuation anomaly index. The arc stability fluctuation index and the heat accumulation and heat fluctuation anomaly index are normalized, and the normalized arc stability fluctuation index and the heat accumulation and heat fluctuation anomaly index are weighted and summed to obtain the extension index.

[0093] Final judgment module: Determines whether to extend the length of the initial test section based on the extension index; the formula for calculating the extension index is: In the formula, To extend the index, and These are the normalized arc stability fluctuation index and the heat accumulation and heat fluctuation anomaly index, respectively. These represent the preset weighting coefficients of the arc stability fluctuation index and the heat accumulation and heat fluctuation anomaly index after multi-normalization processing, respectively. All are greater than 0; normalization methods for removing dimensions include Min-Max normalization, Z-Score standardization, etc., which will not be elaborated here; Settings should be set according to the actual situation, generally They are equal and their sum is 1, for example, It can be 0.5 or 0.5.

[0094] In one embodiment, the extended exponent module includes:

[0095] Arc current time series module: acquires the arc current data at each time point during the uniform welding process in the initial test section to obtain the arc current time series; the arc current time series is a discrete data sequence acquired by a current acquisition device with a sampling frequency of 10kHz;

[0096] Current Variation Sequence Module: Calculates the first-order difference sequence of adjacent sampling points of the arc current time series to obtain the current variation sequence.

[0097] Fluctuation Amplitude Sequence Module: Takes the absolute value of the current change sequence to obtain the fluctuation amplitude sequence, and calculates the mean of the fluctuation amplitude sequence;

[0098] The autocorrelation coefficient module calculates the autocorrelation coefficient based on the previous and next values ​​of the fluctuation amplitude sequence, as well as the mean of the fluctuation amplitude sequence. This coefficient measures the continuity of fluctuations. The formula is as follows: In the formula, The autocorrelation coefficient is... This represents the total number of sampling points in the arc current time series. This represents the mean of the fluctuation range sequence. The first value in the fluctuation amplitude sequence A number, The first value in the fluctuation amplitude sequence A number;

[0099] Interval probability distribution module: Divides the fluctuation amplitude sequence into Divide the sampling points in each interval into equal-width intervals, count the number of fluctuations in each interval, and divide the number of sampling points in each interval by the total number of fluctuations in the fluctuation amplitude sequence to obtain the probability distribution of each interval.

[0100] Entropy module: Calculates entropy based on the probability distribution of each interval, used to measure the complexity of fluctuations. The formula for calculating entropy is: In the formula, The entropy value. Indicates the first The probability distribution of each interval;

[0101] Arc stability fluctuation index module: The arc stability fluctuation index is calculated based on the autocorrelation coefficient and entropy value. The calculation formula is as follows: In the formula, This is the arc stability fluctuation index.

[0102] It should be noted that the data involved in the above calculation process, namely the arc current time series data, was acquired in real time by using a current sensor installed on the welding equipment at a high frequency sampling rate of 10kHz while uniform welding was being performed in the initial test section. The acquisition device recorded the current value of the arc at each time point during the welding process and stored it in the form of a discrete sequence, providing basic data support for subsequent fluctuation analysis, autocorrelation coefficient calculation and entropy value calculation.

[0103] It should be noted that the arc stability fluctuation index is a comprehensive indicator used to measure the characteristics of arc current fluctuations during uniform welding of metal plate folds. It reflects the intensity, continuity, and complexity of arc current fluctuations. This index is obtained by analyzing and calculating the time-series data of arc current collected during the welding process. A higher value indicates poorer arc stability during welding, meaning larger arc current fluctuations, more continuous fluctuations, or complex dynamic behavior. The reason why a higher index means the selected initial test section is too short and needs to be extended for continued welding is that in the early stages of welding, the welding equipment system may not have reached thermal equilibrium or the electrical control system may not have completed dynamic adjustment. This results in the current fluctuations in the initial welding stage not reflecting the true characteristics of the material itself or path disturbances, but rather the instability of the system. Therefore, the feedback data at this stage is not representative. The arc current fluctuation is too large to accurately predict subsequent welding behavior. For example, if the test section is 3cm long and it is in a stage of severe path disturbance or when the welding system has just started, the collected arc current fluctuation may show high amplitude continuous changes, resulting in a large calculated fluctuation index. However, if the welding length is extended to 6cm, the arc fluctuation may tend to stabilize, the entropy value will decrease, the autocorrelation will weaken, and the fluctuation index will eventually decrease, indicating that the system has entered a stable welding state. Therefore, only when the test section length is long enough to cover a representative path area and can eliminate misleading data caused by initial disturbances can stable and reliable welding feedback be obtained. The larger the arc stability fluctuation index, the more likely it is that the current feedback data is still in the dynamic transition zone and has not yet "converged" to a stable state that reflects the real welding behavior. Therefore, the test section should be appropriately extended to obtain more robust data to support the adjustment and judgment of welding process strategies.

[0104] It should be noted that the reason for using the above method to calculate the arc stability fluctuation index, instead of using conventional statistical methods such as root mean square deviation, standard deviation, or simple fluctuation range, is that the above methods cannot simultaneously and comprehensively capture the changing characteristics of arc current fluctuations in the three dimensions of intensity, continuity, and complexity, making it difficult to accurately reflect the dynamic instability behavior in the welding process. By extracting the trend of change through difference sequences, calculating the fluctuation amplitude by taking the absolute value, combining it with the autocorrelation coefficient to measure the temporal continuity of the fluctuation, and further introducing the entropy value based on probability distribution to measure the complexity of the fluctuation, this entire analytical chain can not only accurately identify the violent fluctuations of arc current, but also determine whether the fluctuation has persistence and randomness, thus more sensitively reflecting potential anomalies or system response states in the welding process. For example, although the current change in a certain section of the welding path is large, it is highly random and may only be a short-term disturbance, while another section, although the fluctuation amplitude is small, has a long-term high autocorrelation and a high entropy value, which may indicate structural disturbances or lag in the welding control system. The arc stability fluctuation index constructed by the above method makes the welding state between different test sections more comparable, and ultimately provides a more accurate basis for judging whether the initial test section should be extended.

[0105] In one embodiment, the extended exponent module further includes:

[0106] Thermal image sequence module: During the uniform welding process of the initial test section of the metal plate fold, infrared thermography is used to continuously acquire thermal images of the welding area, resulting in a thermal image sequence consisting of several frames. Each frame is a thermal image containing a two-dimensional temperature distribution;

[0107] Temperature Range Sequence Module: For each frame of a thermal image, the module extracts the difference between the highest and lowest temperatures in that frame to obtain the temperature range for each frame, and then combines the temperature ranges of all frames into a temperature range sequence. ; Width characteristics used to reflect the thermal energy distribution of each frame.

[0108] Thermal centroid amplitude sequence module: calculates the coordinates of the thermal centroid for each frame of thermal image. The thermal centroid coordinates are the temperature-weighted image center position. The Euclidean distance between the thermal centroids of the current frame and the previous frame is calculated as the amplitude of the thermal centroid fluctuation, forming a sequence of thermal centroid fluctuation amplitudes. Used to reflect the positional fluctuations of the heat source in the image;

[0109] Inter-frame temperature perturbation value sequence module: For each frame of thermal image, extract the temperature value sequence along the central cross section of the image, calculate the absolute difference between the temperature value of each pixel on the cross section of each frame and the previous frame, and take the average value as the inter-frame temperature perturbation value to form an inter-frame temperature perturbation value sequence; used to measure the intensity of welding heat flow.

[0110] Abnormal mutation frequency module: For all inter-frame temperature perturbation values ​​in the inter-frame temperature perturbation value sequence, the inter-frame temperature perturbation values ​​that exceed the preset temperature change threshold are recorded as mutation values, and the total number of mutation values ​​is divided by the total number of inter-frame temperature perturbation values ​​to obtain the abnormal mutation frequency;

[0111] Normalization module: Calculates the mean of the temperature range sequence, the mean of the amplitude sequence of the thermocentric center of mass, and the mean of the inter-frame temperature perturbation value sequence. Then, normalizes the mean of the temperature range sequence, the mean of the amplitude sequence of the thermocentric center of mass, the mean of the inter-frame temperature perturbation value sequence, and the frequency of anomalous changes, so that their values ​​are uniformly mapped to a closed interval between zero and one. The resulting four normalized indices are used to characterize the thermal distribution amplitude, the stability of the heat source location, the degree of thermal flow perturbation, and anomalous sudden thermal behavior, respectively.

[0112] The thermal accumulation and thermal fluctuation anomaly index module calculates the average values ​​of the normalized temperature range sequence, the thermal center of mass amplitude sequence, and the inter-frame temperature perturbation value sequence to obtain the thermal accumulation and thermal fluctuation anomaly index. This index is used to comprehensively characterize the overall degree of thermal behavior anomaly in the initial test section during uniform welding. The closer the index value is to one, the more significant the thermal accumulation, heat flow fluctuation, and local instability in the test section.

[0113] It should be noted that the data involved in the calculation of the above-mentioned heat accumulation and thermal fluctuation anomaly index mainly comes from real-time thermal monitoring during the welding process of the preliminary test section of the metal plate fold joint. Specifically, an infrared thermal imager is used to continuously collect non-contact temperature data of the welding area, resulting in a high-frequency two-dimensional thermal image sequence. These thermal images record the temperature distribution of the welding area in pixels. By processing thermal image sequences, multi-dimensional thermal characteristic data such as maximum temperature, minimum temperature, thermal centroid coordinates, and temperature distribution along specific intercepts are extracted. Combined with time series analysis, this comprehensively reflects the accumulation, migration, and fluctuation of heat energy during welding, thereby achieving an accurate quantitative evaluation of welding thermal behavior anomalies. Furthermore, the normalization process involved in the above calculations is achieved by mapping the original values ​​of each indicator to a closed interval between 0 and 1. The specific steps are as follows: First, determine the maximum and minimum values ​​of each indicator, usually based on historical data or theoretical ranges; then, apply a linear normalization formula to the original values ​​of each indicator, i.e., normalized value = (current value - minimum value) ÷ (maximum value - minimum value), so that the values ​​of all indicators are standardized to the same scale range, avoiding comparison and comprehensive evaluation biases caused by differences in dimensions or numerical values. This normalization method is simple and effective, ensuring that different thermal behavior indicators have a balanced contribution weight when calculating the comprehensive heat accumulation and thermal fluctuation anomaly index, thereby improving the accuracy and consistency of the overall assessment.

[0114] It should be noted that the heat accumulation and thermal fluctuation anomaly index is a comprehensive numerical indicator used to measure the stability of heat energy distribution and its abnormal fluctuations within the welding area during uniform-speed welding in the initial test section of a metal plate fold joint. Specifically, it reflects the degree of heat energy accumulation in the welding area, the stability of the heat source location, the amplitude of heat flow fluctuations during welding, and any abnormal thermal behaviors (such as sudden and drastic temperature changes). A higher index value indicates greater non-uniformity and severe thermal fluctuations in the welding area, suggesting complex and unstable thermal behavior within the test section. This may be due to the test section being too short, failing to cover sufficient welding process information, resulting in heat accumulation and thermal fluctuations not reaching a stable state. In other words, a larger index indicates that the thermal behavior has not yet stabilized, the welding thermal field changes drastically, and it is difficult to accurately predict the subsequent overall welding quality. In this case, it is necessary to appropriately extend the length of the test section, using uniform-speed welding over a longer time and spatial range to stabilize heat accumulation and fluctuations, thereby obtaining more accurate and comprehensive feedback on welding thermal behavior. For example, suppose that in a short test section, the thermal centroid fluctuates frequently and the temperature range is large, resulting in a heat accumulation and fluctuation anomaly index as high as 0.85. This indicates that the heat energy distribution is extremely uneven and the heat flow changes drastically, meaning the welding process has not yet reached a stable state. The test section needs to be extended to allow the heat energy to gradually equalize. Conversely, if the test section is long, the heat accumulation and heat fluctuation anomaly index is below 0.2, indicating a relatively stable thermal field, a predictable welding process, and relatively stable quality, thus eliminating the need to extend the test section further. Therefore, this index can effectively determine whether the initial test section sufficiently reflects the thermal dynamic characteristics of the welding process, thereby guiding whether the test section length needs to be extended to improve the accuracy of welding quality prediction.

[0115] It should be noted that the method described above for calculating the heat accumulation and thermal fluctuation anomaly index, rather than traditional single temperature monitoring or simple statistical methods, was chosen primarily because this method comprehensively considers the multi-dimensional dynamic characteristics of the thermal field. This includes the width of the heat energy distribution reflected by temperature ranges, the stability of the heat source location revealed by spatial fluctuations of the thermal centroid, the severity of heat flux changes measured by inter-frame temperature disturbances, and sudden thermal events reflected by the frequency of anomalous mutations. By normalizing these heterogeneous indicators to unify their dimensions and then comprehensively evaluating them, the method can more comprehensively and accurately reflect the thermal dynamic changes and anomalies during the welding process. This multi-level, multi-angle analysis overcomes the shortcomings of single indicators in capturing the complexity and spatiotemporal non-uniformity of the thermal field. This allows the heat accumulation and thermal fluctuation anomaly index to not only be sensitive to the overall heat accumulation trend but also effectively identify local thermal disturbances and anomalous mutations, thereby improving the ability to judge the stability of welding thermal behavior. Furthermore, based on continuous infrared thermal imaging data, this method can dynamically track the evolution of the thermal field in real time, avoiding the shortcomings of traditional discrete temperature measurements that are easily limited by sampling points and environmental interference. In summary, this calculation method, while ensuring numerical normalization and the absence of external unknown factors, not only improves the universality and stability of the indicators, but also enhances the scientific rigor and practicality of the thermal behavior analysis of the welding process, providing strong data support and theoretical basis for accurately judging the thermal dynamic characteristics of the preliminary test section.

[0116] In one embodiment, the final determination module includes:

[0117] The first judgment module compares the extension index with the preset extension index threshold. If the extension index is not less than the preset extension index threshold, it means that the length of the preliminary test section needs to be extended to continue transportation and welding. Welding continues at a constant speed based on the extended preliminary test section. Based on the welding results, it is judged whether the remaining folds to be welded can continue to be welded at a constant speed.

[0118] The second judgment module: If the extension index is less than the preset extension index threshold, it means that it is not necessary to extend the length of the preliminary test section to continue transportation and welding. Based on the welding results of the preliminary test section, it is judged whether the remaining folds to be welded can continue to be welded at a constant speed.

[0119] It should be noted that determining whether to extend the length of the initial test section based on the extension index is achieved by comparing the calculated extension index with a preset extension index threshold. Specifically, when the extension index equals or exceeds this threshold, it means that key indicators fed back during the welding process (such as the arc stability fluctuation index and the heat accumulation and heat fluctuation anomaly index) show that the current length of the initial test section is insufficient to fully cover the complex thermal dynamics and arc behavior fluctuations during the welding process, and there may be welding risks or local instabilities that have not been fully assessed. Therefore, the system will determine that the initial test section needs to be appropriately extended, and uniform welding will continue in the newly added test section area to obtain more comprehensive and stable welding feedback data, ensuring the safety and quality of subsequent welding paths. For example, suppose the initial test section only covers a relatively gentle section of the fold, and a high extension index may indicate that there are obvious arc fluctuations or heat accumulation anomalies during the welding process, suggesting that there are more complex thermal stresses or structural disturbances in some adjacent areas of the fold, thus requiring the test section to be extended for further inspection. When the elongation index is below the threshold, it indicates that the welding feedback data collected in the initial test section is sufficient to represent the current welding conditions, and the welding process is relatively stable. There is no need to extend the test section further; the feedback results from this test section can be used to determine whether the remaining folds to be welded are suitable for continuing with the uniform-speed welding strategy, thus effectively saving test time and resources. In summary, this method based on the elongation index can dynamically adapt to the complexity of the welding process, avoiding resource waste caused by excessively extended test sections and preventing welding quality risks caused by excessively short test sections, thereby improving the intelligence and safety of the entire fold welding process.

[0120] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A uniform-speed welding system for metal plate folds, characterized in that, The system includes: Preliminary selection module: Scans the metal plate to be welded, constructs three-dimensional data of the weld joint, analyzes the three-dimensional data of the weld joint, and extracts the preliminary selection area of ​​the weld joint that can be used for welding tests as a hot spot candidate area; Preliminary test module: Screens hotspot candidate regions and selects the optimal test segment from the hotspot candidate regions as the preliminary test segment; Judgment module: Perform uniform welding on the preliminary test section and determine whether to extend the length of the preliminary test section based on the welding feedback effect; Uniform speed welding module: If it is necessary to extend the length of the preliminary test section, continue uniform speed welding according to the extended preliminary test section, and judge whether the remaining folds to be welded can continue to be welded using uniform speed welding based on the feedback of the welding results. The preliminary selection module includes: Path segmentation module: Divides the three-dimensional path of the weld seam into multiple continuous path segments at equal intervals; Feature vector module: For each path segment, extract its descriptive geometric features and construct a feature vector. The feature vector includes the mean local curvature, the mean normal change angle, and the standard deviation of the fold width of the path segment. Feature change rate module: Calculates the first-order difference between the feature vector of each path segment and the feature vector of the previous path segment, as the feature change rate of the path segment; The variable acceleration module calculates the second-order difference based on the characteristic rate of change of the path segment, which is used as the variable acceleration of the path segment. Disturbance intensity scoring module: The characteristic change rate and change acceleration of each path segment are weighted and summed to obtain the disturbance intensity score of the corresponding path segment; Hotspot Candidate Region Module: Based on the disturbance intensity score of each path segment, the preliminary selection area for welding tests of the fold joint is extracted as the hotspot candidate region.

2. The uniform speed welding system for metal plate folds according to claim 1, characterized in that, The hotspot candidate regions include: Candidate segment module: Compare the disturbance intensity score of each path segment with a preset threshold. If the disturbance intensity score is not less than the preset threshold, the corresponding path segment is recorded as a candidate segment. Filtering module: The fold area with a number of consecutive candidate segments not less than a preset threshold is taken as a hot spot candidate area, and several hot spot candidate areas are obtained.

3. The uniform speed welding system for metal plate folds according to claim 1, characterized in that, The preliminary test module includes: Test Segment Module: For each hotspot candidate region, n consecutive candidate segments are generated as test segments using a sliding window method. The total number of candidate segments contained in the test segment is no greater than the total number of candidate segments in the minimum hotspot candidate region. First probability distribution module: For each test segment in the hotspot candidate region, the disturbance intensity of all path segments in the hotspot candidate region is normalized to obtain the proportion of the disturbance intensity of each path segment in the hotspot candidate region, which is used as the probability distribution of the disturbance intensity of the hotspot candidate region. The second probability distribution module: The disturbance intensity of the path segment within each test segment is also normalized to form a probability distribution of the disturbance intensity within the test segment; Information gain module: Calculates the Jensen-Shannon divergence between the probability distribution of disturbance intensity in the test segment and the probability distribution of disturbance intensity in the hotspot candidate region, and uses it as the information gain for each test segment; Preliminary test segment module: The preset cost scores of all candidate segments included in each test segment are added together to obtain the test cost of the test segment. The value corresponding to the information gain of each test segment is divided by the value corresponding to the test cost to obtain the cost-effectiveness score of each test segment. The test segment with the highest cost-effectiveness score is recorded as the optimal test segment and is used as the preliminary test segment.

4. The uniform speed welding system for metal plate folds according to claim 1, characterized in that... The judgment module includes: The extension index module: The welding feedback results include the arc stability fluctuation index and the heat accumulation and heat fluctuation anomaly index. The arc stability fluctuation index and the heat accumulation and heat fluctuation anomaly index are normalized, and the normalized arc stability fluctuation index and the heat accumulation and heat fluctuation anomaly index are weighted and summed to obtain the extension index. Final judgment module: Determines whether to extend the length of the initial test section based on the extension index.

5. A uniform speed welding system for metal plate folds according to claim 4, characterized in that, The extension index module includes: Arc current time series module: acquires the arc current data at each time point during the uniform welding process in the initial test section, and obtains the arc current time series; Current variation sequence module: Calculates the first-order difference sequence of adjacent sampling points of the arc current time series to obtain the current variation sequence: Fluctuation amplitude sequence module: Takes the absolute value of the current change sequence to obtain the fluctuation amplitude sequence, and calculates the mean of the fluctuation amplitude sequence; The autocorrelation coefficient module calculates the autocorrelation coefficient based on the previous and next values ​​of the fluctuation amplitude sequence, as well as the mean of the fluctuation amplitude sequence. This coefficient measures the continuity of fluctuations. The formula is as follows: In the formula, The autocorrelation coefficient is... This represents the total number of sampling points in the arc current time series. This represents the mean of the fluctuation range sequence. The first value in the fluctuation amplitude sequence A number, The first value in the fluctuation amplitude sequence A number.

6. A uniform speed welding system for metal plate folds according to claim 5, characterized in that, The extension index module also includes: Interval probability distribution module: Divides the fluctuation amplitude sequence into Divide the sampling points in each interval into equal-width intervals, count the number of fluctuations in each interval, and divide the number of sampling points in each interval by the total number of fluctuations in the fluctuation amplitude sequence to obtain the probability distribution of each interval. Entropy module: Calculates entropy based on the probability distribution of each interval, used to measure the complexity of fluctuations. The formula for calculating entropy is: In the formula, The entropy value. Indicates the first The probability distribution of each interval; Arc stability fluctuation index module: The arc stability fluctuation index is calculated based on the autocorrelation coefficient and entropy value. The calculation formula is as follows: In the formula, This is the arc stability fluctuation index.

7. A uniform speed welding system for metal plate folds according to claim 4, characterized in that, The extension index module also includes: Thermal image sequence module: During the uniform welding process of the initial test section of the metal plate fold, infrared thermography is used to continuously acquire thermal images of the welding area, resulting in a thermal image sequence consisting of several frames, each frame of which is a thermal image containing a two-dimensional temperature distribution. Temperature range sequence module: For each frame of thermal image, extract the difference between the highest and lowest temperatures in the corresponding frame image to obtain the temperature range of each frame, and combine the temperature ranges of all frames into a temperature range sequence. Thermocentroid amplitude sequence module: For each frame of thermal image, the coordinates of the thermal centroid are calculated. The coordinates of the thermal centroid are the center position of the image after temperature weighting. The Euclidean distance between the thermal centroids of the current frame and the previous frame is calculated as the amplitude of the thermal centroid, and a sequence of thermal centroid amplitude is formed. Inter-frame temperature perturbation value sequence module: For each frame of thermal image, extract the temperature value sequence along the central cross section of the image, calculate the absolute difference between the temperature value of each pixel on the cross section of each frame and the previous frame, and take the average value as the inter-frame temperature perturbation value, thus forming the inter-frame temperature perturbation value sequence. Abnormal mutation frequency module: For all inter-frame temperature disturbance values ​​in the inter-frame temperature disturbance value sequence, the inter-frame temperature disturbance values ​​that exceed the preset temperature change threshold are recorded as mutation values, and the total number of mutation values ​​is divided by the total number of inter-frame temperature disturbance values ​​to obtain the abnormal mutation frequency.

8. A uniform speed welding system for metal plate folds according to claim 7, characterized in that, The extension index module also includes: Normalization module: Calculates the mean of the temperature range sequence, the mean of the amplitude sequence of the thermocentric beat, and the mean of the inter-frame temperature perturbation value sequence. Then, it normalizes the mean of the temperature range sequence, the mean of the amplitude sequence of the thermocentric beat, the mean of the inter-frame temperature perturbation value sequence, and the frequency of abnormal mutations, so that their values ​​are uniformly mapped to a closed interval of zero to one. The module for thermal accumulation and thermal fluctuation anomaly index calculates the mean of the normalized temperature range sequence, the mean of the amplitude sequence of thermocentric beats, the mean of the inter-frame temperature perturbation value sequence, and the average value of the abnormal mutation frequency to obtain the thermal accumulation and thermal fluctuation anomaly index.

9. A uniform speed welding system for metal plate folds according to claim 4, characterized in that, The final judgment module includes: The first judgment module compares the extension index with the preset extension index threshold. If the extension index is not less than the preset extension index threshold, it means that the length of the preliminary test section needs to be extended to continue uniform welding. The module continues uniform welding based on the extended preliminary test section and judges whether the remaining folds to be welded can continue to be welded using uniform welding based on the welding results. The second judgment module: If the extension index is less than the preset extension index threshold, it means that it is not necessary to extend the length of the preliminary test section to continue uniform welding. Based on the welding results of the preliminary test section, it is judged whether the remaining folds to be welded can continue to be welded at a uniform speed.

Images