A welding temperature dynamic control method and system based on machine vision
Patent Information
- Application Number
- CN202610863674.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-06-15
- Publication Date
- 2026-08-18
AI Technical Summary
[0004]鉴于上述问题，本发明的目的是提供一种基于机器视觉的焊接温度动态控制方法及系统，以解决现有焊接温度控制方法难以实时获取熔池区域温度场的空间分布及动态演变信息，无法准确判别熔池的热平衡状态并据此自适应调整焊接参数，导致焊接质量对工况波动敏感、温度调节易出现振荡或过冲的问题
[0015]从上面的技术方案可知，本发明提供的一种基于机器视觉的焊接温度动态控制方法及系统，通过将熔池区域划分为多个同心环状子区域并构建温度场特征矩阵，实现了对熔池温度空间分布的全景感知；利用帧间温度场特征矩阵的逐元素差分，获得温度场变化梯度矩阵，能够实时追踪温度场的动态演变趋势；结合温度场变化梯度矩阵与温度变化速率判别热平衡状态，提高了对热输入过剩、不足或平衡状态的判断准确性与响应速度；在此基础上自适应调整焊接电流和焊接速度，克服了传统恒定参数法无法适应工况波动以及单点反馈法信息单一、响应滞后的缺陷，有效抑制了焊接温度的振荡与过冲，显著提升了焊接质量的稳定性与一致性。
Smart Images
Figure CN122597376A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent temperature control technology, and in particular to a method and system for dynamic control of welding temperature based on machine vision. Background Technology
[0002] Current welding temperature control methods mostly employ constant process parameters or feedback regulation based on single-point sensors. The constant parameter method cannot adapt to dynamic fluctuations in base material heat dissipation conditions and bevel gaps, easily leading to excessively high molten pool temperatures causing burn-through spatter or excessively low temperatures causing incomplete fusion defects. The single-point feedback method suffers from measurement lag, susceptibility to arc light and spatter interference, and can only obtain the temperature at a local point, making it difficult to comprehensively reflect the spatial temperature distribution of the molten pool area. Control decisions are based on a single factor, and the temperature regulation process is prone to oscillations or overshoot.
[0003] Existing technologies lack the ability to comprehensively utilize the spatial distribution and temporal evolution characteristics of the temperature field in dynamic sensing of the molten pool temperature. This prevents real-time determination of the molten pool's thermal equilibrium state and subsequent adaptive adjustment of welding parameters. Consequently, welding quality becomes highly sensitive to changes in operating conditions, making stable, high-quality continuous welding difficult. Therefore, there is an urgent need for a welding temperature control method capable of constructing the spatial distribution of the molten pool temperature field, tracking its dynamic trends, and adaptively adjusting parameters accordingly. Summary of the Invention
[0004] In view of the above problems, the purpose of this invention is to provide a welding temperature dynamic control method and system based on machine vision, so as to solve the problems that existing welding temperature control methods are unable to obtain the spatial distribution and dynamic evolution information of the temperature field in the molten pool area in real time, cannot accurately determine the thermal balance state of the molten pool and adaptively adjust the welding parameters accordingly, resulting in welding quality being sensitive to working condition fluctuations and temperature regulation being prone to oscillation or overshoot.
[0005] This invention provides a machine vision-based dynamic control method for welding temperature, comprising: Step 1: Divide the continuous frame images of the molten pool area during the welding process into multiple concentric ring-shaped sub-regions, and convert the average gray value of the pixels in each sub-region into temperature value according to the preset gray-level-temperature mapping relationship, thus forming the temperature field feature matrix of the molten pool area. Step 2: Perform element-wise difference between the temperature field feature matrix corresponding to the current frame image and the temperature field feature matrix corresponding to the previous frame image to obtain the temperature field change gradient matrix of the molten pool region. Step 3: Determine the thermal equilibrium state of the molten pool region based on the temperature field gradient matrix and the temperature change rate of the current frame image; Step 4: Based on the judgment of the thermal balance state, adaptively adjust the welding current and welding speed in the molten pool area.
[0006] Preferably, the process of dividing the continuous frame images of the molten pool region during welding into multiple concentric ring-shaped sub-regions is as follows: During the welding process, the molten pool contour of the molten pool region in consecutive frame images is extracted, and the geometric center of the molten pool contour is determined. Using the geometric center as the center, multiple concentric circle boundaries with the same radial width are set in a manner that expands layer by layer from the center to the edge of the molten pool outline. The image region within the molten pool contour is divided into multiple concentric ring-shaped sub-regions, with the annular region between adjacent concentric circle boundaries as a sub-region and the circular region where the center of the circle is located as the innermost sub-region.
[0007] Preferably, the process of converting the average grayscale value of pixels in each sub-region into a temperature value according to a preset grayscale-temperature mapping relationship to form a temperature field feature matrix of the molten pool region is as follows: The arithmetic mean of the gray values of all pixels in each concentric ring-shaped sub-region is used as the average gray value of each sub-region. Based on the preset grayscale-temperature mapping relationship, the temperature value corresponding to the average grayscale value is matched, and the temperature value is determined as the temperature value of each sub-region. Following the order of each sub-region from the inner ring to the outer ring, the temperature values corresponding to each sub-region are arranged in a matrix form. The temperature value of the innermost sub-region is taken as the first element of the matrix, and the temperature value of the outermost sub-region is taken as the last element of the matrix, thus forming the temperature field characteristic matrix of the molten pool region.
[0008] Preferably, the step of performing element-wise difference between the temperature field feature matrix corresponding to the current frame image and the temperature field feature matrix corresponding to the previous frame image to obtain the temperature field change gradient matrix of the molten pool region is as follows: The temperature value at each matrix position in the temperature field feature matrix of the current frame image is recorded as the current temperature value, and the temperature value at the same matrix position in the temperature field feature matrix of the previous frame image is recorded as the prior temperature value. According to the arrangement order of the temperature field change gradient matrix, the difference between the current temperature value and the prior temperature value is used to form the temperature field change gradient matrix of the molten pool region.
[0009] Preferably, the process of determining the thermal equilibrium state of the molten pool region based on the temperature field change gradient matrix and the temperature change rate of the current frame image is as follows: The sum of the absolute values of all elements in the temperature field gradient matrix is taken as the overall temperature change in the molten pool region. Calculate the temperature change rate of the current frame image based on the change in the highest grayscale value between the current frame image and the previous frame image; The thermal equilibrium state of the molten pool region is determined based on the overall temperature change and the rate of temperature change.
[0010] Preferably, the formula for calculating the rate of temperature change is as follows: ; In the formula, Indicates the rate of temperature change. This indicates the time interval between the acquisition of the current frame and the previous frame. This represents the difference matrix between the temperature field feature matrices of the current frame and the previous frame. Describes the norm of a matrix. The dimension of the temperature field characteristic matrix is represented. This represents the two-dimensional coordinate vector of the temperature centroid of the current frame image. This represents the two-dimensional coordinate vector of the temperature centroid of the previous frame image. The Euclidean distance representing the displacement of the center of gravity at temperature. This represents the Euclidean distance between the two-dimensional coordinate vector of the temperature centroid of the previous frame image and the center of the molten pool region. This indicates the smallest positive number whose denominator is zero. This represents the transpose of the difference matrix. This represents the temperature field feature matrix of the current frame image. Represents the trace of a matrix.
[0011] Preferably, the process of determining the thermal equilibrium state of the molten pool region based on the overall temperature change and the rate of temperature change is as follows: The intensity of temperature field change in the molten pool region is determined based on the overall temperature change. If the intensity of the temperature field change increases and the rate of temperature change is positive, then the molten pool region is determined to have an excess heat input trend. If the intensity of temperature field change decreases and the rate of temperature change is negative, then the molten pool region is determined to be experiencing insufficient heat input. If both the intensity and rate of temperature change remain stable, the molten pool region is determined to be in thermal equilibrium.
[0012] Preferably, the process of determining the intensity of temperature field change in the molten pool region based on the overall temperature change is as follows: Determine the total number of elements in the temperature field change gradient matrix based on the dimension parameter of the temperature field change gradient matrix. Divide the overall temperature change by the total number of elements to obtain the average temperature change of each sub-region. The average temperature change amplitude is used as the intensity of temperature field change in the molten pool region.
[0013] Preferably, the process of adaptively adjusting the welding current and welding speed in the molten pool region based on the determination of the thermal equilibrium state is as follows: When an excessive heat input trend is detected, reduce the welding current in the molten pool area and simultaneously increase the welding speed in the molten pool area. When the trend of insufficient heat input is determined, the welding current in the molten pool area is increased, while the welding speed in the molten pool area is decreased at the same time. When the state of thermal equilibrium is determined, the current welding current and welding speed in the molten pool area are maintained unchanged.
[0014] The present invention also provides a machine vision-based dynamic control system for welding temperature, the system comprising: The temperature field construction module is used to divide the continuous frame image of the molten pool area during the welding process into multiple concentric ring-shaped sub-regions, and convert the average gray value of the pixels in each sub-region into temperature value according to the preset gray-temperature mapping relationship, thus forming the temperature field feature matrix of the molten pool area. The gradient calculation module is used to perform element-wise difference between the temperature field feature matrix corresponding to the current frame image and the temperature field feature matrix corresponding to the previous frame image to obtain the temperature field change gradient matrix of the molten pool region. The balance discrimination module is used to determine the thermal balance state of the molten pool region based on the temperature field change gradient matrix and the temperature change rate of the current frame image. The parameter adjustment module is used to adaptively adjust the welding current and welding speed in the molten pool area based on the judgment result of the thermal balance state.
[0015] As can be seen from the above technical solution, the present invention provides a welding temperature dynamic control method and system based on machine vision. By dividing the molten pool area into multiple concentric annular sub-regions and constructing a temperature field feature matrix, a panoramic perception of the spatial distribution of the molten pool temperature is achieved. By using the element-wise difference of the inter-frame temperature field feature matrix, the temperature field change gradient matrix is obtained, which can track the dynamic evolution trend of the temperature field in real time. By combining the temperature field change gradient matrix with the temperature change rate to determine the thermal equilibrium state, the accuracy and response speed of judging the state of excessive, insufficient, or balanced heat input are improved. On this basis, the welding current and welding speed are adaptively adjusted, overcoming the shortcomings of the traditional constant parameter method in adapting to working condition fluctuations and the single-point feedback method in terms of single information and lag response. This effectively suppresses the oscillation and overshoot of welding temperature and significantly improves the stability and consistency of welding quality. Attached Figure Description
[0016] Other objects and results of the invention will become more apparent and readily understood by referring to the following description taken in conjunction with the accompanying drawings, and with a more complete understanding of the invention. In the drawings: Figure 1This is a schematic flowchart of a machine vision-based dynamic control method for welding temperature according to an embodiment of the present invention. Figure 2 This is a functional block diagram of a machine vision-based dynamic control system for welding temperature according to an embodiment of the present invention. Detailed Implementation Existing welding temperature control methods are unable to obtain real-time information on the spatial distribution and dynamic evolution of the temperature field in the molten pool area, and cannot accurately determine the thermal balance state of the molten pool and adaptively adjust welding parameters accordingly. This results in welding quality being sensitive to fluctuations in operating conditions and temperature regulation being prone to oscillations or overshoot.
[0017] To address the aforementioned problems, this invention provides a method and system for dynamic control of welding temperature based on machine vision. The specific embodiments of this invention will be described in detail below with reference to the accompanying drawings.
[0018] To illustrate the machine vision-based dynamic welding temperature control method and system provided by this invention, Figure 1 An exemplary illustration of a machine vision-based dynamic control method for welding temperature is provided in this embodiment of the invention. Figure 2 An exemplary illustration is provided for a machine vision-based dynamic control system for welding temperature according to an embodiment of the present invention.
[0019] The following description of exemplary embodiments is merely illustrative and is in no way intended to limit the invention or its application or use. Techniques and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques and equipment should be considered part of the specification.
[0020] Reference Figure 1 The diagram shown is a flowchart illustrating a machine vision-based dynamic welding temperature control method according to an embodiment of the present invention. In this embodiment, the machine vision-based dynamic welding temperature control method includes: Step 1: Divide the continuous frame images of the molten pool area during the welding process into multiple concentric ring-shaped sub-regions, and convert the average gray value of the pixels in each sub-region into temperature value according to the preset gray-level-temperature mapping relationship, thus forming the temperature field feature matrix of the molten pool area. In this embodiment of the invention, the process of dividing the continuous frame image of the molten pool region during welding into multiple concentric ring-shaped sub-regions is as follows: During the welding process, the molten pool contour of the molten pool region in consecutive frame images is extracted, and the geometric center of the molten pool contour is determined. Using the geometric center as the center, multiple concentric circle boundaries with the same radial width are set in a manner that expands layer by layer from the center to the edge of the molten pool outline. The image region within the molten pool contour is divided into multiple concentric ring-shaped sub-regions, with the annular region between adjacent concentric circle boundaries as a sub-region and the circular region where the center of the circle is located as the innermost sub-region.
[0021] The process of converting the average gray value of pixels in each sub-region into a temperature value according to a preset gray-level-temperature mapping relationship to form a temperature field feature matrix of the molten pool region is as follows: The arithmetic mean of the gray values of all pixels in each concentric ring-shaped sub-region is used as the average gray value of each sub-region. Based on the preset grayscale-temperature mapping relationship, the temperature value corresponding to the average grayscale value is matched, and the temperature value is determined as the temperature value of each sub-region. Following the order of each sub-region from the inner ring to the outer ring, the temperature values corresponding to each sub-region are arranged in a matrix form. The temperature value of the innermost sub-region is taken as the first element of the matrix, and the temperature value of the outermost sub-region is taken as the last element of the matrix, thus forming the temperature field characteristic matrix of the molten pool region.
[0022] Extracting the molten pool contour of the molten pool region from consecutive frames of images is accomplished by sequentially processing each frame of the original welding image. First, the color image is converted to grayscale. Then, Gaussian filtering is applied to the grayscale image. Specifically, a fixed-size two-dimensional Gaussian convolution kernel is defined, and its center is aligned sequentially with each pixel in the image. The sum of the products of the grayscale values of all pixels within the kernel's coverage area and the corresponding weights of the kernel's position is calculated. This sum replaces the grayscale value of the original center pixel, resulting in a smoothed grayscale image. Next, the Otsu thresholding method is applied to the smoothed image. This method iterates through all possible grayscale thresholds, calculates the inter-class variance after classifying the image pixels into foreground and background categories based on each threshold, and selects the grayscale value that maximizes the inter-class variance as the final threshold. A threshold is used to binarize the image, setting pixels with gray values greater than or equal to the threshold to pure white and pixels with gray values less than the threshold to pure black, resulting in a clear black-and-white binary image. In this binary image, the melt pool region appears as a connected block of white pixels. Then, a boundary tracking algorithm is used to extract the edges of this white region. The algorithm starts scanning from the top left corner of the image, finds the first white pixel and uses it as the starting point for contour tracking. Then, with this point as the center, it checks its eight neighboring pixels in a clockwise direction, finds the next white pixel and records it as a contour point. At the same time, the current point is moved to the newly found point, and this process of checking neighboring pixels and recording new boundary points is repeated until the tracking path returns to the starting point. Finally, connecting all the sequentially recorded boundary points forms a closed melt pool contour.
[0023] Determining the geometric center of the molten pool profile involves calculating the arithmetic mean center of all vertices of this closed polygon. Specifically, this process involves obtaining the x-coordinates of all boundary points on the molten pool profile, summing these x-coordinates, and simultaneously obtaining the y-coordinates of all boundary points. The sum of these y-coordinates is then calculated. Finally, the total number of boundary points on the profile is counted, and the sum of the x-coordinates is divided by the total number of boundary points. The result is the x-coordinate of the geometric center. The sum of the y-coordinates is then divided by the total number of boundary points. The final coordinate point obtained is the geometric center of the molten pool profile.
[0024] Using the geometric center as the center, multiple concentric circle boundaries with the same radial width are set in a layer-by-layer manner, expanding from the center towards the edge of the molten pool contour. In practice, the straight-line distance from the geometric center to each boundary point on the molten pool contour needs to be calculated first. By comparing all these distances, the maximum value is found. This maximum value is the distance from the center of the circle to the farthest point of the contour. Then, a fixed radial width value is set as the interval between the concentric circles. Starting from the center, the radius of the first circle is set to this fixed radial width value, the radius of the second circle is set to twice this radial width value, the radius of the third circle is set to three times this radial width value, and so on, continuously increasing the radius to draw virtual circles until the radius value of the virtual circle to be drawn exceeds the previously calculated farthest distance value. At this point, drawing stops, and the circumference of all drawn virtual circles is the required concentric circle boundary.
[0025] Using the annular region between adjacent concentric circle boundaries as a sub-region, and the circular region containing the center of each circle as the innermost sub-region, the image region within the molten pool contour is divided into multiple concentric annular sub-regions. The innermost sub-region is the circular region between the geometric center point and the first concentric circle boundary; this region contains all pixels whose distance to the center is less than the radius of the first circle. The second sub-region is the annular region between the first and second concentric circle boundaries; this region contains all pixels whose distance to the center is greater than or equal to the radius of the first circle and less than the radius of the second circle. Each subsequent sub-region is between... The annular region between the boundaries of two adjacent concentric circles contains all pixels whose distance to the center is greater than or equal to the inner circle radius and less than the outer circle radius. The last sub-region is the annular region between the boundary of the last concentric circle and the edge of the molten pool contour. This region contains all pixels whose distance to the center is greater than or equal to the radius of the last circle and which are also located within the white area of the binary image of the molten pool contour. When dividing the region, it is necessary to calculate the distance to the geometric center for each pixel and determine which annular sub-region it belongs to based on this distance value. At the same time, it is necessary to confirm that the pixel belongs to the white area in the binary image to ensure that it is inside the molten pool.
[0026] The arithmetic mean of the gray values of all pixels in each concentric ring-shaped sub-region is used as the average gray value of each sub-region. For each sub-region, it is necessary to traverse all pixels in the original grayscale image that are determined to belong to the sub-region, read the gray intensity value corresponding to each pixel in turn, and sum these gray values one by one to obtain the sum of the gray values of all pixels in the sub-region. At the same time, the total number of pixels contained in the sub-region is obtained by counting. Finally, the sum of gray values is divided by the total number of pixels, and the quotient obtained by the division operation is the average gray value of the sub-region.
[0027] Based on a pre-defined grayscale-temperature mapping relationship, the temperature value corresponding to the average grayscale value is matched and determined as the temperature value for each sub-region. The pre-defined mapping relationship is a lookup table prepared in advance through experimental calibration. This table uses grayscale values as indexes and stores the corresponding precise temperature values. For the average grayscale value calculated for each sub-region, a sequential search or binary search is performed in the lookup table to find the grayscale value entry that is exactly equal to the average grayscale value. The temperature value recorded in that entry is directly read as the result. If the average grayscale value does not have an exact match in the lookup table but lies between two adjacent grayscale values, linear interpolation is used to calculate the temperature value. The temperature value is calculated by determining the smaller and larger gray values adjacent to the average gray value, as well as their corresponding temperature values. The difference between the average gray value and the smaller gray value is calculated, and the difference between the larger and smaller gray values is also calculated. A scaling factor is obtained by dividing the former difference by the latter. Then, the temperature difference is obtained by subtracting the temperature corresponding to the smaller gray value from the temperature corresponding to the larger gray value. This temperature difference is multiplied by the previously calculated scaling factor to obtain a temperature increment. Finally, this temperature increment is added to the temperature corresponding to the smaller gray value. The final result is the temperature value corresponding to the average gray value, and this temperature value is determined as the temperature value of that sub-region.
[0028] Following the order of the sub-regions from the inner ring to the outer ring, the temperature values corresponding to each sub-region are arranged sequentially into a matrix. The temperature value of the innermost sub-region is used as the first element of the matrix, and the temperature value of the outermost sub-region is used as the last element of the matrix, forming the temperature field feature matrix of the molten pool region. Specifically, a one-dimensional empty array is initialized. The temperature value of the innermost circular sub-region is taken out and placed in the first index position of this array. Then, the temperature value of the first annular sub-region immediately adjacent to it is taken out and placed in the second index position of the array. Next, the temperature value of the second annular sub-region is taken out and placed in the third index position of the array. Strictly following the spatial order of the sub-regions from the center to the outer edge, the temperature value of each sub-region is taken out sequentially and placed in the next empty position of the array until the temperature value of the outermost annular sub-region adjacent to the molten pool outline is taken out and placed in the last index position of the array. At this point, the series of temperature values stored sequentially in this one-dimensional array constitutes the final temperature field feature matrix of the molten pool region.
[0029] Step 2: Perform element-wise difference between the temperature field feature matrix corresponding to the current frame image and the temperature field feature matrix corresponding to the previous frame image to obtain the temperature field change gradient matrix of the molten pool region. In this embodiment of the invention, the process of performing element-wise difference between the temperature field feature matrix corresponding to the current frame image and the temperature field feature matrix corresponding to the previous frame image to obtain the temperature field change gradient matrix of the molten pool region is as follows: The temperature value at each matrix position in the temperature field feature matrix of the current frame image is recorded as the current temperature value, and the temperature value at the same matrix position in the temperature field feature matrix of the previous frame image is recorded as the prior temperature value. According to the arrangement order of the temperature field change gradient matrix, the difference between the current temperature value and the prior temperature value is used to form the temperature field change gradient matrix of the molten pool region.
[0030] From the one-dimensional array storing the temperature field feature matrix of the current frame molten pool region, each array element is accessed one by one in ascending order of array index starting from zero, and the value stored in that element is read. This read value is recorded as the current temperature value.
[0031] From the one-dimensional array storing the temperature field feature matrix of the molten pool region of the previous frame, access and read the value stored at the index position that is exactly the same as when reading the current temperature value. This read value is recorded as the prior temperature value.
[0032] According to the arrangement order of the temperature field change gradient matrix, which is strictly consistent with the element arrangement order of the temperature field feature matrix, for the first array index position, the current temperature value is subtracted from the prior temperature value, and the result is the temperature change corresponding to the first position.
[0033] Next, move to the second array index position, read the current temperature value stored in the current frame feature matrix at the second index position, read the prior temperature value stored in the previous frame feature matrix at the second index position, and perform the subtraction operation of the current temperature value and the prior temperature value corresponding to the second position.
[0034] Following a linear order from the first array index to the last array index, the process of reading the current temperature value, reading the prior temperature value, and performing the subtraction operation is repeated for each index position until the last array index position is processed.
[0035] The result of each subtraction operation is stored sequentially into a newly created one-dimensional array according to the array index order corresponding to the operation. This new array completely stores the calculated difference at all positions, which is the final gradient matrix of the temperature field change in the molten pool region.
[0036] Step 3: Determine the thermal equilibrium state of the molten pool region based on the temperature field gradient matrix and the temperature change rate of the current frame image; In this embodiment of the invention, the process of determining the thermal equilibrium state of the molten pool region based on the temperature field change gradient matrix and the temperature change rate of the current frame image is as follows: The sum of the absolute values of all elements in the temperature field gradient matrix is taken as the overall temperature change in the molten pool region. Calculate the temperature change rate of the current frame image based on the change in the highest grayscale value between the current frame image and the previous frame image; The thermal equilibrium state of the molten pool region is determined based on the overall temperature change and the rate of temperature change.
[0037] The formula for calculating the rate of temperature change is as follows: ; In the formula, Indicates the rate of temperature change. This indicates the time interval between the acquisition of the current frame and the previous frame. This represents the difference matrix between the temperature field feature matrices of the current frame and the previous frame. Describes the norm of a matrix. The dimension of the temperature field characteristic matrix is represented. This represents the two-dimensional coordinate vector of the temperature centroid of the current frame image. This represents the two-dimensional coordinate vector of the temperature centroid of the previous frame image. The Euclidean distance representing the displacement of the center of gravity at temperature. This represents the Euclidean distance between the two-dimensional coordinate vector of the temperature centroid of the previous frame image and the center of the molten pool region. This indicates the smallest positive number whose denominator is zero. This represents the transpose of the difference matrix. This represents the temperature field feature matrix of the current frame image. Represents the trace of a matrix.
[0038] The process of determining the thermal equilibrium state of the molten pool region based on the overall temperature change and the rate of temperature change is as follows: The intensity of temperature field change in the molten pool region is determined based on the overall temperature change. If the intensity of the temperature field change increases and the rate of temperature change is positive, then the molten pool region is determined to have an excess heat input trend. If the intensity of temperature field change decreases and the rate of temperature change is negative, then the molten pool region is determined to be experiencing insufficient heat input. If both the intensity and rate of temperature change remain stable, the molten pool region is determined to be in thermal equilibrium.
[0039] The process of determining the intensity of temperature field change in the molten pool region based on the overall temperature change is as follows: Determine the total number of elements in the temperature field change gradient matrix based on the dimension parameter of the temperature field change gradient matrix. Divide the overall temperature change by the total number of elements to obtain the average temperature change of each sub-region. The average temperature change amplitude is used as the intensity of temperature field change in the molten pool region.
[0040] Initialize an accumulator variable from the one-dimensional array of the temperature field gradient matrix and set its value to zero. Then, access each array element in ascending order of array index, read the value stored in the element, and check if the value is negative. If it is negative, multiply it by negative one to convert it to a positive number. If it is positive, keep the original value unchanged. This converted positive number is the absolute value of the element. Add this absolute value to the accumulator variable, that is, add this absolute value to the current value of the accumulator variable to get a new accumulator value. Repeat this process until all array elements have been accessed and processed. Finally, the sum stored in the accumulator variable is the overall temperature change of the molten pool region.
[0041] From the grayscale image of the melt pool region of the current frame image, create a variable to record the currently found maximum grayscale value and initialize it to zero. Then, traverse each pixel in the grayscale image, reading the grayscale intensity value of each pixel in turn. Compare the read grayscale value with the recorded maximum grayscale value. If the read grayscale value is greater than the recorded maximum grayscale value, update the recorded maximum grayscale value to this larger grayscale value. After the traversal is complete, the recorded maximum grayscale value is the highest grayscale value of the current frame. From the grayscale image of the melt pool region of the previous frame image, use the same traversal and comparison method to find the maximum grayscale value. This maximum grayscale value is the prior highest grayscale value. Subtract the prior highest grayscale value from the current frame's highest grayscale value and perform a subtraction operation to obtain the difference. This difference is the temperature change rate of the current frame image.
[0042] The dimension parameter of the temperature field change gradient matrix refers to the total number of elements contained in this matrix. Since the temperature field change gradient matrix is stored as a one-dimensional array, the total number of elements can be obtained directly by querying the length attribute of the array. If the array data structure does not provide a length attribute, the total number of elements is obtained by traversing the array and counting the number of stored numerical items. That is, starting from the beginning of the array, the counter is incremented by one for each valid value encountered until the end of the array. The final value of the counter is the total number of elements in the temperature field change gradient matrix.
[0043] The overall temperature change is used as the dividend, and the total number of elements is used as the divisor. The division operation is performed by dividing the value of the overall temperature change by the value of the total number of elements and calculating the quotient. This quotient is the average temperature change of each sub-region.
[0044] The average temperature change amplitude is directly assigned to the temperature field change intensity of the molten pool region. Specifically, the calculated average temperature change amplitude is stored in a variable representing the temperature field change intensity, so that the temperature field change intensity is numerically equal to the average temperature change amplitude.
[0045] Compare the temperature field change intensity of the current frame with that of the previous frame. If the temperature field change intensity of the current frame is greater than that of the previous frame, and the temperature change rate is positive (i.e., the change in the highest grayscale value is positive), then the melt pool region is determined to have an excess heat input trend.
[0046] Compare the temperature field change intensity of the current frame with that of the previous frame. If the temperature field change intensity of the current frame is less than that of the previous frame, and the temperature change rate is negative (i.e., the change in the highest grayscale value is negative), then the melt pool region is determined to have insufficient heat input.
[0047] Set a positive number as the intensity change threshold and another positive number as the rate change threshold. Calculate the absolute value of the difference between the temperature field change intensity of the current frame and the temperature field change intensity of the previous frame. If this absolute value is less than the intensity change threshold and the absolute value of the calculated temperature change rate is less than the rate change threshold, then the molten pool region is determined to be in thermal equilibrium.
[0048] Step 4: Based on the judgment of the thermal balance state, adaptively adjust the welding current and welding speed in the molten pool area.
[0049] In this embodiment of the invention, the process of adaptively adjusting the welding current and welding speed in the molten pool region based on the determination result of the thermal balance state is as follows: When an excessive heat input trend is detected, reduce the welding current in the molten pool area and simultaneously increase the welding speed in the molten pool area. When the trend of insufficient heat input is determined, the welding current in the molten pool area is increased, while the welding speed in the molten pool area is decreased at the same time. When the state of thermal equilibrium is determined, the current welding current and welding speed in the molten pool area are maintained unchanged.
[0050] When the molten pool region is determined to have an excessive heat input trend, a command to reduce the current is sent to the current control unit of the welding power supply. This command causes the current control unit to reduce the set value of the output current. Specifically, the current control unit gradually lowers the current reference signal by a preset reduction amount, resulting in a decrease in the actual welding current. At the same time, a command to increase the speed is sent to the speed control unit of the welding motion system. This command causes the speed control unit to increase the set value of the welding speed. Specifically, the speed control unit gradually increases the speed reference signal by a preset increase amount, resulting in an increase in the welding speed.
[0051] When the molten pool area is determined to show a trend of insufficient heat input, an instruction to increase the current is sent to the current control unit of the welding power source. This instruction causes the current control unit to increase the set value of the output current. Specifically, the current control unit gradually increases the current reference signal by a preset increase amount, resulting in an increase in the actual welding current. At the same time, an instruction to decrease the speed is sent to the speed control unit of the welding motion system. This instruction causes the speed control unit to decrease the set value of the welding speed. Specifically, the speed control unit gradually decreases the speed reference signal by a preset decrease amount, resulting in a decrease in the welding speed.
[0052] When the molten pool area is determined to be in thermal equilibrium, no adjustment command is sent to the current control unit of the welding power source. The current control unit keeps the current setting value unchanged, and the actual welding current remains stable. At the same time, no adjustment command is sent to the speed control unit of the welding motion system. The speed control unit keeps the current speed setting value unchanged, and the welding speed remains stable.
[0053] As can be seen from the above embodiments, the machine vision-based dynamic welding temperature control method provided by the present invention achieves panoramic perception of the spatial distribution of the molten pool temperature by dividing the molten pool area into multiple concentric annular sub-regions and constructing a temperature field feature matrix; by using the element-wise difference of the inter-frame temperature field feature matrix, the temperature field change gradient matrix is obtained, which can track the dynamic evolution trend of the temperature field in real time; by combining the temperature field change gradient matrix and the temperature change rate to determine the thermal equilibrium state, the accuracy and response speed of judging the state of excessive, insufficient or balanced heat input are improved; on this basis, the welding current and welding speed are adaptively adjusted, overcoming the shortcomings of the traditional constant parameter method in adapting to working condition fluctuations and the single-point feedback method in terms of single information and lag response, effectively suppressing the oscillation and overshoot of welding temperature, and significantly improving the stability and consistency of welding quality.
[0054] like Figure 2 The diagram shown is a functional block diagram of a welding temperature dynamic control system 100 based on machine vision provided in an embodiment of the present invention, including a temperature field construction module 101, a gradient calculation module 102, a balance discrimination module 103, and a parameter adjustment module 104.
[0055] In this embodiment, the functions of each module are as follows: The temperature field construction module 101 is used to divide the continuous frame image of the molten pool region during the welding process into multiple concentric ring-shaped sub-regions, and convert the average gray value of the pixels in each sub-region into a temperature value according to the preset gray-temperature mapping relationship, thereby forming a temperature field feature matrix of the molten pool region. The gradient calculation module 102 is used to perform element-wise difference between the temperature field feature matrix corresponding to the current frame image and the temperature field feature matrix corresponding to the previous frame image to obtain the temperature field change gradient matrix of the molten pool region. The balance discrimination module 103 is used to determine the thermal balance state of the molten pool region based on the temperature field change gradient matrix and the temperature change rate of the current frame image. The parameter adjustment module 104 is used to adaptively adjust the welding current and welding speed in the molten pool area based on the judgment result of the thermal balance state.
[0056] As can be seen from the above embodiments, the welding temperature dynamic control system based on machine vision provided by the present invention achieves panoramic perception of the spatial distribution of the molten pool temperature by dividing the molten pool area into multiple concentric annular sub-regions and constructing a temperature field feature matrix; by using the element-wise difference of the inter-frame temperature field feature matrix, the temperature field change gradient matrix is obtained, which can track the dynamic evolution trend of the temperature field in real time; by combining the temperature field change gradient matrix and the temperature change rate to determine the thermal equilibrium state, the accuracy and response speed of judging the state of excessive, insufficient or balanced heat input are improved; on this basis, the welding current and welding speed are adaptively adjusted, overcoming the shortcomings of the traditional constant parameter method in adapting to working condition fluctuations and the single-point feedback method in terms of single information and lag response, effectively suppressing the oscillation and overshoot of welding temperature, and significantly improving the stability and consistency of welding quality.
[0057] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.
[0058] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.
Claims
1. A method for dynamic control of welding temperature based on machine vision, characterized in that, The method includes: Step 1: Divide the continuous frame images of the molten pool area during the welding process into multiple concentric ring-shaped sub-regions, and convert the average gray value of the pixels in each sub-region into temperature value according to the preset gray-level-temperature mapping relationship, thus forming the temperature field feature matrix of the molten pool area. Step 2: Perform element-wise difference between the temperature field feature matrix corresponding to the current frame image and the temperature field feature matrix corresponding to the previous frame image to obtain the temperature field change gradient matrix of the molten pool region. Step 3: Determine the thermal equilibrium state of the molten pool region based on the temperature field gradient matrix and the temperature change rate of the current frame image; Step 4: Based on the judgment of the thermal balance state, adaptively adjust the welding current and welding speed in the molten pool area.
2. The welding temperature dynamic control method based on machine vision as described in claim 1, characterized in that, The process of dividing the continuous frame images of the molten pool region during welding into multiple concentric ring-shaped sub-regions is as follows: During the welding process, the molten pool contour of the molten pool region in consecutive frame images is extracted, and the geometric center of the molten pool contour is determined. Using the geometric center as the center, multiple concentric circle boundaries with the same radial width are set in a manner that expands layer by layer from the center to the edge of the molten pool outline. The image region within the molten pool contour is divided into multiple concentric ring-shaped sub-regions, with the annular region between adjacent concentric circle boundaries as a sub-region and the circular region where the center of the circle is located as the innermost sub-region.
3. The welding temperature dynamic control method based on machine vision as described in claim 2, characterized in that, The process of converting the average gray value of pixels in each sub-region into a temperature value according to a preset gray-level-temperature mapping relationship to form a temperature field feature matrix of the molten pool region is as follows: The arithmetic mean of the gray values of all pixels in each concentric ring-shaped sub-region is used as the average gray value of each sub-region. Based on the preset grayscale-temperature mapping relationship, the temperature value corresponding to the average grayscale value is matched, and the temperature value is determined as the temperature value of each sub-region. Following the order of each sub-region from the inner ring to the outer ring, the temperature values corresponding to each sub-region are arranged in a matrix form. The temperature value of the innermost sub-region is taken as the first element of the matrix, and the temperature value of the outermost sub-region is taken as the last element of the matrix, thus forming the temperature field characteristic matrix of the molten pool region.
4. The welding temperature dynamic control method based on machine vision as described in claim 1, characterized in that, The process of performing element-wise difference between the temperature field feature matrix corresponding to the current frame image and the temperature field feature matrix corresponding to the previous frame image to obtain the temperature field change gradient matrix of the molten pool region is as follows: The temperature value at each matrix position in the temperature field feature matrix of the current frame image is recorded as the current temperature value, and the temperature value at the same matrix position in the temperature field feature matrix of the previous frame image is recorded as the prior temperature value. According to the arrangement order of the temperature field change gradient matrix, the difference between the current temperature value and the prior temperature value is used to form the temperature field change gradient matrix of the molten pool region.
5. The method for dynamic control of welding temperature based on machine vision as described in claim 1, characterized in that, The process of determining the thermal equilibrium state of the molten pool region based on the temperature field change gradient matrix and the temperature change rate of the current frame image is as follows: The sum of the absolute values of all elements in the temperature field gradient matrix is taken as the overall temperature change in the molten pool region. Calculate the temperature change rate of the current frame image based on the change in the highest grayscale value between the current frame image and the previous frame image; The thermal equilibrium state of the molten pool region is determined based on the overall temperature change and the rate of temperature change.
6. The welding temperature dynamic control method based on machine vision as described in claim 5, characterized in that, The formula for calculating the rate of temperature change is as follows: ； In the formula, Indicates the rate of temperature change. This indicates the time interval between the acquisition of the current frame and the previous frame. This represents the difference matrix between the temperature field feature matrices of the current frame and the previous frame. Describes the norm of a matrix. The dimension of the temperature field characteristic matrix is represented. This represents the two-dimensional coordinate vector of the temperature centroid of the current frame image. This represents the two-dimensional coordinate vector of the temperature centroid of the previous frame image. The Euclidean distance representing the displacement of the center of gravity at temperature. This represents the Euclidean distance between the two-dimensional coordinate vector of the temperature centroid of the previous frame image and the center of the molten pool region. This indicates the smallest positive number whose denominator is zero. This represents the transpose of the difference matrix. This represents the temperature field feature matrix of the current frame image. Represents the trace of a matrix.
7. The welding temperature dynamic control method based on machine vision as described in claim 5, characterized in that, The process of determining the thermal equilibrium state of the molten pool region based on the overall temperature change and the rate of temperature change is as follows: The intensity of temperature field change in the molten pool region is determined based on the overall temperature change. If the intensity of the temperature field change increases and the rate of temperature change is positive, then the molten pool region is determined to have an excess heat input trend. If the intensity of temperature field change decreases and the rate of temperature change is negative, then the molten pool region is determined to be experiencing insufficient heat input. If both the intensity and rate of temperature change remain stable, the molten pool region is determined to be in thermal equilibrium.
8. The welding temperature dynamic control method based on machine vision as described in claim 7, characterized in that, The process of determining the intensity of temperature field change in the molten pool region based on the overall temperature change is as follows: Determine the total number of elements in the temperature field change gradient matrix based on the dimension parameter of the temperature field change gradient matrix. Divide the overall temperature change by the total number of elements to obtain the average temperature change of each sub-region. The average temperature change amplitude is used as the intensity of temperature field change in the molten pool region.
9. The welding temperature dynamic control method based on machine vision as described in claim 1, characterized in that, The process of adaptively adjusting the welding current and welding speed in the molten pool region based on the determination of the thermal balance state is as follows: When an excessive heat input trend is detected, reduce the welding current in the molten pool area and simultaneously increase the welding speed in the molten pool area. When the trend of insufficient heat input is determined, the welding current in the molten pool area is increased, while the welding speed in the molten pool area is decreased at the same time. When the state of thermal equilibrium is determined, the current welding current and welding speed in the molten pool area are maintained unchanged.
10. A machine vision-based dynamic control system for welding temperature, characterized in that, The system is used to implement the machine vision-based dynamic welding temperature control method according to any one of claims 1-9, the system comprising: The temperature field construction module is used to divide the continuous frame image of the molten pool area during the welding process into multiple concentric ring-shaped sub-regions, and convert the average gray value of the pixels in each sub-region into temperature value according to the preset gray-temperature mapping relationship, thus forming the temperature field feature matrix of the molten pool area. The gradient calculation module is used to perform element-wise difference between the temperature field feature matrix corresponding to the current frame image and the temperature field feature matrix corresponding to the previous frame image to obtain the temperature field change gradient matrix of the molten pool region. The balance discrimination module is used to determine the thermal balance state of the molten pool region based on the temperature field change gradient matrix and the temperature change rate of the current frame image. The parameter adjustment module is used to adaptively adjust the welding current and welding speed in the molten pool area based on the judgment result of the thermal balance state.