Welding pool fusion monitoring method and system

Abstract

The invention relates to the field of molten pool monitoring, in particular to a welding molten pool fusion monitoring method and system. Comprising the steps of obtaining ultrasonic scanning data of a welding pool through a single-point scanning method; according to the size parameter information of the welding pool, first reflection time and second reflection time are determined; determining other maximum value points from the ultrasonic scanning data according to the first reflection time and the second reflection time; determining real defect points from other maximum value points; and echo signal features of the real defect points are input into the trained long-short-term memory network, the defect types and the fusion state are output through the trained long-short-term memory network, and fusion parameters and optimization algorithms of a welding pool are adjusted through the defect types and the fusion state. According to the method, the defect type and the fusion state of the welding pool can be identified, and the welding process of the welding pool can be adjusted and optimized in time.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of molten pool monitoring, in particular to a welding molten pool fusion monitoring method and system. BACKGROUND

[0002] The welding molten pool is a liquid metal area formed under the action of a heat source in a fusion welding process. The molten pool metal flows to the tail under the action of an electric arc, overcoming gravity and surface tension, and forms a weld after cooling. The molten pool temperature is distributed in a gradient, with the highest temperature at the center of the electric arc and decreasing towards the edge. The head of the molten pool is in a melting stage, and the tail begins to crystallize, experiencing a dynamic cycle of heating-melting-cooling. This area has the characteristics of small volume and fast cooling speed, and the chemical components such as carbon, sulfur, and phosphorus need to be strictly controlled.

[0003] In order to identify typical defects such as incomplete fusion, pores, and cracks in the welding process, existing defect recognition methods based on visual information mainly rely on image features of the molten pool surface collected by high-speed cameras or infrared thermographs for judgment. These methods usually include gray level distribution analysis, surface texture feature extraction, edge contour detection, and other technical means of molten pool images. However, in dynamic thermal processing processes such as arc welding and laser welding, the molten pool is actually a transient high-temperature liquid metal area containing complex three-dimensional flow, heat and mass transfer, and phase change behavior. Its surface topography (such as molten pool width and trailing angle) can only reflect part of the process state, while the key factors that determine the welding quality such as penetration depth, internal pore distribution, and dendrite growth direction are hidden in the internal structure of the molten pool.

[0004] The existing two-dimensional image processing methods often ignore the correlation between the internal heat-flow coupling characteristics (such as Marangoni convection and keyhole effect) of the molten pool and the formation of defects, resulting in insufficient detection accuracy of hidden defects such as incomplete fusion and root cracks, and failing to meet the quality control requirements of high-reliability welding. SUMMARY

[0005] The present application provides a welding molten pool fusion monitoring method and system to solve the existing problems.

[0006] The welding molten pool fusion monitoring method of the present application adopts the following technical solution: One embodiment of the present application provides a welding molten pool fusion monitoring method, which comprises the following steps: Obtain ultrasonic scan data of the welding molten pool by a single-point scanning method, the welding molten pool comprising at least one scanning point; Determine a first reflection time and a second reflection time according to size parameter information of the welding molten pool, wherein the first reflection time is the time for an ultrasonic echo signal to reflect back to the surface of the molten pool, and the second reflection time is the time for the ultrasonic echo signal to reflect back to the bottom of the molten pool. determine other extreme value points from the ultrasonic scanning data according to the first reflection time and the second reflection time; determine the real defect points from the other extreme value points; input the echo signal features of the real defect points into the trained long short-term memory network, output the defect type and fusion state through the trained long short-term memory network, and adjust the fusion parameters and optimization algorithm of the welding pool through the defect type and fusion state.

[0007] Optionally, the other extreme value points are determined from the ultrasonic scanning data according to the first reflection time and the second reflection time, and specifically include: determine the local extreme value points for the ultrasonic scanning data of each scanning point through the derivative method; arrange the local extreme value points in ascending order of time to obtain a time ascending sequence; segment the time ascending sequence using the otsu multi-threshold segmentation method to obtain extreme value point segments; calculate the mean value of each extreme value point segment, determine the wave peak of the elements in the first extreme value point segment in the ultrasonic scanning data as the molten pool surface echo, and determine the wave peak of the elements in the second extreme value point segment in the ultrasonic scanning data as the molten pool bottom echo, wherein the first extreme value point segment is the extreme value point segment with the smallest mean value and first reflection time difference, and the second extreme value point segment is the extreme value point segment with the smallest mean value and second reflection time difference; determine the local extreme value points between the molten pool surface echo and the molten pool bottom echo in the ultrasonic scanning data as the other extreme value points.

[0008] Optionally, the real defect points are determined from the other extreme value points, and specifically include: determine the abnormal echo points from the other extreme value points, and determine the ultrasonic scanning data of the abnormal echo points as abnormal echoes; obtain the new ultrasonic scanning data of each abnormal echo point, the new ultrasonic scanning data being obtained by emitting sound waves with a preset sound wave intensity interval for each abnormal echo point, and the sound wave intensity of the preset sound wave intensity interval increasing in turn; determine a change trend value sequence according to the new ultrasonic scanning data; determine the amplitude sequence and the time point sequence of the abnormal echo points in the multiple scanning data of each real defect candidate according to the change trend value sequence; determine the first real defect points and the suspected defect points according to the amplitude sequence and the time point sequence; determine the second real defect points from the suspected defect points; determine the first real defect points and the second real defect points as the real defect points.

[0009] Optionally, the abnormal echo point is determined from other maximum points, specifically comprising: An upper envelope line of the other maximum points is determined using a cubic spline interpolation method, and a fitting point of each point in the upper envelope line is determined by a least square method. An absolute value of a difference between each fitting point and each point in the upper envelope line is calculated to obtain a difference absolute value. A ratio of the difference absolute value and each point in the upper envelope line is calculated to obtain a convexity value; An echo abnormality probability of each point in the upper envelope line is determined according to the convexity value. A point with an echo abnormality probability greater than a preset abnormal threshold is determined as an abnormal echo point.

[0010] Optionally, the change trend value sequence is determined according to the newly added ultrasonic scanning data, specifically comprising: For each abnormal echo point, an amplitude value of each abnormal echo point in each sound wave intensity corresponding to the newly added ultrasonic scanning data is obtained to obtain an amplitude sequence; The change trend value sequence is determined according to the amplitude sequence.

[0011] Optionally, the change trend value sequence is determined according to the amplitude sequence, specifically comprising: An element order of elements of the first n amplitude sequences is taken as an abscissa, and an element of each amplitude sequence is taken as an ordinate to determine a principal component analysis coordinate. A projection vector and a projection value corresponding to each projection vector are determined by a principal component analysis method. An inverse tangent angle value of a ratio of the ordinate and the abscissa of the projection vector with the largest projection value is determined as a first element in the change trend value sequence, wherein n is a preset constant; Each element in the change trend value sequence is sequentially determined to obtain the change trend value sequence.

[0012] Optionally, the amplitude sequence and the time point sequence of the abnormal echo point in the multi-scan data of each real defect candidate are determined according to the change trend value sequence, specifically comprising: The otsu multi-threshold segmentation method is used to segment the change trend value sequence to determine two change trend segmentation segments. The variance of each change trend segmentation segment is calculated, and the sum of the variances of the two change trend segmentation segments is obtained to obtain a pseudo-defect probability value; The real defect candidate probability value is determined according to the pseudo-defect probability value. An abnormal echo point with a real defect candidate probability value greater than a preset defect threshold is determined as a real defect candidate; The multi-scan data of each real defect candidate is obtained, the abnormal echo point in the multi-scan data is determined, and the amplitude sequence and the time point sequence of the abnormal echo point in each multi-scan data are obtained.

[0013] Optionally, the first real defect point and the suspected defect point are determined according to the amplitude sequence and the time point sequence, specifically comprising: The variation coefficient of the amplitude sequence and the standard deviation of the time point sequence are calculated, the abnormal echo points with the variation coefficient less than a preset first abnormal threshold and the standard deviation less than a preset second abnormal threshold are determined as first real defect points, and the abnormal echo points in the multiple scanning data, except the first real abnormal points, are determined as suspected defect points.

[0014] Optionally, second real defect points are determined from the suspected defect points, specifically including: Final scanning data of the suspected defect points under different emission energies are acquired, and abnormal echo points in the final scanning data are determined, the absolute value of the time difference between the time of each abnormal echo point in the final scanning data and the first reflection time is calculated, and a time difference sequence of each abnormal echo point in the final scanning data is obtained. The abnormal echo points with the time difference sequence less than a preset variance threshold are determined as the second real defect points.

[0015] The present application provides a kind of welding pool fusion monitoring system, including memory, processor and computer program stored in memory and can be run on processor, the computer program is implemented when processor executes as the steps of the kind of welding pool fusion monitoring method.

[0016] The technical scheme of the present application has the following advantages: In the embodiment of the present application, by analyzing the ultrasonic echo signals under different emission energies, combining the fusion pool surface echo and the fusion pool bottom echo, the pseudo-defect signals removed from the noise influence are obtained, the real defect scanning points are obtained, and then the signal characteristics of the defect scanning points are input into the neural network, the defect type and the fusion state of the welding pool are identified, which is convenient for further analysis and adjustment. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0018] Figure 1 A flow chart of a welding pool fusion monitoring method provided by an embodiment of the present application is shown in the figure. Figure 2 A schematic diagram of a welding pool is shown in the figure. Figure 3 A schematic diagram of maximum value points of probe point-by-point scanning data is shown in the figure. Figure 4 A structure diagram of a welding pool fusion monitoring system provided by an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0019] To further illustrate the technical means and effectiveness of the present invention in achieving its intended objectives, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail a method for monitoring weld pool fusion according to the present invention, including its specific implementation, structure, features, and effectiveness. In the following description, references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0020] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0021] The specific scheme of the welding pool fusion monitoring method provided by the present invention is described in detail below with reference to the accompanying drawings.

[0022] The present invention provides a method and system for monitoring the fusion of a welding pool. Figure 1 , which shows a flow chart of a welding pool fusion monitoring method provided by one embodiment of the present invention, the method comprising the following steps: S101. Acquire ultrasonic scanning data of a welding pool by a single-point scanning method, where the welding pool includes at least one scanning point.

[0023] In a specific embodiment, Figure 2 As shown, Figure 2 Figure 1 is a schematic diagram of a weld pool. Weld pool fusion is a key process in the welding process where the welded material (base metal) and filler material (such as welding wire) are fully mixed in a high-temperature molten pool to form a dense connection. Its quality directly affects the mechanical properties (such as strength and toughness) and defect rate (such as lack of fusion, porosity, etc.) of the weld. Figure 2 As shown in the figure, the white circle is the welding pool generated during welding. As the welding path advances, the molten pool also advances.

[0024] The ultrasonic sensor collects relatively comprehensive information about the welding pool, and then analyzes the monitored information to obtain the fusion state of the base material and welding wire in the welding pool.

[0025] An ultrasonic sensor uses a piezoelectric chip installed inside the ultrasonic probe, which serves as the ultrasonic sensor for stimulating and receiving ultrasonic signals. During testing, the ultrasonic sensor first loads parameters and controls the probe to move to the starting scanning point. Once the probe is stationary, a sinusoidal wave excitation causes the chip to vibrate, generating pulsed ultrasonic waves. The reflected echo signal is received, converted to digital, and stored as the scan data for that scan point. The probe then moves point by point, repeating this process to generate ultrasonic scan data for each scan point.

[0026] S102, determine the first reflection time and the second reflection time according to the size parameter information of the welding pool, wherein the first reflection time is the time for the ultrasonic echo signal to reflect back to the surface of the welding pool, and the second reflection time is the time for the ultrasonic echo signal to reflect back to the bottom of the welding pool.

[0027] In a specific embodiment, the probe height can be set in advance, and then the expected time T1 (i.e., the first reflection time) for the ultrasonic signal to reach the surface of the welding pool and reflect back to the probe can be calculated according to the distance between the probe and the surface of the welding pool. Similarly, the distance between the probe and the bottom of the welding pool can be obtained by laser ranging, and then the expected time T2 (i.e., the second reflection time) for the ultrasonic signal to reach the bottom of the welding pool and reflect back to the probe can be calculated according to the distance between the probe and the bottom of the welding pool.

[0028] S103, determine other maximum points from the ultrasonic scanning data according to the first reflection time and the second reflection time.

[0029] In this embodiment, the other maximum points are determined from the ultrasonic scanning data according to the first reflection time and the second reflection time, which specifically includes: determining the local maximum points for the ultrasonic scanning data of each scanning point by the derivative method; arranging the local maximum points in ascending order of time to obtain a time ascending sequence; segmenting the time ascending sequence by using the otsu multi-threshold segmentation method to obtain extreme point segments; calculating the mean value of each extreme point segment, determining the wave peak of the elements in the first extreme point segment in the ultrasonic scanning data as the welding pool surface echo, and determining the wave peak of the elements in the second extreme point segment in the ultrasonic scanning data as the welding pool bottom echo, wherein the first extreme point segment is the extreme point segment with the smallest difference between the mean value and the first reflection time, and the second extreme point segment is the extreme point segment with the smallest difference between the mean value and the second reflection time; determining the local maximum points between the welding pool surface echo and the welding pool bottom echo in the ultrasonic scanning data as the other maximum points.

[0030] For example, if the fusion state of the base material and the welding wire in the welding pool is good, there are only two relatively obvious wave peaks in the echo signals of different scanning points, one wave peak represents the welding pool surface echo (the reflection wave of the interface between the solid base material and the liquid welding pool), and the other wave peak represents the welding pool bottom echo (the reflection wave of the interface between the welding pool and the solid welding seam).

[0031] If the fusion state is poor and defects exist, defect echoes will be generated, namely scattered waves caused by pores / unfused materials (appearing as time-delayed burr signals). Therefore, first compare multiple scanning points to obtain relatively fixed signals, which are recorded as the molten pool surface echo signal and the molten pool bottom echo signal.

[0032] For the scanning data of each scanning point (a curved sound wave), all local maximum points on the data are obtained by the derivative method. The local maximum point indicates that the sound wave has been reflected at the corresponding position in the molten pool, which may be caused by defects caused by incomplete fusion or by reflections from the surface and bottom of the molten pool.

[0033] Obtain the time point of each local maximum extreme point on the scanning data of all scanning points, obtain all time points, arrange all time points in ascending order to form an ascending sequence, and obtain multiple extreme point segments (i.e., extreme point segmentation segments) from the ascending sequence through the Otsu multi-threshold segmentation method, and the time points in the same extreme point segment are similar.

[0034] For each extreme point segment, the mean of all time points in the segment is calculated; then the time difference between all means and T1 is calculated in turn, and the peak of the corresponding time point in the scanning signal where the element in the extreme point segment with the smallest time difference is located is recorded as the molten pool surface echo; similarly, the time difference between each mean and T2 can be calculated, and the peak of the corresponding time point in the scanning signal where the element in the extreme point segment with the smallest time difference is located is recorded as the molten pool bottom echo.

[0035] like Figure 3 As shown, Figure 3 This is a schematic diagram of the maximum value point of the probe scanning data point by point. Figure 3 In the figure, point 1, point 2, and point 3 are the time points of the maximum value points in the echo signal obtained by scanning the probe point by point, and the dotted lines represent the time points. The straight line corresponding to T1 represents the T1 time point; the straight line corresponding to T2 represents the T2 time point.

[0036] The echoes corresponding to the other maximum points obtained at this time may include: defect echoes and echoes caused by the electronic noise of the ultrasonic probe. The amplitude of the echo caused by electronic noise is generally small, and then the defect echo is identified by the convexity of each other maximum point. The greater the convexity, the greater the probability that the other maximum point belongs to the defect echo.

[0037] S104. Determine the real defect point from other maximum value points.

[0038] In this embodiment, determining the real defect point from other maximum value points specifically includes: Determine an abnormal echo point from other maximum value points, and determine the ultrasonic scanning data of the abnormal echo point as an abnormal echo; acquire new ultrasound scanning data of each abnormal echo point, the new ultrasound scanning data being obtained by emitting sound waves of preset sound intensity intervals to each abnormal echo point, and sound intensity of the preset sound intensity intervals increasing in turn; determine a change trend value sequence according to the new ultrasound scanning data; determine an amplitude sequence and a time point sequence of the abnormal echo point in the multiple scanning data of each real defect candidate according to the change trend value sequence; determine a first real defect point and a suspected defect point according to the amplitude sequence and the time point sequence; determine a second real defect point from the suspected defect point; determine the first real defect point and the second real defect point as real defect points.

[0039] determine the abnormal echo point from other maximum value points, specifically comprising: determine an upper envelope line of the other maximum value points using a cubic spline interpolation method, and determine a fitting point of each point in the upper envelope line through a least square method, calculate an absolute value of a difference value between each fitting point and each point in the corresponding upper envelope line to obtain a difference absolute value, and calculate a ratio of the difference absolute value and each point in the corresponding upper envelope line to obtain a convexity value; determine an echo abnormality probability of each point in the upper envelope line according to the convexity value, and determine a point with an echo abnormality probability greater than a preset abnormal threshold as an abnormal echo point.

[0040] determine a change trend value sequence according to the new ultrasound scanning data, specifically comprising: for each abnormal echo point, acquire an amplitude corresponding to each sound intensity of each abnormal echo point in the new ultrasound scanning data to obtain an amplitude sequence; determine a change trend value sequence according to the amplitude sequence.

[0041] determine a change trend value sequence according to the amplitude sequence, specifically comprising: determine a principal component analysis coordinate by taking an element order of elements of the first n amplitude sequences as a horizontal coordinate and taking each element of each amplitude sequence as a vertical coordinate, determine a projection vector and a projection value corresponding to each projection vector by a principal component analysis method, and determine an inverse tangent angle value of a ratio of the vertical coordinate and the horizontal coordinate of the projection vector with the largest projection value as a first element in the change trend value sequence, wherein n is a preset constant; determine each element in the change trend value sequence in turn to obtain the change trend value sequence.

[0042] determine an amplitude sequence and a time point sequence of the abnormal echo point in the multiple scanning data of each real defect candidate according to the change trend value sequence, specifically comprising: The otsu multi-threshold segmentation method is used to segment the change trend value sequence, two change trend segmentation sections are determined, the variance of each change trend segmentation section is calculated, and the sum of the variances of the two change trend segmentation sections is obtained to obtain a pseudo-defect probability value; A real defect candidate probability value is determined according to the pseudo-defect probability value, and an abnormal echo point with a real defect candidate probability value greater than a preset defect threshold is determined as a real defect candidate; Multiple scan data of each real defect candidate are obtained, abnormal echo points in the multiple scan data are determined, and an amplitude sequence and a time point sequence of the abnormal echo points in each multiple scan data are obtained.

[0043] A first real defect point and a suspected defect point are determined according to the amplitude sequence and the time point sequence, and specifically include: The coefficient of variation of the amplitude sequence and the standard deviation of the time point sequence are calculated, the abnormal echo point with the coefficient of variation less than a preset first abnormal threshold and the standard deviation less than a preset second abnormal threshold is determined as the first real defect point, and the abnormal echo point in the multiple scan data, except for the first real abnormal point, is determined as the suspected defect point.

[0044] A second real defect point is determined from the suspected defect point, and specifically includes: Final scan data of the suspected defect point under different emission energies are obtained, and abnormal echo points in the final scan data are determined, the absolute value of the time difference between the time of the abnormal echo point in each final scan data and the first reflection time is calculated to obtain a time difference sequence of the abnormal echo point in each final scan data; The abnormal echo point with the time difference sequence less than a preset variance threshold is determined as the second real defect point.

[0045] Exemplarily, for other maximum value points in the scan data of each scan point, the adjacent local maximum value points are connected by a cubic spline interpolation method to form an upper envelope line.

[0046] For each point on the upper envelope line corresponding to the scan data of each scan point, the data before the point is fitted by a least square method to obtain a fitting equation, the time point of the point is input to obtain a fitting point n1 of the point, the absolute value of the difference between the fitting value n1 of the point and the actual value n2 of the point is calculated to obtain a difference absolute value, and the ratio of the difference absolute value to the actual value n2 is calculated to obtain a convexity value. The smaller the convexity value b is, the closer the fitting point is to the actual value, the smaller the convexity of the point is, and the smaller the abnormal probability of the echo of the point is.

[0047] The echo abnormal probability of each point in the upper envelope line is determined according to the convexity value, and the formula that can be used is:

[0048] wherein, represents a convexity value, represents a natural constant, represents an echo abnormality probability.

[0049] echo abnormality probability The greater the value, the greater the probability of the point corresponding to an echo abnormality. The points on the upper envelope line with a value greater than 0.5 are recorded as abnormal echoes.

[0050] Optionally, the preset abnormality threshold value can be set to 0.5, or can be set to other values, which can be set according to actual conditions, and is not specifically limited here.

[0051] Since the transmitted acoustic wave itself will affect the molten pool: for the surface echo, it will cause periodic fluctuations in the amplitude of the surface echo; for the bottom echo, it will cause the amplitude of the bottom echo to decrease, which is easy to misjudge the internal flow of the molten pool as a dynamic defect, such as a false gas hole signal. Therefore, the abnormal echoes obtained not only contain defect echoes, but also contain echoes that are normal but are identified as defects, and it is necessary to remove such echoes from the abnormal echoes to obtain defect echoes, and then analyze the defect echoes to obtain the cause of the defect, and then adjust the parameters to avoid the generation of defects, and obtain a better fusion state of the base material and the welding wire.

[0052] In this embodiment, different intensities of acoustic waves are transmitted to distinguish real defects and false defects. For real defects, for example: the physical characteristics of defects such as pores and unfusion are relatively stable, while ultrasonic disturbance artifacts such as molten pool surface vibration and acoustic flow effect will change significantly with the change of ultrasonic energy.

[0053] For each abnormal echo corresponding to a scanning point, acoustic waves with successively increasing acoustic wave intensities are transmitted at the position, and corresponding new ultrasonic scanning data are obtained in sequence. For the same abnormal echo position, the amplitudes of the position in different new ultrasonic scanning data are obtained, and an amplitude sequence is obtained in the order of successively increasing transmitted acoustic waves.

[0054] ​The first element of the change trend value sequence is obtained by taking the inverse tangent angle value of the ratio of the ordinate to the abscissa of the projection vector corresponding to the maximum projection length. The change trend values of the first 6, the first 7, and the like are obtained by the same method to form the change trend value sequence. In this embodiment, n is 5, and can also be set according to actual conditions, which is not specifically limited here.

[0055] In the change trend value sequence of the real defect, the amplitude of the real defect increases linearly or saturates with the increase of energy, the element values in the change trend sequence corresponding to the linear increase stage are similar, and the element values in the change trend sequence corresponding to the saturation stage are close to 0. The elements in the change trend sequence of the artifact signal differ greatly.

[0056] The otsu threshold segmentation method is used to obtain two change trend segmentation sections, the variances of the two change trend segmentation sections are calculated respectively, and then the sum is obtained to obtain the artifact probability value , The greater the artifact probability value is, the more likely it is that the corresponding scanning point is an artifact.

[0057] The formula used to determine the real defect candidate probability value according to the artifact probability value can be:

[0058] wherein, represents the real defect candidate probability value, represents the artifact probability value.

[0059] Since some artifacts (such as acoustic flow vortexes) may also exhibit amplitude saturation at high energy, the corresponding scanning point of is recorded as a real defect candidate, but needs to be verified whether it is a high-energy artifact.

[0060] Optionally, the preset defect threshold is set to 0.7 here, and can also be set according to actual conditions, which is not specifically limited here.

[0061] For the scanning point corresponding to the real defect candidate, the same position is scanned multiple times in this embodiment to further identify the real defect: the position and size of the real defect are fixed, and multiple frames of detection signals should be highly consistent; the artifact is affected by the dynamic of the molten pool, and the signal fluctuates greatly.

[0062] For each scan data corresponding to each real defect candidate, the position of the abnormal echo in each scan data is calculated, and the amplitude sequence and time point sequence of these positions are obtained.

[0063] The coefficient of variation v of the amplitude sequence is calculated, and the standard deviation m of the time point sequence is obtained. If v<0.15 and m<0.2, the corresponding scan point is a real defect point with high probability, and the points that do not meet the conditions are determined as suspected defect points.

[0064] Optionally, in the embodiment, the preset first abnormal threshold is set to 0.15, and the preset second abnormal threshold is set to 0.2. The preset first abnormal threshold and the preset second abnormal threshold can also be set according to actual conditions, which are not limited here.

[0065] In the obtained real defect candidate, the surface vibration of the molten pool may cause instability of the signal at the real defect point, and then in the embodiment, the interface position of the real defect is fixed, and the echo time does not change with the energy. The artifact may cause delay of the echo time due to the dynamic (vibration, acoustic flow) of the molten pool, and then the real defect point in the suspected defect point is identified by whether the echo time is delayed.

[0066] For the final scan data obtained at different emission energies of each suspected defect point corresponding to the scan point, the abnormal echo position in the final scan data is obtained, and then the corresponding time point is obtained. The time difference between each time point and the time point corresponding to the molten pool surface echo in the data is calculated to form a time difference sequence.

[0067] If the time difference changes little, it is a real defect with high probability, otherwise it is an artifact with high probability. The variance of each time difference sequence is calculated, and the suspected defect point with a variance less than 0.3 is recorded as a real defect point. Thus, all real defect points are obtained, and then the echo signal corresponding to each real defect point can be obtained.

[0068] Optionally, the preset variance threshold is set to 0.3 here, which can be set according to actual requirements, which is not limited here.

[0069] S105, input the echo signal feature of the real defect point into the trained long short-term memory network, output the defect type and fusion state through the trained long short-term memory network, and adjust the fusion parameters and optimization algorithm of the welding molten pool through the defect type and fusion state.

[0070] In one specific embodiment, the echo signal features (such as amplitude, time-domain waveform, spectral characteristics, etc.) corresponding to each real defect point can be extracted. Then, these defect echo signals are input into a trained deep learning neural network (such as a convolutional neural network CNN or a long short-term memory network LSTM) for defect type classification and fusion state evaluation (such as lack of penetration, porosity rate, un-melted area, etc.).

[0071] Based on the output results of the neural network, the dynamic adjustment scheme is generated by further combining a welding process parameter library (such as current, voltage, welding speed, etc.) and an optimization algorithm (such as reinforcement learning or genetic algorithm).

[0072] Thus far, the present application is completed.

[0073] In summary, in the embodiments of the present application, by analyzing the ultrasonic echo signals under different emission energies, combining the pool surface echo and the pool bottom echo, the pseudo-defect signals free from noise influence are obtained, the real defect scanning points are obtained, and then the signal features of the defect scanning points are input into the neural network, the defect type and the fusion state of the welding pool are identified, which is convenient for further analysis and adjustment.

[0074] The present application also proposes a welding pool fusion monitoring system, please refer to Figure 4 which shows a structural diagram of a welding pool fusion monitoring system provided by one embodiment of the present application, and the system comprises a data acquisition module 101, a data processing module 102 and a defect output module 103.

[0075] The data acquisition module 101 is used for acquiring ultrasonic scanning data of the welding pool by a single-point scanning method, and the welding pool comprises at least one scanning point; The data processing module 102 is used for determining a first reflection time and a second reflection time according to the size parameter information of the welding pool, wherein the first reflection time is the time for the ultrasonic echo signal to reflect back to the surface of the pool, and the second reflection time is the time for the ultrasonic echo signal to reflect back to the bottom of the pool; determining other maximum points from the ultrasonic scanning data according to the first reflection time and the second reflection time; and determining real defect points from the other maximum points; The data verification module 103 is used for inputting the echo signal features of the real defect points into a trained long short-term memory network, outputting the defect type and the fusion state through the trained long short-term memory network, and adjusting the fusion parameters of the welding pool and the optimization algorithm through the defect type and the fusion state.

[0076] It should be noted that the system provided in the above examples is only exemplified by the division of the above functional modules. In actual application, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the above described functions. In addition, the welding pool fusion monitoring system and the welding pool fusion monitoring method provided in the above examples belong to the same concept, and the specific implementation process is described in the method embodiment, which will not be repeated here.

[0077] It should be noted that the above-mentioned embodiments of the present application are only for description, and do not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or may be advantageous.

[0078] Each of the embodiments in the specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the difference from other embodiments.

[0079] The above only describes the preferred embodiments of the present application and does not limit the present application. Any modification, equivalent replacement, improvement, etc. within the principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for monitoring welding pool fusion, characterized in that: include: Acquiring ultrasonic scanning data of a weld pool by a single-point scanning method, wherein the weld pool includes at least one scanning point; Determining a first reflection time and a second reflection time based on the size parameter information of the welding molten pool, wherein the first reflection time is the time it takes for the ultrasonic echo signal to reflect back to the surface of the molten pool, and the second reflection time is the time it takes for the ultrasonic echo signal to reflect back to the bottom of the molten pool; determining other maximum points from the ultrasonic scanning data according to the first reflection time and the second reflection time; Determining a true defect point from the other maximum value points; The echo signal characteristics of the real defect point are input into a trained long short-term memory network, the defect type and fusion state are output through the trained long short-term memory network, and the fusion parameters and optimization algorithm of the welding pool are adjusted according to the defect type and the fusion state.

2. A welding pool fusion monitoring method according to claim 1, characterized in that: The determining other maximum points from the ultrasonic scanning data according to the first reflection time and the second reflection time specifically includes: For the ultrasonic scanning data of each scanning point, the local maximum point is determined by the derivative method; Arrange the local maximum points in ascending order of time to obtain a time ascending sequence; The time ascending sequence is segmented using the Otsu multi-threshold segmentation method to obtain extreme point segmentation segments; Calculating the mean of each extreme point segment, determining the peak of the elements in the first extreme point segment in the ultrasonic scanning data as the molten pool surface echo, and determining the peak of the elements in the second extreme point segment in the ultrasonic scanning data as the molten pool bottom echo, wherein the first extreme point segment is the extreme point segment with the smallest difference between the mean and the first reflection time, and the second extreme point segment is the extreme point segment with the smallest difference between the mean and the second reflection time; A local maximum point located between the molten pool surface echo and the molten pool bottom echo in the ultrasonic scanning data is determined as the other maximum point.

3. The method for monitoring welding pool fusion according to claim 1, characterized in that: Determining the real defect point from the other maximum value points specifically includes: Determining an abnormal echo point from the other maximum value points, and determining the ultrasonic scanning data of the abnormal echo point as an abnormal echo; Acquire new ultrasonic scanning data for each abnormal echo point, wherein the new ultrasonic scanning data is obtained by emitting a sound wave of a preset sound wave intensity range to each abnormal echo point, and the sound wave intensity of the preset sound wave intensity range increases sequentially; determining a change trend value sequence according to the newly added ultrasound scanning data; Determine the amplitude sequence and time point sequence of abnormal echo points in multiple scan data of each real defect candidate according to the change trend value sequence; Determining a first real defect point and a suspected defect point according to the amplitude sequence and the time point sequence; Determining a second real defect point from the suspected defect points; The first real defect point and the second real defect point are determined as the real defect points.

4. A welding pool fusion monitoring method according to claim 3, characterized in that: Determining the abnormal echo point from the other maximum value points specifically includes: Determine the upper envelope of the other maximum points using a cubic spline interpolation method, determine the fitting point of each point in the upper envelope using a least squares method, calculate the absolute value of the difference between each fitting point and the corresponding point in the upper envelope to obtain the absolute value of the difference, calculate the ratio of the absolute value of the difference to the corresponding point in the upper envelope to obtain a convexity value; The echo abnormality probability of each point in the upper envelope is determined according to the convexity value, and the point where the echo abnormality probability is greater than a preset abnormality threshold is determined as an abnormal echo point.

5. The method for monitoring welding pool fusion according to claim 3, characterized in that: Determining a change trend value sequence according to the newly added ultrasound scanning data specifically includes: For each abnormal echo point, obtaining the amplitude corresponding to each sound wave intensity of each abnormal echo point in the newly added ultrasonic scanning data to obtain an amplitude sequence; A change trend value sequence is determined according to the amplitude sequence.

6. A welding pool fusion monitoring method according to claim 5, characterized in that: Determining a change trend value sequence according to the amplitude sequence specifically includes: Using the order of elements corresponding to the first n elements of the amplitude sequence as the abscissa, and using the element of each amplitude sequence as the ordinate, determining the principal component analysis coordinate, determining the projection vector and the projection value corresponding to each projection vector by the principal component analysis method, and determining the arc tangent angle value of the ratio of the ordinate to the abscissa of the projection vector with the largest projection value as the first element in the change trend value sequence, wherein n is a preset constant; Each element in the change trend value sequence is determined in sequence to obtain the change trend value sequence.

7. The method for monitoring welding pool fusion according to claim 3, characterized in that: The step of determining the amplitude sequence and time point sequence of abnormal echo points in the multiple scan data of each real defect candidate according to the change trend value sequence specifically includes: Segmenting the change trend value sequence using the Otsu multi-threshold segmentation method, determining two change trend segmentation segments, calculating the variance of each change trend segmentation segment, and obtaining the sum of the variances of the two change trend segmentation segments to obtain a pseudo-defect probability value; Determine a true defect candidate probability value based on the pseudo defect probability value, and determine an abnormal echo point whose true defect candidate probability value is greater than a preset defect threshold as a true defect candidate; Acquire multiple scan data of each real defect candidate, determine abnormal echo points in the multiple scan data, and acquire amplitude sequences and time point sequences of each abnormal echo point in the multiple scan data.

8. The method for monitoring welding pool fusion according to claim 3, characterized in that: The determining of the first real defect point and the suspected defect point according to the amplitude sequence and the time point sequence specifically includes: Calculate the coefficient of variation of the amplitude sequence and the standard deviation of the time point sequence, determine the abnormal echo point whose coefficient of variation is less than a preset first abnormal threshold and whose standard deviation is less than a preset second abnormal threshold as a first true defect point, and determine the abnormal echo points in the multiple scanning data other than the first true abnormal point as suspected defect points.

9. The method for monitoring welding pool fusion according to claim 3, characterized in that: Determining a second real defect point from the suspected defect point specifically includes: Acquire final scanning data of the suspected defect point at different transmission energies, determine abnormal echo points in the final scanning data, calculate the absolute value of the time difference between the time of each abnormal echo point in the final scanning data and the first reflection time, and obtain a time difference sequence of each abnormal echo point in the final scanning data; The abnormal echo point whose time difference sequence is smaller than a preset variance threshold is determined as the second real defect point.

10. A welding pool fusion monitoring system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the computer program is executed by a processor, the steps of a welding pool fusion monitoring method according to any one of claims 1 to 9 are implemented.

Images