A method for online detection of laser welding penetration

Abstract

The present invention is an online detection method for laser welding penetration. The present invention relates to online detection and intelligent control of welding. The present invention selects a thermally excited state mesoscopic signal of a penetration characteristic area in a keyhole as a detection object, obtains a mesoscopic detection signal; preprocesses the mesoscopic detection signal to obtain computer-recognizable type data; calibrates the preprocessed data and its corresponding actual penetration state and divides them into a training set, a test set and a verification set; establishes a neural network model, trains the penetration state recognition model through the data in the training set until the result converges, and then adjusts the model hyperparameters through the verification set to obtain the optimal recognition model for penetration; verifies the reliability of the model through the test set, and performs online detection of laser welding penetration according to the optimal model.

Description

Technical Field

[0001] The invention relates to the technical field of welding online detection and intelligent control, and is an online detection method for laser welding penetration. Background Art

[0002] Laser and laser composite energy field welding is one of the mainstream technologies in the field of intelligent manufacturing. However, during the laser welding process, the keyhole will eject a large amount of smoke, spatter, welding plume and other substances, which will greatly interfere with the energy input of the laser. Therefore, laser and laser composite energy field welding often have unstable penetration quality control in engineering applications, which seriously restricts the development of this technology. Therefore, reliable online penetration detection and quality closed-loop control are one of the important issues in the field of laser welding intelligent manufacturing. However, the characteristic area most directly related to the weld penetration state is the metal vapor recoil pressure zone at the bottom of the laser keyhole. The diameter of its core area is only 0.05-0.4 mm, which belongs to the mesoscopic scale category, and is completely covered by a large amount of ejected materials and surrounding high-radiation signals, which has strong concealment. In addition, the extraction position and detection resolution of the penetration mesoscopic signal also have a great influence on the detection results. In addition, the changes in the mesoscopic signal in the keyhole are also extremely complex, without obvious regularity, and the amplitude changes violently and has strong interference. However, due to the limitations of macroscopic sampling methods, existing detection methods are unable to effectively separate the mesoscopic scale signals with penetration characteristics from other interfering signals such as the radiation signal from the inner wall of the keyhole, resulting in a very low proportion of effective signals in the detection data and increased difficulty in extracting penetration information. At the same time, since existing data analysis methods cannot effectively process complex and violently fluctuating mesoscopic signals, it is difficult to extract welding penetration characteristics and it is difficult to ensure the reliability of online detection.

[0003] These factors have had a great impact on existing detection technologies. The main technical problems are as follows: 1. Existing detection methods are limited by macroscopic sampling methods and cannot effectively obtain mesoscopic characteristic information; 2. Existing signal recognition methods are greatly disturbed by the strong fluctuations of signals in the keyhole, and detection reliability is difficult to guarantee. Summary of the invention

[0004] In order to overcome the shortcomings of the prior art, the present invention proposes an artificial intelligence online detection method for laser welding penetration based on online detection of keyhole mesoscopic features machine recognition, which uses array sensing to collect thermally excited state mesoscopic detection signals from the internal characteristic area of ​​the keyhole, establishes a neural network model through machine learning, and finally calls the trained recognition model to analyze signal characteristics online and extract welding penetration quality information.

[0005] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0006] The present invention provides a method for online detection of laser welding penetration, and the present invention provides the following technical solutions:

[0007] A penetration depth detection device, comprising: a housing, a substrate, a sensor, a narrow-band filter and a three-dimensional fine-tuning mechanism;

[0008] The connection between the substrate and the laser welding head serves to fix the entire device. There are mounting holes on the substrate to respectively fix the device housing, the three-dimensional fine-tuning mechanism and the narrow-band filter carrier. The three-dimensional fine-tuning mechanism is connected to the sensor to adjust the three-dimensional spatial position of the sensor sensing chip. The narrow-band filter is located on the optical path between the optical focusing lens group and the array sensor sensing surface. The optical real image after filtering can be projected onto the array sensor sensing surface.

[0009] A method for online detection of laser welding penetration, the method is based on a penetration detection device, and the method comprises the following steps:

[0010] Step 1: Select the thermally excited state mesoscopic signal of the characteristic melting depth area in the keyhole as the detection object, and obtain the mesoscopic detection signal;

[0011] Step 2: pre-process the mesoscopic detection signal to obtain computer-recognizable type data;

[0012] Step 3: According to the preprocessed data, the data is calibrated and divided into training set, test set and validation set;

[0013] Step 4: Build a neural network model, train the model's weight parameters using the training set data, and then adjust the model's hyperparameters using the validation set until the results converge;

[0014] Step 5: Test the reliability of the model through the test set;

[0015] Step 6: Call the trained recognition model to analyze signal features online and extract welding stability information.

[0016] Preferably, the thermally excited state mesoscopic signal in the characteristic area of ​​the penetration depth in the keyhole is selected as the detection object, and the clear real image of the thermally excited state signal at the bottom of the keyhole is projected onto the sensing surface of the array sensor by using the optical focusing imaging and spectral transmission principles to obtain the mesoscopic detection signal, and the sample data set for analysis is obtained and stored in association with the actual weld penetration depth. The recognition model is trained by machine learning methods, and finally the trained model is called to online identify the mesoscopic signal characteristics to obtain the current penetration depth status information.

[0017] Preferably, the thermally excited state signal at the bottom of the keyhole is a near-infrared signal generated by the laser beam entering the base material, causing the metal at the bottom of the keyhole to melt and evaporate rapidly through intense energy input, accompanied by high-density energy excitation.

[0018] Preferably, the method for collecting mesoscopic detection signals is:

[0019] S1. A high-power optical focusing lens group with at least 0.6-1.5m shooting working distance and 10mm shooting depth of field is used to extract a clear real image of the characteristic area at the bottom of the keyhole from the coaxial optical path of the laser welding head. The sufficiently large depth of field can capture a clear real image of the characteristic area in the fluctuating state without changing the focal length;

[0020] S2. In the near-infrared spectrum, the welding arc, plume, laser beam and other large amounts of welding radiation signals above the keyhole are effectively shielded by narrow-band filtering, so that the thermally excited state signals in the characteristic area can be effectively separated, thereby greatly reducing the proportion of invalid signals in the detection signal;

[0021] S3. Project the real image of the thermally excited state signal onto the sensing surface of an array sensor with a large sensing coverage area and high detection accuracy. The sensing area of ​​the array sensor should be ≥ the feature area to be measured, and the resolution accuracy should be ≤10μm, thereby obtaining mesoscopic signals at different positions in the feature area to be measured.

[0022] Preferably, the penetration state recognition model is obtained by:

[0023] The data within a period of time are superimposed and then averaged to obtain the trend characteristic data within this period of time. When analyzing the next frame of data, the first frame of data must be removed from the superimposed data and the current frame of data must be added to it. The superposition and average processing are then performed while keeping the total number of frames analyzed unchanged. This method obtains the change data set of the quasi-steady-state signal;

[0024] After calibrating the processed data set with the actual welding penetration characteristics, it is divided into training set, test set and validation set;

[0025] The weight parameters of the model are trained through the training set data until the results converge, and then the model hyperparameters are adjusted through the validation set. Finally, the reliability of the model is tested through the test set to obtain the optimal melting depth model.

[0026] Preferably, the single-channel / multi-channel / or all mesoscopic detection signal data collected online are first preprocessed and converted into computer-recognizable type data, and then the trained recognition model is called to perform calculations to obtain the current laser welding penetration value, provide online diagnostic results, or provide a basis for regulating key process parameters for the welding closed-loop control system, such as welding laser power, welding speed, and defocus.

[0027] A laser welding penetration online detection system, the system comprising:

[0028] A data acquisition module, wherein the data acquisition module selects a thermally excited state mesoscopic signal of a characteristic region of a melting depth in a keyhole as a detection object and obtains a mesoscopic detection signal;

[0029] A preprocessing module, wherein the preprocessing module preprocesses the mesoscopic detection signal to obtain computer-recognizable type data;

[0030] A calibration module, wherein the calibration module calibrates the preprocessed data and divides the data into a training set, a test set and a validation set;

[0031] A model building module, wherein the model building module builds a neural network model, trains the weight parameters of the model through the training set data until the results converge, and then adjusts the model hyperparameters through the validation set, and verifies the reliability of the model through the test set;

[0032] An online detection module calls a trained model to perform real-time analysis of the penetration state and extract welding stability information.

[0033] A computer-readable storage medium stores a computer program, which is executed by a processor to implement an online detection method for laser welding penetration.

[0034] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements an online detection method for laser welding penetration when executing the computer program.

[0035] The present invention has the following beneficial effects:

[0036] The present invention uses the thermally excited state signal of the characteristic area at the bottom of the keyhole as the detection signal. First, the generation position of the signal has a large correlation with the penetration depth of laser / laser composite energy field welding, so it can be used as a direct detection signal, which can avoid the influence of interference factors such as environmental humidity, temperature, gas flow field, etc. when using indirect detection signal detection. At the same time, the near-infrared spectrum enhancement characteristics of the signal also support the effective shielding effect of other harmful signals such as welding arc and plume in this spectrum, increase the proportion of effective signals in the detection signal, and reduce the difficulty of signal analysis.

[0037] The present invention proposes a method of projecting the real image of the thermally excited state signal onto the array sensor chip to obtain the mesoscopic signal, which can achieve full coverage recognition of the characteristic area at the bottom of the keyhole. Since the laser keyhole is always in a fluctuating state during the laser welding process, the position of the mesoscopic depth characteristic area will also swing accordingly, so full coverage recognition of the characteristic area at the bottom of the keyhole is very necessary. One of the characteristics of the present invention is that it can adaptively track and identify the depth characteristic area, and can accurately locate the detection signal of the key area of ​​depth. At the same time, the high-resolution characteristics of the array sensor can be used to perform high-resolution recognition of the target mesoscopic area, effectively shielding most of the interference signals and improving the detection reliability.

[0038] The present invention adopts a machine learning method to analyze the mesoscopic detection signal. Through a large amount of data analysis, it avoids individual signals and captures the regular characteristics of the signal. Therefore, it can accurately identify the trend characteristics of the mesoscopic characteristic signal at the bottom of the keyhole and effectively avoid the data analysis problems caused by the complexity of the welding signal, large volatility, and many interference signals.

[0039] The present invention adopts a signal preprocessing method of superimposing and averaging data within a period of time, which can accurately obtain the trend characteristic data within this period of time, thereby effectively ignoring the interference of rapid fluctuations in keyhole depth on penetration detection, and can make an effective judgment on the operating trend of welding penetration. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0041] Figure 1 Schematic diagram of the detection device assembly; 1. Shell of the penetration detection device; 2. Substrate; 3. Array or image sensor; 4. Narrow-band filter; 5. Three-dimensional fine-tuning mechanism

[0042] Figure 2is a flow chart of the method of the present invention;

[0043] Figure 3 This is a flow chart of the artificial intelligence detection method. DETAILED DESCRIPTION

[0044] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0046] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances. Specific embodiment one:

[0048] according to Figures 1 to 3 As shown, the specific optimized technical solution adopted by the present invention to solve the above technical problems is: the present invention relates to an online detection method for laser welding penetration.

[0049] A penetration depth detection device, comprising: a housing, a substrate, a sensor, a narrow-band filter and a three-dimensional fine-tuning mechanism;

[0050] The connection between the substrate and the laser welding head serves to fix the entire device. There are mounting holes on the substrate to respectively fix the device housing, the three-dimensional fine-tuning mechanism and the narrow-band filter carrier. The three-dimensional fine-tuning mechanism is connected to the sensor to adjust the three-dimensional spatial position of the sensor sensing chip. The narrow-band filter is located on the optical path between the optical focusing lens group and the array sensor sensing surface. The optical real image after filtering can be projected onto the array sensor sensing surface.

[0051] A method for online detection of laser welding penetration, the method is based on a penetration detection device, and the method comprises the following steps:

[0052] Step 1: Select the thermally excited state mesoscopic signal of the characteristic melting depth area in the keyhole as the detection object, and obtain the mesoscopic detection signal;

[0053] Step 2: pre-process the mesoscopic detection signal to obtain computer-recognizable type data;

[0054] Step 3: According to the preprocessed data, the data is calibrated and divided into training set, test set and validation set;

[0055] Step 4: Build a neural network model, train the model's weight parameters using the training set data until the results converge, and then adjust the model's hyperparameters using the validation set;

[0056] Step 5: Test the reliability of the model through the test set;

[0057] Step 6: Call the trained recognition model to analyze signal features online and extract welding stability information. Specific embodiment 2:

[0059] The difference between the second embodiment of the present application and the first embodiment is that:

[0060] When the thermally excited state mesoscopic signal of the characteristic melting depth area in the keyhole is selected as the detection object, the specific results are as follows:

[0061] The laser welding penetration online detection method based on the machine recognition of keyhole mesoscopic features is characterized by: proposing an artificial intelligence online detection method that uses array sensing to collect the thermally excited state mesoscopic detection signal of the characteristic area inside the keyhole, and establishing a neural network model through machine learning, and finally calling the trained recognition model to analyze the signal characteristics online and extract the welding penetration quality information. The specific steps are as follows:

[0062] Firstly, the thermally excited state mesoscopic signal in the characteristic melting depth area in the keyhole is selected as the detection object. Secondly, the clear real image of the thermally excited state signal at the bottom of the keyhole is projected onto the sensing surface of the array sensor by using the optical focusing imaging and spectral transmission principle to obtain the mesoscopic detection signal. Then, the model established by the machine learning method is used to identify the characteristics of the mesoscopic signal and obtain the current melting depth state information.

[0063] The thermally excited state signal at the bottom of the keyhole is a near-infrared signal generated by the laser beam entering the base material, which causes the metal at the bottom of the keyhole to melt and evaporate rapidly through intense energy input, accompanied by high-density energy excitation. Since the keyhole is formed under the recoil pressure of metal evaporation, and the thermally excited state signal happens to be generated in the main recoil pressure zone where the keyhole is formed, the strength of the signal can effectively reflect the pressure conditions in the recoil pressure zone, and its trend change can be well consistent with the keyhole depth, thereby being able to reflect the changes in the actual penetration depth of the weld.

[0064] The method for collecting mesoscopic detection signals is as follows: first, a high-power optical focusing lens group with a shooting working distance of at least 0.6-1.5m and a shooting depth of field of 10mm is used to extract a clear real image of the characteristic area at the bottom of the keyhole from the coaxial optical path of the laser welding head. A sufficiently large depth of field can capture a clear real image of the characteristic area under a fluctuating state without changing the focal length. Secondly, in the near-infrared spectrum, the welding plume, laser beam and other large amounts of welding radiation signals above the keyhole are effectively shielded by narrow-band filtering, so that the thermally excited state signals in the characteristic area can be effectively separated, thereby greatly reducing the noise in the detection signal. Then, the real image of the thermally excited state signal is projected onto the sensing surface of an array sensor with a large sensing coverage area and high detection accuracy. The sensing area of ​​the array sensor should be ≥ the feature area to be measured, and the resolution accuracy should be ≤10μm. In this way, the mesoscopic signals at different positions in the feature area to be measured are obtained. This method can not only obtain the thermal excitation signals at all positions in the area to be measured and accurately analyze the position of the key characteristic signal of the melting depth, but also directly extract the mesoscopic signals at the key positions to further increase the proportion of effective signals in the detection data, reduce the amount of data for signal analysis, and provide data guarantee for the next step of detection signal analysis.

[0065] The method for establishing the penetration state recognition model is proposed based on the fluctuation characteristics of the penetration signal. Since the penetration signal is in a state of rapid up and down fluctuations as the keyhole depth changes during the welding process, and the distribution trend of the signal fluctuation in each smaller time period is directly related to the welding penetration, the penetration signal is a quasi-steady-state signal with trend characteristics. It is different from the need to accurately identify each transient feature when identifying the transient signal. When analyzing the quasi-steady-state signal, it is necessary to ignore the transient features and make an effective judgment on the running trend of the signal. Therefore, when training the penetration state recognition model, first, when pre-processing the penetration signal, it is necessary to first superimpose the data within a period of time and then homogenize it to obtain the trend characteristic data within this period of time, and then when analyzing the next frame of data, the first frame of data must be removed from the superimposed data and the current frame of data must be added to it, and then superimposed and homogenized while keeping the total number of frames analyzed unchanged, so as to obtain the change data set of the quasi-steady-state signal. Secondly, after the processed data set is calibrated with the actual welding penetration characteristics, it is divided into a training set, a test set and a verification set. Secondly, a neural network model for regression tasks is constructed using a computer. The penetration state recognition model is trained using the data in the training set until the results converge. The model hyperparameters are adjusted using the validation set. The reliability of the model is tested using the test set. The trained recognition model is called to analyze signal features online and extract welding stability information.

[0066] The method for real-time detection of penetration depth characteristics is to first perform the signal preprocessing on the single-channel / multi-channel / or all mesoscopic detection signal data collected online and convert them into computer-recognizable type data, and then call the trained recognition model to perform calculations to obtain the current laser welding penetration depth value, provide online diagnostic results, or provide a basis for regulating key process parameters for the welding closed-loop control system, such as welding laser power, welding speed, defocus, etc.

[0067] The method for establishing the penetration state recognition model is to convert the analysis sample data collected by the array sensor or image sensor into computer-recognizable type data after certain data processing, and then calibrate the analysis sample with the actual welding penetration characteristics and divide it into a training set, a test set and a verification set, and then use a computer to build a neural network model, train the penetration recognition model with the data in the training set until the result converges, adjust the model hyperparameters through the verification set, verify the reliability of the model through the test set, and call the trained recognition model to analyze the signal characteristics online and extract welding stability information.

[0068] The present invention uses the thermally excited state signal of the characteristic area at the bottom of the keyhole as the detection signal. First, the generation position of the signal has a large correlation with the penetration depth of laser / laser composite energy field welding, so it can be used as a direct detection signal, which can avoid the influence of interference factors such as environmental humidity, temperature, gas flow field, etc. when using indirect detection signal detection. At the same time, the near-infrared spectrum enhancement characteristics of the signal also support the effective shielding effect of other harmful signals such as welding arc and plume in this spectrum, increase the proportion of effective signals in the detection signal, and reduce the difficulty of signal analysis.

[0069] The present invention proposes a method of projecting the real image of the thermally excited state signal onto the array sensor chip to obtain the mesoscopic signal, which can achieve full coverage recognition of the characteristic area at the bottom of the keyhole. Since the laser keyhole is always in a fluctuating state during the laser welding process, the position of the mesoscopic depth characteristic area will also swing accordingly, so full coverage recognition of the characteristic area at the bottom of the keyhole is very necessary. One of the characteristics of the present invention is that it can adaptively track and identify the depth characteristic area, and can accurately locate the detection signal of the key area of ​​depth. At the same time, the high-resolution characteristics of the array sensor can be used to perform high-resolution recognition of the target mesoscopic area, effectively shielding most of the interference signals and improving the detection reliability.

[0070] The present invention adopts a machine learning method to analyze the mesoscopic detection signal. Through a large amount of data analysis, it avoids individual signals and captures the regular characteristics of the signal. Therefore, it can accurately identify the trend characteristics of the mesoscopic characteristic signal at the bottom of the keyhole and effectively avoid the data analysis problems caused by the complexity of the welding signal, large volatility, and many interference signals.

[0071] The present invention adopts a signal preprocessing method of superimposing and averaging data within a period of time, which can accurately obtain the trend characteristic data within this period of time, thereby effectively ignoring the interference of rapid fluctuations in keyhole depth on penetration detection, and can make an effective judgment on the operating trend of welding penetration. Specific embodiment three:

[0073] The difference between the third embodiment of the present application and the second embodiment is that:

[0074] The present invention provides an online detection system for laser welding penetration, the system comprising:

[0075] A data acquisition module, wherein the data acquisition module selects a thermally excited state mesoscopic signal of a characteristic region of a melting depth in a keyhole as a detection object and obtains a mesoscopic detection signal;

[0076] A preprocessing module, wherein the preprocessing module preprocesses the mesoscopic detection signal to obtain computer-recognizable type data;

[0077] A calibration module, wherein the calibration module calibrates the preprocessed data and divides the data into a training set, a test set and a validation set;

[0078] A model building module, wherein the model building module builds a neural network model, trains the weight parameters of the model through the training set data until the results converge, and then adjusts the model hyperparameters through the validation set, and verifies the reliability of the model through the test set;

[0079] An online detection module calls a trained model to perform real-time analysis of the penetration state and extract welding stability information. Specific embodiment four:

[0081] The difference between the fourth embodiment of the present application and the third embodiment is that:

[0082] The present invention provides a computer-readable storage medium on which a computer program is stored. The program is executed by a processor to implement an online detection method for laser welding penetration.

[0083] The specific steps are:

[0084] Firstly, the thermally excited state mesoscopic signal in the characteristic melting depth area in the keyhole is selected as the detection object. Secondly, the clear real image of the thermally excited state signal at the bottom of the keyhole is projected onto the sensing surface of the array sensor by using the optical focusing imaging and spectral transmission principle to obtain the mesoscopic detection signal. Then, the model established by the machine learning method is used to identify the characteristics of the mesoscopic signal and obtain the current melting depth state information.

[0085] The thermally excited state signal at the bottom of the keyhole is a near-infrared signal generated by the laser beam entering the base material, which causes the metal at the bottom of the keyhole to melt and evaporate rapidly through intense energy input, accompanied by high-density energy excitation. Since the keyhole is formed under the recoil pressure of metal evaporation, and the thermally excited state signal happens to be generated in the main recoil pressure zone where the keyhole is formed, the strength of the signal can effectively reflect the pressure conditions in the recoil pressure zone, and its trend change can be well consistent with the keyhole depth, thereby being able to reflect the changes in the actual penetration depth of the weld.

[0086] The method for collecting mesoscopic detection signals is as follows: first, a high-power optical focusing lens group with a shooting working distance of at least 0.6-1.5m and a shooting depth of field of 10mm is used to extract a clear real image of the characteristic area at the bottom of the keyhole from the coaxial optical path of the laser welding head. A sufficiently large depth of field can capture a clear real image of the characteristic area under a fluctuating state without changing the focal length. Secondly, in the near-infrared spectrum, the welding plume, laser beam and other large amounts of welding radiation signals above the keyhole are effectively shielded by narrow-band filtering, so that the thermally excited state signals in the characteristic area can be effectively separated, thereby greatly reducing the noise in the detection signal. Then, the real image of the thermally excited state signal is projected onto the sensing surface of an array sensor with a large sensing coverage area and high detection accuracy. The sensing area of ​​the array sensor should be ≥ the feature area to be measured, and the resolution accuracy should be ≤10μm. In this way, the mesoscopic signals at different positions in the feature area to be measured are obtained. This method can not only obtain the thermal excitation signals at all positions in the area to be measured and accurately analyze the position of the key characteristic signal of the melting depth, but also directly extract the mesoscopic signals at the key positions to further increase the proportion of effective signals in the detection data, reduce the amount of data for signal analysis, and provide data guarantee for the next step of detection signal analysis.

[0087] The method for establishing the penetration state recognition model is proposed based on the fluctuation characteristics of the penetration signal. Since the penetration signal is in a state of rapid up and down fluctuations as the keyhole depth changes during the welding process, and the distribution trend of the signal fluctuation in each smaller time period is directly related to the welding penetration, the penetration signal is a quasi-steady-state signal with trend characteristics. It is different from the need to accurately identify each transient feature when identifying the transient signal. When analyzing the quasi-steady-state signal, it is necessary to ignore the transient features and make an effective judgment on the running trend of the signal. Therefore, when training the penetration state recognition model, first, when pre-processing the penetration signal, it is necessary to first superimpose the data within a period of time and then homogenize it to obtain the trend characteristic data within this period of time, and then when analyzing the next frame of data, the first frame of data must be removed from the superimposed data and the current frame of data must be added to it, and then superimposed and homogenized while keeping the total number of frames analyzed unchanged, so as to obtain the change data set of the quasi-steady-state signal. Secondly, after the processed data set is calibrated with the actual welding penetration characteristics, it is divided into a training set, a test set and a verification set. Secondly, a neural network model for regression tasks is constructed using a computer. The penetration state recognition model is trained using the data in the training set until the results converge. The model hyperparameters are adjusted using the validation set. The reliability of the model is tested using the test set. The trained recognition model is called to analyze signal features online and extract welding stability information.

[0088] The method for real-time detection of penetration depth characteristics is to first perform the signal preprocessing on the single-channel / multi-channel / or all mesoscopic detection signal data collected online and convert them into computer-recognizable type data, and then call the trained recognition model to perform calculations to obtain the current laser welding penetration depth value, provide online diagnostic results, or provide a basis for regulating key process parameters for the welding closed-loop control system, such as welding laser power, welding speed, defocus, etc.

[0089] The method for establishing the penetration state recognition model is to convert the analysis sample data collected by the array sensor or image sensor into computer-recognizable type data after certain data processing, and then calibrate the analysis sample with the actual welding penetration characteristics and divide it into a training set, a test set and a verification set, and then use a computer to build a neural network model, train the penetration recognition model with the data in the training set until the result converges, adjust the model hyperparameters through the verification set, verify the reliability of the model through the test set, and call the trained recognition model to analyze the signal characteristics online and extract welding stability information. Specific embodiment five:

[0091] The difference between the fifth embodiment of the present application and the fourth embodiment is that:

[0092] The present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements an online detection method for laser welding penetration when executing the computer program.

[0093] The specific steps are:

[0094] Firstly, the thermally excited state mesoscopic signal in the characteristic melting depth area in the keyhole is selected as the detection object. Secondly, the clear real image of the thermally excited state signal at the bottom of the keyhole is projected onto the sensing surface of the array sensor by using the optical focusing imaging and spectral transmission principle to obtain the mesoscopic detection signal. Then, the model established by the machine learning method is used to identify the characteristics of the mesoscopic signal and obtain the current melting depth state information.

[0095] The thermally excited state signal at the bottom of the keyhole is a near-infrared signal generated by the laser beam entering the base material, which causes the metal at the bottom of the keyhole to melt and evaporate rapidly through intense energy input, accompanied by high-density energy excitation. Since the keyhole is formed under the recoil pressure of metal evaporation, and the thermally excited state signal happens to be generated in the main recoil pressure zone where the keyhole is formed, the strength of the signal can effectively reflect the pressure conditions in the recoil pressure zone, and its trend change can be well consistent with the keyhole depth, thereby being able to reflect the changes in the actual penetration depth of the weld.

[0096] The method for collecting mesoscopic detection signals is as follows: first, a high-power optical focusing lens group with a shooting working distance of at least 0.6-1.5m and a shooting depth of field of 10mm is used to extract a clear real image of the characteristic area at the bottom of the keyhole from the coaxial optical path of the laser welding head. A sufficiently large depth of field can capture a clear real image of the characteristic area under a fluctuating state without changing the focal length. Secondly, in the near-infrared spectrum, the welding plume, laser beam and other large amounts of welding radiation signals above the keyhole are effectively shielded by narrow-band filtering, so that the thermally excited state signals in the characteristic area can be effectively separated, thereby greatly reducing the noise in the detection signal. Then, the real image of the thermally excited state signal is projected onto the sensing surface of an array sensor with a large sensing coverage area and high detection accuracy. The sensing area of ​​the array sensor should be ≥ the feature area to be measured, and the resolution accuracy should be ≤10μm. In this way, the mesoscopic signals at different positions in the feature area to be measured are obtained. This method can not only obtain the thermal excitation signals at all positions in the area to be measured and accurately analyze the position of the key characteristic signal of the melting depth, but also directly extract the mesoscopic signals at the key positions to further increase the proportion of effective signals in the detection data, reduce the amount of data for signal analysis, and provide data guarantee for the next step of detection signal analysis.

[0097] The method for establishing the penetration state recognition model is proposed based on the fluctuation characteristics of the penetration signal. Since the penetration signal is in a state of rapid up and down fluctuations as the keyhole depth changes during the welding process, and the distribution trend of the signal fluctuation in each smaller time period is directly related to the welding penetration, the penetration signal is a quasi-steady-state signal with trend characteristics. It is different from the need to accurately identify each transient feature when identifying the transient signal. When analyzing the quasi-steady-state signal, it is necessary to ignore the transient features and make an effective judgment on the running trend of the signal. Therefore, when training the penetration state recognition model, first, when pre-processing the penetration signal, it is necessary to first superimpose the data within a period of time and then homogenize it to obtain the trend characteristic data within this period of time, and then when analyzing the next frame of data, the first frame of data must be removed from the superimposed data and the current frame of data must be added to it, and then superimposed and homogenized while keeping the total number of frames analyzed unchanged, so as to obtain the change data set of the quasi-steady-state signal. Secondly, after the processed data set is calibrated with the actual welding penetration characteristics, it is divided into a training set, a test set and a verification set. Secondly, a neural network model for regression tasks is constructed using a computer. The penetration state recognition model is trained using the data in the training set until the results converge. The model hyperparameters are adjusted using the validation set. The reliability of the model is tested using the test set. The trained recognition model is called to analyze signal features online and extract welding stability information.

[0098] The method for real-time detection of penetration depth characteristics is to first perform the signal preprocessing on the single-channel / multi-channel / or all mesoscopic detection signal data collected online and convert them into computer-recognizable type data, and then call the trained recognition model to perform calculations to obtain the current laser welding penetration depth value, provide online diagnostic results, or provide a basis for regulating key process parameters for the welding closed-loop control system, such as welding laser power, welding speed, defocus, etc.

[0099] The method for establishing the penetration state recognition model is to convert the analysis sample data collected by the array sensor or image sensor into computer-recognizable type data after certain data processing, and then calibrate the analysis sample with the actual welding penetration characteristics and divide it into a training set, a test set and a verification set, and then use a computer to build a neural network model, train the penetration recognition model with the data in the training set until the result converges, adjust the model hyperparameters through the verification set, verify the reliability of the model through the test set, and call the trained recognition model to analyze the signal characteristics online and extract welding stability information.

[0100] In the description of this specification, the description of reference terms such as "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples without contradiction. In addition, the terms "first" and "second" are used only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, the features defined as "first" and "second" can explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "N" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined. Any process or method description in the flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code including one or more executable instructions for implementing the steps of a custom logic function or process, and the scope of the preferred embodiment of the present invention includes other implementations, in which the functions may not be performed in the order shown or discussed, including in a substantially simultaneous manner or in a reverse order according to the functions involved, which should be understood by a person skilled in the art of the art to which the embodiments of the present invention belong. The logic and / or steps represented in the flowchart or otherwise described herein, for example, may be considered as a sequenced list of executable instructions for implementing the logic function, and may be specifically implemented in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" may be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection having one or N wirings (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM).In addition, the computer readable medium can even be paper or other suitable medium on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, then editing, interpreting or processing in other suitable ways as necessary, and then storing it in a computer memory. It should be understood that the various parts of the present invention can be implemented with hardware, software, firmware, or a combination thereof. In the above-mentioned embodiment, N steps or methods can be implemented with software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented with hardware, as in another embodiment, any one of the following technologies known in the art or their combination can be used to implement: a discrete logic circuit with a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit with a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0101] The above is only a preferred implementation of a method for online detection of laser welding penetration. The protection scope of a method for online detection of laser welding penetration is not limited to the above embodiment. All technical solutions under this idea belong to the protection scope of the present invention. It should be pointed out that for those skilled in the art, several improvements and changes without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A method for online detection of laser welding penetration, the method being based on a penetration detection device, the device comprising: Housing, substrate, sensor, narrow band filter and three-dimensional fine-tuning mechanism; The substrate is connected with the laser welding head to fix the whole device. The substrate is provided with mounting holes to respectively fix the device housing, the three-dimensional fine-tuning mechanism and the narrow-band filter carrier. The three-dimensional fine-tuning mechanism is connected with the sensor to adjust the three-dimensional spatial position of the sensor sensing chip. The narrow-band filter is located on the optical path between the optical focusing lens group and the array sensor sensing surface. The optical real image after filtering can be projected on the array sensor sensing surface. The method is characterized in that: the method comprises the following steps: Step 1: Select the thermally excited state mesoscopic signal of the characteristic melting depth area in the keyhole as the detection object, and obtain the mesoscopic detection signal; Step 2: pre-process the mesoscopic detection signal to obtain computer-recognizable type data; Step 3: According to the preprocessed data, the data is calibrated and divided into training set, test set and validation set; Step 4: Build a neural network model, train the model's weight parameters using the training set data until the results converge, and then adjust the model's hyperparameters using the validation set; Step 5: Test the reliability of the model through the test set; Step 6: Call the trained recognition model to analyze the signal characteristics online and extract the welding stability information; The acquisition method of mesoscopic detection signal is: S1. A high-power optical focusing lens group with at least 0.6-1.5m shooting working distance and 10mm shooting depth of field is used to extract a clear real image of the characteristic area at the bottom of the keyhole from the coaxial optical path of the laser welding head. The sufficiently large depth of field can capture a clear real image of the characteristic area in the fluctuating state without changing the focal length; S2. In the near-infrared spectrum, the welding arc, plume, laser beam and other large amounts of welding radiation signals above the keyhole are effectively shielded by narrow-band filtering, so that the thermally excited state signals in the characteristic area can be effectively separated, thereby greatly reducing the proportion of invalid signals in the detection signal; S3. Project the real image of the thermally excited state signal onto the sensing surface of an array sensor with a large sensing coverage area and high detection accuracy. The sensing area of ​​the array sensor should be ≥ the feature area to be measured, and the resolution accuracy should be ≤10μm, thereby obtaining mesoscopic signals at different positions in the feature area to be measured.

2. The method according to claim 1, characterized in that: The thermally excited state mesoscopic signal in the characteristic area of ​​the keyhole's penetration depth is selected as the detection object. The clear real image of the thermally excited state signal at the bottom of the keyhole is projected onto the sensing surface of the array sensor by using the optical focusing imaging and spectral transmission principles to obtain the mesoscopic detection signal. The sample data set for analysis is obtained and stored in association with the actual weld penetration depth. The recognition model is trained by machine learning methods, and finally the trained model is called to online identify the mesoscopic signal characteristics to obtain the current penetration depth status information.

3. The method according to claim 2, characterized in that: The thermally excited state signal at the bottom of the keyhole is a near-infrared signal generated when the laser beam enters the base material and the metal at the bottom of the keyhole is rapidly melted and evaporated through intense energy input, accompanied by high-density energy excitation.

4. The method according to claim 1, characterized in that: The identification model is specifically: The data within a period of time are superimposed and then averaged to obtain the trend characteristic data within this period of time. When analyzing the next frame of data, the first frame of data must be removed from the superimposed data and the current frame of data must be added to it. The superposition and average processing are then performed while keeping the total number of frames analyzed unchanged. This method obtains the change data set of the quasi-steady-state signal; After calibrating the processed data set with the actual welding penetration characteristics, it is divided into training set, test set and validation set; The weight parameters of the model are trained through the training set data until the results converge, and then the model hyperparameters are adjusted through the validation set. Finally, the reliability of the model is tested through the test set to obtain the optimal melting depth model.

5. The method according to claim 4, characterized in that: The single-channel / multi-channel / or all mesoscopic detection signal data collected online are first converted into computer-recognizable type data through the signal preprocessing, and then the trained recognition model is called to perform calculations to obtain the current laser welding penetration value, provide online diagnosis results, or provide a basis for regulating key process parameters for the welding closed-loop control system, such as welding laser power, welding speed, and defocus.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the method according to any one of claims 1 to 5.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 5 is implemented.

Images