An Online Detection Method and System for Laser Welding Quality

Abstract

The present invention discloses an on-line detection method and system for laser welding quality, which relates to the technical field of laser welding quality detection. The method includes emitting laser and active infrared light to a welding workpiece, and the welding workpiece generates radiation light and active reflected light signals; decomposing the radiation light into light signals of different bands, and converting them into corresponding electrical signals together with the active reflected light signals; after preprocessing the electrical signals, performing empirical mode decomposition to obtain a corresponding number of IMF components and residual components; using the welding quality classification results of the welding workpiece together with the above-mentioned light signals, IMF components and residual components to jointly train a welding quality classifier, and using the trained welding quality classifier to detect the welding quality of the workpiece to be welded. The present invention overcomes the disadvantages of visual detection being easily interfered by strong reflected light and having a complex processing algorithm, and discriminates the stability of the welding process in real time through the state changes of the radiation light and active reflected light signals, realizing rapid and accurate discrimination of welding quality.

Description

Technical Field

[0001] The present invention relates to the technical field of laser welding quality detection, and more specifically, to an online detection method and system for laser welding quality. Background Art

[0002] Laser welding is a precise and efficient welding method using a high-energy density laser beam as a heat source. Its main process is that a high-energy density laser beam is emitted by a laser, transmitted through an optical fiber, and focused on the surface of the workpiece material by a lens. Then, the workpiece material absorbs the laser energy, locally melts or even vaporizes, and then transfers heat to the inside of the workpiece by heat conduction to form a molten pool, and radiates multiple signals such as plasma, metal vapor, radiation light signal, and radiation sound signal. A large number of studies have shown that the quality of laser welding is closely related to the signals generated during the above welding process. When welding defects such as pores, slag inclusions, incomplete penetration, lack of fusion, cracks, pits, undercuts, and weld beads are formed during the welding process, the above signals will also exhibit different characteristics. The detection of the laser welding process has always attracted the attention of researchers. Various sensors such as acoustic sensors, photodiodes, spectrometers, X-rays, and vision sensors are used in the detection of the laser welding process. During the laser welding process, collecting the radiation light signal as the detection original signal has a wider universality; for the photoelectric sensor, it has the advantages of response time and can identify colors. By collecting the radiation light signal and the reflected light signal of the active light during the laser processing process, and then through the decomposition and conversion processing of the above signals, the characterization data reflecting the welding quality during the laser processing process can be obtained; finally, through the discrimination system for discrimination, the online real-time detection of the laser welding process is realized to ensure the quality of laser welding. At present, most of the commonly used welding process detections use a high-speed camera to photograph the welding molten pool and observe the state change of the molten pool to judge the stability of the welding state. However, the high-speed camera used in this method is expensive, and the strong reflected light interference during the laser welding process will have a great impact on image acquisition. Online image acquisition occupies a large number of computer data input channels, and the corresponding image processing algorithm is complex and time-consuming, which is not conducive to online real-time transmission of feedback signals to the computer to control the process parameters of the laser beam.

[0003] The prior art discloses an online detection method for laser welding penetration. According to the special state characteristics when the laser welding penetration changes, by sequentially adopting three processing methods of special spectral band signal separation, specific welding area signal separation, and welding defect probability distribution characteristic signal separation for the conventional optical mixed signal, the severe interference in the mixed signal is effectively removed, and the reliable welding penetration characteristic information is simplified and separated layer by layer to realize the accurate online identification of laser welding penetration. The detection steps of this application are complex, the detection time is long, it is easily interfered by strong reflected light, and the detection result is inaccurate. Summary of the Invention

[0004] To overcome the defect that the prior art is vulnerable to strong reflected light interference, resulting in inaccurate detection of welding quality, the present invention provides an on-line detection method and system for laser welding quality. By collecting the optical signals during the welding process, converting the optical signals into electrical signals and then performing signal processing, accurate detection of welding quality is achieved; moreover, the amount of signal data to be processed is small, the time consumed for signal processing is short, and the detection time is shortened.

[0005] To solve the above technical problems, the technical solution of the present invention is as follows:

[0006] The present invention provides an on-line detection method for laser welding quality, including:

[0007] S1: Emitting laser light to the welding workpiece, and the welding workpiece absorbs the laser light to generate radiation light; emitting active infrared light to the welding workpiece, and the welding workpiece generates an active reflected light signal;

[0008] S2: Decomposing the radiation light into several optical signals of different bands;

[0009] S3: Converting the several optical signals of different bands and the active reflected light signal into corresponding electrical signals;

[0010] S4: Preprocessing the electrical signals to obtain preprocessed electrical signals;

[0011] S5: Respectively performing empirical mode decomposition on the preprocessed electrical signals to obtain several IMF components and residual components corresponding to each preprocessed electrical signal;

[0012] S6: Obtaining the morphological characteristics of the welding workpiece after laser welding, and classifying the welding quality of the welding workpiece according to the morphological characteristics after welding to obtain a welding quality classification result;

[0013] S7: Composing the several optical signals of different bands and the active reflected light signal and their corresponding several IMF components and residual components into a feature data set, and inputting the feature data set and the welding quality classification result of the welding workpiece into a preset welding quality classifier for training to obtain a trained welding quality classifier;

[0014] S8: Obtaining the several optical signals of different bands and the active reflected light signal of the workpiece to be welded during the welding process, and inputting them into the trained welding quality classifier for detection to obtain a welding quality discrimination result of the workpiece to be welded.

[0015] Preferably, the several optical signals of different bands include visible light signals, laser reflected light signals and infrared light signals; among them, the band of the visible light signal is 390-780 nanometers, the band of the laser reflected light signal is 780-1100 nanometers, and the band of the infrared light signal is greater than 1100 nanometers.

[0016] Preferably, in the step S4, the electrical signal is preprocessed to obtain a preprocessed electrical signal. The specific method is as follows:

[0017] The visible light electrical signal, the laser reflection electrical signal, the infrared electrical signal, and the active reflection electrical signal are all sequentially subjected to a pre-amplification operation, a filtering operation, and a main amplification operation to obtain the preprocessed visible light electrical signal, laser reflection electrical signal, infrared electrical signal, and active reflection electrical signal.

[0018] The visible light electrical signal, the laser reflection electrical signal, the infrared electrical signal, and the active reflection electrical signal are all very weak and will be buried in the noise signal. First, a pre-amplification operation is performed on the visible light electrical signal, the laser reflection electrical signal, the infrared electrical signal, and the active reflection electrical signal, and the included noise signal is amplified at the same time, which is beneficial to filtering out the noise signal cleanly during the filtering operation. After the filtering operation filters out the noise signal, the pure visible light electrical signal, laser reflection electrical signal, infrared electrical signal, and active reflection electrical signal are amplified again to a voltage amplitude that meets the requirements of subsequent step processing.

[0019] Preferably, the specific steps of the step S5 are as follows:

[0020] The visible light electrical signal, the laser reflection electrical signal, the infrared electrical signal, and the active reflection electrical signal are sequentially used as the original voltage signal sequence, and the following steps are executed;

[0021] S5.1: Calculate all the extreme points of the original voltage signal sequence, fit the lower envelope at the minimum extreme points and the upper envelope at the maximum extreme points, and calculate the mean value of the upper and lower envelopes;

[0022] S5.2: Subtract the mean value of the upper and lower envelopes from the original voltage signal sequence to obtain a new voltage signal sequence;

[0023] S5.3: Determine whether the new voltage signal sequence meets the IMF condition; if it meets the IMF condition, use the new voltage signal sequence as the first decomposed IMF component c1(t) of the original voltage signal sequence; otherwise, use the new voltage signal sequence as the original voltage signal sequence, and repeat steps S5.1 - S5.3 until the mean value of the upper and lower envelopes tends to zero, and use the corresponding new voltage signal sequence at this time as the first decomposed IMF component c1(t) of the original voltage signal sequence;

[0024] S5.4: Subtract the first decomposed IMF component c1(t) from the original voltage signal sequence to obtain a remaining voltage signal sequence; use the remaining voltage signal sequence as the original voltage signal sequence, and repeat steps S5.1 - S5.3 until a series of IMF components c are obtained n(t) and an indecomposable remainder term, taking the indecomposable remainder term as the residual component r(t), where c n (t) represents the nth decomposed IMF component, n = 2, 3…, N, then the original voltage signal sequence is expressed as the sum of several IMF components and a residual component.

[0025] Preferably, the specific method of step S5.1 is:

[0026] Calculate all the extreme points of the original voltage signal sequence, use the interpolation method to fit the lower envelope at the minimum points and the upper envelope at the maximum points, and calculate the mean value of the upper and lower envelopes:

[0027]

[0028] In the formula, m(t) represents the mean value of the upper and lower envelopes, emint(t) represents the lower envelope, and emax(t) represents the upper envelope.

[0029] Preferably, in step S5.4, the fact that the original voltage signal sequence is expressed as the sum of several IMF components and a residual component is specifically:

[0030]

[0031] In the formula, x(t) represents the original voltage signal sequence, c i (t) represents the ith decomposed IMF component, r(t) represents the residual component, i = 1, 2,…, N.

[0032] Preferably, the specific method of step S6 is:

[0033] Obtain the weld width uniformity of the welded workpiece after laser welding, and compare the weld width uniformity with a preset uniformity threshold; determine the welding quality of the welded workpiece with a weld width uniformity greater than the preset uniformity threshold as a low-quality welded workpiece, and determine the welding quality of the welded workpiece with a weld width uniformity equal to or less than the preset uniformity threshold as a high-quality welded workpiece.

[0034] The present invention also provides an on-line detection system for laser welding quality, including

[0035] A laser emitter for emitting laser to the welded workpiece, and the welded workpiece absorbs the laser to generate radiant light;

[0036] An active light emitter for emitting active infrared light to the welded workpiece, and the welded workpiece generates an active reflected light signal;

[0037] A spectroscopic device for decomposing the radiant light into several optical signals of different bands;

[0038] An optoelectronic converter for converting optical signals and active reflected optical signals in several different bands into corresponding electrical signals;

[0039] A preprocessing unit for preprocessing the electrical signals to obtain preprocessed electrical signals;

[0040] An empirical mode decomposition unit for respectively performing empirical mode decomposition on the preprocessed electrical signals to obtain several IMF components and residual components corresponding to each preprocessed electrical signal;

[0041] A welding quality classification unit for obtaining the morphological characteristics of a welded workpiece after laser welding and classifying the welding quality of the welded workpiece based on the morphological characteristics after welding to obtain a welding quality classification result;

[0042] A welding quality classifier training unit for forming a feature data set from optical signals and active reflected optical signals in several different bands and their corresponding several IMF components and residual components, and inputting the welding quality classification result of the welded workpiece into a preset welding quality classifier for training to obtain a trained welding quality classifier;

[0043] A welding quality detection unit for obtaining optical signals and active reflected optical signals in several different bands of a workpiece to be welded during the welding process, and inputting them into the trained welding quality classifier for detection to obtain a welding quality discrimination result of the workpiece to be welded.

[0044] Preferably, the spectroscopic device includes a plate placement table, a first convex lens, a second convex lens, a third convex lens, a fourth convex lens, a first mirror, a second mirror, a third mirror, a first optical sensor, a second optical sensor, and a third optical sensor;

[0045] The spectroscopic device decomposes the radiation light into optical signals in several different bands, including visible light optical signals, laser reflected light optical signals, and infrared light optical signals;

[0046] The plate placement table is used to place the welded workpiece. After the welded workpiece absorbs the laser to generate radiation light, the radiation light passes through the first convex lens for focusing, is reflected by the first mirror, and then reflected by the second mirror, and is decomposed into visible light and remaining radiation light; the visible light passes through the second convex lens for focusing onto the first optical sensor to form a visible light optical signal; after the remaining radiation light passes through the second mirror, it is reflected by the third mirror and decomposed into laser reflected light and infrared light; the laser reflected light passes through the third convex lens for focusing onto the second optical sensor to form a laser reflected light optical signal; the infrared light passes through the third mirror and is focused onto the third optical sensor through the fourth convex lens to form an infrared light optical signal;

[0047] The welded workpiece generates active reflected light. After the active reflected light is focused by the first convex lens and reflected by the first mirror, it sequentially passes through the second mirror and the third mirror, and is focused by the fourth convex lens onto the third optical sensor, forming an active reflected light signal.

[0048] Preferably, the working wavelength of the first mirror is greater than 1100 nm, the working wavelength of the second mirror is 780 - 1100 nm, and the working wavelength of the third mirror is 390 - 780 nm.

[0049] Compared with the prior art, the beneficial effects of the technical solution of the present invention are:

[0050] The present invention collects the radiation light generated by the welded workpiece during the welding process, decomposes the radiation light into optical signals of different wavelength bands, and simultaneously emits active infrared light to the welded workpiece to make up for the weak situation of the infrared light signal wavelength band during laser welding; then converts the optical signals of different wavelength bands and the active reflected light signal into corresponding electrical signals, performs empirical mode decomposition after preprocessing, and obtains several IMF components and residual components corresponding to each preprocessed electrical signal. When defects occur during the welding process, the optical signal and the corresponding several IMF components and residual components will also show different characteristics; the optical signal of the welded workpiece and the corresponding several IMF components and residual components are used as a feature data set, which is matched with the welding quality of the welded workpiece, and a preset welding quality classifier is jointly trained; finally, several optical signals of different wavelength bands and active reflected light signals during the welding process of the workpiece to be welded are collected, and the trained welding quality classifier can be used to detect the welding quality of the workpiece to be welded. The present invention determines the stability of the welding process by obtaining the characteristics of the optical signal, IMF components and residual components of the welded workpiece, overcomes the disadvantages of visual inspection being easily interfered by strong reflected light and the processing algorithm being complex, and quickly and accurately realizes the detection of welding quality and screens out qualified welded workpieces. Description of the Drawings

[0051] Figure 1 It is a flowchart of an on - line detection method for laser welding quality described in Embodiment 1.

[0052] Figure 2 It is a schematic flow diagram of an on - line detection method for laser welding quality described in Embodiment 2.

[0053] Figure 3 It is a schematic structural diagram of an on - line detection system for laser welding quality described in Embodiment 3.

[0054] Figure 4 It is a schematic structural diagram of the spectroscopic device described in Embodiment 3.

[0055] 1-plate placement table, 2-first convex lens, 3-second convex lens, 4-third convex lens, 5-fourth convex lens,

[0056] 6 - first reflector, 7 - second reflector, 8 - third reflector, 9 - first light sensor, 10 - second light sensor, 11 - third light sensor. DETAILED DESCRIPTION

[0057] The drawings are for illustrative purposes only and should not be construed as limiting the present patent;

[0058] In order to better illustrate the present embodiment, some parts in the drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product;

[0059] It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0060] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0061] Example 1

[0062] This embodiment provides an online detection method for laser welding quality. Figure 1 As shown, including:

[0063] S1: emit laser to the welding workpiece, the welding workpiece absorbs the laser to generate radiation light; emit active infrared light to the welding workpiece, the welding workpiece generates active reflected light signal;

[0064] S2: decompose the radiation light into several optical signals of different wavelength bands;

[0065] S3: converting optical signals of several different wavelength bands and active reflected optical signals into corresponding electrical signals;

[0066] S4: preprocessing the electrical signal to obtain a preprocessed electrical signal;

[0067] S5: performing empirical mode decomposition on the preprocessed electrical signals respectively to obtain a number of IMF components and residual components corresponding to each preprocessed electrical signal;

[0068] S6: obtaining the morphological features of the welded workpiece after laser welding, and classifying the welding quality of the welded workpiece according to the morphological features after welding to obtain a welding quality classification result;

[0069] S7: A plurality of optical signals of different wavelength bands and active reflected optical signals and their corresponding plurality of IMF components and residual components are combined into a feature data set, and the data is input into a preset welding quality classifier for training together with the welding quality classification result of the welding workpiece to obtain a trained welding quality classifier;

[0070] S8: Obtain optical signals and active reflected optical signals of a workpiece to be welded in several different wavelength bands during the welding process, input them into a trained welding quality classifier for detection, and obtain the discrimination result of the welding quality of the workpiece to be welded.

[0071] In the specific implementation process, in this embodiment, the radiation light generated by the welding workpiece during the welding process is collected, and the radiation light is decomposed into optical signals in different wavelength bands. At the same time, active infrared light is emitted to the welding workpiece to make up for the weak infrared light signal wavelength band during laser welding. Then, the optical signals in different wavelength bands and the active reflected optical signals are converted into corresponding electrical signals, and after preprocessing, empirical mode decomposition is performed to obtain several IMF components and residual components corresponding to each preprocessed electrical signal. When defects occur during the welding process, the optical signals and the corresponding several IMF components and residual components will also show different characteristics. The optical signals of the welding workpiece and the corresponding several IMF components and residual components are used as a feature data set, which is matched with the welding quality of the welding workpiece, and a preset welding quality classifier is jointly trained. Finally, collect the optical signals and active reflected optical signals of the workpiece to be welded in several different wavelength bands during the welding process, and the welding quality of the workpiece to be welded can be detected by using the trained welding quality classifier. In this embodiment, the stability of the welding process is judged by obtaining the characteristics of the optical signals, IMF components and residual components of the welding workpiece, which is not easily interfered by strong reflected light, and the welding quality detection is realized quickly and accurately, and qualified welding workpieces are screened out.

[0072] Embodiment 2

[0073] This embodiment provides an on-line detection method for laser welding quality, as Figure 2 shown, including:

[0074] S1: Emit laser to the welding workpiece, and the welding workpiece absorbs the laser to generate radiation light; emit active infrared light to the welding workpiece, and the welding workpiece generates an active reflected optical signal;

[0075] S2: Decompose the radiation light into three optical signals in different wavelength bands, including visible light optical signal, laser reflected light optical signal and infrared light optical signal; the wavelength band of the visible light optical signal is 390 - 780 nanometers, the wavelength band of the laser reflected light optical signal is 780 - 1100 nanometers, and the wavelength band of the infrared light optical signal is greater than 1100 nanometers;

[0076] S3: Convert the three optical signals in different wavelength bands and the active reflected optical signal into corresponding electrical signals;

[0077] S4: Preprocess the electrical signals to obtain preprocessed electrical signals; the specific method is:

[0078] Pre-amplification operations, filtering operations, and main amplification operations are sequentially performed on visible optoelectronic signals, laser reflection optoelectronic signals, infrared optoelectronic signals, and active reflection optoelectronic signals to obtain preprocessed visible optoelectronic signals, laser reflection optoelectronic signals, infrared optoelectronic signals, and active reflection optoelectronic signals;

[0079] Visible optoelectronic signals, laser reflection optoelectronic signals, infrared optoelectronic signals, and active reflection optoelectronic signals are all very weak and will be buried in noise signals; pre-amplification operations are first performed on visible optoelectronic signals, laser reflection optoelectronic signals, infrared optoelectronic signals, and active reflection optoelectronic signals, and the included noise signals are amplified simultaneously, which is beneficial to filtering out the noise signals cleanly during the filtering operation; after the filtering operation filters out the noise signals, the pure visible optoelectronic signals, laser reflection optoelectronic signals, infrared optoelectronic signals, and active reflection optoelectronic signals are amplified again to the voltage amplitude required for subsequent step processing.

[0080] S5: Empirical mode decomposition is respectively performed on the preprocessed electrical signals to obtain several IMF components and residual components corresponding to each preprocessed electrical signal; the specific steps are as follows:

[0081] Visible optoelectronic signals, laser reflection optoelectronic signals, infrared optoelectronic signals, and active reflection optoelectronic signals are sequentially used as the original voltage signal sequences, and the following steps are executed;

[0082] S5.1: Calculate all extreme points of the original voltage signal sequence, use the interpolation method to fit the lower envelope line at the minimum extreme points and the upper envelope line at the maximum extreme points, and calculate the mean value of the upper and lower envelope lines:

[0083]

[0084] In the formula, m(t) represents the mean value of the upper and lower envelope lines, emint(t) represents the lower envelope line, and emax(t) represents the upper envelope line;

[0085] S5.2: Subtract the mean value of the upper and lower envelope lines from the original voltage signal sequence to obtain a new voltage signal sequence;

[0086] S5.3: Determine whether the new voltage signal sequence meets the IMF conditions; if it meets the IMF conditions, use the new voltage signal sequence as the first decomposed IMF component c1(t) of the original voltage signal sequence; otherwise, use the new voltage signal sequence as the original voltage signal sequence, and repeat steps S5.1 - S5.3 until the mean value of the upper and lower envelope lines tends to zero, and use the corresponding new voltage signal sequence at this time as the first decomposed IMF component c1(t) of the original voltage signal sequence;

[0087] The IMF conditions are as follows:

[0088] a. The sum of the number of local maxima and the number of local minima must be equal to or at most differ by one from the number of zero crossings;

[0089] b. The mean value of the upper envelope defined by local maxima and the lower envelope defined by local minima is zero.

[0090] S5.4: Subtract the first decomposed IMF component c1(t) from the original voltage signal sequence to obtain the remaining voltage signal sequence; take the remaining voltage signal sequence as the original voltage signal sequence, and repeat steps S5.1 - S5.3 until a series of IMF components c n (t) and an indecomposable remainder are obtained. Take the indecomposable remainder as the residual component r(t), where c n (t) represents the nth decomposed IMF component, n = 2, 3…, N. Then the original voltage signal sequence is expressed as the sum of several IMF components and the residual component:

[0091]

[0092] In the formula, x(t) represents the original voltage signal sequence, c i (t) represents the ith decomposed IMF component, r(t) represents the residual component, i = 1, 2,…, N.

[0093] S6: Obtain the morphological characteristics of the welded workpiece after laser welding, and classify the welding quality of the welded workpiece based on the morphological characteristics after welding to obtain the welding quality classification result; specifically:

[0094] Obtain the weld width uniformity of the welded workpiece after laser welding, and compare the weld width uniformity with a preset uniformity threshold; determine the welding quality of the welded workpiece with a weld width uniformity greater than the preset uniformity threshold as a low-quality welded workpiece, and determine the welding quality of the welded workpiece with a weld width uniformity equal to or less than the preset uniformity threshold as a high-quality welded workpiece.

[0095] S7: Combine optical signals of several different bands, active reflected optical signals, and their corresponding several IMF components and residual components to form a feature dataset, and input it together with the welding quality classification result of the welded workpiece into a preset welding quality classifier for training to obtain a trained welding quality classifier;

[0096] The preset welding quality classifier is a support vector machine classifier.

[0097] S8: Obtain optical signals of several different bands and active reflected optical signals of the workpiece to be welded during the welding process, and input them into the trained welding quality classifier for detection to obtain the welding quality discrimination result of the workpiece to be welded.

[0098] Embodiment 3

[0099] This embodiment provides an on - line detection system for laser welding quality, as Figure 3 shown, including

[0100] A laser emitter, which is used to emit laser to the welding workpiece, and the welding workpiece absorbs the laser to generate radiation light;

[0101] An active light emitter, which is used to emit active infrared light to the welding workpiece, and the welding workpiece generates an active reflected light signal;

[0102] A spectroscopic device, which is used to decompose the radiation light into several optical signals of different bands;

[0103] A photoelectric converter, which is used to convert several optical signals of different bands and the active reflected light signal into corresponding electrical signals;

[0104] A pre - processing unit, which is used to pre - process the electrical signal to obtain a pre - processed electrical signal;

[0105] An empirical mode decomposition unit, which is used to perform empirical mode decomposition on the pre - processed electrical signal respectively to obtain several IMF components and residual components corresponding to each pre - processed electrical signal;

[0106] A welding quality classification unit, which is used to obtain the morphological characteristics of the welding workpiece after laser welding, and classify the welding quality of the welding workpiece according to the morphological characteristics after welding to obtain a welding quality classification result;

[0107] A welding quality classifier training unit, which is used to form a feature data set by combining several optical signals of different bands, the active reflected light signal, and their corresponding several IMF components and residual components, and input them together with the welding quality classification result of the welding workpiece into a preset welding quality classifier for training to obtain a trained welding quality classifier;

[0108] A welding quality detection unit, which is used to obtain several optical signals of different bands and the active reflected light signal of the workpiece to be welded during the welding process, and input them into the trained welding quality classifier for detection to obtain a welding quality discrimination result of the workpiece to be welded.

[0109] As Figure 4 shown, the spectroscopic device includes a plate placement table 1, a first convex lens 2, a second convex lens 3, a third convex lens 4, a fourth convex lens 5, a first reflector 6, a second reflector 7, a third reflector 8, a first optical sensor 9, a second optical sensor 10, and a third optical sensor 11;

[0110] The spectroscopic device decomposes the radiation light into several optical signals of different bands, including visible light optical signals, laser reflected light optical signals, and infrared light optical signals;

[0111] The plate placing table 1 is used to place welding workpieces. After the welding workpieces absorb laser to generate radiation light, the radiation light is successively focused by the first convex lens 2, reflected by the first reflecting mirror 6, and reflected by the second reflecting mirror 7, and decomposed into visible light and remaining radiation light. The visible light is focused by the second convex lens 3 onto the first optical sensor 9 to form a visible light signal. After the remaining radiation light passes through the second reflecting mirror 7, it is reflected by the third reflecting mirror 8 and decomposed into laser reflected light and infrared light. The laser reflected light is focused by the third convex lens 4 onto the second optical sensor 10 to form a laser reflected light signal. After the infrared light passes through the third reflecting mirror 8, it is focused by the fourth convex lens 5 onto the third optical sensor 11 to form an infrared light signal.

[0112] The welding workpieces generate active reflected light. After the active reflected light is focused by the first convex lens 2 and reflected by the first reflecting mirror 6, it successively passes through the second reflecting mirror 7 and the third reflecting mirror 8, and is focused by the fourth convex lens 5 onto the third optical sensor 11 to form an active reflected light signal.

[0113] In this embodiment, the operating wavelength of the first reflecting mirror 6 is greater than 1100 nm, the operating wavelength of the second reflecting mirror 7 is 780 - 1100 nm, and the operating wavelength of the third reflecting mirror 8 is 390 - 780 nm.

[0114] The same or similar reference numerals correspond to the same or similar components.

[0115] The terms describing the positional relationship in the drawings are only for illustrative purposes and should not be construed as a limitation of this patent.

[0116] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, rather than limitations on the implementation manners of the present invention. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all implementation manners here. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included in the protection scope of the claims of the present invention.

Claims

1. An on-line detection method for laser welding quality, characterized in that, Including: S1: Emitting a laser to the welding workpiece, and the welding workpiece absorbs the laser to generate radiation light; emitting active infrared light to the welding workpiece, and the welding workpiece generates an active reflected light signal; S2: Decomposing the radiation light into several optical signals of different bands; The several optical signals of different bands include visible light signals, laser reflected light signals, and infrared light signals; among them, the band of the visible light signal is 390 - 780 nanometers, the band of the laser reflected light signal is 780 - 1100 nanometers, and the band of the infrared light signal is greater than 1100 nanometers; S3: Converting the several optical signals of different bands and the active reflected light signal into corresponding electrical signals; S4: Preprocessing the electrical signals to obtain preprocessed electrical signals; S5: Performing empirical mode decomposition on the preprocessed electrical signals respectively to obtain several IMF components and residual components corresponding to each preprocessed electrical signal; Taking the visible light electrical signal, the laser reflected light electrical signal, the infrared light electrical signal, and the active reflected light electrical signal as the original voltage signal sequence in turn, and performing the following steps: S5.1: Calculating all extreme points of the original voltage signal sequence, fitting a lower envelope at the minimum extreme points and an upper envelope at the maximum extreme points, and calculating the mean of the upper and lower envelopes; S5.2: Subtracting the mean of the upper and lower envelopes from the original voltage signal sequence to obtain a new voltage signal sequence; S5.3: Determine whether the new voltage signal sequence meets the IMF condition; if it meets the IMF condition, use the new voltage signal sequence as the first decomposed IMF component of the original voltage signal sequence ; Otherwise, use the new voltage signal sequence as the original voltage signal sequence, repeat steps S5.1 - S5.3 to make the mean values of the upper and lower envelope lines tend to zero, and use the corresponding new voltage signal sequence at this time as the first decomposed IMF component of the original voltage signal sequence ; S5.4: Subtract the first decomposed IMF component from the original voltage signal sequence , to obtain the residual voltage signal sequence; take the residual voltage signal sequence as the original voltage signal sequence, and repeat steps S5.1 - S5.3 until a series of IMF components and an undecomposable remainder are obtained, and take the undecomposable remainder as the residual component , where represents the -th decomposed IMF component, , then the original voltage signal sequence is expressed as the sum of several IMF components and the residual component; S6: Obtaining the morphological characteristics of the welding workpiece after laser welding, and classifying the welding quality of the welding workpiece according to the morphological characteristics after welding to obtain a welding quality classification result; S7: Combining the several optical signals of different bands and the active reflected light signal and their corresponding several IMF components and residual components to form a feature dataset, and inputting the feature dataset and the welding quality classification result of the welding workpiece into a preset welding quality classifier for training to obtain a trained welding quality classifier; S8: Obtaining the several optical signals of different bands and the active reflected light signal of the workpiece to be welded during the welding process, and inputting them into the trained welding quality classifier for detection to obtain a welding quality discrimination result of the workpiece to be welded.

2. The on-line detection method for laser welding quality according to claim 1, characterized in that, In the step S4, when preprocessing the electrical signals to obtain preprocessed electrical signals, the specific method is: Performing a pre - amplification operation, a filtering operation, and a main amplification operation on the visible light electrical signal, the laser reflected light electrical signal, the infrared light electrical signal, and the active reflected light electrical signal in turn to obtain the preprocessed visible light electrical signal, laser reflected light electrical signal, infrared light electrical signal, and active reflected light electrical signal.

3. The on-line detection method for laser welding quality according to claim 1, characterized in that, The specific method of the step S5.1 is: Calculating all extreme points of the original voltage signal sequence, using the interpolation method to fit a lower envelope at the minimum extreme points and an upper envelope at the maximum extreme points, and calculating the mean of the upper and lower envelopes: In the formula, represents the mean of the upper and lower envelopes, represents the lower envelope, represents the upper envelope.

4. The on-line detection method for laser welding quality according to claim 1, characterized in that, In the step S5.4, the original voltage signal sequence is expressed as the sum of several IMF components and residual components specifically as: In the formula, represents the original voltage signal sequence, represents the th decomposed IMF component, represents the residual component, .

5. The on-line detection method for laser welding quality according to claim 1, characterized in that, The specific method of the step S6 is: Obtain the morphological characteristics of the welded workpiece after laser welding, and compare the weld width uniformity with a preset uniformity threshold; determine the welding quality of the welded workpiece with a weld width uniformity greater than the preset uniformity threshold as a low-quality welded workpiece, and determine the welding quality of the welded workpiece with a weld width uniformity equal to or less than the preset uniformity threshold as a high-quality welded workpiece.

6. An on-line detection system for laser welding quality, characterized in that, Comprising a laser emitter for emitting laser light to the welded workpiece, and the welded workpiece absorbs the laser light to generate radiation light; an active light emitter for emitting active infrared light to the welded workpiece, and the welded workpiece generates an active reflected light signal; a spectroscopic device for decomposing the radiation light into several optical signals of different bands; the several optical signals of different bands include visible light optical signals, laser reflected light optical signals, and infrared light optical signals; wherein the band of the visible light optical signal is 390 - 780 nanometers, the band of the laser reflected light optical signal is 780 - 1100 nanometers, and the band of the infrared light optical signal is greater than 1100 nanometers; a photoelectric converter for converting the several optical signals of different bands and the active reflected light signal into corresponding electrical signals; a preprocessing unit for preprocessing the electrical signals to obtain preprocessed electrical signals; an empirical mode decomposition unit for respectively performing empirical mode decomposition on the preprocessed electrical signals to obtain several IMF components and residual components corresponding to each preprocessed electrical signal, including: successively taking the visible light electrical signal, the laser reflected light electrical signal, the infrared light electrical signal, and the active reflected light electrical signal as the original voltage signal sequence, and performing the following steps: calculate all extreme points of the original voltage signal sequence, fit a lower envelope line at the minimum extreme points and an upper envelope line at the maximum extreme points, and calculate the mean of the upper and lower envelope lines; subtract the mean of the upper and lower envelope lines from the original voltage signal sequence to obtain a new voltage signal sequence; Determine whether the new voltage signal sequence meets the IMF conditions; if it meets the IMF conditions, use the new voltage signal sequence as the first decomposed IMF component of the original voltage signal sequence ; otherwise, use the new voltage signal sequence as the original voltage signal sequence, repeat the above steps to make the mean of the upper and lower envelope lines tend to zero, and use the corresponding new voltage signal sequence at this time as the first decomposed IMF component of the original voltage signal sequence ; Subtract the first decomposed IMF component from the original voltage signal sequence , to obtain the remaining voltage signal sequence; take the remaining voltage signal sequence as the original voltage signal sequence, and repeat the above steps until a series of IMF components and an undecomposable remainder are obtained, and take the undecomposable remainder as the residual component , where represents the th decomposed IMF component, , then the original voltage signal sequence is expressed as the sum of several IMF components and the residual component; a welding quality classification unit for obtaining the morphological characteristics of the welded workpiece after laser welding, and classifying the welding quality of the welded workpiece based on the morphological characteristics after welding to obtain a welding quality classification result; a welding quality classifier training unit for forming a feature data set from the several optical signals of different bands and the active reflected light signal and their corresponding several IMF components and residual components, and inputting them together with the welding quality classification result of the welded workpiece into a preset welding quality classifier for training to obtain a trained welding quality classifier; a welding quality detection unit for obtaining the several optical signals of different bands and the active reflected light signal of the workpiece to be welded during the welding process, and inputting them into the trained welding quality classifier for detection to obtain a welding quality discrimination result of the workpiece to be welded.

7. The on-line detection system for laser welding quality according to claim 6, wherein, The spectroscopic device includes a plate placement table (1), a first convex lens (2), a second convex lens (3), a third convex lens (4), a fourth convex lens (5), a first reflector (6), a second reflector (7), a third reflector (8), a first optical sensor (9), a second optical sensor (10), and a third optical sensor (11); The spectroscopic device decomposes the radiation light into several optical signals of different bands, including visible light optical signals, laser reflected light optical signals, and infrared light optical signals; The plate placing table (1) is used to place welding workpieces. After the welding workpieces absorb laser to generate radiation light, the radiation light is successively focused by the first convex lens (2), reflected by the first reflector (6), and reflected by the second reflector (7), and is decomposed into visible light and remaining radiation light; the visible light is focused by the second convex lens (3) onto the first optical sensor (9) to form a visible light signal; after the remaining radiation light passes through the second reflector (7), it is reflected by the third reflector (8) and decomposed into laser reflected light and infrared light; the laser reflected light is focused by the third convex lens (4) onto the second optical sensor (10) to form a laser reflected light signal; after the infrared light passes through the third reflector (8), it is focused by the fourth convex lens (5) onto the third optical sensor (11) to form an infrared light signal. The welding workpiece generates active reflected light. After the active reflected light is focused by the first convex lens (2) and reflected by the first reflector (6), it successively passes through the second reflector (7) and the third reflector (8), and is focused by the fourth convex lens (5) onto the third optical sensor (11) to form an active reflected light signal.

8. The on-line detection system for laser welding quality according to claim 7, wherein, The working wavelength of the first reflector (6) is greater than 1100 nm, the working wavelength of the second reflector (7) is 780 - 1100 nm, and the working wavelength of the third reflector (8) is 390 - 780 nm.

Images