Method and system for monitoring cutting penetration state in water-jet guided laser processing process based on sound signal blind source separation

Abstract

The invention provides a monitoring method and system for a cutting penetration state in a water-jet guided laser processing process based on sound signal blind source separation, and the monitoring method comprises the steps: S1, in the water-jet guided laser processing process, collecting a plurality of observation sound signals in real time, the observation sound signals being mixed sound signals composed of a plurality of independent sound source signals; s2, estimating and solving a plurality of independent sound source signals by using a blind source separation algorithm according to the plurality of collected observation sound signals; s3, the frequency spectrums of the multiple independent sound source signals are compared with sound signals generated during material laser processing, so that sound signals generated during water-guided laser processing due to material removal are determined from the frequency spectrums; and S4, the water-jet guided laser machining process is judged according to the relation that sound signals generated due to material removal change along with time. Compared with a traditional coaxial visual monitoring scheme, the method is not affected by factors such as water flow and light intensity, and can be applied to monitoring of the water-guided laser machining process.

Description

Technical Field

[0001] The present invention relates to the technical field of water-guided laser processing, and in particular to a method and system for monitoring the cutting penetration state during water-guided laser processing based on blind source separation of acoustic signals. Background Art

[0002] Water-guided laser processing, a new laser processing application, demonstrates excellent adaptability to the processing of new materials through its unique processing mechanism. This technology couples a high-energy laser into an ultra-fine water jet and directs it onto the surface of the material being processed. This removes the material through laser-matter interaction, significantly reducing thermal deformation and heat concentration in the material being processed.

[0003] Water-guided laser processing, because its efficiency is significantly lower than that of traditional laser processing, often requires multiple repetitions of the processing trajectory. Insufficient repetitions can lead to defects such as partial incomplete penetration of the material. While traditional laser processing can rapidly penetrate the material by simply increasing the power, water-guided laser processing is not feasible. Increasing the power of the coupled laser beam can actually disrupt the stability of the water jet, resulting in a decrease in processing efficiency. Therefore, it is necessary to monitor the current material processing progress to ensure complete material processing.

[0004] Traditional coaxial visual monitoring solutions use high-speed cameras to capture the light intensity reflection signal of the processing area, but they have significant limitations in practical applications. Traditional processing monitoring methods do not consider the influence of water flow. However, during water-guided laser processing, the atomized droplets produced by the water jet and the shielding gas induce light scattering, resulting in a sharp drop in the image signal-to-noise ratio. This is especially true in deep hole processing scenarios, where the optical path is completely obscured by the water medium, resulting in high feature extraction error rates. More importantly, the dynamic processing environment easily distorts the light intensity signal, making it difficult to establish a stable process mapping model. Therefore, a suitable monitoring method for water-guided laser processing progress is needed. Summary of the Invention

[0005] Based on the above problems existing in the prior art, the present invention aims to solve the technical problem that the prior art lacks a monitoring means suitable for the progress of water-guided laser processing.

[0006] The present invention provides a method for monitoring the cutting penetration state during water-guided laser processing based on blind source separation of acoustic signals, which comprises the following steps:

[0007] S1: During the water-guided laser processing, multiple observation sound signals are collected in real time. The observation sound signals are mixed sound signals composed of multiple independent sound source signals.

[0008] S2: Based on the collected multiple observation sound signals, a blind source separation algorithm is used to estimate and solve multiple independent sound source signals;

[0009] S3: comparing the multiple independent sound source signals with the sound signals generated when the laser processes the material, so as to confirm the sound signals generated by the material removal during the water-conducting laser processing;

[0010] S4: judging whether cutting is penetrating during the water-guided laser processing process based on the relationship between the sound signal generated by material removal and time.

[0011] In some embodiments, step S2 specifically includes the following steps:

[0012] S21: Construct a blind source separation mathematical model based on the mathematical representation relationship between the independent sound source signal and the observed sound signal, thereby constructing a signal separation matrix;

[0013] S22: Construct an evaluation function based on the criterion of maximizing negative entropy, and use negative entropy to measure the non-Gaussian degree of the separated signal;

[0014] S23: Construct an iterative optimization process based on FastICA, substitute the collected multiple observation sound signals into it, and iteratively calculate the separation signal matrix corresponding to the extreme value of the evaluation function. This matrix is ​​the separation signal to be solved.

[0015] In some embodiments, in step S21, the blind source separation mathematical model is a linear mixture model.

[0016] In some other embodiments, before the m collected observation sound signals are substituted into the FastICA-based iterative optimization process in step S23, the m collected observation sound signals need to be preprocessed, including de-meaning, whitening and principal component analysis.

[0017] In some embodiments, when the collected multiple observation sound signals are substituted into the iterative optimization process based on FastICA in step S23, only a section of the multiple observation sound signals is intercepted and substituted into the iterative optimization process based on FastICA.

[0018] In some embodiments, step S3 includes the following steps:

[0019] S31: performing spectrum analysis on the multiple independent sound source signals to obtain frequency domain graphs of multiple separated signals;

[0020] S32: comparing the frequency domain graphs of the plurality of independent sound source signals with the frequency characteristics of the sound signals generated when the laser processes the material, so as to confirm the sound signals generated by the material removal during the water-conducting laser processing.

[0021] In some embodiments, step S4 includes the following steps:

[0022] S41: Processing the frequency domain graph of the sound signal generated by material removal to obtain a time domain graph;

[0023] S42: Performing short-time energy analysis on the obtained time domain graph to obtain a graph of the energy of the sound signal generated by material removal changing with time, which reflects the water-guided laser processing process.

[0024] The present invention also provides a water-guided laser processing process monitoring system for implementing the aforementioned method for monitoring the cutting penetration state during the water-guided laser processing process based on blind source separation of acoustic signals, which comprises:

[0025] Multiple sound collection elements, each used to collect in real time an observation sound signal generated during the water-guided laser processing process, wherein the observation sound signal is a mixed sound signal composed of multiple independent sound source signals, and the number of the sound collection elements is not less than the number of the independent sound source signals;

[0026] a signal separation module, which establishes communication connections with the plurality of sound collection elements respectively, for receiving a plurality of observation sound signals collected by the plurality of sound collection elements, and estimating and solving a plurality of separated signals based on a blind source separation algorithm;

[0027] a comparison and confirmation module electrically connected to the signal separation module and configured to receive the plurality of separated signals; the comparison and confirmation module being configured to compare the plurality of separated signals with the sound signals generated during the laser processing of the material, so as to confirm that the sound signal is generated by the removal of the material during the water-conducting laser processing;

[0028] The judgment module is in communication with the comparison and confirmation module, and is used to receive the sound signal generated by material removal, and can judge the water-guided laser processing process according to the relationship between the energy of the sound signal generated by material removal and time.

[0029] In some embodiments, the water-guided laser processing process monitoring system further includes a signal amplifier, a data acquisition card, and an industrial computer. A plurality of the sound collection elements are connected in parallel and electrically connected to the signal amplifier, the data acquisition card, and the industrial computer in sequence. The industrial computer is provided with the signal separation module, the comparison and confirmation module, and the judgment module.

[0030] In some embodiments, at least one sound collecting element is provided at the generating location of each independent sound source signal.

[0031] Beneficial effects:

[0032] The present invention provides a method and system for monitoring the cutting penetration status during water-guided laser processing based on blind source separation of acoustic signals. By collecting multiple observation sound signals and using a blind source separation algorithm to solve for multiple independent sound source signals, the separated multiple independent sound source signals are compared with the sound signals generated during laser processing. The sound signals generated by material removal during water-guided laser processing can be obtained, and the progress of the water-guided laser processing can be determined based on the sound signals generated by material removal. Compared with traditional coaxial visual monitoring solutions, this method is not affected by factors such as water flow and light intensity and can be applied to monitoring the progress of water-guided laser processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions in the embodiments or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0034] Figure 1 This is a principle block diagram of a water-guided laser processing process monitoring system provided by an embodiment of the present invention;

[0035] Figure 3 1 is a flow chart of a method for monitoring the cutting penetration status during water-guided laser processing based on blind source separation of acoustic signals provided by an embodiment of the present invention;

[0036] Figure 2 Schematic diagram of the arrangement of the sound collection element in the water-guided laser processing equipment according to an embodiment of the present invention;

[0037] Figure 4 It is a flowchart of a blind source separation algorithm in an embodiment of the present invention;

[0038] Figure 5 This is a time domain waveform diagram of a segment of the original signal intercepted from the four collected observation signals in Example 1 of the present invention;

[0039] Figure 6 In the first embodiment of the present invention, a blind source separation algorithm is used to separate Figure 5 The time domain waveforms of the 4 separated signals obtained after processing the 4 segments of the original signal;

[0040] Figure 7 This is a frequency domain diagram of the four-channel signals obtained after performing spectrum analysis on the four-channel signals separated in the first embodiment of the invention;

[0041] Figure 8 This is a time domain waveform diagram of the sound signal generated when the laser processes the material during the water-guided laser processing in the first embodiment of the invention;

[0042] Figure 9 1 is a schematic diagram showing how the energy of the sound signal generated when the laser processes the material during the water-guided laser processing in the first embodiment of the invention changes with time;

[0043] Figure numerals: 1. Sound collection element; 1-1. Sound collection element one; 1-2. Sound collection element two; 1-3. Sound collection element three; 1-4. Sound collection element four; 2. Signal separation module; 3. Comparison and confirmation module; 4. Judgment module; 5. Signal amplifier; 6. Data acquisition card; 10. Industrial computer. DETAILED DESCRIPTION

[0044] The following descriptions of the embodiments refer to the accompanying drawings to illustrate specific embodiments in which the present invention may be implemented.

[0045] The present invention provides a water-guided laser processing process monitoring system, the principle block diagram of which is as follows: Figure 1 As shown, the water-guided laser processing process monitoring system includes multiple sound collection elements 1, a signal separation module 2, a comparison and confirmation module 3, and a judgment module 4. The sound collection element 1 is used to collect observation sound signals generated during the water-guided laser processing process in real time. The observation sound signal is a mixed sound signal composed of multiple independent sound source signals. The number of the sound collection elements 1 is not less than the number of the independent sound source signals, and at least one sound collection element 1 is provided at the location where each independent sound source signal occurs (i.e., the independent sound source position). Specifically, the sound collection element 1 is an acoustic sensor. The signal separation module 2 establishes communication connections with the multiple sound collection elements 1 to receive the sound signals. The plurality of observed sound signals collected by the plurality of sound collecting elements 1 are collected, and a plurality of separated signals (i.e., independent sound source signals) are estimated and solved based on a blind source separation algorithm. The comparison and confirmation module 3 is electrically connected to the signal separation module 2 for receiving the plurality of separated signals. The comparison and confirmation module 3 is configured to compare the plurality of separated signals with the sound signals generated during laser material processing to determine the sound signals generated by material removal during water-conducting laser processing. The judgment module 4 is communicatively connected to the comparison and confirmation module 3 for receiving the sound signals generated by material removal and determining the progress of the water-conducting laser processing based on the relationship between the energy of the sound signals generated by material removal and time.

[0046] In some embodiments, the water-guided laser processing process monitoring system further includes a signal amplifier 5, a data acquisition card 6, and an industrial computer 10. A plurality of the sound collection elements 1 are connected in parallel and electrically connected to the signal amplifier 5, the data acquisition card 6, and the industrial computer 10 in sequence. The data acquisition card 6 is connected to the industrial computer 10 via a USB2.0 data cable. The industrial computer 10 is provided with the signal separation module 2, the comparison and confirmation module 3, and the judgment module 4.

[0047] The acoustic environment of water-guided laser processing equipment during operation is complex. This includes broadband turbulent noise generated by the impact of the high-pressure water jet on the material surface, characteristic processing sounds caused by phase changes caused by the interaction between the laser and the material, periodic exhaust noise from the air compressor, and mechanical vibrations from the reciprocating motion of the pneumatic booster pump. The frequency bands of these different signals overlap and interfere with each other, necessitating the separation of the desired acoustic signal from the laser-processed material from this mixed signal.

[0048] It should be noted that this embodiment only lists one type of acoustic environment composition during operation of the water-guided laser processing equipment. The acoustic environment includes: broadband turbulent noise generated by the impact of the high-pressure water jet on the material surface, characteristic processing sound caused by the interaction between the laser and the material causing the material phase change, periodic exhaust noise from the air compressor, and mechanical vibration sound from the reciprocating motion of the pneumatic booster pump, which contains four independent sound source signals. Due to the different structures of different water-guided laser processing equipment, water-guided laser processing equipment with complex structures may generate more complex acoustic environments during operation. Such acoustic environments may have more types of independent sound source signals. Therefore, the four independent sound source signals listed in this embodiment do not constitute a limitation on the number of independent sound source signals. In actual use, the types of independent sound source signals may include five, six, or even more.

[0049] Since the acoustic environment of the water-guided laser processing equipment in this embodiment includes the above four independent sound source signals during operation, the number of the sound collecting elements 1 in this embodiment of the present invention is four, and the number of the signal amplifiers 5 is two; the signal amplifier 5 is a two-channel amplifier, and the four sound collecting elements 1 are connected in parallel in pairs and are respectively connected to the two signal amplifiers 5, and the two signal amplifiers 5 are connected in parallel and are electrically connected to the data acquisition card 6. Figure 2 As shown, the four sound collecting elements 1 are respectively named as sound collecting element 1-1, sound collecting element 2 1-2, sound collecting element 3 1-3, and sound collecting element 4 1-4. In order to make the sound signals collected by each sound collecting element 1 more distinguishable from those collected by other sound collecting elements, in this embodiment, each sound collecting element 1 should be arranged as close as possible to its corresponding sound source. Therefore, the sound collecting element 1-1 is arranged on the outside of the chassis of the water-guided laser processing equipment to collect external environmental noise; the sound collecting element 2 1-2 is arranged on the inside of the top of the chassis to collect the processing acoustic signal of the upper surface of the sample (referring to the workpiece or processing material); the sound collecting element 3 1-3 is arranged on the inside of the front of the chassis to collect the water jet sound signal; and the sound collecting element 4 1-4 is arranged on the inside of the back of the chassis to collect the overall sound field environment signal inside the chassis.

[0050] Based on this, the present invention also provides a method for monitoring the cutting penetration state during water-guided laser processing based on blind source separation of acoustic signals. The method is implemented based on the water-guided laser processing process monitoring system. The process of the method for monitoring the cutting penetration state during water-guided laser processing based on blind source separation of acoustic signals is as follows: Figure 3 As shown, it includes the following steps:

[0051] S1: During the water-guided laser processing, multiple observation sound signals are collected in real time. The observation sound signals are mixed sound signals composed of multiple independent sound source signals.

[0052] S2: Based on the collected multiple observation sound signals, a blind source separation algorithm is used to estimate and solve multiple independent sound source signals;

[0053] S3: comparing the multiple independent sound source signals with the sound signals generated when the laser is processing the material, so as to confirm the sound signals generated by material removal during the water-conducting laser processing (the sound signals generated when the laser is processing the material are the sound signals generated by material removal);

[0054] S4: judging whether cutting is penetrating during the water-guided laser processing process based on the relationship between the sound signal generated by material removal and time.

[0055] The following combination Figure 3-Figure 9 The method for monitoring the cutting penetration state during the water-guided laser processing process based on blind source separation of acoustic signals is described.

[0056] Step S2 is based on the assumption of statistical independence of the source signals. Its core is to separate the source signals that meet statistical independence from the linearly mixed observed sound signals. Assuming that the source signals are non-Gaussian distributed and statistically independent of each other, an evaluation function is constructed to quantify the independence between the separated signals, and an iterative optimization algorithm is used to solve the separation matrix. Step S2 specifically includes the following steps:

[0057] S21: Construct a blind source separation mathematical model based on the mathematical representation relationship between the independent sound source signal and the observed sound signal, thereby constructing a signal separation matrix;

[0058] Specifically, ignoring the influence of factors such as time delay and sound field reverberation, the blind source separation mathematical model is a linear mixing model. The model is described as follows: multiple independent sound source signals that are statistically independent of each other are linearly mixed and received by multiple sound collection elements 1. Each collected observation sound signal is regarded as a linear combination of these multiple independent sound source signals, as shown in Formula 1.1:

[0059]

[0060] In this formula, h ji(i=1,2,...,n;j=1,2,...,m) represents the mixing parameters; s i is the source signal; x j is the acquisition signal. Equation 1.1 can be expressed in vector form as:

[0061] X=HS(1.2)

[0062] In this formula, H∈R n , represents the signal mixing matrix; S∈R n , represents the source signal matrix; X∈R n , represents the acquisition signal matrix.

[0063] The fundamental purpose of the blind source separation algorithm is to separate the source signal matrix S as much as possible under the premise that the collected signal X is known but the signal mixing matrix H and the source signal matrix S are unknown. To achieve this goal, it is necessary to construct a signal separation matrix W. After the collected signal X is transformed by the separation matrix W, an n-dimensional vector matrix Y is obtained, which is the estimated source signal, denoted as Y = [y1, y2, ..., y n ] T .

[0064] Therefore, the solution to the ICA problem is:

[0065] Y=WX=WHS(1.3)

[0066] When WH=I (I is an n×n unit matrix) in Equation 1.3, then Y=S, which means that the source signal is successfully separated from the observed sound signal.

[0067] S22: Construct an evaluation function based on the criterion of maximizing negative entropy, and use negative entropy to measure the non-Gaussian degree of the separated signal;

[0068] The evaluation function is used to evaluate the independence of the separated signals. By selecting an appropriate evaluation function, the computational efficiency of the subsequent iterative algorithm can be improved. This embodiment constructs an evaluation function based on the criterion of maximizing negative entropy, and measures the non-Gaussian degree of the separated signals by negative entropy.

[0069] For any signal y i , whose negative entropy is defined as:

[0070] J(y i )=H(y gauss )-H(y i )(1.4)

[0071] Signal y i Any linear transformation of will not lead to the change of negative entropy, and the negative entropy value is not positive. According to formula 1.9, only the signal y i Negative entropy is 0 only when the Gaussian distribution is satisfied. iWhen the negative entropy of the output signal Y reaches its maximum value, it is considered that the independence of the signal is the greatest, that is, the purpose of signal separation is achieved. The relationship between the negative entropy of the output signal Y and its mutual information is:

[0072]

[0073] According to formula 1.10, when the negative entropy of each component of signal Y and When is the maximum, the mutual information of each component I(Y) is the minimum. Therefore, the evaluation function ρ(Y) of the independent component analysis algorithm based on negative entropy can be obtained as follows:

[0074]

[0075] S23: Construct an iterative optimization process based on FastICA, substitute the collected multiple observation sound signals into it, and iteratively calculate the corresponding separation signal matrix when the evaluation function takes the extreme value. This matrix is ​​the separation signal to be solved. The process of step S23 is as follows Figure 4 shown.

[0076] After determining the evaluation function, an iterative algorithm can be applied to solve the separation signal matrix. This embodiment adopts a fixed-point algorithm based on negative entropy. This method obtains the projection vector through a fixed-point iteration method, so that the non-Gaussianity of the component of the signal projection on the projection vector is maximized, and thus the iterative convergence speed is extremely fast.

[0077] The evaluation function is constructed using the criterion of maximizing negative entropy. Negative entropy can be approximately expressed as:

[0078] J(y i )≈k[E{G(y i )}-E{G(y gauss )}] 2 (1.7)

[0079] In this formula, y gauss is a random variable with zero mean, G(.) is a non-quadratic function, and k is a constant. To maximize the negative entropy J(Y), let y i =h i T v, at this time h i Taking the derivative we get:

[0080]

[0081] When the negative entropy J(Y) reaches its maximum value, E{G(Y)} also reaches its maximum value, and the following conditions are met:

[0082] F(h i )=E[vg(h i T v)]-βh i=0 (1.9)

[0083] g(.) is the derivative of the function G(.), β is the Lagrange multiplier, and Newton's iteration method yields:

[0084]

[0085] Since E[vv in formula (1.9) T g'(h i T (k)v)]≈E[vv T ]E[g'(h i T (k)v)]=E[g'(h i T (k)v)], so Equation (1.9) can be simplified to:

[0086]

[0087] Formula 1.10 is the FastICA iterative algorithm process constructed based on the negative entropy maximization criterion. After repeated iterative calculations, the separation signal matrix can be finally obtained.

[0088] Compared to traditional ICA iterative algorithms, such as the conventional gradient method and the stochastic gradient method, which suffer from slow convergence and sensitivity to iteration step size, the negative entropy-based fixed-point algorithm (FastICA) is used for blind signal separation of multi-channel acoustic signals. This algorithm directly optimizes the projection vectors through a fixed-point iterative strategy to maximize the non-Gaussianity of the signal components (the negative entropy criterion), achieving superlinear convergence (up to quadratic convergence) while ensuring separation accuracy, significantly improving computational efficiency. Compared to the limitations of principal component analysis (PCA), which relies on second-order statistical decorrelation, and joint approximate diagonalization (JADE), which requires the construction of high-order tensors, FastICA, based on the non-Gaussian assumption, can separate source signals with complex statistical dependencies without predefined signal distributions, demonstrating greater robustness in acoustic scenarios with low signal-to-noise ratios and unknown mixing matrices.

[0089] Independent component analysis effectively decouples the processed signal from the processing sound signal and ambient equipment noise, ensuring consistent and reliable processing results and reducing processing errors caused by misidentification. A blind source separation algorithm, in real time, resolves the nonlinear coupling relationship between water jet turbulence noise, air compressor noise, pneumatic booster pump noise, and laser impact noise. It dynamically separates the processing sound signal using the negentropy maximization principle, demonstrating strong dynamic anti-interference capabilities.

[0090] Based on the fixed point iterative algorithm (FastICA), independent signals are separated from the aliased signal matrix, which not only ensures signal accuracy but also ensures fast convergence speed and improves algorithm efficiency.

[0091] Furthermore, in order to optimize the computational efficiency of the ICA algorithm, before the collected multiple observation sound signals are substituted into the iterative optimization process based on FastICA in step S23, the collected multiple observation sound signals need to be preprocessed, including de-averaging, whitening processing and principal component analysis.

[0092] The signal de-averaging process is to satisfy the zero mean constraint in the constraint condition that there is a unique solution for the blind source separation algorithm. N ] T , the signal de-averaging processing formula is:

[0093]

[0094] Signal whitening is used to eliminate correlations between the components of the acquired signal. Currently, there are two methods for obtaining the whitening matrix: performing a linear iterative transformation on the mixing matrix and obtaining it through eigenvalue decomposition of the acquired signal matrix. Considering the real-time requirements of the algorithm, this embodiment uses eigenvalue decomposition to obtain the whitening matrix.

[0095] Specifically, let R x is the correlation matrix of the collected signal. According to the basic properties of the matrix, the matrix R x There exists an eigenvalue decomposition that satisfies:

[0096] R x =Q∑ 2 Q T (1.12)

[0097] In formula 1.12, R x The eigenvalues ​​of are λ1,λ2,...,λ n ,∑ 2 It is a diagonal matrix, and the elements on the diagonal are λ1 2 ,λ2 2 ,...,λ n 2 , the column vector of the matrix Q is the standard orthogonal eigenvector corresponding to the eigenvalue, then the whitening matrix T can be obtained:

[0098] T=∑ -1 Q T (1.13)

[0099] The purpose of principal component analysis is to reduce the dimensionality and remove noise of the signal while retaining the original signal. Its principle is to obtain a set of new features of the same number from the original features of the original acquired signal through linear transformation, and this set of new features contains the main information of the original signal features.

[0100] The principal component analysis process needs to go a step further on the basis of signal whitening processing, and the eigenvalue λ i (i=1,2,...,n) are arranged in descending order, and η is defined as the information contribution rate of the signal:

[0101]

[0102] In this formula, the numerator is the sum of the first m eigenvalues ​​after the acquisition signal matrix is ​​sorted, and the denominator is the sum of all eigenvalues. The value of m is determined according to the contribution rate η. Usually, η ≥ 0.8 is set, and the processed signal can be obtained:

[0103]

[0104] Where V=[v1,v2,...,v m ],v i is the eigenvalue λ i The corresponding eigenvector; The m×n variable matrix is ​​constructed from the first m most informative eigenvectors. At this point, the original n variables are converted into m variables, and these m variables are sufficient to describe the information described by the original n variables, removing the overlapping information.

[0105] In some embodiments, step S3 includes the following steps:

[0106] S31: performing spectrum analysis on the multiple independent sound source signals to obtain frequency domain graphs of multiple separated signals;

[0107] S32: comparing the frequency domain graphs of the plurality of independent sound source signals with the frequency characteristics of the sound signals generated when the laser processes the material, so as to confirm the sound signals generated by the material removal during the water-conducting laser processing.

[0108] In some embodiments, step S4 includes the following steps:

[0109] S41: Processing the frequency domain graph of the sound signal generated by material removal to obtain a time domain graph;

[0110] S42: Performing short-time energy analysis on the obtained time domain graph to obtain a graph of the energy of the sound signal generated by material removal changing with time, which reflects the water-guided laser processing process.

[0111] Example 1

[0112] This embodiment uses NdFeB permanent magnet material as the processing material and monitors its through-hole processing process in a water-guided laser processing system. The main process is as follows:

[0113] First, four acoustic sensors are used to directly collect the sound signals generated during the processing as the observation sound signals. A section of each of the four observation sound signals is intercepted to obtain the original signal time domain waveform as shown in the following figure: Figure 5 shown.

[0114] Secondly, the blind source separation algorithm is used to Figure 5 The four segments of original signals shown in the figure are processed, and the time domain waveforms of the four separated signals are obtained as follows Figure 6 shown.

[0115] Due to the nature of the blind source separation algorithm, the amplitudes of the four separated signals have changed, and the order of the four separated signals has also changed, making them unable to correspond one-to-one with the original signals. From the time domain waveform, it is impossible to clearly distinguish the specific meaning of each separated signal. It is not possible to effectively establish a judgment basis based solely on the time domain waveform information of the four separated signals. Therefore, the spectrum analysis of the four separated signals is performed, and the frequency domain diagram of the four signals after processing is as follows: Figure 7 shown.

[0116] exist Figure 7 In the figure, based on the equipment characteristics of the water-guided laser processing system, the differences between the four separated signals can be distinguished based on the spectrum of the separated signals. Specifically, the spectrum energy of the first separated signal is concentrated in the low frequency band, with a peak at 35Hz, which is similar to the sound signal of an air compressor; the spectrum energy of the second separated signal is also concentrated in the low frequency band, with a peak at 100Hz, which is the sound of the high-pressure water system; the overall distribution of the third separated signal is relatively uniform, which is the sound of the water jet; the fourth separated signal has large peaks at 1000Hz, 2000Hz, and 3000Hz, which is the sound signal generated by laser processing materials; therefore, the time domain waveform of the fourth separated signal can more realistically reflect the entire actual processing process.

[0117] The sound signals of the entire processing process are processed and the time domain waveform of the sound signals generated when the laser processing material is separated is shown as follows: Figure 8 As shown. The sound signal generated by the separated laser processing material is analyzed in a short time, as shown Figure 9 As shown in the figure, at 4s, the signal energy rises sharply. This is because the laser starts to output at this time, the surface of the material is ablated, the light-heat conversion triggers a violent phase change, the metal vapor and plasma expand instantaneously to produce high-frequency pressure pulses, and the water jet contacts the high-temperature molten pool at this time, triggering a transient cavitation effect, accompanied by a micro-explosion when the slag is peeled off; from 4 to 34s, the overall energy shows a downward trend. This is because as the laser gradually penetrates into the material, the sound energy is absorbed by the processed material itself; at 34s, the energy rises sharply again, and the perforation is successful at this time. When the high-pressure water flow is ejected from the bottom slit, a whistling sound is caused, and at the same time, the water jet penetrates and hits the base.

[0118] In summary, it can be determined that the sound signal separated by the ICA algorithm can be used to characterize the laser processing process, confirm the sound signal characteristics during water-conducting laser processing of materials, and the feasibility of perforation monitoring characterization based on this.

[0119] It should be noted that although the present invention is disclosed above with specific embodiments, the above embodiments are not intended to limit the present invention. Ordinary technicians in this field can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be based on the scope defined by the claims.

Claims

1. A method for monitoring the cutting penetration status during water-guided laser processing based on blind source separation of acoustic signals, characterized in that: The following steps are involved: S1: During the water-guided laser processing, multiple observation sound signals are collected in real time. The observation sound signals are mixed sound signals composed of multiple independent sound source signals. S2: Based on the collected multiple observation sound signals, a blind source separation algorithm is used to estimate and solve multiple independent sound source signals; S3: comparing the multiple independent sound source signals with the sound signals generated when the laser processes the material, so as to confirm the sound signals generated by the material removal during the water-conducting laser processing; S4: judging whether cutting is penetrating during the water-guided laser processing process based on the relationship between the sound signal generated by material removal and time.

2. The method for monitoring the cutting penetration status during water-guided laser processing based on blind source separation of acoustic signals according to claim 1, characterized in that: Step S2 specifically includes the following steps: S21: Construct a blind source separation mathematical model based on the mathematical representation relationship between the independent sound source signal and the observed sound signal, thereby constructing a signal separation matrix; S22: Construct an evaluation function based on the criterion of maximizing negative entropy, and use negative entropy to measure the non-Gaussian degree of the separated signal; S23: Construct an iterative optimization process based on FastICA, substitute the collected multiple observation sound signals into it, and iteratively calculate the separation signal matrix corresponding to the extreme value of the evaluation function. This matrix is ​​the separation signal to be solved.

3. The method for monitoring the cutting penetration status during water-guided laser processing based on blind source separation of acoustic signals according to claim 2, characterized in that: In step S21, the blind source separation mathematical model is a linear mixture model.

4. The method for monitoring the cutting penetration status during water-guided laser processing based on blind source separation of acoustic signals according to claim 2, characterized in that: Before the collected multiple observation sound signals are substituted into the iterative optimization process based on FastICA in step S23, the collected multiple observation sound signals need to be preprocessed, including de-averaging, whitening processing and principal component analysis.

5. The method for monitoring the cutting penetration status during water-guided laser processing based on blind source separation of acoustic signals according to claim 2, characterized in that: When the collected multiple observation sound signals are substituted into the iterative optimization process based on FastICA in step S23, only a section of the multiple observation sound signals is intercepted and substituted into the iterative optimization process based on FastICA.

6. The method for monitoring the cutting penetration status during water-guided laser processing based on blind source separation of acoustic signals according to claim 1, characterized in that: Step S3 includes the following steps: S31: performing spectrum analysis on the multiple independent sound source signals to obtain frequency domain graphs of the multiple independent sound source signals; S32: comparing the frequency domain graphs of the plurality of independent sound source signals with the frequency characteristics of the sound signals generated when the laser processes the material, so as to confirm the sound signals generated by the material removal during the water-conducting laser processing.

7. The method for monitoring the cutting penetration status during water-guided laser processing based on blind source separation of acoustic signals according to claim 6, characterized in that: Step S4 includes the following steps: S41: Processing the frequency domain graph of the sound signal generated by material removal to obtain a time domain graph; S42: Performing short-time energy analysis on the obtained time domain graph to obtain a graph of the energy of the sound signal generated by material removal changing with time, which reflects the water-guided laser processing process.

8. A water-guided laser processing process monitoring system for implementing the method for monitoring the cutting penetration state in a water-guided laser processing process based on blind source separation of acoustic signals as claimed in any one of claims 1 to 7, characterized in that: include: Multiple sound collection elements, each used to collect in real time an observation sound signal generated during the water-guided laser processing process, wherein the observation sound signal is a mixed sound signal composed of multiple independent sound source signals, and the number of the sound collection elements is not less than the number of the independent sound source signals; a signal separation module, which establishes communication connections with the plurality of sound collection elements respectively, for receiving a plurality of observation sound signals collected by the plurality of sound collection elements, and estimating and solving a plurality of separated signals based on a blind source separation algorithm; a comparison and confirmation module electrically connected to the signal separation module and configured to receive the plurality of separated signals; the comparison and confirmation module being configured to compare the plurality of separated signals with the sound signals generated during the laser processing of the material, so as to confirm that the sound signal is generated by the removal of the material during the water-conducting laser processing; The judgment module is in communication with the comparison and confirmation module, and is used to receive the sound signal generated by material removal, and can judge the water-guided laser processing process according to the relationship between the energy of the sound signal generated by material removal and time.

9. The water-guided laser processing process monitoring system according to claim 8, characterized in that: It also includes a signal amplifier, a data acquisition card and an industrial computer. A plurality of the sound collection elements are connected in parallel and electrically connected to the signal amplifier, the data acquisition card and the industrial computer in sequence. The industrial computer is equipped with the signal separation module, the comparison and confirmation module and the judgment module.

10. The water-guided laser processing process monitoring system according to claim 8, characterized in that: At least one sound collecting element is provided at the generating location of each independent sound source signal.

Images