Damage acoustic emission source positioning method based on fruit fly optimization independent variational mode decomposition

Abstract

The invention discloses a damage acoustic emission source positioning method based on fruit fly optimization independent variational mode decomposition, which comprises the steps of acoustic emission signal acquisition and sample selection: acquiring laser cladding acoustic emission signals of a metal plate under different process parameters, and selecting a verification data set from the laser cladding acoustic emission signals; preprocessing the original acoustic emission signal: preprocessing the original acoustic emission signal in a laser cladding experiment; signal decomposition and source positioning: decomposing the preprocessed acoustic emission signal, and finding the coordinate of an acoustic emission source by using the fruit fly optimization independent variational mode decomposition method; optimizing the positioning result of the acoustic emission source: dynamically updating the threshold according to the attenuation characteristic of the acoustic emission signal, and optimizing the positioning result of the acoustic emission source; according to the structural damage acoustic emission source positioning method based on fruit fly optimization independent variational mode decomposition, the independent variational mode decomposition method is optimized, and the fruit fly optimization algorithm is combined, so that the acoustic emission source positioning precision is remarkably improved.

Description

Technical Field

[0001] The invention relates to the field of damage acoustic emission source positioning, and in particular to a damage acoustic emission source positioning method based on fruit fly optimized independent variational mode decomposition. Background Art

[0002] With the continuous development of engineering technology, the complexity of mechanical equipment and structures has gradually increased, and it has become particularly important to ensure their safety and stability. In this context, structural health monitoring (SHM) technology has gradually become one of the core technologies to ensure the safety of equipment and extend its service life. As a non-destructive testing method, acoustic emission (AE) technology has been widely used in health monitoring and fault diagnosis of metal structures, composite materials, concrete structures and various mechanical equipment because it can monitor and locate damage and defects inside materials or on the surface of equipment in real time. Acoustic emission technology obtains damage information of materials or structures by detecting stress waves caused by physical processes such as crack propagation, corrosion, and fatigue. Since the AE signal can reflect the progress of damage inside the material, its real-time and sensitivity give it a unique advantage in structural health monitoring. However, the source location problem of AE signals remains a challenging research direction.

[0003] At present, the methods for locating acoustic emission sources mainly include: time domain analysis method, frequency domain analysis method, and time-frequency domain analysis method. In practical applications, the time domain analysis method is affected by factors such as multipath propagation, noise interference, and signal attenuation, and the positioning accuracy is easily reduced, and the calculation complexity is high, especially in large-scale monitoring systems, and its real-time and accuracy are difficult to guarantee. The frequency domain analysis method relies on accurate preprocessing and feature selection for the extraction of signal features. When the signal is more complex or there are multiple interferences, how to extract more recognizable feature information is still a difficult problem. In addition, when facing multi-modal and multi-scale complex signals, frequency domain analysis often faces problems such as mode overlap or inaccurate mode selection, resulting in inaccurate positioning results. Time-frequency domain analysis method: At present, many time-frequency analysis methods are dedicated to acoustic emission source positioning. These representative methods include wavelet transform (WT), ensemble empirical mode decomposition (EEMD), local mean decomposition (LMD) and intrinsic time scale decomposition (ITD). However, these methods have some obvious shortcomings when processing acoustic emission signals. For example, wavelet transform can preserve the local edge features of the signal, but due to some defects in the selected threshold function (such as discontinuity and large errors), the variance of the denoising result is large. In practical applications, EEMD and LMD will have disadvantages such as modal aliasing and end effect, which will bring great difficulties to the extraction of acoustic emission signal features. In addition, since the operation process of the inherent time scale decomposition adopts linear transformation, it is easy to have problems such as "waveform burrs" and "curve distortion" when processing nonlinear acoustic emission signals.

[0004] In addition, in order to solve the shortcomings of traditional time domain and frequency domain analysis methods, variational mode decomposition (VMD) is proposed as a new signal processing method. VMD effectively solves the difficulties in nonlinear and non-stationary signal analysis by decomposing complex signals into multiple modes with different frequency bands. When decomposing signals, VMD can select the number of modes in an adaptive manner and accurately process different modes to extract the potential characteristic information of the signal. Although VMD has shown good performance in processing complex signals, it still faces certain limitations. For example, the mode selection problem and mode overlap problem of VMD, especially in a high-noise environment, may lead to inaccurate mode decomposition, thereby affecting the subsequent source localization accuracy. In addition, the performance of VMD is strictly limited by its key parameters (i.e., modulus K and penalty parameter a). Improper setting of the modulus will cause over-decomposition or under-decomposition of the signal, thereby affecting the accuracy of signal decomposition. Therefore, the present application proposes a fruit fly optimized independent variational mode decomposition method to solve the problem that the current VMD key parameters are difficult to accurately determine and improve the signal decomposition capability of VMD. Summary of the invention

[0005] In order to overcome the shortcomings and deficiencies of the prior art, the present invention provides a method for locating damage acoustic emission sources based on fruit fly optimized independent variational mode decomposition.

[0006] The technical solution adopted by the present invention is a method for locating damage acoustic emission sources based on fruit fly optimized independent variational mode decomposition, the method comprising:

[0007] Step S1: Acoustic emission signal collection and sample selection: Collect the laser cladding acoustic emission signals of the metal plate under different process parameters, and select the verification data set from the laser cladding acoustic emission signals;

[0008] Step S2: Preprocessing of the original acoustic emission signal: In the laser cladding experiment, the original acoustic emission signal is preprocessed using an analog bandpass filter with a frequency range of 40 kHz to 450 kHz;

[0009] Step S3: signal decomposition and source location: decompose the preprocessed acoustic emission signal and find the coordinates of the acoustic emission source using the proposed fruit fly optimized independent variational mode decomposition method;

[0010] Step S4: Optimization of acoustic emission source localization results: According to the attenuation characteristics of the acoustic emission signal, the threshold is dynamically updated to optimize the acoustic emission source localization results; in this step, the time difference between each sensor and all acoustic emission events is first extracted based on the fixed threshold A0; the coordinates of the acoustic emission source (x s ,y s ), and calculate the coordinates and the i-th sensor (x i ,y i ) i , the expression is:

[0011]

[0012] The updated dynamic threshold DA of the i-th sensor i It is expressed as follows:

[0013]

[0014] Among them, d min =min(d i ), SA(·) is the attenuation function of the amplitude.

[0015] The optimal time difference of each sensor is obtained by using the updated threshold, and then substituted into the proposed fruit fly optimized independent variational mode decomposition algorithm to optimize the coordinates of the acoustic emission source;

[0016] Step S5: Comparative analysis with existing methods: Through comparative analysis with existing acoustic emission source localization methods, the effectiveness of the proposed structural damage acoustic emission source localization method based on fruit fly optimized independent variational mode decomposition is verified.

[0017] Furthermore, the fruit fly optimization independent variational mode decomposition includes:

[0018] Step A1: fitness function construction;

[0019] Step A2: optimal IMF component determination;

[0020] Step A3: Chaos operation;

[0021] Step A4: fitness value calculation;

[0022] Step A5: Iterative judgment.

[0023] Furthermore, the step A1: constructing a fitness function constructs a fitness function. Based on the definition of information entropy, the marginal spectral entropy Hp is used to measure the uncertainty and frequency complexity of the signal in the frequency domain. The expression is:

[0024]

[0025] Among them, H p represents the marginal spectral entropy value, p is the value used to measure the probability of a specific event in the system, P i represents the probability of the amplitude corresponding to the i-th frequency, h(i) represents the marginal spectrum corresponding to the i-th IMF component, N represents the total number of events, i represents the index of the event, and p i is the probability of event i occurring; the marginal spectral entropy corresponding to the IMF component is normalized, that is, H E represents the normalized entropy value, L is the length of the marginal spectrum h(i), and E represents the sign of entropy.

[0026] Furthermore, in step A2: determining the optimal IMF component, the minimum value of the marginal spectral entropy corresponding to the IMF component is taken as the optimal IMF component, that is, the marginal spectral entropy corresponding to the IMF component is called the local minimum value, and the expression is:

[0027]

[0028] Where L is the length of the marginal spectrum h(i), i represents the index of the event, represents the entropy value of the IMF component, min represents the minimization operation, and p i is the probability of event i occurring, and IMF represents the intrinsic mode component.

[0029] 5. The method for locating the damage acoustic emission source based on fruit fly optimized independent variational mode decomposition according to claim 2, characterized in that the step A3: chaos operation, initializing the number of fruit flies N, the maximum number of iterations T, and the upper limit T max At the same time, the relevant parameters of chaos are initialized, and two fruit fly populations with a population size of N are randomly obtained, namely U a and U K , and set the fitness value Fitbest≈0. a and U K Each component of is mapped to a chaotic variable and And complete the chaos operation.

[0030] Furthermore, in step A4: fitness value calculation, take a K-scale IMF component u k (t), k = 1, 2, ..., K composed of the sampling signal x(t), t = 1, 2, ..., t m Initialize it and perform Hilbert transform on IMF components of different scales to obtain the analytical signal U k (t), the expression is:

[0031]

[0032] Where t represents time, δ(t) is the Dirac function, and U k (t) represents the output analytical signal, represents a complex term, j is an imaginary unit, indicating that the signal has a complex part, and u k (t) represents the spectrum expression of the input signal, k represents u k (t) index, the symbol * indicates the convolution operation. k (t) and the estimated center frequency Modulate the spectrum uk(t) to the corresponding root frequency band, that is:

[0033]

[0034] Among them, F k (f) represents the signal u obtained after transformation k (t) is represented in the frequency domain, f represents the frequency domain characteristics of the signal, e is the base of the natural logarithm, j is the imaginary unit, ω k represents the center frequency. u is estimated by calculating the L2 norm of the signal gradient k The bandwidth of (t). The constraints are introduced and the optimal variational model is established, which is specifically expressed as follows:

[0035]

[0036] Among them, ω k Indicates u k (t), x(t) represents the input signal, and K is the set constant value. The quadratic penalty factor β and Lagrangian multiplier γ(t) are introduced to construct the extended Lagrangian function, which is expressed as:

[0037]

[0038] Where L(·) represents the Lagrangian function. Therefore, all IMF components can be calculated as follows:

[0039]

[0040] in, represents the updated value of the kth IMF component in the (n+1)th iteration, where n represents the number of iterations. represents the frequency domain representation of the target signal x(t), represents the i-th signal component u i (t) is represented in the frequency domain, represents the frequency domain representation of the Lagrange multiplier γ(t), ω is the angular frequency, and β is the quadratic penalty factor. Select N effective IMF components to construct the input observation signal, and whiten and zero-center it. At the same time, a kernel function k(·) is given, and the signal estimation vector S = {s1, s2, …s n}'s Gram matrix G=(G1,G2,…G m ). Set λ = (G1, G2, ... G m ) is the maximum eigenvalue of formula (13), and the expression is:

[0041]

[0042] The fitness value is calculated according to the above process, and the optimal concentration value and its corresponding position are obtained.

[0043] Furthermore, the step A5: iterative judgment, according to the maximum number of iterations preset in step A1, judge whether the current number of iterations meets the stop condition. If the stop condition is met, complete steps A1-A5 and output the optimal parameters of the proposed fruit fly optimized independent variational mode decomposition algorithm. Otherwise, return to step A2 to continue iterating until the stop condition is met.

[0044] Beneficial effects:

[0045] This application proposes a method for locating a damage acoustic emission source based on fruit fly optimized independent variational mode decomposition, which enhances the performance of acoustic emission signal feature extraction and improves the accuracy of acoustic emission source positioning. In order to effectively solve the problems mentioned in the background technology, this application proposes a method for locating a damage acoustic emission source based on fruit fly optimized independent variational mode decomposition, and explains the beneficial effects one by one, as follows:

[0046] 1. Optimized acoustic emission source localization accuracy: The fruit fly optimized independent variational mode decomposition (FO-IVMD) method proposed in the present invention shows significant advantages in processing high-noise, nonlinear, and non-stationary acoustic emission signals. By introducing the fruit fly optimization algorithm and dynamically adjusting the key parameters of VMD (such as the modulus K and the penalty parameter a), the common problems of traditional VMD methods in processing complex signals, such as inaccurate mode selection, mode overlap, and mode over-decomposition or under-decomposition, are effectively overcome, thereby improving the accuracy of signal decomposition. Through more precise signal decomposition, this method can more accurately capture and extract potential feature information in acoustic emission signals, thereby improving the accuracy of acoustic emission source localization.

[0047] 2. Reduce the noise interference and signal attenuation problems of traditional methods: The independent variational mode decomposition method based on fruit fly optimization can reduce the impact of noise by optimizing the key parameters of mode decomposition, and can more accurately extract effective signal features under various signal interference and attenuation conditions, thereby effectively improving the stability and reliability of source positioning.

[0048] 3. Dynamic threshold optimization improves source localization accuracy: The present invention optimizes the localization results of the acoustic emission source by dynamically updating the threshold according to the attenuation characteristics of the acoustic emission signal. This innovative strategy enables source localization to adapt to different signal attenuation conditions, avoiding the problem that the fixed threshold in the traditional method cannot adapt to signal changes, thereby further improving the localization accuracy. Through this dynamic optimization mechanism, the present method can adjust and optimize the source localization results in real time, especially under complex process conditions, showing high robustness and accuracy that is superior to traditional methods.

[0049] 4. Independent mode decomposition avoids mode aliasing and end effect: The FO-IVMD method proposed in the present invention avoids mode aliasing and end effect in traditional methods by optimizing the mode selection process, ensuring more accurate signal decomposition, thereby effectively improving the accuracy of source positioning.

[0050] 5. Solve the problem of inaccurate modal decomposition in high-noise environments: The present invention can automatically optimize the key parameters of VMD through the fruit fly optimization algorithm to ensure the accuracy of modal decomposition. By optimizing the modal parameters, the present invention can effectively reduce the impact of high-noise environments on the accuracy of acoustic emission signal decomposition and ensure the reliability of source localization results.

[0051] 6. It has broad application prospects and promotion value: The fruit fly optimized independent variational mode decomposition method of the present invention not only has high acoustic emission source positioning accuracy, but also can be applied to damage detection and source positioning problems in various complex working environments and conditions, and has broad application prospects. This method can be widely used in health monitoring in the fields of metal structures, composite materials, concrete structures, etc., especially in laser cladding, welding, corrosion and other processes, and can effectively provide high-precision acoustic emission source positioning, providing strong support for structural health monitoring, fault diagnosis and safety assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 is a general step flow chart of the present invention;

[0053] Figure 2 This is a specific framework diagram of the fruit fly optimized independent variational mode decomposition algorithm of the present invention. DETAILED DESCRIPTION

[0054] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application may be combined with each other. The present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0055] like Figure 1 As shown, a method for locating damage acoustic emission sources based on fruit fly optimization independent variational mode decomposition includes:

[0056] Step S1: Acoustic emission signal collection and sample selection: Collect the laser cladding acoustic emission signals of the metal plate under different process parameters, and select the verification data set from the laser cladding acoustic emission signals;

[0057] Specifically, this step includes first collecting acoustic emission signals of laser cladding of metal plates under different process parameters through experiments or monitoring equipment. Acoustic emission signals are stress waves generated by physical phenomena such as crack propagation and corrosion inside the material, and these signals can reflect the damage process of the material. After collecting the signals, a sample data set suitable for verification is selected from these acoustic emission signals. The selection of the verification data set should ensure that different working conditions and fault states are covered to ensure the diversity and accuracy of subsequent model training and verification.

[0058] Step S2: Preprocessing of the original acoustic emission signal: In the laser cladding experiment, the original acoustic emission signal is preprocessed using an analog bandpass filter with a frequency range of 40 kHz to 450 kHz;

[0059] Specifically, this step includes preprocessing the collected original acoustic emission signals to remove noise and irrelevant signal components to ensure the validity of the data. Specifically, in the laser cladding experiment, an analog bandpass filter with a frequency range of 40kHz to 450kHz is used to preprocess the acoustic emission signals. This frequency range is selected because the acoustic emission signals in the laser cladding process are mainly concentrated in this frequency range. The filter can effectively remove low-frequency and high-frequency interference components, retain effective signal characteristics, and enhance the accuracy of subsequent analysis.

[0060] Step S3: signal decomposition and source location: decompose the preprocessed acoustic emission signal and find the coordinates of the acoustic emission source using the proposed fruit fly optimized independent variational mode decomposition method;

[0061] Specifically, this step includes using the proposed fruit fly optimized independent variational mode decomposition method to perform modal decomposition on the signal to extract useful signal components. The proposed fruit fly optimized independent variational mode decomposition method can decompose complex signals into multiple IMF components, each of which represents a different frequency component in the signal. Through this decomposition method, the detailed features in the acoustic emission signal can be effectively captured, and then the source location analysis can be performed. Using these decomposed modal components, by solving the optimal acoustic emission source coordinates (x s ,y s ) to determine the location of the acoustic emission source.

[0062] Step S4: Optimization of acoustic emission source localization results: According to the attenuation characteristics of the acoustic emission signal, the threshold is dynamically updated to optimize the acoustic emission source localization results; in this step, the time difference between each sensor and all acoustic emission events is first extracted based on the fixed threshold A0; the coordinates of the acoustic emission source (x s ,y s ), and calculate the coordinates and the i-th sensor (x i ,y i ) i , the expression is:

[0063]

[0064] Therefore, the updated dynamic threshold DA of the i-th sensor is i It is expressed as follows:

[0065]

[0066] Among them, d min =min(d i ), SA(·) is the attenuation function of the amplitude.

[0067] The optimal time difference of each sensor is obtained by using the updated threshold, and then substituted into the proposed fruit fly optimized independent variational mode decomposition algorithm to optimize the coordinates of the acoustic emission source;

[0068] Specifically, this step includes optimizing the localization result of the acoustic emission source by utilizing the attenuation characteristics of the acoustic emission signal. First, the time difference between each sensor and all acoustic emission events is extracted based on a fixed threshold A0. Then, the coordinates (x s ,y s ), and calculate the coordinates and the i-th sensor (x i ,y i ) i According to these calculation results, the amplitude attenuation function (d min =min(d i )) to dynamically update the threshold DA of each sensor i , in order to more accurately adjust the parameters of source localization. The updated dynamic threshold can better adapt to different signal attenuation characteristics, thereby further optimizing the localization accuracy of the acoustic emission source. Finally, using these updated thresholds, the coordinates of the acoustic emission source are optimized again through the fruit fly optimization algorithm to ensure the most accurate localization results.

[0069] Step S5: Comparative analysis with existing methods: Through comparative analysis with existing acoustic emission source localization methods, the effectiveness of the proposed structural damage acoustic emission source localization method based on fruit fly optimized independent variational mode decomposition is verified.

[0070] Specifically, this step includes verifying the effectiveness of the proposed method for locating structural damage acoustic emission sources based on fruit fly optimized independent variational mode decomposition by comparative analysis with existing acoustic emission source localization methods. Through experimental or simulation data, the method is compared with traditional methods (such as PAC acoustic emission acquisition system, Newton method and multiple cross-correlation method based on Geiger algorithm, deep learning and Bayesian method) in terms of positioning accuracy, robustness and computational efficiency. The results of the comparative analysis can fully demonstrate the advantages of this method in complex signal processing and source localization accuracy, and verify its feasibility and effectiveness in practical applications.

[0071] Fruit fly optimization independent variational mode decomposition, including:

[0072] Step A1: fitness function construction;

[0073] Step A2: optimal IMF component determination;

[0074] Step A3: Chaos operation;

[0075] Step A4: fitness value calculation;

[0076] Step A5: Iterative judgment.

[0077] like Figure 2 From the flow of the fruit fly optimization independent variation mode decomposition algorithm shown in the figure, we can see that first, the maximum value of the population size and the number of iterations is set. At this stage, the basic parameters of the algorithm need to be initialized, including setting the population size and the maximum value of the number of iterations. These parameters have an important impact on the performance of the algorithm, and reasonable settings can ensure the convergence and solution effect of the algorithm. Then, the fruit fly population is initialized. In this step, the fruit fly population is randomly initialized, and the random direction and distance of the individual are given. Each fruit fly individual represents a potential solution and is generated through the initialization process. Secondly, the random direction and distance of the individual are given. In this stage, a random direction and distance are assigned to each fruit fly to determine their initial position in the search space. Thirdly, the marginal spectral entropy is selected as the judgment value of the objective function to compare and optimize the fitness of the fruit fly. Subsequently, the taste concentration fitness value of the population is calculated. In this link, the algorithm calculates the taste concentration fitness value of each fruit fly individual. These fitness values ​​reflect the degree of pros and cons of the individual in the search space and are the criteria selected in the fruit fly optimization process. Next, a new population gathering position is obtained. After calculation, the fruit fly algorithm updates the population position and finds a new population gathering position by evaluating the taste fitness value. This new position represents a better search area. Again, check whether the stopping conditions are met, such as reaching the maximum number of iterations or finding a sufficiently good solution. If the conditions are met, the algorithm enters the end stage; otherwise, it continues to execute. Then, the optimal internal parameters of the proposed method are output. If the termination conditions are met, the algorithm will output the optimal internal parameters, that is, the relevant parameters of the current optimal solution. These parameters are used for subsequent optimization tasks. Finally, run the optimal fruit fly optimization independent decomposition model. After outputting the optimal parameters, run the fruit fly optimization independent variational mode decomposition algorithm to perform the final model solution and optimization to complete the task objectives.

[0078] Step A1: Fitness function construction: construct a fitness function. Based on the definition of information entropy, use the marginal spectral entropy Hp to measure the uncertainty and frequency complexity of the signal in the frequency domain. The expression is:

[0079]

[0080] Among them, H p represents the marginal spectral entropy value, p is the value used to measure the probability of a specific event in the system, P i represents the probability of the amplitude corresponding to the i-th frequency, h(i) represents the marginal spectrum corresponding to the i-th IMF component, N represents the total number of events, i represents the index of the event, and p i is the probability of event i occurring; the marginal spectral entropy corresponding to the IMF component is normalized, that is, HE represents the normalized entropy value, L is the length of the marginal spectrum h(i), and E represents the sign of entropy.

[0081] Step A2: Determine the optimal IMF component. Take the minimum value of the marginal spectral entropy corresponding to the IMF component as the optimal IMF component. That is, the marginal spectral entropy corresponding to the IMF component is called the local minimum, and the expression is:

[0082]

[0083] Where L is the length of the marginal spectrum h(i), i represents the index of the event, represents the entropy value of the IMF component, min represents the minimization operation, and p i is the probability of event i occurring, and IMF represents the intrinsic mode component.

[0084] Step A3: Chaos operation, initializing the number of fruit flies N, the maximum number of iterations T, and the upper limit T max At the same time, the relevant parameters of chaos are initialized, and two fruit fly populations with a population size of N are randomly obtained, namely U a and U K , and set the fitness value Fitbest≈0. a and U K Each component of is mapped to a chaotic variable and And complete the chaos operation.

[0085] Step A4: Calculate the fitness value, take a K-scale IMF component u k (t), k = 1, 2, ..., K composed of the sampling signal x(t), t = 1, 2, ..., t m Initialize it and perform Hilbert transform on IMF components of different scales to obtain the analytical signal U k (t), the expression is:

[0086]

[0087] Where t represents time, δ(t) is the Dirac function, and U k (t) represents the output analytical signal, represents a complex term, j is an imaginary unit, indicating that the signal has a complex part, and u k (t) represents the spectrum expression of the input signal, k represents u k (t) index, the symbol * indicates the convolution operation. k (t) and the estimated center frequency The spectrum u k (t) modulated to the corresponding root frequency band, that is:

[0088]

[0089] Among them, F k (f) represents the signal u obtained after transformation k (t) is represented in the frequency domain, f represents the frequency domain characteristics of the signal, e is the base of the natural logarithm, j is the imaginary unit, ω k represents the center frequency. u is estimated by calculating the L2 norm of the signal gradient k The bandwidth of (t). The constraints are introduced and the optimal variational model is established, which is specifically expressed as follows:

[0090]

[0091] Among them, ω k Indicates u k (t), x(t) represents the input signal, and K is the set constant value. The quadratic penalty factor β and Lagrangian multiplier γ(t) are introduced to construct the extended Lagrangian function, which is expressed as:

[0092]

[0093] Where L(·) represents the Lagrangian function. Therefore, all IMF components can be calculated as follows:

[0094]

[0095] in, represents the updated value of the kth IMF component in the (n+1)th iteration, where n represents the number of iterations. represents the frequency domain representation of the target signal x(t), represents the i-th signal component u i (t) is represented in the frequency domain, represents the frequency domain representation of the Lagrange multiplier γ(t), ω is the angular frequency, and β is the quadratic penalty factor. Select N effective IMF components to construct the input observation signal, and whiten and zero-center it. At the same time, a kernel function k(·) is given, and the signal estimation vector S = {s1, s2, …s n}'s Gram matrix G=(G1,G2,…G m ). Set λ = (G1, G2, ... G m ) is the maximum eigenvalue of formula (13), and the expression is:

[0096] According to the above process and the fitness value calculated in step 3, the optimal concentration value and its corresponding position are obtained.

[0097]

[0098] Step A5: Iteration judgment: According to the maximum number of iterations preset in step A1, determine whether the current number of iterations meets the stop condition. If the stop condition is met, complete steps A1-A5 and output the optimal parameters of the proposed fruit fly optimized independent variational mode decomposition algorithm. Otherwise, return to step A2 to continue iterating until the stop condition is met.

[0099] This application proposes a method for locating a damage acoustic emission source based on fruit fly optimized independent variational mode decomposition, which enhances the performance of acoustic emission signal feature extraction and improves the accuracy of acoustic emission source positioning. In order to effectively solve the problems mentioned in the background technology, this application proposes a method for locating a damage acoustic emission source based on fruit fly optimized independent variational mode decomposition, and explains the beneficial effects one by one, as follows:

[0100] Optimized acoustic emission source localization accuracy: The fruit fly optimized independent variational mode decomposition (FO-IVMD) method proposed in the present invention shows significant advantages in processing high-noise, nonlinear, and non-stationary acoustic emission signals compared to traditional time domain, frequency domain, and time-frequency domain analysis methods. By introducing the fruit fly optimization algorithm and dynamically adjusting the key parameters of VMD (such as the modulus K and the penalty parameter a), the common problems of traditional VMD methods when processing complex signals, such as inaccurate mode selection, mode overlap, and mode over-decomposition or under-decomposition, are effectively overcome, thereby improving the accuracy of signal decomposition. Through more precise signal decomposition, this method can more accurately capture and extract potential feature information in acoustic emission signals, thereby improving the accuracy of acoustic emission source localization.

[0101] Reduce noise interference and signal attenuation problems of traditional methods: Traditional time domain analysis methods often lead to low positioning accuracy due to factors such as multipath propagation, noise interference and signal attenuation, especially in large-scale monitoring systems where real-time and accuracy are difficult to guarantee. Frequency domain analysis methods rely on signal preprocessing and feature selection. In the case of complex signals or multiple interferences, it is often difficult to extract recognizable feature information, and even face problems such as mode overlap or inaccurate mode selection. In comparison, the independent variational mode decomposition method based on fruit fly optimization can reduce the impact of noise by optimizing the key parameters of mode decomposition, and can more accurately extract effective signal features in the case of multiple signal interference and attenuation, thereby effectively improving the stability and reliability of source positioning.

[0102] Dynamic threshold optimization improves source localization accuracy: The present invention optimizes the localization results of the acoustic emission source by dynamically updating the threshold according to the attenuation characteristics of the acoustic emission signal. This innovative strategy enables source localization to adapt to different signal attenuation conditions, avoiding the problem that the fixed threshold in the traditional method cannot adapt to signal changes, thereby further improving the localization accuracy. Through this dynamic optimization mechanism, the present method can adjust and optimize the source localization results in real time, especially under complex process conditions, showing high robustness and accuracy that is superior to traditional methods.

[0103] Independent mode decomposition avoids mode aliasing and end effect: Common signal decomposition methods, such as wavelet transform (WT), EEMD and LMD, often face problems such as mode aliasing and end effect when processing acoustic emission signals. These problems will seriously affect the extraction of signal features and thus affect the accuracy of source positioning. The FO-IVMD method proposed in the present invention avoids mode aliasing and end effect in traditional methods by optimizing the mode selection process, ensuring more accurate signal decomposition, thereby effectively improving the accuracy of source positioning.

[0104] Solve the problem of inaccurate modal decomposition in high-noise environments: In high-noise environments, traditional VMD methods are prone to inaccurate modal selection or modal overlap, affecting subsequent analysis. The present invention uses the fruit fly optimization algorithm to automatically optimize the key parameters of VMD to ensure the accuracy of modal decomposition. By optimizing the modal parameters, the present invention can effectively reduce the impact of high-noise environments on the decomposition accuracy of acoustic emission signals and ensure the reliability of source location results.

[0105] It has broad application prospects and promotion value: The fruit fly optimized independent variational mode decomposition method of the present invention not only has high acoustic emission source positioning accuracy, but also can be applied to damage detection and source positioning problems in various complex working environments and conditions, and has broad application prospects. This method can be widely used in health monitoring in the fields of metal structures, composite materials, concrete structures, etc., especially in laser cladding, welding, corrosion and other processes, and can effectively provide high-precision acoustic emission source positioning, providing strong support for structural health monitoring, fault diagnosis and safety assessment.

[0106] In summary, the present invention achieves a significant improvement in the accuracy of acoustic emission source positioning by optimizing the independent variational mode decomposition method and combining it with the fruit fly optimization algorithm, solves many technical problems in traditional methods, and provides an efficient and accurate solution for complex signal processing and acoustic emission source positioning in engineering practice.

[0107] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various equivalent changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for locating damage acoustic emission sources based on fruit fly optimized independent variational mode decomposition, characterized in that: The method includes: Step 1: Acoustic emission signal collection and sample selection: Collect the acoustic emission signals of laser cladding of metal plates under different process parameters, and select the verification data set from the laser cladding acoustic emission signals; Step 2: Preprocessing of the original AE signal: In the laser cladding experiment, an analog bandpass filter with a frequency range of 40kHz to 450kHz was used to preprocess the original AE signal; Step 3: Signal decomposition and source location: Decompose the preprocessed AE signal and use the proposed fruit fly optimized independent variational mode decomposition method to find the coordinates of the AE source; Step 4: Optimization of acoustic emission source localization results: According to the attenuation characteristics of the acoustic emission signal, the threshold is dynamically updated to optimize the acoustic emission source localization results. In this step, the time difference between each sensor and all acoustic emission events is first extracted based on the fixed threshold A0. The coordinates of the acoustic emission source (x s ,y s ), and calculate the coordinates and the i-th sensor (x i ,y i ) i , the expression is: The updated dynamic threshold DA of the i-th sensor i The expression is: Among them, d min =min(d i ), SA(·) is the attenuation function of the amplitude; The optimal time difference of each sensor is obtained by using the updated threshold, and then substituted into the proposed fruit fly optimized independent variational mode decomposition algorithm to optimize the coordinates of the acoustic emission source; Step 5: Comparative analysis with existing methods: Through comparative analysis with existing acoustic emission source localization methods, the effectiveness of the proposed structural damage acoustic emission source localization method based on fruit fly optimized independent variational mode decomposition is verified.

2. The method for locating damage acoustic emission sources based on fruit fly optimized independent variational mode decomposition according to claim 1, characterized in that: The fruit fly optimization independent variational mode decomposition includes: Step A1: fitness function construction; Step A2: optimal IMF component determination; Step A3: Chaos operation; Step A4: fitness value calculation; Step A5: Iterative judgment.

3. The method for locating damage acoustic emission sources based on fruit fly optimized independent variational mode decomposition as claimed in claim 2, characterized in that: Step A1: fitness function construction, construct a fitness function, based on the definition of information entropy, using marginal spectral entropy H p To measure the uncertainty and frequency complexity of the signal in the frequency domain, the expression is: Among them, H p represents the marginal spectral entropy value, p is the value used to measure the probability of a specific event in the system, P i represents the probability of the amplitude corresponding to the i-th frequency, h(i) represents the marginal spectrum corresponding to the i-th IMF component, N represents the total number of events, i represents the index of the event, and p i is the probability of event i occurring; the marginal spectral entropy corresponding to the IMF component is normalized, that is, H E represents the normalized entropy value, L is the length of the marginal spectrum h(i), and E represents the sign of entropy.

4. The method for locating damage acoustic emission sources based on fruit fly optimized independent variational mode decomposition as claimed in claim 2, characterized in that: The step A2: determining the optimal IMF component, taking the minimum value of the marginal spectral entropy corresponding to the IMF component as the optimal IMF component, that is, the marginal spectral entropy corresponding to the IMF component is called the local minimum value, and the expression is: Where L is the length of the marginal spectrum h(i), i represents the index of the event, represents the entropy value of the IMF component, min represents the minimization operation, and p i is the probability of event i occurring, and IMF represents the intrinsic mode component.

5. The method for locating damage acoustic emission sources based on fruit fly optimized independent variational mode decomposition as claimed in claim 2, characterized in that: Step A3: Chaotic operation, initializing the number of fruit flies N, the maximum number of iterations T, and the upper limit T max At the same time, the relevant parameters of chaos are initialized, and two fruit fly populations with a population size of N are randomly obtained, namely U a and U K , and set the fitness value Fitbest≈0, and set U a and U K Each component of is mapped to a chaotic variable and And complete the chaos operation.

6. The method for locating damage acoustic emission sources based on fruit fly optimized independent variational mode decomposition as claimed in claim 2, characterized in that: Step A4: Calculate the fitness value, take a K-scale IMF component u k (t), k = 1, 2, ..., K composed of the sampling signal x(t), t = 1, 2, ..., t m , initialize and perform Hilbert transform on IMF components of different scales to obtain the analytical signal U k (t), the expression is: Where t represents time, δ(t) is the Dirac function, and U k (t) represents the output analytical signal, represents a complex term, j is an imaginary unit, indicating that the signal has a complex part, and u k (t) represents the spectrum expression of the input signal, k represents u k (t) index, the symbol * represents the convolution operation, fusion U k (t) and the estimated center frequency The spectrum u k (t) is modulated to the corresponding root frequency band, and the expression is: Among them, F k (f) represents the signal u obtained after transformation k (t) is represented in the frequency domain, f represents the frequency domain characteristics of the signal, e is the base of the natural logarithm, j is the imaginary unit, ω k Represents the center frequency, and u is estimated by calculating the L2 norm of the signal gradient k (t), introduce constraints and establish the optimal variational model, the expression is: Among them, ω k Indicates u k (t), x(t) represents the input signal, K is the set constant value, and the quadratic penalty factor β and Lagrangian multiplier γ(t) are introduced to construct the extended Lagrangian function, which is expressed as: Where L(·) represents the Lagrangian function, and all IMF components are calculated as follows: in, represents the updated value of the kth IMF component in the (n+1)th iteration, where n represents the number of iterations. represents the frequency domain representation of the target signal x(t), represents the i-th signal component u i (t) is represented in the frequency domain, represents the frequency domain representation of the Lagrange multiplier γ(t), ω is the angular frequency, β is the quadratic penalty factor, and N effective IMF components are selected to construct the input observation signal, which is then whitened and zero-centered. At the same time, a kernel function k(·) is given and the signal estimation vector S = {s1, s2, …s n }'s Gram matrix G=(G1,G2,…G m ), set λ = (G1, G2, ... G m ) is the maximum eigenvalue of formula (13), and the expression is: The fitness value is calculated according to the above process, and the optimal concentration value and its corresponding position are obtained.

7. The method for locating damage acoustic emission sources based on fruit fly optimized independent variational mode decomposition as claimed in claim 2, characterized in that: The step A5: iterative judgment, according to the maximum number of iterations preset in step A1, judge whether the current number of iterations meets the stopping condition. If the stopping condition is met, complete steps A1-A5 and output the optimal parameters of the proposed fruit fly optimized independent variational mode decomposition algorithm. Otherwise, return to step A2 to continue iterating until the stopping condition is met.

Images