W-Cu alloy laser ultrasonic detection method based on GA-EE-VMD and longitudinal wave enhancement algorithm

Abstract

The invention discloses a W-Cu alloy laser ultrasonic detection method based on GA-EE-VMD and a longitudinal wave enhancement algorithm, and the method specifically comprises the steps: 1, collecting a W-Cu alloy at a T temperature through employing a laser ultrasonic system, and obtaining an ultrasonic signal; 2, performing decomposition, reconstruction and de-noising processing on the acquired ultrasonic signals by adopting a GA-EE-VMD method to obtain K modal functions; 3, calculating instantaneous amplitudes and frequencies of different IMFs components, and carrying out weighted enhancement processing to obtain an enhanced longitudinal wave signal; and 4, the thickness of the W-Cu alloy at the high temperature is calculated through the enhanced longitudinal wave signals. According to the invention, the identification precision of the longitudinal wave is improved, so that the thickness detection precision is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of nondestructive testing, and in particular relates to a W-Cu alloy laser ultrasonic testing method based on GA-EE-VMD and longitudinal wave enhancement algorithm. Background Art

[0002] Tungsten-copper (W-Cu) alloys combine the high density, high melting point, low thermal expansion coefficient, and high strength of W with the excellent thermal and electrical conductivity and high fracture toughness of Cu. These composite materials have been widely used in civilian applications such as high-voltage electrical contacts, welding electrodes, and heat sinks. They are particularly well-known in the aerospace industry, particularly in the manufacture of aircraft engine turbine blades, nuclear power units, and spacecraft reentry capsules. Traditionally, W-Cu alloys can be produced through infiltration, liquid-phase sintering, and powder metallurgy. However, these conventional manufacturing processes are limited by cost, insufficient density, and geometric complexity.

[0003] To address these issues, a new technology, known as additive manufacturing (AM), has been widely researched and applied to the production of W-Cu alloys. It is a laser-based, layer-by-layer building process that uses a computer to construct objects with complex geometries from powder. Compared to traditional manufacturing methods, AM offers attractive advantages by eliminating material waste and reducing costs. However, since AM technology uses layer-by-layer cumulative printing, there is a chance of defects during the printing process, which can affect the fatigue resistance and service life of the part. Consequently, defect detection during the AM process has received increasing attention.

[0004] AM is in an extreme environment of high temperature and high pressure, making conventional detection methods difficult to implement. However, non-contact non-destructive testing technology has great potential. Due to the advancement of image acquisition technology, digital image correlation methods are widely used for detection in high-temperature environments. Although they have a wide measurement range and high detection efficiency, their biggest limitation is that they cannot detect defects inside objects. Computed tomography methods take X-ray images by rotation and reconstruct the object. This method can detect surface and internal defects, but the detection process is time-consuming and is limited by the size of the object, so it is not suitable for online inspection. Ultrasonic-based electromagnetic acoustic transducers use ultrasonic waves and electromagnetic sounds to detect at high temperatures, but the probe must be close to the object being measured and cannot detect small defects. Summary of the Invention

[0005] Purpose of the invention: In order to solve the problems existing in the above-mentioned prior art, the present invention provides a W-Cu alloy laser ultrasonic detection method based on GA-EE-VMD and longitudinal wave enhancement algorithm.

[0006] Technical solution: The present invention discloses a W-Cu alloy laser ultrasonic testing method based on GA-EE-VMD and longitudinal wave enhancement algorithm, which specifically includes the following steps:

[0007] Step 1: Using a laser ultrasonic system to collect W-Cu alloy at a temperature T to obtain an ultrasonic signal; T is greater than a preset temperature threshold;

[0008] Step 2: Use the GA-EE-VMD method to decompose, reconstruct and denoise the collected ultrasonic signal to obtain K modal functions y k (t);

[0009] Step 3: Calculate the instantaneous amplitude and frequency of different IMFs components and perform weighted enhancement processing to obtain the enhanced longitudinal wave signal L(t);

[0010] Step 4: Calculate the thickness of the W-Cu alloy at high temperature using the enhanced longitudinal wave signal.

[0011] Furthermore, in step 1, linear scanning and area scanning are performed on the surface of the W-Cu alloy to generate ultrasonic signals.

[0012] Furthermore, step 2 is specifically as follows: using genetic algorithm GA as the optimization algorithm for optimizing variational mode decomposition VMD parameters, using envelope entropy EE as the fitness function, using GA algorithm to search for the optimal decomposition number K and penalty coefficient α in the VMD decomposition process, and then performing VMD decomposition, selecting appropriate IMF components for reconstruction to achieve ultrasonic signal denoising.

[0013] Furthermore, the step 2 is specifically as follows:

[0014] The quadratic penalty factor α and the Lalangrange factor λ are used to transform the variational constraint problem of VMD into the following augmented Lagrangian expression:

[0015]

[0016] Among them, y k is the kth mode function after VMD decomposition, t represents time, j is the imaginary unit, y(t) is the original signal, is the Hilbert transform; w k is the center frequency of the kth mode function;

[0017] The above expressions are solved using the alternating direction multiplier method by alternatingly updating the mode function, center frequency and Lagrange factor;

[0018] When using genetic algorithm to search parameters K and α, the expression of fitness function is:

[0019]

[0020] Wherein, EE is the envelope entropy, N represents the total number of sampling points, and ε(j1) is the normalized probability distribution sequence of the envelope signal amplitude. The envelope signal is obtained by Hilbert transforming K modal function components.

[0021] Furthermore, the step 3 is specifically as follows: each IMF component y k (k) are all Hilbert transformed to obtain the corresponding analytical signal z k (t), and then calculate the instantaneous frequency ω of each component by the following formula k (t) and instantaneous amplitude A k (t) can be expressed as follows:

[0022]

[0023] Where t represents time and φ(t) represents phase;

[0024] Enhance y by weighting the two features k (k) The corresponding longitudinal wave signal L k (t):

[0025] L k (t) = ω k (t)*A k (t)

[0026] Get the total longitudinal wave signal:

[0027]

[0028] Beneficial effects:

[0029] 1. The present invention proposes a method for improving the signal-to-noise ratio of laser ultrasonic signals at high temperatures, which is beneficial to solving the problem of low signal-to-noise ratio caused by high temperature and non-contact detection, and improves the detection accuracy of defects based on the enhanced imaging resolution.

[0030] 2. Considering that the longitudinal wave energy used for thickness detection is small and difficult to identify, a longitudinal wave enhancement algorithm based on instantaneous frequency and amplitude weighting was invented. By enhancing the longitudinal wave energy, the recognition accuracy of the longitudinal wave is improved, thereby improving the detection accuracy of the sample thickness at high temperature. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 Flowchart of the GA-EE-VMD algorithm of the present invention;

[0032] Figure 2 This is the laser ultrasonic experimental system of the present invention;

[0033] Figure 3Five W-Cu alloy samples with different thicknesses prepared in the present invention;

[0034] Figure 4 This is the GA-EE-VMD decomposition process of the present invention;

[0035] Figure 5 Comparison of B-scan images before and after denoising by GA-EE-VMD in the present invention; (a) is the B-scan image before denoising; (b) is the B-scan image after denoising;

[0036] Figure 6 The figure shows the B-scan comparison of longitudinal waves before and after enhancement in the present invention; (a) is the B-scan image before enhancement; (b) is the B-scan image after enhancement. DETAILED DESCRIPTION

[0037] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0038] This paper proposes a laser ultrasonic signal noise reduction and longitudinal wave enhancement method for surface defect and thickness detection in W-Cu alloys. First, a laser ultrasonic system is used to scan defective W-Cu alloys of varying thicknesses at 500°C. The collected ultrasonic signals are then subjected to GA-EE-VMD decomposition and reconstruction to remove noise, and the defect size is determined based on the imaging results. Secondly, considering that longitudinal waves are difficult to identify due to their low energy, a longitudinal wave enhancement algorithm based on instantaneous frequency and amplitude weighting is developed. The instantaneous amplitude and frequency of different IMFs components are calculated and weighted for enhancement. Finally, the actual thickness of the alloy is calculated based on the arrival time of the longitudinal waves.

[0039] The present invention is specifically:

[0040] S1: Using a laser ultrasonic system to collect crack defects in W-Cu alloy with a thickness of 5-8mm at 500℃;

[0041] S2: Perform GA-EE-VMD decomposition, reconstruction and denoising on the collected ultrasonic signal;

[0042] S3: Calculate the instantaneous amplitude and frequency of different IMFs components and perform weighted enhancement processing;

[0043] S4: Calculate the thickness of W-Cu alloy at high temperature using the enhanced longitudinal wave signal.

[0044] The step S1 is specifically as follows:

[0045] In this example, a laser ultrasonic experimental system including a high-temperature heating furnace was built, and five defective W-Cu alloys with different thicknesses were prepared. B-scan (linear scan) and C-scan (area scan) modes were used to excite the surfaces to generate ultrasonic signals.

[0046] The step S2 is specifically as follows:

[0047] A GA (genetic algorithm) was selected as the optimization algorithm for VMD (variational mode decomposition) parameters, with EE (envelope entropy) as the fitness function. The GA algorithm was used to search for the optimal parameters K and α (the number of decompositions, K, and α is the penalty coefficient) during the VMD decomposition process. After determining the optimal parameters, VMD decomposition was performed, and appropriate IMF components were selected for reconstruction to achieve ultrasonic signal denoising. The crack size was determined using B-scan or C-scan images.

[0048] The GA-EE-VMD denoising algorithm proposed in the present invention is as follows:

[0049] VMD decomposes the ultrasonic signal into K components with a center frequency of w k The IMF (intrinsic mode function) of the variational constraint problem is expressed as:

[0050]

[0051] in, represents the time derivative, which is used to measure the bandwidth of the mode. y(t) is the original signal, y k (t) is the kth mode after VMD decomposition, k y k The sum of (t) is equal to y(t), t is the time variable, j is the imaginary unit, is the Hilbert transform, is the Hilbert transform kernel function.

[0052] Among them, y k is the kth mode, {y k}={y1,y2,y3,…,y k},w k is the center frequency of the kth mode.

[0053] The above constrained optimization problem can be transformed into an unconstrained optimization problem using the quadratic penalty factor α and the Lagrangian factor λ(t). Its augmented Lagrangian expression is as follows:

[0054]

[0055] Use the alternating direction method of multipliers (ADMM) to update y by alternating kn+1 ,w k n+1 ,λ(t) n+1 Find the saddle point of the above equation. n represents the number of iterations, y k n+1 ,w k n+1 ,λ(t) n+1 are the mode function, center frequency, and Lagrange multiplier at the n+1th iteration respectively; from the above, it can be seen that in order to adaptively select K and α, the present invention adopts GA optimization algorithm to select key parameters.

[0056] GA is a typical heuristic search algorithm used to solve optimization and search problems. This algorithm draws on biological genetics and evolutionary theory to solve optimization problems. GA primarily involves chromosome encoding, population initialization, setting a fitness function, and genetic operations (including selection, crossover, and mutation). It possesses high global optimization capabilities and search efficiency. Through this mechanism, the algorithm is able to find the optimal parameter combination within a given VMD parameter space.

[0057] When using the GA optimization algorithm to search for K and α parameters, a suitable fitness function must be established. EE can reflect the uncertainty and complexity of the signal. When there is more noise and less defect feature information in the IMF, the EE value is larger. Conversely, the smaller the EE value, the greater the possibility that the signal carries defect information. For a given original signal, VMD decomposes it into K sub-signals y k ,{y k}={y1,y2,y3,…,y k} is a collection of modes. The expression of EE is as follows:

[0058]

[0059] Wherein, EE is the envelope entropy, N represents the total number of sampling points, and ε(j1) is the normalized probability distribution sequence of the envelope signal amplitude. The envelope signal is obtained by Hilbert transforming K modal function components.

[0060] Therefore, the overall GA-EE-VMDVMD optimization process is as follows Figure 1 As shown, the main steps are as follows:

[0061] Step 1: Collect laser ultrasonic signals and implement the GA-EE-VMD algorithm.

[0062] Step 2: Set the number of iterations to 20, the population size to 10, and the crossover probability to 0.8. Based on previous experimental experience, set the VMD penalty factor α to [1000, 6000] and the number of decompositions K to [3, 9].

[0063] Step 3: EE is the fitness function, which uses the GA algorithm to continuously optimize the K and α parameters and determine whether the termination conditions are met.

[0064] Step 4: If the termination condition is met, the optimal parameter solution is obtained; otherwise, continue to the next iteration.

[0065] The step S3 is specifically as follows:

[0066] A laser ultrasonic signal can be expressed as follows after GA-EE-VMD decomposition:

[0067]

[0068] Then, each IMF component y k (t) are all Hilbert transformed to obtain the corresponding analytical signal z k (t). The Hilbert transform converts the real-valued signal into a complex form, which can extract the instantaneous amplitude and instantaneous frequency of each mode, thereby obtaining a clearer longitudinal wave signal. The Hilbert transform formula is as follows:

[0069] z k (t) = y k (t)+j·H[y k (t)]

[0070] =A k (t)e jφ(t)

[0071] Where H[·] is the Hilbert operator, e jφ(t) Describes the rotation of the complex signal in time, φ(t) represents the phase, A k (t) represents the instantaneous amplitude.

[0072] Therefore, the instantaneous frequency ω of each component is k (t) and instantaneous amplitude A k (t) can be expressed as follows:

[0073]

[0074] In order to enhance the longitudinal wave signal, the instantaneous frequency and amplitude of each component signal are obtained by Hilbert transform, and the longitudinal wave signal L is enhanced by weighting the two features. k (t), the formula is as follows:

[0075] L k (t) = ω k (t)*A k (t)

[0076] It should be noted that the enhanced signal L k(t) is a single frequency IMF component. The overall signal is reconstructed by combining multiple enhanced signals. The formula is as follows:

[0077]

[0078] Where L(t) is the enhanced reconstructed signal, and K represents the number of decompositions.

[0079] The step S4 is specifically as follows:

[0080] The enhanced longitudinal wave signal is used to generate a B-scan image. The arrival time of the longitudinal wave is determined by combining the time domain signal and the B-scan image, and then the thickness of the W-Cu alloy is calculated.

[0081] like Figure 2 As shown, this embodiment uses additively manufactured W-Cu alloy specimens with thicknesses of 5 mm, 6 mm, 7 mm, 8 mm, and 9 mm as test objects, and the implementation process specifically includes the following steps:

[0082] S1: Establishing laser ultrasound experimental system;

[0083] The LU system uses a Dawa-200 Nd:YAG pulsed laser with an output wavelength of 1064nm, a repetition frequency of 1-20Hz, and a pulse width of 8ns. The maximum energy of a single pulse is about 200mJ. A QUARTET-FH laser interferometer with an energy of 500mW, a wavelength of 532nm, and a detection bandwidth of 0-10MHz is used to receive ultrasonic waves. A VA-NT1510Y digital galvanometer is used to achieve X and Y direction scanning, and a NIPCI5114 oscilloscope card with a sampling frequency of 125MHz is used to digitize the signal. The single-chip microcomputer control circuit mainly ensures the synchronization of the laser frequency with the movement of the digital galvanometer and the acquisition card. The SXL-1002 electric furnace is used to heat the sample. The laser ultrasonic system is as follows: Figure 2 shown.

[0084] S2: Preparation of W-Cu alloy specimens for additive manufacturing at high temperature;

[0085] Five groups of W-Cu alloy specimens with thickness of 5 to 9 mm, width of 50 mm and length of 300 mm were prepared by laser selective melting technology, and a crack defect with length of 1 mm and width of 5 mm was processed in each group. Figure 3 The sample was placed in an electric furnace, the heating temperature was set to 500°C, the heating time was 30 minutes, and the heat preservation time was 20 minutes. After the sample was heated, the furnace door was opened for LU testing.

[0086] S3: denoising the ultrasonic signal using the GA-EE-VMD algorithm;

[0087] The GA-EE-VMD algorithm was executed on the acquired high-temperature ultrasonic signal. The number of iterations in the decomposition algorithm was set to 20, the population size was set to 10, and the crossover probability was set to 0.8. Based on previous experimental experience, the VMD penalty factor α was set to [1000, 6000] and the decomposition number K was set to [3, 9]. Different IMF components were compared and the IMF component with the maximum signal-to-noise ratio was obtained for reconstruction. Figure 4 (a) is the time domain signal of 7 components after GA-EE-VMD decomposition. Figure 4 (b) in the figure is the frequency domain signal of 7 components. Figure 5 It is the B-scan image after decomposition and reconstruction.

[0088] S4: Calculate the instantaneous amplitude and frequency of different IMFs and perform weighted enhancement processing;

[0089] Based on S3, GA-EE-VMD decomposition of the ultrasonic signal can obtain 7 IMF components. Then the instantaneous amplitude and instantaneous frequency of each component are calculated and weighted. Selecting appropriate weighted components for reconstruction can obtain the enhanced ultrasonic signal, such as Figure 6 As shown, the enhanced signal significantly highlights the energy of the longitudinal wave.

[0090] S5: Thickness measurement by enhanced longitudinal waves

[0091] According to the enhanced longitudinal wave time domain signal and B-scan image, the arrival time of the longitudinal wave can be accurately obtained, and the actual thickness of the W-Cu alloy can be measured in combination with the longitudinal wave velocity at high temperature.

[0092] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any appropriate manner without contradiction. To avoid unnecessary repetition, the present invention will not further describe various possible combinations.

Claims

1. W-Cu alloy laser ultrasonic testing method based on GA-EE-VMD and longitudinal wave enhancement algorithm, characterized by: The specific steps include: Step 1: Using a laser ultrasonic system to collect W-Cu alloy at a temperature T to obtain an ultrasonic signal; T is greater than a preset temperature threshold; Step 2: Use the GA-EE-VMD method to decompose, reconstruct and denoise the collected ultrasonic signal to obtain K modal functions y k (t); Step 3: Calculate the instantaneous amplitude and frequency of different IMFs components and perform weighted enhancement processing to obtain the enhanced longitudinal wave signal L(t); Step 4: Calculate the thickness of the W-Cu alloy at high temperature using the enhanced longitudinal wave signal.

2. The W-Cu alloy laser ultrasonic testing method based on GA-EE-VMD and longitudinal wave enhancement algorithm according to claim 1 is characterized in that: In the step 1, the surface of the W-Cu alloy is excited by linear scanning and area scanning to generate ultrasonic signals.

3. The W-Cu alloy laser ultrasonic testing method based on GA-EE-VMD and longitudinal wave enhancement algorithm according to claim 1 is characterized in that: The step 2 is specifically as follows: using the genetic algorithm GA as the optimization algorithm for optimizing the variational mode decomposition VMD parameters, using the envelope entropy EE as the fitness function, using the GA algorithm to search for the optimal decomposition number K and penalty coefficient α in the VMD decomposition process, then performing VMD decomposition, and selecting the appropriate IMF component for reconstruction to achieve ultrasonic signal denoising.

4. The W-Cu alloy laser ultrasonic testing method based on GA-EE-VMD and longitudinal wave enhancement algorithm according to claim 1 is characterized in that: The step 2 is specifically as follows: The quadratic penalty factor α and the Lalangrange factor λ are used to transform the variational constraint problem of VMD into the following augmented Lagrangian expression: Among them, y k is the kth mode function after VMD decomposition, t represents time, j is the imaginary unit, y(t) is the original signal, is the Hilbert transform; w k is the center frequency of the kth mode function; The above expressions are solved using the alternating direction multiplier method by alternatingly updating the mode function, center frequency and Lagrange factor; When using genetic algorithm to search parameters K and α, the expression of fitness function is: Wherein, EE is the envelope entropy, N represents the total number of sampling points, and ε(j1) is the normalized probability distribution sequence of the envelope signal amplitude. The envelope signal is obtained by Hilbert transforming K modal function components.

5. The W-Cu alloy laser ultrasonic testing method based on GA-EE-VMD and longitudinal wave enhancement algorithm according to claim 1 is characterized in that: The step 3 is specifically as follows: k (t) are all Hilbert transformed to obtain the corresponding analytical signal z k (t), and then calculate the instantaneous frequency ω of each component by the following formula k (t) and instantaneous amplitude A k (t) can be expressed as follows: Where t represents time and φ(t) represents phase; Enhance y by weighting the two features k (t) The corresponding longitudinal wave signal L k (t): L k (t)=ω k (t)*A k (t) Get the total longitudinal wave signal:

Images