Automated ultrasonic phased array detection device and detection method thereof

Abstract

The present invention relates to an automated ultrasonic phased array detection device and a detection method thereof, comprising a moving seat, an ultrasonic phased array probe, a transducer, a probe clamping frame, a wheel encoder, and a phased array controller. The ultrasonic phased array probe comprises a metal probe and a filter. The metal probe is connected to the phased array controller through a transducer, and the metal probe is used to emit an ultrasonic beam; the metal probe has a piezoelectric chip array, and each piezoelectric chip of the piezoelectric chip array works independently; the filter is fixedly installed at the bottom of the metal probe for filtering; the wheel encoder is used to record the scanning position information of the ultrasonic phased array probe on the workpiece to be tested, and transmit it to the phased array controller to determine whether the defect scanning detection area completely covers the workpiece to be tested. The present invention collects the position information of the metal probe scanning through the wheel encoder, and the phased array controller controls the moving seat to scan the undetected area in combination with the size information of the workpiece to be tested, thereby realizing automated defect scanning detection.

Description

Technical Field

[0001] The invention belongs to the technical field of nondestructive testing, and in particular relates to an automated ultrasonic phased array testing device and a testing method thereof. Background Art

[0002] Carbon fiber composite material is a high-performance composite material with carbon fiber as reinforcement and resin as matrix. It has excellent properties such as light weight, high strength, high modulus, high temperature resistance and corrosion resistance. It is widely used in aerospace, military, automobile, construction and other fields. However, the manufacturing process of carbon fiber composite materials is very complicated. There are both chemical reactions and physical changes. Many slight differences in process parameters will cause many defects, making the product quality obviously discrete. In actual engineering, inspection personnel need to take accurate and effective measures to repair the defects according to the type of defects. Therefore, it is of great significance to accurately and reliably identify the internal defects of carbon fiber composite materials.

[0003] Many non-destructive testing methods are used for defect detection of carbon fiber composite materials, among which the most mature and widely used detection method is ultrasonic testing. Ultrasonic testing detects defects in materials based on the reflection, transmission and energy attenuation of ultrasonic waves caused by differences in acoustic properties between different materials. It has the advantages of high detection sensitivity, fast detection speed, strong penetration and harmlessness to the body. When traditional ultrasonic testing is used to detect materials, the ultrasonic beam diffuses outward to a certain extent, resulting in insufficient energy of the sound beam for defects inside the material, and thus the probability of detecting tiny defects inside the material is low.

[0004] The development of focused probes has overcome the disadvantage of unconcentrated sound beam energy. The sound field energy emitted by the focused probe is concentrated, with high resolution and high sensitivity. However, the sound beam of the focused probe is very thin and the defect detection rate is very low. Therefore, based on the development of focused probes, ultrasonic phased array nondestructive testing has set up a large number of array elements to improve the range and defect detection rate of defect detection. Compared with traditional ultrasonic nondestructive testing technology, the ultrasonic sound beam of ultrasonic phased array nondestructive testing can scan workpieces at multiple angles and directions, and can perform all-round scanning of workpieces with complex shapes. It has good identifiability and improves the detection speed. However, the defect detection accuracy of existing ultrasonic phased array equipment is insufficient, and the degree of automation is insufficient, which makes the detection results inaccurate and the scanning is not comprehensive enough. Summary of the invention

[0005] Based on the above-mentioned shortcomings and deficiencies in the prior art, one of the objects of the present invention is to at least solve one or more of the above-mentioned problems in the prior art. In other words, one of the objects of the present invention is to provide an automated ultrasonic phased array detection device and a detection method thereof that meet one or more of the above-mentioned needs.

[0006] In order to achieve the above-mentioned object of the invention, the present invention adopts the following technical solutions:

[0007] An automated ultrasonic phased array detection device comprises a moving seat, an ultrasonic phased array probe, a transducer, a probe clamping frame, a wheel encoder, and a phased array controller. The probe clamping frame is arranged on the moving seat and is used to clamp the ultrasonic phased array probe.

[0008] The ultrasonic phased array probe comprises a metal probe and a filter. The metal probe is connected to the phased array controller through a transducer. The metal probe is used to emit an ultrasonic beam. The metal probe has a piezoelectric chip array. Each piezoelectric chip of the piezoelectric chip array works independently and is used to deflect and focus the ultrasonic beam. The filter is fixedly installed at the bottom of the metal probe and is used to filter the deflected ultrasonic beam. The filtered ultrasonic beam performs defect scanning and detection on the workpiece to be tested. The filter is also used to filter the target ultrasonic beam reflected after the defect scanning and detection. The filtered target ultrasonic beam is focused by controlling the piezoelectric chip array through the transducer. The focused ultrasonic beam is converted into an electrical signal by the transducer and input into the phased array controller. The phased array controller converts the received electrical signal into a detection image.

[0009] The wheel encoder is arranged on the moving seat, and is used to record the scanning position information of the ultrasonic phased array probe on the workpiece to be measured, and transmit it to the phased array controller to determine whether the defect scanning detection area completely covers the workpiece to be measured.

[0010] As a preferred solution, the ultrasonic beam deflection is implemented as follows:

[0011] The piezoelectric chips of the piezoelectric chip array increase the delay time of the excitation from left to right in sequence, so that the delay time is an arithmetic progression, thereby realizing the deflection of the ultrasonic beam.

[0012] As a preferred solution, the ultrasonic beam focusing is achieved by:

[0013] The transducer excites the piezoelectric chips at the left and right ends of the piezoelectric chip array, and then excites other piezoelectric chips in sequence along the direction of the center position of the piezoelectric chip array by continuously increasing the delay time, forming a wavefront pointing to the center of the target curvature, thereby achieving ultrasonic beam focusing.

[0014] As a preferred solution, the workpiece to be tested is a carbon fiber composite material.

[0015] The present invention also provides a detection method of the automated ultrasonic phased array detection device as described in any of the above schemes, comprising the following process:

[0016] The size parameters of the workpiece to be measured are input into the phased array controller, which controls the movement of the mobile seat and the metal probe to emit an ultrasonic beam. The ultrasonic beam is deflected, scanned and focused through the transducer and the piezoelectric chip array. The focused ultrasonic beam is converted into an electrical signal through the transducer and transmitted to the phased array controller, which converts the electrical signal into a detection image. At the same time, the wheel encoder records the position information of the metal probe scanned on the workpiece to be measured and transmits it to the phased array controller.

[0017] As a preferred solution, the phased array controller converts the electrical signal into a detection image, comprising the following steps:

[0018] (1) Feature extraction of electrical signals is performed by feature pooling, and characteristic parameters of defects are extracted from three perspectives: time domain, frequency domain, and time-frequency domain. Time domain characteristic parameters, frequency domain characteristic parameters, and time-frequency domain characteristic parameters are obtained respectively;

[0019] (2) The time domain feature parameters and frequency domain feature parameters are fused in a weighted manner to obtain feature parameters f 1. Then the characteristic parameters f 1 is fused with the time-frequency domain feature parameters through feature selection to obtain the feature parameters f 2;

[0020] (3) The characteristic parameters are optimized by BP neural network based on genetic algorithm f 2 Perform intelligent recognition, identify the type and location of defects and convert them into images.

[0021] As a preferred solution, in step (1), the characteristic parameters of the defect are extracted from three perspectives: time domain, frequency domain and time-frequency domain, including:

[0022] By analyzing the waveform characteristics of the electrical signal, time domain analysis is performed to obtain the time domain characteristic parameters f a ;

[0023] The electrical signal is converted into the frequency domain using the fast Fourier transform, and then the spectrum of the signal is analyzed in the frequency domain to obtain the frequency domain characteristic parameters. f b ;

[0024] The fourth-order Daubechies wavelet function is used to decompose the electrical signal into three layers of wavelet packets to obtain eight frequency band signals. S j , j The value is an integer from 1 to 8; then calculate the energy corresponding to the signal in each frequency band E j , extract the characteristic energy spectrum in the time-frequency domain T jAnd the time-frequency domain feature parameters are obtained through feature selection f c ;

[0025] ;

[0026] ;

[0027] in, S j ( t )for t Frequency band signal at time S j The signal value of Frequency band signal S j No. k The amplitude of discrete points, k The value range is 1 to N integer, N Frequency band signal S j The total number of discrete points.

[0028] As a preferred solution, in step (2), the characteristic parameter f 1 is:

[0029] ;

[0030] in, w a and w b are the weights of the time domain characteristic parameters and the frequency domain characteristic parameters respectively;

[0031] The characteristic parameters f 1 is fused with the time-frequency domain feature parameters through feature selection to obtain the feature parameters f The process of 2 includes:

[0032] The characteristic parameters f 1 and time-frequency domain characteristic parameters f c Combine to get the feature vector f =( f 1, f c ), and the feature matrix f To standardize:

[0033] ;

[0034] in, J For the eigenvector f The cost function of The cost functionJ In the feature vector f The amount of change, is the feature vector f The weight change of

[0035] Calculate the cost function J The Hessian matrix H :

[0036] ;

[0037] Pair Matrix H Perform eigenvalue decomposition to obtain two eigenvalues ​​and their corresponding eigenvectors. Take the eigenvector corresponding to the larger eigenvalue as the target eigenvector. f Project it onto the target feature vector to obtain a feature parameter set; replace the feature vector with the feature vector composed of the feature parameter set f , repeat the above process to obtain a new feature parameter set; until the number of feature parameters in the accumulated feature parameter set reaches the target number of feature parameters, the feature parameter set is obtained. f 2.

[0038] As a preferred solution, in step (3), the BP neural network optimized based on genetic algorithm is: optimizing the initial weights and thresholds of the BP neural network by genetic algorithm, and training the BP neural network by Levenberg-Marquardt algorithm.

[0039] As a preferred solution, after displaying the detection image, the phased array controller calculates the position information of the metal probe input by the wheel encoder that has been scanned on the workpiece to be tested, and then combines the size parameters of the workpiece to be tested to determine whether there is an unscanned area; if so, the mobile seat is controlled to move to the unscanned area for defect scanning and detection; if not, the phased array controller controls the mobile seat to stop running and the detection is completed.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] (1) The automated ultrasonic phased array detection device of the present invention collects the position information of the metal probe scanning through the wheel encoder and transmits it to the phased array controller. The phased array controller controls the moving seat to scan the undetected area in combination with the size information of the workpiece to be detected, thereby avoiding missed detection and realizing automated defect scanning and detection of the workpiece to be detected;

[0042] (2) The metal probe of the present invention is equipped with a filter, which can filter out clutter signals and amplify defect signals when receiving ultrasonic beams, so that the signals received by the phased array controller are more accurate;

[0043] (3) The present invention extracts characteristic parameters of defects from three perspectives: time domain, frequency domain, and time-frequency domain. The characteristic parameters of the time domain and frequency domain are first fused by weighting. The obtained characteristic parameters are then fused with the characteristic parameters of the time-frequency domain by feature selection to obtain a characteristic parameter with more comprehensive information. The fused characteristic parameters are then intelligently identified by a BP neural network optimized based on a genetic algorithm. The type and specific location of the defect can be obtained more accurately, thereby effectively improving the recognition rate of the defect. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is a schematic structural diagram of an automated ultrasonic phased array detection device according to Embodiment 1 of the present invention;

[0045] Figure 2 is a signal processing flow chart of a phased array controller according to Embodiment 1 of the present invention;

[0046] Figure 3 is a recognition rate graph extracted by time domain feature parameters in Example 1 of the present invention;

[0047] Figure 4 is a recognition rate graph extracted by frequency domain feature parameters in Example 1 of the present invention;

[0048] Figure 5 The recognition rate diagram extracted by time-frequency domain feature parameters in Example 1 of the present invention;

[0049] Figure 6 This is a recognition rate diagram of embodiment 1 of the present invention obtained by extracting feature parameters in the time domain, frequency domain, and time-frequency domain fusion. DETAILED DESCRIPTION

[0050] In order to more clearly illustrate the embodiments of the present invention, the specific implementation methods of the present invention will be described below with reference to the accompanying drawings. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those of ordinary skill in the art, other accompanying drawings and other implementation methods can be obtained based on these accompanying drawings without creative work.

[0051] Embodiment 1:

[0052] like Figure 1 As shown, the automated ultrasonic phased array detection device of this embodiment includes a moving seat 1, a wheel encoder 2, a probe clamping frame 3, an ultrasonic phased array probe consisting of a metal probe 4 and a filter 5, a transducer 7, a phased array controller 8, and a motor 9.

[0053] The mobile seat 1 of this embodiment is used to realize the movement of the entire automated ultrasonic phased array detection device, and realize the all-round defect scanning detection of the carbon fiber composite material 6 to be tested. Specifically, a driving wheel and a driven wheel are provided below the mobile seat 1, and the driving wheel is driven by a motor 9 to realize the movement of the mobile seat 1. The specific structure of the mobile seat can refer to the prior art and will not be repeated here. Among them, the motor 9 is controlled to be turned on or off by the phased array controller 8, thereby realizing the movement control of the mobile seat 1.

[0054] In addition, the wheel encoder 2 is installed in the middle of the moving base 1, and is used to record the scanning position information of the ultrasonic phased array probe on the carbon fiber composite material 6 to be tested, and transmit it to the phased array controller 8 to determine whether the defect scanning detection area completely covers the workpiece to be tested;

[0055] The probe clamping frame 3 of the present embodiment is a U-shaped structure, which is convenient for clamping the ultrasonic phased array probe; the probe clamping frame 3 is fixedly mounted on the front side of the moving seat 1 .

[0056] The ultrasonic phased array probe of this embodiment realizes the emission and reception of ultrasonic beams by controlling the emission timing of ultrasonic beams; the metal probe 4 included therein is connected to the phased array controller 8 via the transducer 7, and the metal probe 4 is used to emit ultrasonic beams; wherein the metal probe 4 has a piezoelectric chip array 41, and each piezoelectric chip of the piezoelectric chip array 41 works independently to realize the deflection and focusing of the ultrasonic beam.

[0057] Specifically, the ultrasonic beam deflection of the present embodiment is implemented as follows: the piezoelectric chips of the piezoelectric chip array increase the delay time of the excitation from left to right in sequence, so that the delay time is an arithmetic progression, thereby realizing the ultrasonic beam deflection; after the deflection of the ultrasonic beam is completed, the ultrasonic beam enters the interior of the carbon fiber composite material 6 to be tested to detect defects.

[0058] The filter 5 of this embodiment is fixedly installed on the bottom of the metal probe 4 by bolts, and is used to filter the deflected ultrasonic beam, and the filtered ultrasonic beam is used to perform defect scanning and detection on the workpiece to be tested; the filter 5 is also used to filter the target ultrasonic beam reflected after the defect scanning and detection, and the filtered target ultrasonic beam is focused by controlling the piezoelectric chip array 41 through the transducer 7, and the focused ultrasonic beam is converted into an electrical signal through the transducer 7 and input into the phased array controller 8; the phased array controller 8 converts the received electrical signal into a defect detection image, and the type and position of the defect can be located from the defect detection image.

[0059] After the ultrasonic beam detects the defect, the echo beam passes through the filter 5 to filter out the clutter, and then passes through the piezoelectric chip array to focus the ultrasonic beam, which is received by the transducer 7 and converted into an electrical signal. Specifically, the ultrasonic beam focusing of this embodiment is implemented as follows: the transducer excites the piezoelectric chips at the left and right ends of the piezoelectric chip array, and then continuously increases the delay time along the direction of the center position of the piezoelectric chip array to sequentially excite other piezoelectric chips, forming a wavefront pointing to the center of curvature of the target, thereby achieving ultrasonic beam focusing. After the ultrasonic beam focusing is completed, the focused ultrasonic beam is received by the transducer 7, and then the focused ultrasonic beam is converted into an electrical signal by the transducer 7 and input into the phased array controller 8.

[0060] The specific detection method of the automated ultrasonic phased array detection device according to this embodiment includes the following process:

[0061] The size parameters of the carbon fiber composite material 6 to be tested are input into the phased array controller 8, the phased array controller 8 controls the movement of the movable seat 1, and controls the metal probe 4 to emit an ultrasonic beam, and controls the deflection, scanning and focusing of the ultrasonic beam of the scanning echo through the transducer 7 and the piezoelectric chip array 41, and converts the focused ultrasonic beam into an electrical signal through the transducer 7 and transmits it to the phased array controller 8, and the phased array controller 8 converts the electrical signal into a detection image; at the same time, the wheel encoder 2 records the position information of the metal probe on the workpiece to be tested, and transmits it to the phased array controller 8; after displaying the detection image, the phased array controller 8 calculates the position information of the metal probe 4 input by the wheel encoder on the carbon fiber composite material 6 to be tested, and then combines the size parameters of the carbon fiber composite material 6 to be tested to determine whether there is an unscanned area; if so, the movable seat 1 is controlled to travel to the unscanned area for defect scanning and detection; if not, the phased array controller 8 controls the movable seat 1 to stop running, and the detection is completed.

[0062] like Figure 2 As shown, the phased array controller of this embodiment converts the electrical signal into a detection image, including the following steps:

[0063] (1) The electrical signal is extracted by feature pooling, and the characteristic parameters of the defect are extracted from the time domain, frequency domain and time-frequency domain. The time domain characteristic parameters are obtained respectively. f a , frequency domain characteristic parameters f b and time-frequency domain characteristic parameters f c ;

[0064] Specifically, the time domain characteristic parameters are obtained by analyzing the waveform characteristics of the electrical signal. f a ;

[0065] The electrical signal is converted to the frequency domain using the fast Fourier transform, and then the spectrum of the signal is analyzed in the frequency domain to obtain the frequency-related characteristics that are not easy to find in the time domain, that is, the frequency domain characteristic parameters. f b ;

[0066] The fourth-order Daubechies wavelet function is used to decompose the electrical signal into three layers of wavelet packets to obtain eight frequency band signals. S j , j The value is an integer from 1 to 8; then calculate the energy corresponding to the signal in each frequency band E j , extract the characteristic energy spectrum in the time-frequency domain T j And the time-frequency domain feature parameters are obtained through feature selection f c ;

[0067] ;

[0068] ;

[0069] in, S j ( t )for t Frequency band signal at time S j The signal value of Frequency band signal S j No. k The amplitude of discrete points, k The value range is 1 to N integer, N Frequency band signal S j The total number of discrete points.

[0070] (2) The time domain characteristic parameters f a , frequency domain characteristic parameters f b The fusion feature parameters are obtained by weighted fusion f 1. Then the characteristic parameters f 1 and time-frequency domain characteristic parameters f c Through feature selection, the target feature parameters are obtained. f 2;

[0071] Specifically, the characteristic parameters f 1 is:

[0072] ;

[0073] in, w a and w b are the weights of the time domain characteristic parameters and the frequency domain characteristic parameters respectively;

[0074] The characteristic parameters f 1 is fused with the time-frequency domain feature parameters through feature selection to obtain the feature parameters f The process of 2 includes:

[0075] The characteristic parameters f 1 and time-frequency domain characteristic parameters f c Combine to get the feature vector f =( f 1, f c ), and the eigenvector f is normalized:

[0076] ;

[0077] in, J is the cost function for the feature vector f, The cost function J The change in the eigenvector f, is the weight change of the feature vector f;

[0078] Calculate the cost function J The Hessian matrix H :

[0079] ;

[0080] Pair Matrix H Perform eigenvalue decomposition to obtain two eigenvalues ​​and their corresponding eigenvectors. Take the eigenvector corresponding to the larger eigenvalue as the target eigenvector. f Project to the target feature vector to obtain the feature parameter set;

[0081] Replace the feature vector with the feature vector composed of the feature parameter set f , repeat the above process to obtain a new feature parameter set;

[0082] Continue to iterate the above process until the number of feature parameters in the accumulated feature parameter set reaches the target number of feature parameters, that is, the target feature parameter f 2.

[0083] (3) Optimize the target feature parameters through the BP neural network based on genetic algorithm f 2. Perform intelligent recognition, identify the type and location of defects and convert them into detection images;

[0084] Specifically, the BP neural network optimized based on genetic algorithm is as follows: the initial weights and thresholds of the BP neural network are optimized by genetic algorithm, so that the optimized BP neural network can better predict samples, and the Levenberg-Marquardt algorithm is used to train the BP neural network to classify defects of different sizes and depths of carbon fiber composite materials, further improving the convergence speed.

[0085] like Figures 3 to 6 As shown, this embodiment is based on the carbon fiber composite material to be tested, taking inclusions, holes, delamination and normal defects as examples, the phased array collected a total of 120 delamination defect signals of different sizes, 30 samples for each signal, and performed time domain feature extraction, frequency domain feature extraction, time-frequency domain feature extraction and feature extraction based on time domain, frequency domain, and time-frequency domain fusion of the received signals, respectively. The feature parameters extracted above are respectively input into the BP neural network based on genetic optimization for recognition. The results show that the signal recognition rate through time domain feature extraction is 75.8%, the signal recognition rate through frequency domain feature extraction is 80%, the signal recognition rate through time-frequency domain feature extraction is 84.2%, and the signal recognition rate through the feature extraction based on time domain, frequency domain, and time-frequency domain fusion of the present embodiment reaches 100%.

[0086] Embodiment 2:

[0087] The automated ultrasonic phased array detection device of this embodiment is different from that of Embodiment 1 in that:

[0088] The workpiece to be tested is not limited to the carbon fiber composite material of Example 1, and any other existing workpiece that can be tested for defects by ultrasonic beam can be used;

[0089] For other structures and detection methods, please refer to Example 1.

[0090] The above description is only a detailed description of the preferred embodiments and principles of the present invention. For ordinary technicians in this field, according to the ideas provided by the present invention, there will be changes in the specific implementation methods, and these changes should also be regarded as the protection scope of the present invention.

Claims

1. A detection method of an automated ultrasonic phased array detection device, characterized in that: The automated ultrasonic phased array detection device comprises a moving seat, an ultrasonic phased array probe, a transducer, a probe clamping frame, a wheel encoder, and a phased array controller. The probe clamping frame is arranged on the moving seat and is used to clamp the ultrasonic phased array probe. The ultrasonic phased array probe comprises a metal probe and a filter. The metal probe is connected to the phased array controller via a transducer and is used to emit an ultrasonic beam. The metal probe has a piezoelectric chip array. Each piezoelectric chip of the piezoelectric chip array works independently and is used to deflect and focus the ultrasonic beam. The filter is fixedly installed at the bottom of the metal probe and is used to filter the deflected ultrasonic beam. The filtered ultrasonic beam performs defect scanning and detection on the workpiece to be tested. The filter is also used to filter the target ultrasonic beam reflected after the defect scanning and detection. The filtered target ultrasonic beam is focused by controlling the piezoelectric chip array through the transducer. The focused ultrasonic beam is converted into an electrical signal through the transducer and input into the phased array controller. The phased array controller converts the received electrical signal into a detection image. The wheel encoder is arranged on the moving seat, and is used to record the scanning position information of the ultrasonic phased array probe on the workpiece to be measured, and transmit it to the phased array controller to determine whether the defect scanning detection area completely covers the workpiece to be measured; The detection method The process includes: The size parameters of the workpiece to be measured are input into the phased array controller, which controls the movement of the moving seat and controls the metal probe to emit an ultrasonic beam. The ultrasonic beam is deflected, scanned and focused through the transducer and the piezoelectric chip array. The focused ultrasonic beam is converted into an electrical signal through the transducer and transmitted to the phased array controller, which converts the electrical signal into a detection image. At the same time, the wheel encoder records the position information of the metal probe on the workpiece to be measured, and transmits it to the phased array controller. The phased array controller converts the electrical signal into a detection image, comprising the following steps: (1) Feature extraction of electrical signals is performed by feature pooling, and characteristic parameters of defects are extracted from three perspectives: time domain, frequency domain, and time-frequency domain. Time domain characteristic parameters, frequency domain characteristic parameters, and time-frequency domain characteristic parameters are obtained respectively; (2) The time domain feature parameters and frequency domain feature parameters are fused in a weighted manner to obtain feature parameters f 1. Then the characteristic parameters f 1 is fused with the time-frequency domain feature parameters through feature selection to obtain the feature parameters f 2; (3) The characteristic parameters are optimized by BP neural network based on genetic algorithm f 2. Perform intelligent recognition, identify the type and location of defects and convert them into images; In the step (1), characteristic parameters of the defect are extracted from three perspectives: time domain, frequency domain and time-frequency domain, including: By analyzing the waveform characteristics of the electrical signal, time domain analysis is performed to obtain the time domain characteristic parameters ; The electrical signal is converted into the frequency domain using the fast Fourier transform, and then the spectrum of the signal is analyzed in the frequency domain to obtain the frequency domain characteristic parameters. f b ; The fourth-order Daubechies wavelet function is used to decompose the electrical signal into three layers of wavelet packets to obtain eight frequency band signals. S j , j The value is an integer from 1 to 8; then calculate the energy corresponding to the signal in each frequency band E j , extract the characteristic energy spectrum in the time-frequency domain T j And the time-frequency domain feature parameters are obtained through feature selection f c ; ； ； in, S j ( t )for t Frequency band signal at time S j The signal value of Frequency band signal S j No. k The amplitude of discrete points, k The value range is 1 to N integer, N Frequency band signal S j The total number of discrete points; In step (2), the characteristic parameter f 1 is: ； in, and are the weights of the time domain characteristic parameters and the frequency domain characteristic parameters respectively; The characteristic parameters f 1 is fused with the time-frequency domain feature parameters through feature selection to obtain the feature parameters f The process of 2 includes: The characteristic parameters f 1 and time-frequency domain characteristic parameters f c Combine to get the feature vector f =( f 1, f c ), and for the eigenvector f To standardize: in, J For the eigenvector f The cost function of The cost function J In the feature vector f The amount of change, is the feature vector f The weight change of Calculate the cost function J The Hessian matrix H : ； Pair Matrix H Perform eigenvalue decomposition to obtain two eigenvalues ​​and their corresponding eigenvectors. Take the eigenvector corresponding to the larger eigenvalue as the target eigenvector. f Project it onto the target feature vector to obtain a feature parameter set; replace the feature vector with the feature vector composed of the feature parameter set f , repeat the above process to obtain a new feature parameter set; until the number of feature parameters in the accumulated feature parameter set reaches the target number of feature parameters, the feature parameter set is obtained. f 2.

2. The detection method according to claim 1, characterized in that: In the step (3), the BP neural network optimized based on the genetic algorithm is: optimizing the initial weights and thresholds of the BP neural network by the genetic algorithm, and training the BP neural network by the Levenberg-Marquardt algorithm.

3. The detection method according to claim 1 or 2, characterized in that: After displaying the inspection image, the phased array controller calculates the position information of the metal probe input by the wheel encoder on the workpiece to be tested, and then combines the size parameters of the workpiece to be tested to determine whether there is an unscanned area; if so, the phased array controller controls the moving seat to move to the unscanned area for defect scanning and detection; if not, the phased array controller controls the moving seat to stop running and the detection is completed.

4. The detection method according to claim 1, characterized in that: The ultrasonic beam deflection is achieved by: The piezoelectric chips of the piezoelectric chip array increase the delay time of the excitation from left to right in sequence, so that the delay time is an arithmetic progression, thereby realizing the deflection of the ultrasonic beam.

5. The detection method according to claim 4, characterized in that: The ultrasonic beam focusing is achieved by: The transducer excites the piezoelectric chips at the left and right ends of the piezoelectric chip array, and then excites other piezoelectric chips in sequence along the direction of the center position of the piezoelectric chip array by continuously increasing the delay time, forming a wavefront pointing to the center of the target curvature, thereby achieving ultrasonic beam focusing.

6. The detection method according to claim 1, 4 or 5, characterized in that: The workpiece to be tested is a carbon fiber composite material.

Images