Mechanical arm ultrasonic guided wave multi-defect detection path planning method based on data driving

Abstract

The invention relates to a mechanical arm ultrasonic guided wave multi-defect detection path planning method based on data driving, relates to the technical field of nondestructive detection and automatic detection, and solves the technical problem that automatic defect detection for a large plate structure and a large curved surface structure does not exist in the prior art. A system applicable to the method comprises a control and test system and a tested piece. According to the invention, the mechanical arm is used for carrying the electromagnetic acoustic sensor to scan a large plate and a large curved surface structure, so that detection, positioning and identification of various different defects are realized; the use of a coupling agent can be reduced, the detection time efficiency is improved, the detection precision is improved, the detection accuracy is realized, and the accuracy of positioning and identifying defects is improved; the scanning path of the mechanical arm is optimally designed, so that the detection efficiency is improved; and the collected data serve as an original data set to construct a deep learning model and an integrated CNN model, and mechanical arm scanning detection path planning of multiple defects in a large component and a large curved surface structure is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of nondestructive testing and automatic testing, and in particular to a data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method. Background Art

[0002] Large plate structures and large curved structures are widely used in the fields of automobiles, high-speed railways, aerospace, etc. Delamination and cracks generated during the processing of these large structures, as well as microcracks and corrosion generated during service, will seriously affect their performance, and in severe cases, will cause the entire device or equipment to fail. At present, the detection of defects in these large structures mostly adopts traditional ultrasonic non-destructive testing methods. During the detection process, the application of coupling agents and the contact between the sensor and the test piece will seriously affect the accuracy of the detection. At the same time, most traditional ultrasonic non-destructive testing methods use manual detection methods, which are time-consuming and labor-intensive for the detection of large structures, and it is difficult to guarantee positioning accuracy and detection accuracy. Therefore, it is necessary to propose an automated detection method for defects in large components to improve timeliness, ensure detection accuracy and positioning accuracy. Summary of the Invention

[0003] The present invention aims to solve the technical problem that there is no automatic defect detection method for large plate structures and large curved surface structures in the prior art, and provides a data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method.

[0004] In order to solve the above technical problems, the technical solutions of the present invention are as follows:

[0005] A data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method includes the following steps: First, based on the reference defect signal and noise signal, the signals collected N times are post-processed and analyzed to distinguish the noise signal from the defect signal. At the same time, the signal amplitude thresholds for different defects on the test piece are set. N is the number of times the signal is collected, and its value can be any positive integer. Secondly, the computer controls the robotic arm carrying the electromagnetic acoustic sensor to perform a global scan of the test piece according to the preset scanning path, and the collected signals are recorded; Compare and analyze the recorded signal with the noise signal and defect signal distinguished in the previous step to determine whether the recorded signal is a defect signal. If so, record the detection position of the defect and perform multi-position detection based on the detection position distribution with the detection position as the center and a certain distance from the defect as the radius. If not, continue scanning the test piece according to the preset scanning path and repeat the above steps. Determine whether the robot arm has completed the global scan. If so, record the locations and types of all defects and the time taken for the scan. If not, continue scanning the test piece according to the preset scan path and repeat the above steps until the global scan is completed. Determine whether all candidate scanning paths have been scanned. If so, compare the scanning times to determine the optimal scanning path. If not, select a different scanning path and repeat the scanning process. The recorded noise signals and defect signals are sorted as the original data set, and the original data set is divided into a training set, a validation set, and a test set; Construct a deep learning model and an integrated CNN model, and use the training set data to train the integrated CNN model; apply the test set data to verify the trained deep learning model, and compare the obtained results with the original data set results to determine whether the defect location and type are accurate.

[0006] In the above technical solution, the detection position distribution is specifically as follows: the defect position is , defect detection position to Located with The distance is On the circumference, they are arranged in a clockwise or counterclockwise direction with equal spacing. to Represents the defect detection position number, where The value of is any positive integer; The deep learning model outputs a classification prediction result, which is the probability of a defect in a specific area, the specific location and type of the defect; The integrated CNN model includes two parts: a CNN model and a regression model; wherein the CNN model is used to extract the data features of the defects, and the regression model is used to establish the relationship between the data features and the defect locations; the integrated CNN model is trained using the training set data; The test set data is used to verify the trained deep learning model, and the obtained results are compared with the results of the original data set to determine whether the defect location and type are accurate. If so, the proposed deep learning model is applied to other test pieces for scanning and inspection. Otherwise, the proposed deep learning model and the integrated CNN model should be optimized to improve the output accuracy.

[0007] In the above technical solution, the preset scanning path of the robot arm is specifically as follows: the scanning direction is first in the positive horizontal direction; when the preset scanning distance is reached, the robot arm scans in the positive vertical direction. distance; then, the scanning direction is the reverse direction of the horizontal direction; when the preset scanning distance is reached, the forward scanning direction is vertically Then, the scanning direction is adjusted to the horizontal positive direction again, and the above steps are repeated until the global scan is completed, wherein the forward and reverse preset scanning distances should be less than or equal to the size of the test piece along the scanning direction.

[0008] In the above technical solution, the preset scanning path of the robot arm is specifically: the robot arm scans in a clockwise or counterclockwise direction, with the end point of the specified scanning area as the scanning starting position, and the scanning step is Scan horizontally in the forward direction for the step increment. When the scanning distance reaches the length of the specified scanning area, turn right 90 degrees and scan at the step increment. Perform vertical reverse scanning for step increments. When the scanning distance reaches the width of the specified scanning area, turn right 90 degrees for the second time to scan the step Perform horizontal reverse scanning for step increments. When the scanning distance reaches the length of the specified scanning area, turn right 90 degrees for the third time to scan the step Scan vertically in the forward direction for the step increment. When the scanning distance reaches the end point of the scanning area specified by the distance, it is a scanning step. When the scan distance before each next turn decreases, the scan step is reduced. ,in is the scanning step, The value of is any numeric value.

[0009] In the above technical solution, the preset scanning path of the robot arm is specifically: randomly selecting Y positions within the specified scanning area to test the test piece and record the collected signals; where Y is the number of selected positions and is any positive integer. In the above technical solution, the deep learning model specifically includes: an input layer, a convolution layer, a pooling layer, an activation layer, and an output layer;

[0010] Among them, the input layer serves as the data entry of the model, is responsible for receiving the original data, and converting it into a tensor format that the model can process, providing initial input for subsequent layers; the convolution layer extracts local features of the input data through filters, captures the local correlation of the data, and reduces model parameters through parameter sharing to improve computational efficiency; the pooling layer samples the feature map output by the convolution layer, and reduces the feature dimension and the amount of computation by retaining the maximum value or average value of the local area, while enhancing the model's robustness to slight deformations of the input data; the activation layer breaks the limitations of the linear model by introducing nonlinear transformations, enabling the model to learn complex nonlinear relationships and improve its ability to express complex patterns; the output layer serves as the result outlet of the model and outputs the final result according to the task type.

[0011] In the above technical solution, the tested piece is a plate structure or a large curved surface structure.

[0012] In the above technical solution, the tested piece is a large-sized metal plate specimen, and its specific size is 2 m×2 m×3 mm.

[0013] In the above technical solution, the types of defects include: regular type and irregular type; wherein the regular type includes: rectangle, circle and polygon.

[0014] In the above technical solution, the test system to which the method of the present invention is applicable includes: a control and test system, a test piece; The control and test system includes: a computer, a robotic arm, an electromagnetic acoustic sensor, a signal generator / receiver, a power amplifier, and a signal amplifier; wherein the electromagnetic acoustic sensor includes: a transmitting sensor and a receiving sensor; The transmitting sensor is used to transmit a first ultrasonic guided wave; the receiving sensor is used to receive a second ultrasonic guided wave;

[0015] The computer is connected to the signal generator / receiver and the robotic arm respectively; the signal generator / receiver is connected to the power amplifier and the signal amplifier respectively; the electromagnetic acoustic sensor is connected to the power amplifier, the signal amplifier and the robotic arm respectively;

[0016] The computer is used to control the signal generator / receiver and the operation of the robotic arm;

[0017] The robotic arm is used to control the movement position of the electromagnetic acoustic sensor; the electromagnetic acoustic sensor is used to detect the test piece;

[0018] A signal generator / receiver is used to generate and / or receive signals;

[0019] The power amplifier is used to amplify the power of the signal generated by the signal generator / receiver;

[0020] The signal amplifier is used to amplify the signal detected and returned by the electromagnetic acoustic sensor.

[0021] The present invention has the following beneficial effects:

[0022] The data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method of the present invention uses a robotic arm carrying an electromagnetic acoustic sensor to scan large plates and large curved surface structures, thereby realizing the detection, positioning and identification of a variety of different defects; it can reduce the use of coupling agents, greatly improve detection time efficiency, improve detection accuracy and achieve detection accuracy, as well as improve the accuracy of defect positioning and identification; optimize the scanning path of the robotic arm to improve detection efficiency; use the collected data as the original data set to construct a deep learning model and an integrated CNN model to realize robotic arm scanning detection path planning for multiple defects inside large components and large curved surface structures. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0024] Figure 1 This is a structural diagram of a test system applicable to the data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method of the present invention.

[0025] Figure 2 A schematic diagram of a robotic arm scanning path.

[0026] Figure 3 A schematic diagram of another robotic arm scanning path.

[0027] Figure 4 Schematic diagram of another robotic arm scanning path.

[0028] Figure 5 Schematic diagram of detection location distribution.

[0029] Figure 6 Schematic diagram of noise signal.

[0030] Figure 7 Schematic diagram of defect signal.

[0031] Figure 8 It is a schematic diagram of the detection process of the data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method of the present invention.

[0032] Figure 9 This is a workflow diagram of the deep learning model.

[0033] The reference numerals in the figures indicate:

[0034] 1: Control and test system;

[0035] 11: Computer; 12: Robotic arm; 13: Electromagnetic acoustic sensor; 14: Signal generator / receiver; 15: Power amplifier; 16: Signal amplifier;

[0036] 131: transmitting sensor; 132: receiving sensor; 133: first ultrasonic guided wave; 134: second ultrasonic guided wave;

[0037] 2: Large-size metal sheet specimen; 21: Defect. DETAILED DESCRIPTION

[0038] The present invention will be described in detail below with reference to the accompanying drawings.

[0039] The data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method of the present invention is applicable to the test system, such as Figure 1 As shown, it includes: a control and test system 1, and a test piece; the test piece is a large-sized metal plate specimen 2, and its specific size is 2 m×2 m×3 mm.

[0040] in:

[0041] The control and test system 1 includes a computer 11, a robotic arm 12, an electromagnetic acoustic sensor 13, a signal generator / receiver 14, a power amplifier 15, and a signal amplifier 16. The electromagnetic acoustic sensor 13 is primarily composed of two parts: a transmitting sensor 131 and a receiving sensor 132. The transmitting sensor 131 is used to emit a first ultrasonic guided wave 133; the receiving sensor 132 is used to receive a second ultrasonic guided wave 134.

[0042] The computer 11 is connected to the signal generator / receiver 14 and the robotic arm 12 respectively; the signal generator / receiver 14 is connected to the power amplifier 15 and the signal amplifier 16 respectively; the electromagnetic acoustic sensor 13 is connected to the power amplifier 15, the signal amplifier 16, and the robotic arm 12 respectively;

[0043] The computer 11 is used to control the operation of the signal generator / receiver 14 and the robotic arm 12;

[0044] The robotic arm 12 is used to control the movement of the electromagnetic acoustic sensor 13; the electromagnetic acoustic sensor 13 is used to detect the large-sized metal plate specimen 2;

[0045] The signal generator / receiver 14 is used to generate and / or receive signals;

[0046] The power amplifier 15 is used to amplify the power of the signal generated by the signal generator / receiver 14;

[0047] The signal amplifier 16 is used to amplify the signal detected and returned by the electromagnetic acoustic sensor 13;

[0048] Specifically, the signal generated by the signal generator / receiver 14 is amplified and processed by the power amplifier 15, and then emitted through the transmitting sensor 131 to form a first ultrasonic guided wave 133; the second ultrasonic guided wave 134 is received by the receiving sensor 132, amplified and processed by the signal amplifier 16, and then received by the signal generator / receiver 14.

[0049] There is a defect 21 on the large-sized metal plate specimen 2. There are many types of defects 21, such as Figure 1 Regular types drawn in the image include rectangles, circles, triangles, and polygons, and irregular types such as corrosion and those generated during processing.

[0050] The data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method of the present invention is as follows: Figure 8 As shown (only the outline of the steps of the method is shown in the figure), the following steps are included:

[0051] First, based on the reference defect signal and noise signal, the signals collected N times are post-processed and analyzed to distinguish the noise signal (such as Figure 6 as shown) and defect signals (as Figure 7 As shown), signal amplitude thresholds for different defects 21 are set at the same time, and signal amplitudes are calibrated for different types of defects 21, such as corrosion, cracks, and delamination defects 21; N is the number of times the signal is collected, and its value can be any positive integer. In this embodiment, N = 1000 is taken as an example;

[0052] Secondly, the computer 11 controls the robot arm 12 carrying the electromagnetic acoustic sensor 13 to scan the path diagram (such as Figure 2 As shown in the figure, a global scan is performed on the large-size metal sheet specimen 2 under test; wherein the scanning step is , The value of can be any value, here is = 2 mm as an example, and record the collected signals;

[0053] Compare and analyze the recorded signal with the calibrated signals of different types of defects 21 to determine whether the recorded signal is a defect signal; if so, record the detection position of the defect 21, and use the detection position as the center and the distance between the defect 21 and the detected position as the center. The distance is the radius, according to the detection position distribution (such as Figure 5 as shown) for multiple positions ( Position) detection, here is = 9 as an example; the specific distribution of detection positions is: taking defect 21 as , detect position to Located with The distance is On the circumference, they are arranged in a clockwise or counterclockwise direction with equal spacing. to Represents the detection position number of defect 21, where The value of can be any positive integer, here is = 9 as an example; otherwise, the robot arm 12 follows the preset scanning path (such as Figure 2 As shown in FIG2 , continue scanning the large-size metal sheet specimen 2 and repeat the above steps; Figure 2 As shown, the preset scanning path of the robot arm 12 is specifically as follows: the scanning direction is first in the positive horizontal direction; when the preset scanning distance is reached, the scanning direction is in the positive vertical direction. distance; then, the scanning direction is the reverse direction of the horizontal direction; when the preset scanning distance is reached, the forward scanning direction is vertically Then, the scanning direction is adjusted to the horizontal positive direction again, and the above steps are repeated until the global scan is completed, wherein the forward and reverse preset scanning distances should be less than or equal to the size of the test piece along the scanning direction; wherein The value of satisfies: it can be any positive integer, and The value range should be ≥ wavelength to ensure that a clear reflection signal of defect 21 can be obtained.

[0054] Determine whether the robot arm 12 has completed the global scan. If so, record the positions of all defects 21 and the time taken for the scan. If not, follow the preset scan path (such as Figure 2 Continue scanning the test piece and repeat the above steps until the global scan is completed.

[0055] Determine whether all alternative scanning paths are completed ( Figure 2-4 ) scan, if yes, compare the scan time and determine the optimal scan path, if not, select a different scan path and repeat the scan detection process.

[0056] Next, build a deep learning model and an integrated CNN model;

[0057] The workflow of building a deep learning model, such as Figure 9 As shown (only the outline of the steps of the method is shown in the figure), it specifically includes the following steps:

[0058] First, the recorded noise signals and defect signals are sorted and used as the original data set. The original data set is then divided into three parts: training set, validation set, and test set, and the ratio is set according to the requirements.

[0059] Construct a deep learning model, which specifically includes: input layer, convolution layer, pooling layer, activation layer and output layer; the input layer serves as the data entry of the model, is responsible for receiving the original data, and converting it into a tensor format that the model can process, providing initial input for subsequent layers; the convolution layer mainly extracts local features of the input data through filters, captures the local correlation of the data, and reduces model parameters through parameter sharing to improve computational efficiency; the pooling layer samples the feature map output by the convolution layer, and reduces the feature dimension and the amount of computation by retaining the maximum value or average value of the local area, while enhancing the model's robustness to slight deformations of the input data; the activation layer mainly introduces nonlinear transformations to break the limitations of the linear model, enabling the model to learn complex nonlinear relationships and improve its ability to express complex patterns; the output layer serves as the result exit of the model, outputs the final result according to the task type, and can output classification prediction results based on the model, i.e., the probability of the presence of defect 21 in a specific area, the specific location of defect 21, and the specific type of defect 21;

[0060] Constructing an integrated CNN model, which mainly includes two parts: a CNN model and a regression model; wherein the CNN model is used to extract data features of defect 21, and the regression model is used to establish a relationship between the data features and the location of defect 21; using the training set data to train the integrated CNN model;

[0061] The test set data is used to verify the trained model, and the obtained results are compared with the results of the original data set to determine whether the location and type of defect 21 are accurate. If so, the scanning path efficiency is compared and the optimal scanning path is selected. The proposed deep learning model can then be applied to other test samples for scanning and detection. Otherwise, the proposed deep learning model and the integrated CNN model should be optimized to improve their output accuracy.

[0062] In the data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method of the present invention, the preset and alternative scanning paths include: a serpentine structure, a meandering structure, and a scattered point structure; specifically, Figure 2 As shown, in the above specific embodiment, the scanning path of the robot arm 12 in the data-driven robot arm ultrasonic guided wave multi-defect detection path planning method of the present invention is a serpentine structure, wherein the scanning step is , The value of can be any value. The preset or alternative scanning path of the robot arm 12 scanning path also includes Figure 3 and Figure 4 Specifically, Figure 3 The scanning path shown is as follows: the robot scans in a clockwise or counterclockwise direction, with the end point of the specified scanning area as the scanning starting position, and the scanning step is Scan horizontally in the forward direction for the step increment. When the scanning distance reaches the length of the specified scanning area, turn right 90 degrees and scan at the step increment. Perform vertical reverse scanning for step increments. When the scanning distance reaches the width of the specified scanning area, turn right 90 degrees for the second time to scan the step Perform horizontal reverse scanning for step increments. When the scanning distance reaches the length of the specified scanning area, turn right 90 degrees for the third time to scan the step Scan vertically in the forward direction for the step increment. When the scanning distance reaches the end point of the scanning area specified by the distance, it is a scanning step. When the scan distance before each next turn decreases, the scan step is reduced. ,in is the scanning step, The value of can be any number. Figure 4 The scanning path shown is specifically as follows: randomly selecting multiple test positions, randomly selecting Y positions in the specified scanning area to test the device under test, and recording the collected signals; where Y is the number of selected positions, which is any positive integer.

[0063] In order to accurately determine the location of the defect 21, in other embodiments, The value of should satisfy: is a positive integer, and ≥6.

[0064] In the above specific embodiments, the test piece is a large-sized metal plate test piece 2; in other specific embodiments, the test piece may also be a plate structure of other materials or a large curved surface structure.

[0065] In the above specific embodiment, the sensor involved is the electromagnetic acoustic sensor 13; in other specific embodiments, the sensor may also be: an eddy current sensor, a traditional piezoelectric ceramic sensor, etc.

[0066] The data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method of the present invention uses a robotic arm carrying an electromagnetic acoustic sensor to scan large plates and large curved surface structures, thereby realizing the detection, positioning and identification of a variety of different defects; it can reduce the use of coupling agents, greatly improve detection time efficiency, improve detection accuracy and achieve detection accuracy, as well as improve the accuracy of defect positioning and identification; optimize the scanning path of the robotic arm to improve detection efficiency; use the collected data as the original data set to construct a deep learning model and an integrated CNN model to realize robotic arm scanning detection path planning for multiple defects inside large components and large curved surface structures.

[0067] Obviously, the above embodiments are merely examples for clarity of explanation and are not intended to limit the implementation methods. Those skilled in the art will readily appreciate that other variations or modifications based on the above descriptions are possible. It is not necessary and impossible to enumerate all implementation methods here. Obvious variations or modifications arising therefrom remain within the scope of protection of the present invention.

Claims

1. A data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method, characterized in that: The following steps are involved: First, based on the reference defect signal and noise signal, the signals collected N times are post-processed and analyzed to distinguish the noise signal from the defect signal, and the signal amplitude thresholds for different defects (21) on the test piece are set; N is the number of times the signal is collected, and its value is any positive integer; Secondly, the computer (11) controls the robotic arm (12) carrying the electromagnetic acoustic sensor (13) to perform a global scan of the test piece according to a preset scanning path, and records the collected signals; The recorded signal is compared with the noise signal and the defect signal after the previous step of differentiation to determine whether the recorded signal is a defect signal. If so, the detection position where the defect (21) is detected is recorded, and the detection position is used as the center and a certain distance from the defect (21) is used as the radius to perform multi-position detection according to the detection position distribution; if not, the test piece is scanned according to the preset scanning path and the above steps are repeated; Determine whether the robot arm (12) has completed the global scan, and if so, record the positions and types of all defects (21) and the time taken for the scan detection; if not, continue to scan the test piece according to the preset scan path, and repeat the above steps until the global scan is completed; Determine whether all candidate scanning paths have been scanned. If so, compare the scanning times to determine the optimal scanning path. If not, select a different scanning path and repeat the scanning process. The recorded noise signals and defect signals are sorted as the original data set, and the original data set is divided into a training set, a validation set, and a test set; A deep learning model and an integrated CNN model are constructed, and the integrated CNN model is trained using the training set data; the test set data is used to verify the trained deep learning model, and the obtained results are compared with the results of the original data set to determine whether the position and type of the defect (21) are accurate.

2. The data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method according to claim 1 is characterized in that: The specific distribution of detection locations is as follows: The defect (21) is located at , defect (21) detection position to Located with The distance is On the circumference, they are arranged in a clockwise or counterclockwise direction with equal spacing. to represents the detection position number of defect (21), where The value of is any positive integer; The deep learning model outputs a classification prediction result, which is the probability of whether there is a defect (21) in a specific area, the specific location and type of the defect (21); The integrated CNN model includes two parts: a CNN model and a regression model; wherein the CNN model is used to extract the data features of the defect (21), and the regression model is used to establish the relationship between the data features and the position of the defect (21); the integrated CNN model is trained using the training set data; The test set data is used to verify the trained deep learning model, and the obtained results are compared with the results of the original data set to determine whether the location and type of the defect (21) are accurate. If so, the proposed deep learning model is applied to other test pieces for scanning and detection. If not, the proposed deep learning model and the integrated CNN model should be optimized to improve the output accuracy.

3. The data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method according to claim 1 is characterized in that: The preset scanning path of the robotic arm (12) is specifically: The scanning direction is first in the horizontal positive direction; when the preset scanning distance is reached, the vertical positive direction is scanned. distance; then, the scanning direction is the reverse direction of the horizontal direction; when the preset scanning distance is reached, the forward scanning direction is vertically distance; then, the scanning direction is adjusted to the positive direction of the horizontal direction again, and the above steps are repeated until the global scan is completed; The scanning step is the preset forward and reverse scanning distances, which should be less than or equal to the size of the test piece along the scanning direction.

4. The data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method according to claim 1 is characterized in that: The preset scanning path of the robotic arm (12) is specifically: The robot arm (12) scans in a clockwise or counterclockwise direction, with the end point of the specified scanning area as the scanning starting position and the scanning step distance Scan horizontally in the forward direction for the step increment. When the scanning distance reaches the length of the specified scanning area, turn right 90 degrees and scan at the step increment. Perform vertical reverse scanning for step increments. When the scanning distance reaches the width of the specified scanning area, turn right 90 degrees for the second time to scan the step Perform horizontal reverse scanning for step increments. When the scanning distance reaches the length of the specified scanning area, turn right 90 degrees for the third time to scan the step Scan vertically in the forward direction for the step increment. When the scanning distance reaches the end point of the scanning area specified by the distance, it is a scanning step. When the scan distance before each next turn decreases, the scan step is reduced. ,in is the scanning step, The value of is any numeric value.

5. The data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method according to claim 1 is characterized in that: The preset scanning path of the robotic arm (12) is specifically: Randomly select Y positions within the specified scanning area to test the device under test and record the collected signals; where Y is the number of selected positions and is any positive integer.

6. The data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method according to claim 1 is characterized in that: The deep learning model specifically includes: input layer, convolution layer, pooling layer, activation layer and output layer; Among them, the input layer serves as the data entry of the model, is responsible for receiving the original data, and converting it into a tensor format that the model can process, providing initial input for subsequent layers; the convolution layer extracts local features of the input data through filters, captures the local correlation of the data, and reduces model parameters through parameter sharing to improve computational efficiency; the pooling layer samples the feature map output by the convolution layer, and reduces the feature dimension and the amount of computation by retaining the maximum value or average value of the local area, while enhancing the model's robustness to slight deformations of the input data; the activation layer mainly breaks the limitations of the linear model by introducing nonlinear transformations, enabling the model to learn complex nonlinear relationships and improve its ability to express complex patterns; the output layer serves as the result outlet of the model and outputs the final result according to the task type.

7. The data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method according to claim 1 is characterized in that: The tested piece is a plate structure or a large curved surface structure.

8. The data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method according to claim 7 is characterized in that: The tested piece is a large-sized metal plate specimen (2), whose specific dimensions are 2 m×2 m×3 mm.

9. The data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method according to claim 1 is characterized in that: The types of defects (21) include regular types and irregular types; regular types include rectangles, circles, and polygons.

10. The data-driven robotic arm ultrasonic guided wave multi-defect detection path planning method according to any one of claims 1 to 9, characterized in that: The test system applicable to the method includes: a control and test system (1), a test piece; The control and test system (1) includes: a computer (11), a robotic arm (12), an electromagnetic acoustic sensor (13), a signal generator / receiver (14), a power amplifier (15), and a signal amplifier (16); wherein the electromagnetic acoustic sensor (13) includes: a transmitting sensor (131) and a receiving sensor (132); The transmitting sensor (131) is used to transmit a first ultrasonic guided wave (133); the receiving sensor (132) is used to receive a second ultrasonic guided wave (134); The computer (11) is connected to the signal generator / receiver (14) and the robotic arm (12) respectively; the signal generator / receiver (14) is connected to the power amplifier (15) and the signal amplifier (16) respectively; the electromagnetic acoustic sensor (13) is connected to the power amplifier (15), the signal amplifier (16), and the robotic arm (12) respectively; The computer (11) is used to control the operation of the signal generator / receiver (14) and the robotic arm (12); The mechanical arm (12) is used to control the movement position of the electromagnetic acoustic sensor (13); the electromagnetic acoustic sensor (13) is used to detect the test piece; A signal generator / receiver (14) is used to generate and / or receive signals; The power amplifier (15) is used to amplify the power of the signal generated by the signal generator / receiver (14); The signal amplifier (16) is used to amplify the signal detected and returned by the electromagnetic acoustic sensor (13).

Images