Metal shaft ultrasonic nondestructive testing method based on intelligent sensing system

Abstract

The application discloses a metal shaft ultrasonic nondestructive flaw detection method based on an intelligent sensing system and belongs to the technical field of metal component nondestructive detection. The method emits ultrasonic signals to a metal shaft to be detected which has been subjected to surface cleaning treatment through a phased array ultrasonic detector subjected to calibration processing, obtains echo signals, images after preprocessing of the echo signals, generates a three-dimensional image of the metal shaft, extracts metal shaft damage feature data based on the three-dimensional image of the metal shaft, inputs the metal shaft damage feature data into a metal shaft damage intelligent classification model to obtain a metal shaft flaw detection result, and generates a metal shaft flaw detection report accordingly. The application guarantees detection precision through surface cleaning and instrument calibration, realizes accurate identification and classification of internal damage of the metal shaft by combining three-dimensional imaging, feature extraction and an intelligent classification model, has the advantages of high detection efficiency and good accuracy, and can be effectively applied to nondestructive flaw detection of metal shafts.

Description

Technical Field

[0001] This invention relates to the field of nondestructive testing technology for metal components, and specifically to an ultrasonic nondestructive testing method for metal shafts based on an intelligent sensing system. Background Technology

[0002] As a critical component in mechanical systems, the accurate detection of internal damage to metal shafts is essential for the safe operation of equipment. Currently, ultrasonic non-destructive testing (NDT) is a common method for detecting damage to metal shafts. However, traditional ultrasonic NDT methods are not precise enough for the surface treatment of metal shafts; residual oil, rust, or oxide scale can easily lead to distorted echo data. Furthermore, the calibration process of testing instruments lacks standardized procedures, directly affecting the reliability of the test results. In addition, existing technologies mostly employ two-dimensional imaging or simple three-dimensional reconstruction, which is insufficient to comprehensively present the spatial distribution characteristics of internal damage in metal shafts. Moreover, the extraction of damage features relies on manual analysis, resulting in low efficiency and high subjectivity. Additionally, the classification of damage types is often based on experience or simple algorithms, which cannot accurately identify complex damage morphologies, thus limiting the accuracy of detection.

[0003] Therefore, achieving efficient pretreatment of metal shaft surfaces, accurate calibration of testing instruments, three-dimensional visualization of damage, and intelligent classification have become key technical challenges in improving the accuracy and efficiency of non-destructive testing of metal shafts. To address this, we propose an ultrasonic non-destructive testing method for metal shafts based on an intelligent sensing system. Summary of the Invention

[0004] The purpose of this invention is to provide an ultrasonic non-destructive testing method for metal shafts based on an intelligent sensing system, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] An ultrasonic non-destructive testing method for metal shafts based on an intelligent sensing system includes the following steps:

[0007] S1. The surface of the metal shaft to be tested is cleaned, and the phased array ultrasonic tester is calibrated using a standard test block.

[0008] S2. An ultrasonic signal is emitted to the surface-cleaned metal shaft to be tested using a calibrated phased array ultrasonic detector, and the echo signal is received.

[0009] S3. Preprocess the echo signal, and then perform imaging processing on the preprocessed echo signal to generate a three-dimensional image of the inside of the metal shaft.

[0010] S4. Perform image analysis on the three-dimensional image inside the metal shaft to extract the damage feature data of the metal shaft. At the same time, construct an intelligent classification model for metal shaft damage and input the damage feature data of the metal shaft into the intelligent classification model for metal shaft damage to obtain the metal shaft flaw detection results.

[0011] S5. Store the metal shaft flaw detection results and metal shaft damage characteristic data in the database, and generate a metal shaft flaw detection report at the same time.

[0012] Preferably, the method for surface cleaning of the metal shaft to be tested is as follows:

[0013] The surface of the metal shaft to be tested was cleaned using an ultrasonic cleaner with a frequency of 40kHz for 10 to 30 minutes. The high-frequency vibration removed oil and rust from the surface of the metal shaft. At the same time, the areas of the metal shaft surface with oxide scale were subjected to gradient grinding using a mechanical grinding device. After grinding, the surface roughness was measured with a surface roughness measuring instrument to ensure that the surface roughness of the areas of the metal shaft with oxide scale was less than or equal to 6.3μm. The surface cleanliness was ensured by visual inspection combined with a white cotton cloth wiping test. After wiping, no visible stains were found on the cloth. Based on this, the surface-cleaned metal shaft to be tested met the requirements of ultrasonic testing.

[0014] Preferably, the method for calibrating the phased array ultrasonic testing instrument using a standard test block is as follows:

[0015] A standard test block with the same material and specifications as the metal shaft to be tested is selected. The standard test block is prefabricated with artificial defects of known size and location using wire cutting technology. The phased array ultrasonic probe of the phased array ultrasonic testing instrument is attached to the test surface of the standard test block using coupling agent and machine oil. The parameters of the phased array ultrasonic testing instrument, including gain, attenuation and probe zero point, are adjusted to generate an excitation signal with a pulse width of 0.1 to 1 μs and a voltage of 50 to 200 V to drive the phased array ultrasonic probe to emit ultrasonic waves and receive the echo signal.

[0016] When the phased array ultrasonic detector displays an echo amplitude of 75% to 85% of the full screen, and the echo time has an error of less than or equal to 1% from the theoretical echo time, the current gain, attenuation, and probe delay parameters are recorded as reference detection parameters.

[0017] If the conditions are not met, adjust the gain in 0.5dB steps or recalibrate the probe zero point until the calibration requirements are met. The gain adjustment range is 0 to 100dB.

[0018] Preferably, the method of transmitting ultrasonic signals to a surface-cleaned metal shaft using a calibrated phased array ultrasonic detector is as follows:

[0019] Based on the reference detection parameters including gain, attenuation and probe zero point, the transmission frequency, pulse width and excitation voltage of the phased array ultrasonic detector are set through the operation interface of the phased array ultrasonic detector to complete the detector parameter configuration;

[0020] The calibrated phased array ultrasonic probe is fixed to a six-axis rotatable robotic arm using a magnetic clamp. The six-axis rotatable robotic arm's motion trajectory is circumferential. The scanning method combines continuous rotation with axial linear movement at a speed of 5 to 50 mm / s. During the scanning process, the coupling pressure between the probe and the metal shaft surface is monitored in real time by a pressure sensor to ensure uniform coupling agent layer thickness. At the same time, a high-speed data acquisition card is used to acquire echo signals in real time at a sampling rate of 100 to 500 MS / s. The echo signals are transmitted to a database via gigabit Ethernet for storage. The stored data includes timestamps, phased array ultrasonic probe position coordinates, and echo amplitude values.

[0021] Preferably, the method for data preprocessing of the echo signal is as follows:

[0022] The echo signal is filtered using an 8th-order Butterworth bandpass filter. The passband range is automatically adjusted according to the metal shaft material. The attenuation slope at the cutoff frequency is greater than or equal to 40dB / decade to filter out low-frequency environmental noise below 0.5MHz and high-frequency electrical interference above 25MHz. The echo signal after filtering out low-frequency noise and high-frequency interference is decomposed into three layers using a db4 wavelet basis for noise reduction. The detail coefficients of each layer are processed by a soft threshold function. When reconstructing the signal, the Symlet boundary extension algorithm is used to avoid edge distortion. Based on this, the preprocessed echo signal is obtained.

[0023] The Symlet boundary extension algorithm is a dedicated extension method for Symlet wavelets to address edge distortion problems when processing signals of finite length.

[0024] Preferably, the process of generating a three-dimensional image of the interior of the metal shaft from the preprocessed echo signal using synthetic aperture focusing technology is as follows:

[0025] The preprocessed echo signals are grouped according to the position coordinates of the phased array ultrasonic probe. Delay compensation is performed on each group of signals. The focused pixel intensity values ​​are generated through a coherent superposition algorithm until the pixel intensity values ​​of all pixels in the entire metal shaft detection area are obtained. Based on this, a three-dimensional image of the metal shaft interior with a resolution of 0.5mm×0.5mm×0.5mm is constructed, and the image depth range covers the entire wall thickness of the metal shaft.

[0026] The synthetic aperture focusing technology is an ultrasonic detection technology that simulates the focusing effect of a large-aperture probe, performs phase compensation and coherent superposition on ultrasonic echo signals collected by a small-sized phased array probe at different positions, thereby achieving high-resolution imaging.

[0027] The coherent superposition algorithm is a signal processing method based on wave theory.

[0028] Preferably, the method for performing image analysis on the three-dimensional image inside the metal shaft to extract damage feature data of the metal shaft is as follows:

[0029] The damage characteristic data of the metal shaft includes axial range, radial range, depth, circumferential distribution range, volume, surface area, aspect ratio and roundness;

[0030] The 3D image inside the metal shaft is binarized using the Otsu 3D adaptive thresholding segmentation algorithm to segment the 3D image inside the metal shaft into damaged regions. and background area For the damaged area voxel coordinates of each voxel point The corresponding physical coordinates are obtained by calculating using coordinate system transformation formulas. Statistical analysis of damaged areas Physical coordinates of all voxel points The maximum value of the Z coordinate is obtained. Minimum Z-coordinate Maximum absolute value of Y coordinate and the minimum absolute value of the Y coordinate Based on this, and Difference and and By subtracting, the axial range is obtained. and radial range Among them, the minimum absolute value of the Y-coordinate depth ;

[0031] The Otsu 3D adaptive threshold segmentation algorithm is a 3D extension of the classic Otsu 2D threshold segmentation algorithm, used to achieve accurate separation of defect areas and background areas;

[0032] The coordinate system transformation formula is:

[0033] ;

[0034] in, , and It is the resolution of the three-dimensional image inside the metal shaft. These are the origin voxel coordinates of the three-dimensional image inside the metal shaft;

[0035] For the damaged area Physical coordinates of all voxel points The circumferential angles were calculated separately using the circumferential angle formula. Statistical analysis of the circumferential angles of all voxel points Minimum value of circumferential angle and the maximum value of the circumferential angle Based on this, the circumferential distribution interval is obtained. and the circumferential distribution range and axial range The aspect ratio is calculated using the aspect ratio formula. ;

[0036] The formula for the circumferential angle is:

[0037] ;

[0038] in, These are the physical coordinates of the voxel point on the XOY plane;

[0039] For the damaged area The volume was obtained by processing the data using the three-dimensional voxel integral method and the Marching Cubes algorithm, respectively. and surface area and volume and surface area The roundness is calculated using the roundness formula. ;

[0040] The formula for roundness is:

[0041] ;

[0042] in, It is the damaged area volume, It is the damaged area Surface area;

[0043] The three-dimensional voxel integration method is a volume calculation method based on three-dimensional image voxel units;

[0044] The Marching Cubes algorithm is an algorithm for extracting isosurfaces from three-dimensional volume data.

[0045] Preferably, the method for constructing the intelligent classification model for metal shaft damage and obtaining the metal shaft flaw detection results is as follows:

[0046] Historical metal shaft damage feature data were extracted from the detection database as a sample set. The sample set was divided into a training set and a test set in an 8:2 ratio. The metal shaft flaw detection results of the training set and the test set were manually labeled. The training set with labeled metal shaft flaw detection results was input into a convolutional neural network (CNN) for training, and the test set with labeled metal shaft flaw detection results was input into the trained CNN to obtain the predicted metal shaft flaw detection results. The number of samples where the predicted metal shaft flaw detection results were inconsistent with the metal shaft flaw detection results labeled in the test set was counted. If the number of samples was less than 0.04 of the number of samples in the test set, the intelligent classification model for metal shaft damage was obtained. Otherwise, the trained CNN was optimized using the Adam optimizer, and the optimized CNN was retrained.

[0047] Inputting the metal shaft damage feature data into the intelligent classification model of metal shaft damage yields metal shaft flaw detection results including cracks, porosity, inclusions, or no defects.

[0048] The Convolutional Neural Network (CNN) is a multi-layer neural network model based on deep learning.

[0049] Preferably, the method for generating a metal shaft flaw detection report is as follows:

[0050] A template-based automatic generation method is adopted, which integrates the metal shaft flaw detection results, metal shaft damage feature data, benchmark detection parameters and internal 3D images of the metal shaft into a preset report template through the report generation module;

[0051] The PDF template is designed using LaTeX syntax and includes a cover page, parameter page, defect analysis page, and conclusion page.

[0052] The Word template uses an XML schema to define bookmark fields, automatically fills in feature parameters through a data mapping engine, and embeds 3D images into OLE objects using Base64 encoding.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] 1. This invention achieves fine pretreatment of the metal shaft surface through ultrasonic cleaning combined with mechanical grinding, effectively eliminating the interference of oil, rust and oxide scale on the echo signal. At the same time, it uses a standardized calibration process with standard test blocks to ensure the accuracy of the phased array ultrasonic testing instrument parameters, thus ensuring the reliability of the test data from the source.

[0055] 2. This invention utilizes synthetic aperture focusing technology to construct a three-dimensional image of the interior of a metal shaft, breaking through the limitations of traditional two-dimensional imaging and comprehensively presenting the spatial distribution characteristics of damage. Furthermore, through three-dimensional adaptive threshold segmentation and automated extraction of feature parameters, it avoids the subjectivity and inefficiency of manual analysis. At the same time, based on a convolutional neural network, it constructs an intelligent classification model to achieve accurate identification of damage types and significantly improve the classification accuracy of complex damage morphologies.

[0056] 3. This invention realizes a templated report generation and data storage mechanism, further improving the automation and standardization of the testing process. The overall testing efficiency and accuracy are superior to existing technologies, and it has significant engineering application value. Attached Figure Description

[0057] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1 This is a flowchart of the method steps of the present invention. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be described in detail below. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other implementation methods obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0060] Examples, such as Figure 1 As shown, a method for ultrasonic non-destructive testing of metal shafts based on an intelligent sensing system includes the following steps:

[0061] S1. Perform surface cleaning on the metal shaft to be tested, and simultaneously calibrate the phased array ultrasonic testing instrument using a standard test block; S2. Emitter ultrasonic signals are emitted from the calibrated phased array ultrasonic testing instrument to the surface-cleaned metal shaft to be tested, and echo signals are received.

[0062] S3. Preprocess the echo signal, and then perform imaging processing on the preprocessed echo signal to generate a three-dimensional image of the inside of the metal shaft.

[0063] S4. Perform image analysis on the three-dimensional image inside the metal shaft to extract the damage feature data of the metal shaft. At the same time, construct an intelligent classification model for metal shaft damage and input the damage feature data of the metal shaft into the intelligent classification model for metal shaft damage to obtain the metal shaft flaw detection results.

[0064] S5. Store the metal shaft flaw detection results and metal shaft damage characteristic data in the database, and generate a metal shaft flaw detection report at the same time.

[0065] Furthermore, the working principle of the present invention will be illustrated below through embodiments:

[0066] The 45# steel metal shaft with a diameter of 100mm and a length of 500mm to be inspected underwent surface pretreatment. A 40kHz ultrasonic cleaner was used with deionized water for 20 minutes to remove surface oil and rust. For the shoulder area with oxide scale, a diamond wheel mechanical grinding device was used for gradient grinding, from coarse to fine and then to high-precision grinding. The surface roughness after grinding was measured to be 4.5μm using a TR200 roughness meter. A wiping test with a white cotton cloth showed no visible stains on the cloth. Meanwhile... A standard test block with the same material and specifications as the metal shaft to be tested is selected. The standard test block contains a pre-fabricated transverse hole with a diameter of 1 mm and a depth of 5 mm. A 5 MHz phased array ultrasonic probe with 64 elements is attached to the test surface of the test block using machine oil coupling agent. The gain of the detector is adjusted to 60 dB and the attenuation to 10 dB to generate an excitation signal with a pulse width of 0.5 μs and a voltage of 100 V. When the echo amplitude reaches 80% of the full screen and the echo time error is 0.8%, the current parameters are recorded as the reference test parameters.

[0067] Based on the benchmark detection parameters, the detector's transmission frequency was set to 5MHz and the pulse width to 0.5μs. The probe was fixed to the six-axis robotic arm using a magnetic clamp, and a rotation speed of [missing information] was used. Circumferential The composite scanning method combines continuous rotation with axial linear movement at a speed of 20 mm / s. During the scanning process, the coupling pressure is monitored in real time by a pressure sensor, and the echo signal is acquired using a high-speed data acquisition card with a sampling rate of 200 MS / s. The data includes timestamps, probe coordinates, and amplitude values, and is stored in the database via gigabit Ethernet.

[0068] The echo signal was filtered by an 8th-order Butterworth bandpass filter with a passband of 1 to 20 MHz and an attenuation slope of 40 dB / decade to remove low-frequency environmental noise and high-frequency electrical interference. Then, a 3-level decomposition was performed using a db4 wavelet basis, noise was reduced by a soft thresholding function with a threshold coefficient of 1.5, and the signal was reconstructed using the Symlet boundary extension algorithm. The preprocessed echo signal was grouped according to the probe position and time compensation was performed. The intensity value of the focused pixel was generated by the coherent superposition algorithm to construct a 3D image of the metal shaft with a resolution of 0.5 mm × 0.5 mm × 0.5 mm, covering the entire wall thickness of the metal shaft.

[0069] The Otsu 3D adaptive thresholding segmentation algorithm is used to binarize the 3D image inside the metal shaft, separating the damaged region Ω from the background region. The voxel coordinates are then converted to physical coordinates using a coordinate transformation formula. The coordinate transformation formula is... and All values ​​are 0.5mm. The Z-coordinate range is statistically determined to be [100, 150]mm, resulting in an axial range L of 50mm. The minimum absolute value of the Y-coordinate, min|y|, is 10mm, indicating a depth d of 10mm. The circumferential angle of the voxel is calculated using the circumferential angle formula, yielding the circumferential distribution range. The aspect ratio AR is calculated to be 1.19 based on the axial range. At the same time, the volume V is calculated to be 20 mm³ using the three-dimensional voxel integration method, and the surface area S is extracted to be 35 mm² using the Marching Cubes algorithm. Substituting these values ​​into the roundness formula, the roundness C is calculated to be 0.68.

[0070] 1000 sets of historical metal shaft damage feature data were extracted from the database and divided into training and testing sets in an 8:2 ratio. Damage types were manually labeled and input into a convolutional neural network (CNN) for training. The CNN consists of 3 convolutional layers and 2 fully connected layers. When the classification error rate of the testing set is less than 4%, a metal shaft damage intelligent classification model is obtained. Inputting the current metal shaft damage feature data, the output metal shaft flaw detection result is "axial crack, length 50mm, depth 10mm".

[0071] The flaw detection results and benchmark detection parameters of the metal shaft are stored in a MySQL database. A templated PDF report is generated, which includes a cover page, a parameter page, a defect analysis page, and a conclusion page. The conclusion page displays the flaw detection results of the metal shaft, the defect analysis page displays the embedded 3D image of the metal shaft, and the parameter page displays the damage characteristic data and benchmark detection parameters of the metal shaft.

[0072] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them; modifications to the technical solutions described in the foregoing embodiments, or equivalent substitutions of some of the technical features, do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for ultrasonic non-destructive testing of metal shafts based on an intelligent sensing system, characterized in that, Includes the following steps: S1. The surface of the metal shaft to be tested is cleaned, and the phased array ultrasonic tester is calibrated using a standard test block. S2. An ultrasonic signal is emitted to the surface-cleaned metal shaft to be tested using a calibrated phased array ultrasonic detector, and the echo signal is received. S3. Preprocess the echo signal, and then perform imaging processing on the preprocessed echo signal to generate a three-dimensional image of the inside of the metal shaft. S4. Perform image analysis on the three-dimensional image of the metal shaft's interior to extract metal shaft damage feature data, including axial range, radial range, depth, circumferential distribution range, volume, surface area, aspect ratio, and roundness. Simultaneously, construct a metal shaft damage intelligent classification model and input the metal shaft damage feature data into the model to obtain the metal shaft flaw detection results. The method for extracting the metal shaft damage feature data is as follows: The 3D image inside the metal shaft is binarized using the Otsu 3D adaptive thresholding segmentation algorithm to segment the 3D image inside the metal shaft into damaged regions. and background area For the damaged area voxel coordinates of each voxel point The corresponding physical coordinates are obtained by calculating using coordinate system transformation formulas. Statistical analysis of damaged areas Physical coordinates of all voxel points The maximum value of the Z coordinate is obtained. Minimum Z-coordinate Maximum absolute value of Y coordinate and the minimum absolute value of the Y coordinate Based on this, and Difference and and By subtracting, the axial range is obtained. and radial range The minimum absolute value of the Y coordinate depth ; For the damaged area Physical coordinates of all voxel points The circumferential angles were calculated separately using the circumferential angle formula. Statistical analysis of the circumferential angles of all voxel points Minimum value of circumferential angle and the maximum value of the circumferential angle Based on this, the circumferential distribution interval is obtained. and the circumferential distribution range and axial range The aspect ratio is calculated using the aspect ratio formula. ; For the damaged area The volume was obtained by processing the data using the three-dimensional voxel integral method and the Marching Cubes algorithm, respectively. and surface area and volume and surface area The roundness is calculated using the roundness formula. ; The method for constructing the intelligent classification model for metal shaft damage and obtaining the metal shaft flaw detection results is as follows: Historical metal shaft damage feature data were extracted from the detection database as a sample set. The sample set was divided into a training set and a test set in an 8:2 ratio. The metal shaft flaw detection results of the training set and the test set were manually labeled. The training set with labeled metal shaft flaw detection results was input into a convolutional neural network (CNN) for training, and the test set with labeled metal shaft flaw detection results was input into the trained CNN to obtain the predicted metal shaft flaw detection results. The number of samples where the predicted metal shaft flaw detection results were inconsistent with the metal shaft flaw detection results labeled in the test set was counted. If the number of samples was less than 0.04 of the number of samples in the test set, the intelligent classification model for metal shaft damage was obtained. Otherwise, the trained CNN was optimized using the Adam optimizer, and the optimized CNN was retrained. Inputting the metal shaft damage feature data into the intelligent classification model of metal shaft damage yields metal shaft flaw detection results including cracks, porosity, inclusions, or no defects. S5. Store the metal shaft flaw detection results and metal shaft damage characteristic data in the database, and generate a metal shaft flaw detection report at the same time.

2. The ultrasonic non-destructive testing method for metal shafts based on an intelligent sensing system according to claim 1, characterized in that, The method for surface cleaning of the metal shaft to be tested: An ultrasonic cleaner is used to clean the surface of the metal shaft to be tested, removing oil and rust. At the same time, the areas of the metal shaft surface with oxide scale are polished using a mechanical polishing device, resulting in a surface-cleaned metal shaft.

3. The ultrasonic non-destructive testing method for metal shafts based on an intelligent sensing system according to claim 2, characterized in that, The method for calibrating a phased array ultrasonic testing instrument using a standard test block: A standard test block with the same material and specifications as the metal shaft to be tested is selected. The standard test block is prefabricated with artificial defects of known size and location. The phased array ultrasonic probe of the phased array ultrasonic tester is attached to the surface of the standard test block. The parameters of the phased array ultrasonic tester are adjusted, and an excitation signal is generated to drive the phased array probe to emit ultrasonic signals. At the same time, the echo signal is received. If the echo amplitude of the echo signal displayed by the phased array ultrasonic tester is between 75% and 85% of the full screen, and the error between the echo time and the theoretical echo time is less than or equal to 1%, then the parameters of the phased array ultrasonic tester are recorded as the reference test parameters. Otherwise, adjust the parameters of the phased array ultrasonic detector and recalibrate the phased array ultrasonic detector.

4. The ultrasonic non-destructive testing method for metal shafts based on an intelligent sensing system according to claim 3, characterized in that, The method of transmitting ultrasonic signals to a surface-cleaned metal shaft using a calibrated phased array ultrasonic detector: Based on the benchmark detection parameters, the parameters of the phased array ultrasonic detector are determined, resulting in a calibrated phased array ultrasonic detector. The phased array ultrasonic probe of the calibrated phased array ultrasonic detector is fixed on a rotatable robotic arm. Using a scanning method combining circumferential and axial directions, ultrasonic signals are emitted to the surface-cleaned metal shaft, and echo signals are received in real time.

5. The ultrasonic non-destructive testing method for metal shafts based on an intelligent sensing system according to claim 4, characterized in that, The method for data preprocessing of echo signals: A bandpass filter is used to filter out low-frequency noise and high-frequency interference from the echo signal. The signal-to-noise ratio of the echo signal after filtering out low-frequency noise and high-frequency interference is improved by wavelet denoising algorithm, thereby obtaining the preprocessed echo signal.

6. The ultrasonic non-destructive testing method for metal shafts based on an intelligent sensing system according to claim 5, characterized in that, The method for generating a three-dimensional image of the interior of a metal shaft: The preprocessed echo signal is used to generate a three-dimensional image of the interior of the metal shaft using synthetic aperture focusing technology.

7. The ultrasonic non-destructive testing method for metal shafts based on an intelligent sensing system according to claim 1, characterized in that, The method for generating a flaw detection report for a metal shaft: A template-based automatic generation method is adopted to integrate the metal shaft flaw detection results, metal shaft damage feature data, benchmark detection parameters and internal three-dimensional images of the metal shaft into a preset report template to generate a metal shaft flaw detection report. The preset report template is a PDF template or a Word template.

Images