Rubber-lined pipeline defect detection method and detection system

Abstract

The invention discloses a rubber-lined pipeline defect detection method and system, and the method comprises the steps: arranging an ultrasonic probe array, obtaining large-area scanning information of a rubber-lined pipeline, collecting an ultrasonic signal, an echo signal and a resonance signal, generating a resonance signal curve, and reflecting a mapping image of the interior of the rubber-lined pipeline. And calculating a half-power bandwidth and a damping characteristic parameter, generating a corresponding characteristic vector through the identification model, identifying and positioning defects in the rubber-lined pipeline, evaluating the severity of the defects, and outputting an identification result. Through the structural design of arranging a plurality of array elements of the array probe, the coverage range is increased, ultrasonic wave emission and receiving are carried out on the rubber-lined pipeline from different angles and positions, the probability of discovering small-area debonding is improved, in addition, the machine learning technology is combined, an intelligent recognition model is trained, rich features are provided, and the detection accuracy is improved. The capability of identifying different types of defects is enhanced, the sensitivity to different types of defects is improved, and the risk of missing detection of small-area debonding is reduced.

Description

Technical Field

[0001] The present invention relates to the field of nuclear power technology, specifically to the field of non-destructive testing technology, and in particular to a rubber-lined pipe defect detection method and detection system. Background Art

[0002] In the nuclear power sector, the reliability of piping systems is crucial. Rubber-lined pipes are widely used in nuclear power facilities to transport a variety of media. Due to the complex operating environment of nuclear power plants, pipelines are constantly exposed to high temperatures, high pressures, and rubber lining quality issues, making them susceptible to defects such as debonding, wear, and cracks in the rubber lining. If these defects are not discovered and addressed promptly, they can lead to pipeline leaks, compromising the safe operation of nuclear power plants. Therefore, rapid and effective defect detection for rubber-lined pipes is crucial.

[0003] Ultrasonic testing is a common method for defect detection in rubber-lined pipes. It utilizes the characteristics of ultrasonic waves that, when propagating through different media, they will reflect, refract, and scatter when encountering interfaces. When there are defects in rubber-lined pipes, such as debonding, bubbles, cracks, etc., the propagation of ultrasonic waves at these defects will be abnormal. The location, size, and nature of the defects can be determined by analyzing the ultrasonic signal. However, the rubber lining of rubber-lined pipes is located inside the pipe, and debonding usually occurs at the bonding interface between the lining and the metal pipe substrate, which cannot be directly observed from the outside of the pipe. Traditional single-point ultrasonic testing methods use the reflection, refraction, and scattering of ultrasonic waves in different media to determine the presence, location, and range of debonding. However, this method is suitable for detecting debonding over larger areas and may miss small debonding areas. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to address the existing difficulty in detecting internal defects in rubber-lined pipes, particularly debonding defects. To this end, a method for detecting defects in rubber-lined pipes is provided. An array probe, composed of multiple elements, can simultaneously transmit and receive ultrasonic waves, enabling rapid scanning and detection over large areas or over long distances. This significantly improves detection efficiency compared to point-by-point detection with a single probe. Simultaneously, machine learning techniques are combined to train an intelligent recognition model. After preprocessing and feature extraction of the data to be tested, the data is input into the trained model. Based on the learned features and classification rules, the model determines the presence of defects, as well as information such as the defect type and location. This improves the accuracy and efficiency of rubber-lined pipe defect detection and enables automatic defect identification and classification.

[0005] The technical solution adopted by the present invention to solve the technical problem is: a method for detecting defects in rubber-lined pipes, S1, ultrasonic array layout and signal transmission: arranging an ultrasonic probe array, the ultrasonic probe array including multiple array elements, the multiple array elements simultaneously transmitting ultrasonic signals to the rubber-lined pipe to obtain scanning information of a large area of ​​the rubber-lined pipe;

[0006] S2. Signal acquisition: Acquiring the ultrasonic signal emitted by the ultrasonic probe array, the echo signal reflected by the rubber-lined pipe, and the resonance signal generated by the in-phase superposition of the echo signals;

[0007] S3. Image formation: Based on the received resonance signal, a resonance signal curve and a mapping image reflecting the interior of the rubber-lined pipe are generated, wherein the mapping image includes the metal layer, the adhesive layer, and the rubber layer of the rubber-lined pipe, and the resonance signal maps the structural distribution of the metal layer, the adhesive layer, and the rubber layer;

[0008] S4. Data processing: Based on the resonance signal curve, spectrum data is acquired and half-power bandwidth is calculated, and damping characteristic parameters of the rubber-lined pipe are calculated based on the half-power bandwidth;

[0009] S5. Model Identification and Evaluation: Inputting the ultrasonic signal, echo signal, resonance signal curve, mapping image, half-power bandwidth, damping characteristic parameters, and material characteristic parameters of the rubber-lined pipe into an identification model, the identification model generating corresponding feature vectors and identifying and locating defects in the rubber-lined pipe based on the feature vectors, obtaining an identification result, and evaluating the severity of the defects;

[0010] S6. Output result: The recognition result includes the mapping image, defect type, defect location, and severity assessed by the recognition model.

[0011] Preferably, in step S1, the step of arranging the ultrasonic probe array includes: arranging the number, arrangement and spacing of array elements in the ultrasonic probe array according to the size of the rubber-lined pipe.

[0012] Preferably, the arrangement includes linear arrangement and matrix arrangement.

[0013] Preferably, the step of arranging the ultrasonic probe array further comprises: presetting the defect type of the rubber-lined pipe, and setting the ultrasonic frequency and emission angle of the ultrasonic probe array in combination with the material type of the rubber-lined pipe.

[0014] Preferably, the emission angle is adjustable.

[0015] Preferably, in step S4 data processing, the step of acquiring spectrum data and calculating half-power bandwidth based on the resonance signal curve includes:

[0016] The resonance spectrum of the resonance signal curve is obtained, and the resonance peaks and their corresponding half-power points f1 and f2, and the first-order resonance frequency fr in the resonance spectrum are obtained. Then, the half-power bandwidth b=(f2-f1) / fr.

[0017] Preferably, the damping characteristic parameters include a first attenuation coefficient, a damping ratio, a quality factor, and a loss factor, wherein the loss factor, the damping ratio, the quality factor, and the half-power bandwidth have the following relationship:

[0018] b=η=2ξ=Q -1 ;

[0019] Where η is the loss factor, ξ is the damping ratio, and Q is the quality factor.

[0020] Preferably, the first attenuation coefficient δ is calculated by the formula A(t)=αe βt and formula Calculate, where A is the amplitude of the ultrasonic signal, α is the initial amplitude, t is the propagation distance, β is the second attenuation coefficient of the ultrasonic signal, and c is the sound velocity of the ultrasonic signal in the rubber layer.

[0021] Preferably, before step S5, the method further comprises: performing a preliminary prediction on the internal structure of the rubber-lined pipe according to the resonance signal curve and the mapping image, and judging whether the adhesive layer is debonded from the metal layer and / or rubber layer.

[0022] A rubber-lined pipe defect detection system includes an ultrasonic probe and an identification system. The ultrasonic probe is installed on the rubber-lined pipe, and the ultrasonic probe includes multiple array elements. The array elements are arranged linearly or in a matrix on the rubber-lined pipe. The array elements emit ultrasonic signals. The identification system receives the ultrasonic signals and the echo signals and resonance signals generated by the ultrasonic signals passing through the metal layer, adhesive layer and rubber layer of the rubber-lined pipe. The identification system processes and calculates the ultrasonic signals, echo signals and resonance signals to generate resonance signal curves and mapping images, and classifies and identifies the defect types of the rubber-lined pipe and assesses the severity of the defects.

[0023] The implementation of the present invention has the following beneficial effects:

[0024] 1. The rubber-lined pipe defect detection method and detection system provided by the present invention, through the structural design of the layout of multiple array elements of the array probe, can simultaneously transmit and receive ultrasonic waves, realize multi-channel simultaneous detection, increase the coverage range, and enable rapid scanning detection of large areas or long distances, thereby improving detection efficiency. In addition, ultrasonic waves can be transmitted and received from different angles and positions on the rubber-lined pipe, increasing the probability of detecting small-area debonding.

[0025] 2. Based on the collected ultrasonic signals, echo signals and resonance signals, rich ultrasonic data can be obtained. Combined with signal processing and imaging algorithms, high-quality mapping images reflecting the interior of the rubber-lined pipe can be generated, intuitively displaying the internal structure and defect information of the inspected object, making it easier for inspectors to identify and analyze defects more clearly.

[0026] 3. Combined with machine learning technology, intelligent recognition models are trained to provide rich features, thereby enhancing the ability to identify different types of defects, increasing sensitivity to different types of defects, and reducing the risk of missed detection of small-area debonding. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solution of the present invention, the present invention will be further described below with reference to the accompanying drawings and embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be considered as limiting the scope. A person of ordinary skill in the art can derive other relevant drawings based on these drawings without inventive effort. In the drawings:

[0028] Figure 1 Schematic diagram of a rubber-lined pipe defect detection method according to the present invention;

[0029] Figure 2 This is a signal propagation diagram of a rubber-lined pipe defect detection method according to the present invention;

[0030] Figure 3 It is a schematic diagram of a frequency sweep curve of the present invention;

[0031] Figure 4-5 are the resonance peaks corresponding to the range of 4.5 to 6 MHz obtained by the metal plate and the rubber adhesive plate of the present invention;

[0032] Figure 6-7 It is the time domain signal of the metal plate and rubber adhesive plate corresponding to the resonance frequency of 4.70MHz;

[0033] Figure 8 is the half-power bandwidth diagram corresponding to the resonance peak;

[0034] Figure 9 It is a structural schematic diagram of a rubber-lined pipeline defect detection system of the present invention. DETAILED DESCRIPTION

[0035] In order to provide a clearer understanding of the technical features, objectives, and effects of the present invention, specific embodiments of the present invention are now described in detail with reference to the accompanying drawings. In the following description, it should be understood that the directions or positional relationships indicated by "upper," "lower," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., are based on the directions or positional relationships shown in the accompanying drawings and are constructed and operated in specific directions. These directions are merely for the purpose of facilitating the description of the present technical solution and do not necessarily require the devices or components referred to to have specific directions. Therefore, they should not be construed as limitations on the present invention.

[0036] It should also be noted that, unless otherwise clearly specified and limited, terms such as "installed", "connected", "connected", "fixed", and "set" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integrated connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the internal connection of two elements or the interaction relationship between two elements. When an element is referred to as being "on" or "under" another element, the element can be "directly" or "indirectly" located on the other element, or there may be one or more intervening elements. The terms "first", "second", etc. are only for the convenience of describing the present technical solution, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first", "second", etc. may explicitly or implicitly include one or more of the features. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to the specific circumstances.

[0037] Example 1:

[0038] Ultrasonic testing is a common method for defect detection in rubber-lined pipes. It exploits the fact that when ultrasonic waves propagate through different media, they are reflected, refracted, and scattered at interfaces. When defects exist in rubber-lined pipes, such as debonding, bubbles, or cracks, ultrasonic waves propagate abnormally at these locations. The location, size, and nature of the defects can be determined by analyzing the ultrasonic signal.

[0039] like Figure 1 As shown, the rubber-lined pipe defect detection method provided in this embodiment includes the following steps:

[0040] S1. Ultrasonic array layout and signal transmission: Layout the ultrasonic probe array, which contains multiple array elements. Multiple array elements simultaneously transmit ultrasonic signals to the rubber-lined pipe.

[0041] S2. Signal acquisition: Acquire the ultrasonic signal emitted by the ultrasonic probe array, the echo signal reflected by the rubber-lined pipe, and the resonance signal generated by the in-phase superposition of the echo signal.

[0042] S3. Image formation: Based on the received resonance signal, a resonance signal curve and a mapping image reflecting the interior of the rubber-lined pipe are generated.

[0043] S4. Data processing: Based on the resonance signal curve, obtain the spectrum data and calculate the half-power bandwidth, and calculate the damping characteristic parameters of the rubber-lined pipe based on the half-power bandwidth.

[0044] S5. Model identification and evaluation: The ultrasonic signal, echo signal, resonance signal curve, mapping image, half-power bandwidth, and damping characteristic parameters are input into the identification model. The identification model generates the corresponding feature vector and identifies and locates the defects in the rubber-lined pipe based on the feature vector. The identification result is obtained and the severity of the defect is evaluated.

[0045] S6. Output results: The recognition results include the mapping image, defect type, defect location, and severity assessed by the recognition model.

[0046] It should be noted that in step S1, an array ultrasonic probe is used to scan along the circumference or axial direction of the pipe to achieve multi-channel simultaneous detection. Through the structural design of the array probe, such as the number, arrangement, size, and frequency of the array elements, it can adapt to the detection requirements of rubber-lined pipes with different materials, pipe diameters, wall thicknesses, rubber lining materials, and thicknesses. Specialized probes designed for different pipe diameters and rubber lining thicknesses can achieve rapid scanning and detection of rubber-lined pipes with larger areas or longer distances. At the same time, compared with conventional probes, it can transmit and receive ultrasonic waves on rubber-lined pipes from different angles and positions, increasing the detection coverage and the probability of detecting small-area debonding.

[0047] like Figure 2 As shown, the rubber-lined pipe 100 includes a metal layer 2, an adhesive layer 3 and a rubber layer 4. An ultrasonic probe array 1 is provided on the outer wall of the metal layer. The ultrasonic probe array 1 includes a plurality of array elements 5. The array elements 5 are arranged linearly and the spacing between the array elements 5 is equal. The array elements 5 are excited simultaneously to generate a long pulse ultrasonic signal 7 inside the metal layer. The ultrasonic signal 7 passes through the metal layer 2 and the rubber layer 4 to generate an echo signal 8, an echo signal 12 and a resonance signal 10 and a resonance signal 13 generated by the resonance phenomenon.

[0048] In other possible implementations, the emission and receiving time of each array element can be controlled to achieve flexible focusing and deflection of the sound beam, which can more accurately locate defects and improve the measurement accuracy of defect position and size. At the same time, the influence of human factors on the detection results is reduced, making the detection process more stable and the detection results highly repeatable, which is conducive to multiple detections and comparative analysis of the rubber-lined pipe 100.

[0049] Furthermore, the ultrasonic probe array 1 is composed of multiple small-sized array elements 5, such as small-sized chips, and the array elements 5 are arranged linearly or in a matrix. The array elements 5 in multiple ultrasonic probe arrays 1 are simultaneously excited to perform large-area scanning of the rubber-lined pipe, which can achieve rapid scanning and detection of a larger area or a longer distance. Compared with the point-by-point detection of a single probe, the detection efficiency is greatly improved.

[0050] It should be noted that, based on the diameter, wall thickness, rubber lining thickness and other parameters of the rubber-lined pipe 100, an optimization algorithm such as a genetic algorithm is used to optimize the number, arrangement, spacing, excitation frequency and other parameters of the ultrasonic probe array 1, and determine the optimal probe layout scheme. By optimizing the layout and parameters of the ultrasonic probe array 1, ultrasonic information of the rubber-lined pipe 100 can be obtained more comprehensively and accurately. For example, Figure 1 The ultrasonic probe array 1 shown is arranged linearly and is provided with at least 12 array elements 5. The 12 array elements are arranged at equal intervals along the axial direction of the pipeline. In this way, continuous detection can be performed in the longitudinal direction of the pipeline, which is convenient for discovering axially distributed defects. The detection results of adjacent probes can be compared and analyzed to improve the accuracy of detection.

[0051] It is understood that for small-diameter rubber-lined pipes 100, a compact ultrasonic probe array 1 can be used to improve detection resolution. For example, for small pipes (under DN100), the array elements 5 can be arranged in a linear arrangement, with the ultrasonic probe array 1 detecting in a circumferential direction. Furthermore, because small-sized pipes generally have thinner walls and higher resonant frequencies, a higher excitation frequency is required to transmit ultrasonic signals.

[0052] It is understood that for large-diameter rubber-lined pipes 100, a sparse ultrasonic probe array 1 layout is used to expand the detection range. For example, for large-sized pipes, the array elements 5 can be arranged in a matrix, the ultrasonic probe array 1 detection direction is set to the axial direction, and a lower excitation frequency is set.

[0053] It should be noted that the detection direction of the ultrasonic probe array 1 can be set according to the detection location and requirements. The axial layout is used to detect the rubber lining condition of the rubber-lined pipe 100 in the axial direction, and is used to detect debonding defects distributed along the length of the rubber-lined pipe 100, such as axial debonding at the pipe connection near the weld of the rubber-lined pipe. The circumferential layout is used to detect the rubber lining condition of the rubber-lined pipe in the circumferential direction, and is used to detect debonding conditions on the entire circumference.

[0054] Furthermore, this embodiment can also preset the defect type of the rubber-lined pipe 100 and set the ultrasonic frequency and emission angle of the ultrasonic probe array 1 based on the material type and detection location of the rubber-lined pipe 100 to enhance the detection capability of different defect types. Specifically, based on user experience and prior knowledge, combined with the area where the rubber-lined pipe 100 is located, the defect type can be estimated in advance. For example, if the defect is prone to rubber lining damage and substrate corrosion, the defect is predicted to be a pitting corrosion pit, and the probe is set to emit ultrasonic signals at an oblique angle to increase echo intensity. If the defect type is primarily rubber lining blistering, vertical emission is primarily used.

[0055] Furthermore, the ultrasonic probe array 1 can adjust the emission angle, adjusting the sound wave emission angle to improve the defect detection rate and detection resolution. Specifically, the emission angle can be set based on parameters such as the characteristics of the rubber-lined pipe 100 itself, the predicted defect type, and the set ultrasonic frequency.

[0056] It should be understood that the emission angle can be adjusted based on the material type, shape, and size of the rubber-lined pipe 100. For example, for small-sized pipes, a smaller emission angle may be required to better detect internal defects, allowing the ultrasonic wave to be fully reflected and refracted when propagating within the rubber-lined pipe 100, thereby detecting defects. For large-sized pipes, the emission angle may need to be appropriately increased to allow the ultrasonic wave to penetrate deeper into the rubber-lined pipe 100 for detection. For complex and difficult-to-reach areas, the array layout and number of ultrasonic probe arrays 1 can be combined. For example, if the surface structure is relatively small, the number of array elements 5 can be reduced to accommodate the surface of the structure. If the array elements 5 cannot reach the surface of the structure, such as curved surfaces or corners, oblique sound wave detection can be used.

[0057] Furthermore, different defects require detection at different depths inside the rubber-lined pipe. In this embodiment, the detection position of the rubber-lined pipe and the estimated defect type need to be considered when adjusting the emission angle. For example, if there is internal debonding, to detect defects at a deeper position inside the rubber-lined pipe 100, the emission angle needs to be adjusted according to the specific situation to ensure that the ultrasonic wave can reach the target depth and effectively reflect the echo.

[0058] Specifically, the ultrasonic probe array 1 can be mounted on a rotatable or tiltable bracket, and the angle between the ultrasonic probe array 1 and the surface of the rubber-lined pipe 100 can be changed by rotating or tilting the bracket, thereby adjusting the incident angle.

[0059] In other possible implementations, the time delay of each array element 5 in transmitting ultrasonic waves may be controlled to change the direction of the synthesized beam, thereby adjusting the incident angle.

[0060] Furthermore, because ultrasonic waves of different frequencies have different propagation speeds and attenuation characteristics in different materials, this embodiment can also calculate a suitable emission angle based on the laws of ultrasonic refraction and reflection, combined with parameters such as the sound velocity of the rubber-lined pipe 100 material, to ensure that the ultrasonic wave propagates along the desired path in the rubber-lined pipe 100 and achieve the best detection effect.

[0061] In a specific implementation, the emission angle can be calculated according to Snell's Law. Specifically, the ultrasonic wave for detecting the rubber-lined pipe 100 passes through two media, air and rubber-lined pipe 100, and the refraction and reflection laws of the ultrasonic wave conform to the formula: c1sinθ1=c2sinθ2, where c1 is the sound velocity of the ultrasonic wave in the incident medium (such as air), c2 is the sound velocity of the ultrasonic wave in the rubber-lined pipe material, θ1 is the incident angle, that is, the emission angle θ2 is the refraction angle. In application, the sound velocity of the ultrasonic wave in the air and the rubber-lined pipe 100 material, as well as the desired refraction angle (for example, in order to make the ultrasonic wave propagate specifically in the pipe to achieve the best detection effect) are usually known, and the incident angle of the probe can be calculated by the above formula.

[0062] This embodiment can flexibly adjust the arrangement and combination of the array elements 5 according to the shape of the rubber-lined pipe 100 and the inspection requirements, adapting to defect inspection in various complex locations, such as curved surfaces and corners. Furthermore, it has strong adaptability to the inspection environment and can perform inspections in areas with limited space or that are difficult to access.

[0063] It should be noted that in step S2, the ultrasonic signals emitted by the ultrasonic probe array 1, the echo signals reflected by the rubber-lined pipe 100, and the resonance signals generated by the in-phase superposition of the echo signals are collected. The collected data is then subjected to preprocessing operations such as denoising, filtering, and normalization. The echo signals are formed when the ultrasonic waves emitted by the array element 5 are reflected from different media interfaces (such as the interface between the rubber layer and the metal pipe wall, or defect interfaces) in the rubber-lined pipe 100. When the transmitted ultrasonic pulse propagates within the rubber-lined pipe 100 and encounters these interfaces, part of the sound wave is reflected back and received by the probe, converted into an electrical signal, i.e., the echo signal. The generation of the resonance signal depends on the echo signals. When the phases of multiple echoes are consistent, in-phase superposition occurs, resulting in resonance, which in turn generates the resonance signal. The echo signals directly reflect the location and reflection of different interfaces within the pipeline, allowing for a preliminary determination of the presence and approximate location of defects. The resonance signal focuses more on reflecting the overall acoustic and damping characteristics of the pipeline. By analyzing parameters such as the frequency, amplitude, and half-power bandwidth of the resonance signal, we can further understand the structural health of the rubber-lined pipeline, such as the bonding quality and thickness variation of the rubber layer. Combining the two can more comprehensively and accurately detect defects in rubber-lined pipelines.

[0064] like Figure 2As shown, after the signals collected in step S2 are processed, each element 5 in the ultrasonic probe array 1 forms a one-to-one corresponding resonance signal curve. By analyzing the characteristic changes in the resonance signal, the attenuation of the sound waves in the rubber-lined pipe 100 can be inferred, and the internal structure and defect conditions of the rubber-lined pipe 100 can be understood. For example, if the resonance signal amplitude decreases and the half-power bandwidth becomes wider, it may mean that the rubber-lined pipe 100 has a defect, resulting in increased sound wave attenuation.

[0065] It should be noted that the ultrasonic probe array 1 of this embodiment generates a corresponding resonance signal curve graph for each array element 5 arranged therein, and each array element 5 generates a corresponding resonance signal curve graph. Specifically, Figure 2 As shown, the array elements 5 are all array elements of the ultrasonic probe array 1. For further explanation, two array elements 5 are selected for illustration. In the figure, the metal layer 2 and the rubber layer 4 of the rubber-lined pipe 100 below the array element 5 on the left side are not debonded, that is, when the adhesive layer 3 is not damaged, most of the ultrasonic signals in the metal layer 2 enter the rubber layer 4 through the adhesive layer 3 to generate a projected ultrasonic signal 9, and a small amount of the ultrasonic signal 9 passes through the boundary between the metal layer 2 and the adhesive layer 3 to generate a reflected echo signal 8. Therefore, the resonance signal 10 received by the array element 5 at this position has low energy and high attenuation. The resonance signal curve is shown in the resonance signal 10 in the figure, which means that the damping at the position detected by the array element 5 is large and the resonance effect is weak. The signal fluctuation amplitude of the generated resonance signal curve is small, the curve is relatively flat, the signal fluctuation change is not drastic, and there is no obvious large-scale oscillation. On the right side, debonding occurs between the metal layer 2 and the rubber layer 4 of the rubber-lined pipe below an array element 5 at position 11, that is, when the adhesive layer 3 is damaged, most of the ultrasonic signals in the metal layer 2 generate echo signals 12 at the boundary between the metal layer 2 and the adhesive layer 3. Therefore, the resonance signal 13 with strong energy and low attenuation can be received at the position of the array element 5. The generated resonance signal 13 has a large fluctuation amplitude and the curve shows a clear "envelope" shape. The middle part oscillates violently, the signal intensity is high, and it gradually decays towards both sides.

[0066] It should be noted that Figure 2 The figure only shows the resonance signal graphs corresponding to two of the array elements 5; the resonance signal graphs corresponding to the other array elements 5 are not shown. Furthermore, in other possible implementations, the resonance signals of the array elements 5 may form one or more continuous resonance graphs based on their positional relationships. For example, the resonance signals of all array elements 5 arranged linearly may form one continuous resonance signal graph, while array elements 5 in a matrix array may generate multiple continuous resonance signal graphs based on the array layout.

[0067] Furthermore, based on the collected ultrasonic signals, echo signals and resonance signals, rich ultrasonic data can be obtained. Combined with signal processing and imaging algorithms, high-quality two-dimensional or three-dimensional images reflecting the interior of the rubber-lined pipe 100 can be generated, intuitively displaying the internal structure and defect information of the inspected object, making it easier for inspectors to identify and analyze defects more clearly. Figure 2 As shown in FIG, the ultrasonic signals, echo signals and resonance signals collected by each array element 5 generate corresponding hierarchical structures according to the reflection, refraction and scattering characteristics of the sound waves by different materials in the metal layer 2, the adhesive layer 3 and the rubber layer 4. In the mapping image, it can be intuitively seen whether the adhesive layer 3 has debonded, as shown in FIG. Figure 2 As shown, the debonding location 11 of the adhesive layer 3 can be directly observed based on the mapping image. Specifically, the ultrasonic signal, echo signal, and resonance signal information can provide internal information of the rubber-lined pipe 100 from multiple angles, achieving information complementarity, providing a more comprehensive understanding of the internal conditions of the rubber-lined pipe 100, and providing richer features, thereby enhancing the ability to identify different types of defects and increasing sensitivity to different types of defects.

[0068] More specifically, each array element 5 forms a corresponding segmented image based on its collected signal mapping. Based on the positional relationship, each array element 5 continuously forms a mapping image reflecting the overall structure of the rubber-lined pipe 100 at the detection location. As the ultrasonic probe array 1 moves, a mapping image of the corresponding rubber-lined pipe 100 is generated.

[0069] It should be noted that the mapping image can be a two-dimensional image or a three-dimensional image. The mapping image can be continuously displayed on the screen, mapping the structural distribution of the metal layer 2, the adhesive layer 3, and the rubber layer 4, converting abstract signals into intuitive images, so that the user can observe the internal structure of the rubber-lined pipe 100 and identify internal defects and deformations.

[0070] It should be noted that debonding is a defect that is difficult to detect in the rubber-lined pipe 100. The existing technology is suitable for detecting large-area debonding, but it is easy to miss small-area debonding or shallow debonding, and the detection result has low accuracy.

[0071] This embodiment also incorporates a machine recognition model to improve detection accuracy. The machine recognition model calculates the location, shape, and size of defects and assesses the defect condition. On one hand, the machine learning model automatically extracts complex features from the ultrasonic image data, generates a mapping image based on the layout of the array elements 5, and uses this mapping image to identify debonding defect features that are difficult for the human eye to detect, such as specific textures and grayscale variations, thereby more accurately determining the presence of small-area debonding. On the other hand, by training the machine learning model with a large amount of annotated data, the recognition model can learn the manifestation patterns of different types of debonding defects. As the amount of data increases and training deepens, the model can continuously optimize its ability to identify small-area debonding, reducing the probability of false positives and missed detections.

[0072] Specifically, after the signal collected in step S2 undergoes imaging processing in step S3, it undergoes preliminary data processing in step S4. Based on the resonance signal curve, spectrum data is acquired and the half-power bandwidth is calculated. The damping characteristic parameters of the rubber-lined pipe are then calculated based on the half-power bandwidth. Specifically, the damping characteristic parameters include a first attenuation coefficient, a damping ratio, a quality factor, and a loss factor, wherein the first attenuation coefficient reflects the attenuation coefficient of the rubber-lined pipe.

[0073] More specifically, by obtaining the resonance spectrum of the resonance signal curve, the resonance peak in the resonance spectrum and its corresponding half-power points f1 and f2, and the first-order resonance frequency f r , then the half-power bandwidth b=(f2-f1) / f r The damping characteristic parameters include the first attenuation coefficient δ, the damping ratio ξ, the quality factor Q and the loss factor η, where b = η = 2ξ = Q -1 The first attenuation coefficient δ is given by the formula A(t)=αe βt , where A is the amplitude of the ultrasonic signal, α is the initial amplitude, t is the propagation distance, and β is the second attenuation coefficient of the ultrasonic signal. Then the first attenuation coefficient c is the sound velocity of the ultrasonic signal in the rubber layer.

[0074] Furthermore, the ultrasonic signal, echo signal, resonance signal curve, mapping image, half-power bandwidth, damping characteristic parameters and material characteristic parameters of the rubber-lined pipe are input into the recognition model as data sources, and the ultrasonic signal, echo signal, resonance signal curve, mapping image, half-power bandwidth, damping characteristic parameters and material characteristic parameters form multimodal data, wherein the ultrasonic signal, echo signal, resonance signal curve, mapping image, half-power bandwidth, and damping characteristic parameters are data generated or processed based on the ultrasonic signal, and the characteristics of each data are extracted respectively. The material characteristic parameters are the characteristic parameters of the rubber-lined pipe 100 itself, including the material, thickness, size and heat treatment status of the rubber-lined pipe.

[0075] Furthermore, feature extraction is performed on various modal data to generate corresponding feature vectors. Defects in rubber-lined pipes are identified and located based on the feature vectors, resulting in identification results and an assessment of the severity of the defects. For example, time domain features (such as peak amplitude, time delay, and waveform energy) and frequency domain features (such as main frequency, harmonic distribution, and attenuation coefficient) are extracted from ultrasonic signals, echo signals, and resonance signals. Feature fusion is then performed to create a structured feature vector. The extracted features are then classified or regressed through pooling layers and fully connected layers to identify and analyze the defect type, location, size, etc., and the severity of the defect is assessed.

[0076] More specifically, the recognition model in this embodiment is based on prior knowledge. It is trained using a large amount of prior data and contains a wealth of data on defect size, defect type, and material. By comparing image features, rubber lining defects will show distinct, strong resonance peaks. By standardizing the defect contour scale and comparing contour features in the database, defect identification can be achieved. This is combined with other raw data, and refined recognition techniques, such as comparing signal attenuation and half-wavewidth, accurately identify and locate defects in pipelines. Evaluation criteria are set based on the recognition results. For example, if the rubber lining is identified as bubbling or partially damaged, but the base metal material is not corroded, the evaluation result is mild. If the rubber lining is identified as damaged and the base metal is potentially corroded, the evaluation result is moderate, requiring user attention. If the recognition results clearly detect pitting, the condition is assessed as severe.

[0077] It should be noted that before feature extraction, the ultrasonic echo signal and the resonance signal are respectively subjected to pre-processing operations such as amplification and filtering, and noise removal to improve signal quality.

[0078] It should be noted that before model identification, this embodiment can perform a preliminary prediction of the internal structure of the rubber-lined pipe 100 based on the resonance signal curve and the mapping image, either individually or in combination, to determine whether the adhesive layer is debonded from the metal layer and / or rubber layer. This preliminary prediction can reveal potential patterns or trends in the ultrasonic, resonance, and echo signals, providing guidance for further feature engineering and model optimization.

[0079] Specifically, based on the preliminary prediction results, the recognition model can more easily capture defect-related information during feature extraction, reduce interference from noise and irrelevant information, enhance defect-related features, and strengthen the model's ability to capture key information. In other implementations, the preliminary prediction results can be used as auxiliary input features, forming a multidimensional feature matrix with the original data, improving the clarity of the recognition model's decision boundaries.

[0080] It should be noted that the output results of the recognition model include the mapping image, defect type, defect location and severity assessed by the recognition model, and the recognition results are visualized so that users can intuitively understand the detection results.

[0081] Example 2:

[0082] This embodiment is based on the first embodiment. The ultrasonic probe array 1 includes a transmitter, a receiver, and a signal processor. The transmitter includes multiple array elements 5, which form an array layout along the surface of the rubber-lined pipe 100 and transmit ultrasonic signals. The receiver receives echo signals and resonance signals returned by the metal layer 2, adhesive layer 3, and rubber layer 4 of the rubber-lined pipe 100. The signal processor generates a corresponding resonance signal curve and mapping image based on the received echo signals and resonance signals. Based on the resonance signal curve, the signal processor obtains spectrum data and calculates the half-power bandwidth. The damping characteristic parameters of the rubber-lined pipe are calculated based on the half-power bandwidth. The initial defect prediction of the rubber-lined pipe 100 is also performed based on the resonance signal curve, mapping image, and calculated half-power bandwidth and damping characteristic parameters.

[0083] The signal processor also includes a recognition model that inputs the ultrasonic signal, echo signal, resonance signal curve, mapping image, half-power bandwidth, damping characteristic parameters, and material characteristic parameters of the rubber-lined pipe as data sources. The recognition model generates corresponding feature vectors and uses these feature vectors to identify and locate defects in the rubber-lined pipe, obtaining recognition results and assessing the severity of the defects. The recognition model integrates multiple parameters for analysis and automatically extracts complex relationships between features, achieving more accurate and intelligent debonding defect identification and reducing the errors and uncertainties associated with human judgment.

[0084] It should be noted that the recognition model includes a built-in defect recognition method based on rich test data and trained based on data such as defect size, defect type, and material. By preparing rubber-lined pipe samples with different diameters and prefabricated defect sizes, signals of different prefabricated defects are collected on the sample pipes to form a defect signal comparison database.

[0085] It should be noted that rubber layers of varying thickness and material have varying degrees of influence on damping. Therefore, during training, the damage to the rubber layer can be assessed by measuring the plate's damping characteristics. The characteristic value of the mechanical structural system damping coefficient of the rubber-lined pipe 100 can be measured using ultrasonic methods.

[0086] Specifically, the half-power bandwidth b can be calculated by the resonance method, and the calculation process includes the following steps:

[0087] 1. Convert the time domain signal of the ultrasonic reflected wave into the frequency domain power spectrum through FFT (Fast Fourier Transform);

[0088] 2. Use a Butterworth bandpass filter or wavelet transform to remove high-frequency noise and eliminate DC bias or baseline drift introduced by the signal acquisition system;

[0089] 3. Use the local maximum method, derivative method or polynomial fitting (such as Savitzky-Golay) to locate the center frequency of the resonance peak;

[0090] 4. If multiple peaks overlap, decomposition should be performed based on the symmetry of the peak shape (such as Lorentz distribution or Gaussian distribution);

[0091] 5. For each resonance peak’s power spectrum curve, take the maximum power value of the peak (normalized to 0 dB) and set the threshold to the peak power minus 3 dB;

[0092] 6. Find the first frequency points on the left and right sides of the peak whose power is equal to or lower than the threshold, and record them as f1 and f2.

[0093] Furthermore, the resonance spectrum is obtained by frequency scanning, and the half-power points f1 and f2 near the resonance peak (i.e., the point where the amplitude is 0.707 times the peak value) are taken, and the first-order resonance frequency f is taken. r , then the half-power bandwidth b=(f2-f1) / f r Specifically, a frequency domain scanning curve is obtained through a frequency sweep curve: a vertically incident ultrasonic probe is placed on one side of the metal plate, and the ultrasonic pulse duration is set to 10μs. At this time, the bottom surface reflection echoes overlap with each other. When the phases are consistent, the echoes are superimposed in the same phase, which greatly increases the signal amplitude. In other words, resonance occurs. The spectrum signal reflecting the resonance phenomenon can be obtained through frequency domain scanning.

[0094] For example, a rubber-lined pipe 100 with a damaged rubber layer 4 is used as a training sample for training the recognition model. The damaged rubber-lined pipe 100 has an exposed metal layer 2 and an intact rubber layer 4. During training, a 5MHz probe is first placed on the exposed metal plate portion, and the frequency domain scanning range is set to 1 to 9MHz. The following is obtained: Figure 3 The frequency sweep curve is shown in the figure. The resonance peaks of each order can be clearly observed in the figure. The distance between two adjacent resonance peaks represents the first-order resonance frequency. In order to avoid background noise interference and obtain a relatively pure resonance peak signal, the 4.5-6MHz range is intercepted for detailed frequency domain scanning for subsequent analysis. Figure 3 It is shown that three resonance peaks can be obtained in the range of 4.5 to 6 MHz and the time domain signal at the resonance frequency of 4.70 MHz is selected. The half-power bandwidth corresponding to the three resonance peaks of bare metal and rubber adhesive plate (not debonded) in this range is calculated by identification model calculation. Then, the resonance peaks of the defective part (bare metal plate) and the intact part (rubber adhesive plate) of the rubber-lined pipe sample are analyzed and compared. Figure 4 and Figure 5 As shown in FIG. 1 , the resonance peaks corresponding to the range of 4.5 to 6 MHz are obtained for the metal layer 2 (bare metal plate) and the rubber layer 4 (rubber adhesive plate). Figure 6 and Figure 7 These are the time domain signals of metal layer 2 (bare metal plate) and rubber layer 4 (rubber adhesive plate) at a resonance frequency of 4.70 MHz. Figure 8 is the half-power bandwidth measured. Figure 4 and Figure 5 It can be seen that the resonance peak of the rubber adhesive plate is significantly lower than that of the bare metal plate, and the resonance peak is wider. Figure 6 and Figure 7 It can be seen that the time domain signal amplitude of the rubber adhesive plate is significantly reduced compared to the bare metal plate, and the attenuation is more obvious. Figure 8 It can be seen that the half-power bandwidth of the bare metal plate is generally lower than that of the rubber-bonded plate. The recognition model can identify the debonding defects on the rubber-lined pipe samples by learning the patterns and regularities of the resonance peaks and half-power bandwidths.

[0095] Furthermore, the recognition model extracts time-domain features (such as peak amplitude, time delay, and waveform energy) and frequency-domain features (such as dominant frequency, harmonic distribution, and attenuation coefficient) from the ultrasonic resonance reflection signal. A structured feature vector is created by combining the material properties of the rubber-lined pipe 100 (such as alloy type and heat treatment status) with defect parameters (size and type). Local features (such as defect-induced distortion of the reflected pulse waveform) are then automatically extracted to capture the temporal dependencies of the ultrasonic resonance signal (such as the periodic reflection characteristics of the fault echo). The features output by the CNN and LSTM are combined to perform classification (defect type) and regression (defect size and severity).

[0096] Example 3:

[0097] This embodiment is different from the first and second embodiments. Figure 9 As shown, a rubber-lined pipe defect detection system 700 is provided, including an ultrasonic probe 72 and an identification system 73, wherein the ultrasonic probe 72 is used on the rubber-lined pipe 71. Specifically, the rubber-lined pipe 71 includes a metal layer, an adhesive layer and a rubber layer. The ultrasonic probe 72 includes a plurality of array elements 702, and the array elements 702 are arranged linearly or in a matrix on the metal layer. The array elements 702 emit ultrasonic signals. The identification system 73 is electrically or wirelessly connected to the ultrasonic probe 72. The identification system 73 receives the ultrasonic signal and the echo signal and resonance signal generated by the ultrasonic signal passing through the metal layer, the adhesive layer and the rubber layer of the rubber-lined pipe 71. The identification system 73 processes and calculates the ultrasonic signal, the echo signal and the resonance signal to generate a resonance signal curve and a mapping image, and classifies and identifies the defect type of the rubber-lined pipe 71 and evaluates the severity of the defect.

[0098] In this embodiment, the recognition system 73 can be integrated into the ultrasonic probe 72 or independently provided. The recognition system 73 can be provided with a display unit to display the recognition results, including the mapping image, defect type, defect location, and severity assessed by the recognition model.

[0099] Furthermore, identification system 73 incorporates a machine learning-based defect recognition algorithm. Based on extensive test data, this algorithm trains built-in defect recognition methods based on defect size, type, and material. This allows it to accurately identify and locate defects in pipelines and assess their severity during actual inspections. This defect recognition algorithm is currently available and will not be further described here.

[0100] Specifically, the recognition system 73 uses ultrasonic signals, echo signals, resonance signals, resonance signal curves, and mapping images as data sources to form multimodal data to provide rich rubber-lined pipe information, capture comprehensive data features, make recognition results more accurate, and enhance the generalization ability and robustness of the recognition system.

[0101] It can be understood that the above embodiments only express the preferred implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the patent scope of the present invention. It should be pointed out that for ordinary technicians in this field, without departing from the concept of the present invention, the above technical features can be freely combined, and several deformations and improvements can be made, all of which fall within the scope of protection of the present invention. Therefore, all equivalent changes and modifications made to the scope of the claims of the present invention should fall within the scope of coverage of the claims of the present invention.

Claims

1. A method for detecting defects in rubber-lined pipes, characterized in that: The following steps are involved: S1. Ultrasonic array layout and signal transmission: Layout an ultrasonic probe array, wherein the ultrasonic probe array includes multiple array elements, and the multiple array elements simultaneously transmit ultrasonic signals to the rubber-lined pipe to obtain scanning information of a large area of ​​the rubber-lined pipe; S2. Signal acquisition: Acquiring the ultrasonic signal emitted by the ultrasonic probe array, the echo signal reflected by the rubber-lined pipe, and the resonance signal generated by the in-phase superposition of the echo signals; S3. Image formation: Based on the received resonance signal, a resonance signal curve and a mapping image reflecting the interior of the rubber-lined pipe are generated, wherein the mapping image includes the metal layer, the adhesive layer, and the rubber layer of the rubber-lined pipe, and the resonance signal maps the structural distribution of the metal layer, the adhesive layer, and the rubber layer; S4. Data processing: Based on the resonance signal curve, spectrum data is acquired and half-power bandwidth is calculated, and damping characteristic parameters of the rubber-lined pipe are calculated based on the half-power bandwidth; S5. Model Identification and Evaluation: Inputting the ultrasonic signal, echo signal, resonance signal curve, mapping image, half-power bandwidth, damping characteristic parameters, and material characteristic parameters of the rubber-lined pipe into an identification model, the identification model generating corresponding feature vectors and identifying and locating defects in the rubber-lined pipe based on the feature vectors, obtaining an identification result, and evaluating the severity of the defects; S6. Output result: The recognition result includes the mapping image, defect type, defect location, and severity assessed by the recognition model.

2. The method for detecting defects in rubber-lined pipes according to claim 1, characterized in that: In step S1, the step of laying out the ultrasonic probe array includes: The number, arrangement and spacing of the array elements in the ultrasonic probe array are arranged according to the size of the rubber-lined pipe.

3. The method for detecting defects in rubber-lined pipes according to claim 2, characterized in that: The arrangement includes linear arrangement and matrix arrangement.

4. The method for detecting defects in rubber-lined pipes according to claim 2, characterized in that: The step of arranging the ultrasonic probe array further includes: The defect type of the rubber-lined pipe is preset, and the ultrasonic frequency and emission angle of the ultrasonic probe array are set in combination with the material type of the rubber-lined pipe.

5. The method for detecting defects in rubber-lined pipes according to claim 4, characterized in that: The emission angle is adjustable.

6. The method for detecting defects in rubber-lined pipes according to claim 1, characterized in that: In step S4 data processing, the step of acquiring spectrum data and calculating half-power bandwidth based on the resonance signal curve includes: Obtain the resonance spectrum of the resonance signal curve, obtain the resonance peak in the resonance spectrum and its corresponding half-power points f1 and f2, and the first-order resonance frequency f r , then the half-power bandwidth b=(f2-f1) / f r .

7. The method for detecting defects in rubber-lined pipes according to claim 6, characterized in that: The damping characteristic parameters include a first attenuation coefficient, a damping ratio, a quality factor, and a loss factor, wherein the loss factor, the damping ratio, the quality factor, and the half-power bandwidth have the following relationship: b=η=2ξ=Q -1 ; Where η is the loss factor, ξ is the damping ratio, and Q is the quality factor.

8. The method for detecting defects in rubber-lined pipes according to claim 7, characterized in that: The first attenuation coefficient is δ by the formula A(t) = αe βt and formula Calculate, where A is the amplitude of the ultrasonic signal, α is the initial amplitude, t is the propagation distance, β is the second attenuation coefficient of the ultrasonic signal, and c is the sound velocity of the ultrasonic signal in the rubber layer.

9. The method for detecting defects in rubber-lined pipes according to claim 1, characterized in that: Before step S5, the method further includes: A preliminary prediction is made on the internal structure of the rubber-lined pipe based on the resonance signal curve and the mapping image to determine whether the adhesive layer is debonded from the metal layer and / or rubber layer.

10. A rubber-lined pipe inspection system, applied to the rubber-lined pipe defect detection method according to any one of claims 1 to 9, characterized in that: The invention comprises an ultrasonic probe and an identification system. The ultrasonic probe is arranged on the rubber-lined pipe. The ultrasonic probe comprises a plurality of array elements. The array elements are arranged linearly or in a matrix on the rubber-lined pipe. The array elements transmit ultrasonic signals. The identification system receives the ultrasonic signals and the echo signals and resonance signals generated by the ultrasonic signals passing through the metal layer, adhesive layer and rubber layer of the rubber-lined pipe. The identification system processes and calculates the ultrasonic signals, echo signals and resonance signals to generate resonance signal curves and mapping images, and classifies and identifies the defect types of the rubber-lined pipe and evaluates the severity of the defects.

Images