A defect positioning method and system based on three-dimensional visualization of a heating surface tube

Abstract

The present disclosure provides a defect positioning method and system for three-dimensional visualization based on a heating surface tube, which comprises the following steps: obtaining heating surface tube arrangement data, soot blower nozzle cold-state installation data, multi-channel detection response and scanning displacement; constructing a three-dimensional surface model of the heating surface tube arrangement and generating a tube wall unfolding response image; converting the detection data into spatialized data with axial position, circumferential position and tube segment sequence relationship; extracting the weighted center of the abnormal response candidate area from the tube wall unfolding response image; identifying a hot-state blowing boundary inclined band phase error model based on the axial and circumferential misalignment of the weighted centers of adjacent tube segments with the tube segment sequence; determining the three-dimensional positioning point, attribution distance and inclined band normalized coordinates of the defect; and finally generating a three-dimensional visualization result containing the defect position and the hot-state blowing boundary inclined band, thereby effectively reducing the defect positioning deviation caused by uneven scanning speed, dense tube segment arrangement and insufficient two-dimensional display information.

Description

Technical Field

[0001] This invention relates to the field of image data processing and three-dimensional visualization modeling technology, and in particular to a defect localization method and system based on three-dimensional visualization of heated surface tube detection. Background Technology

[0002] In thermal power plant boilers, the tube banks of superheaters and reheaters are subjected to long-term high-temperature flue gas scouring, steam heat exchange, and periodic soot blowing. These tube banks are numerous and densely packed, with varying degrees of obstruction and accessibility for maintenance between different sections. Particularly in the boundary area covered by long telescopic steam soot blowers, the tube wall damage response may exhibit a gradually staggered, oblique banded distribution in both axial and circumferential directions due to the effects of nozzle rotation and sweeping, hot-state drooping of the nozzle tubes, thermal expansion and bulging of the tube screen, and obstruction by adjacent tube sections. During unit shutdown and maintenance, inspection personnel typically need to perform multi-channel inspections in these areas, generating a large amount of inspection response data. Accurately mapping the inspection responses to specific tube sections, axial positions, and circumferential orientations in the three-dimensional space of the heat-receiving tube banks becomes a key technical issue affecting the accuracy of defect location and the reliability of maintenance judgment.

[0003] Existing inspection results for heated surface tubes are typically presented in the form of two-dimensional waveform diagrams, single-tube unfolded diagrams, C-scan diagrams, or inspection lists. Inspectors mainly determine the location of defects based on tube segment numbers, scanning start points, axial distances, and experience. This method is difficult to express the spatial misalignment between adjacent tube segments in the sootblower's spray coverage boundary area, and it is also difficult to distinguish between continuous damage formed by the same spray action boundary and multiple isolated abnormal responses. When spatial offset occurs after the tube screen is in hot operation, the scanning speed is uneven, or the scanning benchmarks for different maintenance cycles are inconsistent, the two-dimensional inspection results are prone to inaccurate attribution of defective tube segments, deviations in circumferential position marking, and errors in cross-cycle defect alignment, which in turn affect the three-dimensional visualization of heated surface tube defects and subsequent maintenance decisions. Summary of the Invention

[0004] To address the aforementioned technical problems, embodiments of this application provide a defect localization method based on three-dimensional visualization of heated surface tube inspection, comprising the following steps: S1. Obtain the structural data of the heated surface tube bank and the cold installation data of the soot blower nozzle, construct a three-dimensional surface model of the heated surface tube bank, and determine the reference coordinates of the cold soot blowing boundary. S2. Obtain the multi-channel detection response and scanning displacement, match the multi-channel detection response with the scanning displacement, and generate a pipe wall unfolding response image; S3. Extract candidate regions of abnormal responses from the expanded response image of the pipe wall and calculate the weighted center of the candidate regions of abnormal responses. S4. Based on the axial and circumferential misalignment of the weighted center of adjacent pipe segments with the pipe segment sequence, identify the phase misalignment model of the hot-state jetting boundary oblique zone. S5. Determine the three-dimensional location point, the assigned distance, and the normalized coordinates of the oblique zone of the hot-state jetting boundary and the reference coordinates of the cold-state soot blowing boundary based on the hot-state jetting boundary oblique zone misalignment model and the cold-state soot blowing boundary reference coordinates. S6. Based on the three-dimensional defect location points, the assigned distance, the oblique zone normalized coordinates, and the three-dimensional surface model of the heated surface tube array, generate three-dimensional visualized defect location results.

[0005] In conjunction with the first aspect, a three-dimensional surface model of the heated tube bank is constructed, and the reference coordinates of the cold-state soot blowing boundary are determined, including: The heat-receiving surface tube bank structure data includes tube segment number, tube diameter, tube centerline, pitch and arrangement sequence; the sootblower nozzle cold installation data includes nozzle outlet reference point and spray direction. A three-dimensional surface model of each heated tube is established based on the tube coreline, tube diameter, and arrangement sequence. Determine the cold injection centerline based on the nozzle exit reference point and the injection direction; Calculate the nearest interaction position between the cold jet centerline and the three-dimensional surface model of each heated surface tube; The axial coordinates and circumferential angles corresponding to the most recent action position are used as the reference coordinates for the cold soot blowing boundary.

[0006] Combining the first aspect, multi-channel detection responses and scanning displacements are acquired, and the multi-channel detection responses are correlated with the scanning displacements to generate a tube wall unfolding response image, including: The scanning displacement sequence recorded by the servo encoder is obtained, and the detection time is mapped to the axial position of the heated surface tube according to the scanning displacement sequence. Acquire the multi-channel detection response sequence and map each detection channel to the circumferential angle of the heated surface tube; Using the axial position and circumferential angle of the heated surface tube as image coordinates, and the multi-channel detection response value as the image grayscale value, a tube wall unfolding response image is formed; The response values ​​in the tube wall unfolding response image are normalized based on the median and median absolute deviation to obtain the normalized tube wall unfolding response image.

[0007] In conjunction with the first aspect, candidate regions of abnormal responses are extracted from the unfolded response image of the pipe wall, and the weighted centers of the candidate regions of abnormal responses are calculated, including: Candidate regions for abnormal responses are obtained from local abnormal regions with continuous distribution of response values ​​in the tube wall unfolding response image; The response weights are obtained by summing the response values ​​within each candidate region of an anomaly response. Calculate the area of ​​the candidate region for each candidate region of an anomaly response. Based on the response values ​​of each image location within the candidate region of abnormal response, the axial weighted center and the circumferential weighted center are calculated, and the axial weighted center and the circumferential weighted center are used as candidate centers for identifying the phase-shifting model of the hot jet boundary band.

[0008] Combining the first aspect, based on the axial and circumferential misalignment of the weighted centers of adjacent pipe segments caused by the pipe segment sequence, a hot-state jetting boundary oblique zone phase misalignment model is identified, including: The candidate center is converted into a tube-screen unfolding coordinate system consisting of axial position and circumferential arc length position; For adjacent pipe sections near the sootblower's spray coverage boundary, an oblique-band phase-shifting model is established that varies with the pipe section sequence. The oblique-band phase-shifting model includes axial phase-shifting components and circumferential phase-shifting components. Candidate centers are selected from the candidate regions of abnormal responses of each heated surface tube, and the initial parameters of the oblique-band phase-shifting model are fitted with the goal of minimizing the distance error from the candidate center to the oblique-band phase-shifting model after weighting by response weight. A phase-in-phase model is established, which indicates that the candidate centers of adjacent pipe segments do not generate axial phase-out components and circumferential phase-out components with the pipe segment sequence. Calculate the sum of squared residuals for the skew-band out-of-phase model and the in-phase model, respectively; The information criterion values ​​for the oblique-band out-of-phase model and the in-phase model are calculated based on the sum of squared residuals, the number of model parameters, and the number of candidate centers involved in the fitting, respectively. When the information criterion value of the oblique zone phase-shifting model is less than that of the in-phase model, the oblique zone phase-shifting model is used as the hot-state jet boundary oblique zone phase-shifting model.

[0009] In conjunction with the first aspect, the three-dimensional location point, attribution distance, and normalized coordinates of the oblique zone of the hot-state blowing boundary are determined based on the hot-state blowing boundary oblique phase misalignment model and the cold-state blowing boundary reference coordinates, including: Based on the predicted axial position and predicted circumferential arc length position of the hot-state jetting boundary band misalignment model on the corresponding pipe section, and the reference coordinates of the cold-state soot blowing boundary, the misalignment offset of the hot-state jetting band relative to the cold-state soot blowing boundary is calculated. Based on the axial weighted center and circumferential weighted center of the candidate region of abnormal response, three-dimensional location points of defects are generated in the three-dimensional surface model of the heated surface tube bank. Calculate the attribution distance between the candidate regions of the abnormal response and the phase-shifted model of the thermal jet boundary band; The oblique zone direction vector is formed based on the axial and circumferential phase-shifting components of the hot-state jetting boundary oblique zone phase-shifting model. The pipe segment sequence, axial weighting center, and circumferential arc length position of the candidate region of abnormal response are used to form three-dimensional pipe screen parameter points; The three-dimensional tube screen parameter points are projected onto the oblique band direction vector to obtain the oblique band normalized coordinates.

[0010] Combining the first aspect, based on the three-dimensional defect location points, attribution distances, oblique zone normalized coordinates, and the three-dimensional surface model of the heated pipe array, a three-dimensional visualized defect location result is generated, including: The predicted axial position and predicted circumferential angle of the hot-state jetting boundary band phase-shifting model on each pipe segment are mapped to the three-dimensional surface model of the heated pipe row to generate the hot-state jetting boundary band marking points. Connect the marked points of the hot-state jetting boundary slope band of adjacent pipe sections to generate the hot-state jetting boundary slope band; The defect location is displayed based on the three-dimensional positioning points of the defect, and the correlation strength between the defect and the hot-blown boundary slope is generated based on the attribution distance. The smaller the attribution distance, the greater the correlation strength. The output includes a 3D visualization of the defect location, including the defect's 3D location points, the hot-blown boundary band, the phase shift, the attribution distance, the correlation strength, and the normalized coordinates of the band.

[0011] In conjunction with the first aspect, when there are multiple detection cycles, defects in different detection cycles are aligned according to the normalized coordinates of the slant band; the area of ​​the candidate region of abnormal response and the detection time corresponding to the normalized coordinates of the same slant band are obtained; the defect expansion rate is obtained based on the difference in the area of ​​the candidate region of abnormal response and the difference in the detection time between adjacent detection cycles; and the defect expansion rate is added to the three-dimensional visualization defect localization result.

[0012] A second aspect of this disclosure provides a defect localization system based on three-dimensional visualization of heated surface tube inspection, comprising: The model baseline construction module is used to acquire the structural data of the heated surface tube bank and the cold installation data of the sootblower nozzles, construct a three-dimensional surface model of the heated surface tube bank, and determine the baseline coordinates of the cold sootblowing boundary. The tube wall unfolding imaging module is used to acquire multi-channel detection response and scanning displacement, and to generate tube wall unfolding response image by mapping the multi-channel detection response to the scanning displacement. An abnormal region extraction module is used to extract candidate regions of abnormal response from the unfolded response image of the pipe wall and to calculate the weighted center of the candidate regions of abnormal response. The oblique-band phase misalignment identification module is used to identify the oblique-band phase misalignment model of the hot-state jetting boundary based on the axial and circumferential misalignment of the weighted center of adjacent pipe segments as the pipe segment sequence occurs. The defect 3D location module is used to determine the defect 3D location point, the assigned distance, and the normalized coordinates of the oblique band based on the hot-state jetting boundary oblique phase misalignment model and the cold-state soot blowing boundary reference coordinates. The visualization module is used to generate three-dimensional visualized defect location results based on the three-dimensional defect location points, the assigned distance, the oblique zone normalized coordinates, and the three-dimensional surface model of the heated surface tube array.

[0013] A third aspect of this disclosure provides an electronic device comprising: One or more processors; A storage unit for storing one or more programs that, when executed by one or more processors, enable the one or more processors to implement the method.

[0014] Beneficial Effects: This disclosure provides a defect localization method and system based on three-dimensional visualization of heated surface tube inspection. By acquiring structural data of the heated surface tube bank, cold installation data of the sootblower nozzles, and multi-channel detection response and scanning displacement, a three-dimensional surface model of the heated surface tube bank is constructed and a tube wall unfolding response image is generated. The detection data is transformed into spatialized data with axial position, circumferential position, and tube segment sequence relationship. Simultaneously, the weighted center of the abnormal response candidate region is extracted from the tube wall unfolding response image. Based on the axial and circumferential misalignment of the weighted center of adjacent tube segments with the tube segment sequence, the hot-state blowing boundary oblique phase misalignment model is identified, thereby determining the three-dimensional location of the defect. By using the point, assigned distance, and normalized coordinates of the slope, a three-dimensional visualization result containing the defect location and the slope of the hot-state spray boundary is generated. This accurately maps the detected abnormal response to the specific pipe segment and three-dimensional spatial coordinates, effectively reducing defect location deviations caused by uneven scanning speed, dense pipe segment arrangement, and insufficient two-dimensional display information. It distinguishes between continuous damage and isolated abnormal responses formed by the same spray boundary, reduces errors in defect pipe segment assignment, circumferential labeling deviation, and cross-cycle alignment error, and enables maintenance personnel to intuitively view the spatial distribution and changing trends of defects. This significantly improves the accuracy of spatial location and cross-cycle alignment of defects in heated surface pipe rows and the reliability of maintenance judgment. Attached Figure Description

[0015] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 This is a flowchart illustrating a defect localization method based on three-dimensional visualization of heated surface tube inspection, according to an embodiment of this application. Figure 2 This is a structural diagram of a defect localization system based on three-dimensional visualization for detecting heated surface tubes, according to an embodiment of this application. Figure 3This is a schematic diagram of the pipe wall unfolding response image and abnormal response candidate region extraction in an embodiment of this application; Figure 4 The electronic device described in this application is an embodiment of the present application. Detailed Implementation

[0017] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0018] As indicated in this application and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not specifically singular and may include plural forms. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0019] While this application makes various references to certain modules of the systems according to embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The modules described are merely illustrative, and different aspects of the systems and methods may use different modules.

[0020] Flowcharts are used in this application to illustrate the operations performed by the system according to embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them.

[0021] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0022] like Figure 1 The diagram shows a flowchart of a defect localization method based on three-dimensional visualization of heated surface tube inspection according to an embodiment of the application, including the following steps: S1. Obtain the structural data of the heated surface tube bank and the cold installation data of the soot blower nozzle, construct a three-dimensional surface model of the heated surface tube bank, and determine the reference coordinates of the cold soot blowing boundary. S2. Obtain the multi-channel detection response and scanning displacement, match the multi-channel detection response with the scanning displacement, and generate a pipe wall unfolding response image; S3. Extract candidate regions of abnormal responses from the expanded response image of the pipe wall and calculate the weighted center of the candidate regions of abnormal responses. S4. Based on the axial and circumferential misalignment of the weighted center of adjacent pipe segments with the pipe segment sequence, identify the phase misalignment model of the hot-state jetting boundary oblique zone. S5. Determine the three-dimensional location point, the assigned distance, and the normalized coordinates of the oblique zone of the hot-state jetting boundary and the reference coordinates of the cold-state soot blowing boundary based on the hot-state jetting boundary oblique zone misalignment model and the cold-state soot blowing boundary reference coordinates. S6. Based on the three-dimensional defect location points, the assigned distance, the oblique zone normalized coordinates, and the three-dimensional surface model of the heated surface tube array, generate three-dimensional visualized defect location results.

[0023] Specifically, this embodiment takes the high-temperature reheater tube panel of a 660MW supercritical thermal power generating unit boiler as the implementation object. The tube panel is equipped with a long telescopic steam soot blower. The implementation scenario is that during the unit shutdown and maintenance, multi-channel eddy current non-destructive testing is carried out on the heated surface tube panel of the soot blower spray coverage boundary area. It should be noted that the hot-state blowing boundary in this embodiment does not refer to the blowing boundary obtained directly through real-time measurement during boiler operation. Instead, it refers to the equivalent blowing action boundary during operation, which is obtained by fitting the axial and circumferential misalignment distribution of the abnormal response candidate region between adjacent heated surface tubes, combined with the reference coordinates of the cold-state soot blowing boundary, based on the abnormal response candidate region obtained during unit shutdown and maintenance. This equivalent blowing action boundary during operation is used to characterize the positional trend of actual damage formed on the tube wall surface by the combined effects of nozzle rotation sweeping, hot-state drooping of the gun tube, outward bulging of the tube screen due to thermal expansion, and shielding by adjacent tube sections during the boiler's hot-state operation. The hot-state blowing boundary oblique phase misalignment model referred to later in this embodiment is an equivalent boundary model fitted based on the defect spatial distribution results reflected by the shutdown detection data, and does not require the addition of hot-state real-time measurement equipment during boiler operation.

[0024] By combining boiler design drawing extraction with on-site verification using 3D laser scanning, the structural data of the target tube panel's heating surface tube array and the cold-state installation data of the sootblower nozzles were obtained. The heating surface tube array structural data includes tube segment numbers, the designed or measured outer diameter of a single heating surface tube, the 3D coordinate sequence of the tube centerline of a single heating surface tube, the transverse pitch and longitudinal pitch of the tube panel, and the arrangement sequence of tube segments along the sootblower's spray direction. The tube segment number is used to uniquely identify the corresponding heating surface tube, and the designed or measured outer diameter is used to determine the outer radius of the tube wall in the 3D surface model. The cold-state installation data of the sootblower nozzles includes the 3D coordinates of the nozzle outlet reference point when the sootblower is fully inserted into the furnace and in cold-state positioning, and the unit vector of the nozzle's designed spray direction. For each heated surface tube with a segment number, a parametric three-dimensional surface model is constructed based on the obtained three-dimensional coordinate sequence of the tube centerline and the tube diameter parameters. Specifically, let the tube centerline of the i-th heated surface tube be a spatially continuous curve, and its parametric expression is: , In the formula, Let be the centerline of the i-th heated surface tube; The coordinates represent the axial length along the tangent to the tube's core, with the origin set as the starting reference point for scanning the tube end of this heated surface tube. The value range is from 0 to the effective scanning length of the heated surface tube; , , These are the X, Y, and Z coordinate components of the pipe centerline in the three-dimensional coordinate system of the boiler furnace. This three-dimensional coordinate system takes the lower left corner of the boiler furnace as the origin, the furnace width direction as the X-axis, the furnace depth direction as the Y-axis, and the furnace height direction as the Z-axis, which conforms to the general design coordinate system specifications of the boiler industry. Let the outer radius of the i-th heated surface tube be... , The value of is half of the designed or measured outer diameter of the heated surface tube. Based on the Frenet frame of the space curve, the parameterized equation of the tube wall surface is constructed, that is, the parameterized expression of the three-dimensional surface point of the tube wall of the i-th heated surface tube is: , In the formula, Let the axial coordinate of the i-th heated surface tube be... and circumferential angle Three-dimensional surface points of the pipe wall below; The circumferential angle of the tube wall is defined as follows: the 0° reference point is set at the center of the fire-facing surface of the tube screen, and the direction of angle increase is counterclockwise rotation along the positive direction of the tube axis. The value range is from 0° to 360°. Let be the principal normal unit vector of the tube centerline at the axial coordinate s; Let be the subnormal unit vector of the pipe centerline at the axial coordinate s. Both the principal normal and the subnormal are perpendicular to the tangential unit vector of the pipe centerline, and the three are orthogonal to each other, forming an orthogonal coordinate system in the circumferential direction of the pipe wall. For straight pipe sections or heated surface pipe sections with curvature lower than the modeling accuracy requirements, since the principal normal direction in the Frenet frame may not be unique or may be numerically reversed, this embodiment uses the center direction of the fire-facing surface of the pipe screen as the circumferential zero-degree reference direction. This circumferential zero-degree reference direction is projected onto a plane perpendicular to the tangential direction of the pipe axis to obtain the circumferential reference direction at the current axial position. Then, the tangential direction of the pipe axis is cross-multiplied with this circumferential reference direction to obtain another circumferential orthogonal direction, thus forming a stable circumferential orthogonal coordinate system of the pipe wall. For cases where normal reversal occurs between adjacent axial sampling points... Furthermore, the circumferential reference direction of the previous axial sampling point is re-orthogonalized after parallel movement to obtain the circumferential reference direction of the current axial sampling point. Through this processing method, it can be ensured that a continuous and consistent three-dimensional surface model of the pipe wall can be established for straight pipe sections, low-curvature pipe sections, and curved pipe sections, and that the definition of the circumferential angle is continuous on the same heated surface pipe. Through this parameterized equation, any point on the pipe wall with any axial position and any circumferential angle can be uniquely mapped to three-dimensional spatial coordinates, completing the construction of the three-dimensional surface model of a single heated surface pipe, and then splicing the three-dimensional surface model of the entire pipe array according to the pipe section number and arrangement sequence.

[0025] The cold-state spray centerline is constructed based on the cold-state installation data of the sootblower nozzle; assuming the cold-state spray centerline of the sootblower nozzle is a spatial straight line, its parameterized expression is: , In the formula, The center line of the cold-state spray of the sootblower nozzle; The three-dimensional coordinates of the nozzle outlet reference point in the above-mentioned three-dimensional coordinate system of the boiler furnace; Design a unit vector for the nozzle's spray direction; For linear parameters along the injection direction, The range of values ​​covers the effective spray range of the sootblower design; the spatial position of the spray centerline of the sootblower in a cold state can be fully expressed through this parametric equation; The nearest interaction position between the cold-state injection centerline and the 3D surface model of each heated surface tube is calculated. This nearest interaction position is the point with the minimum spatial distance between the cold-state injection centerline and the tube wall surface. It is used to characterize the action reference of the cold-state design injection boundary of the sootblower on a single heated surface tube. Specifically, this is achieved by optimizing the solution to minimize the spatial distance. That is, for the i-th heated surface tube, the expression for solving its cold-state sootblowing boundary reference coordinates is: , In the formula, The Euclidean norm is used to calculate the straight-line distance between two spatial points, thereby quantifying the spatial proximity of the injection centerline to the pipe wall surface. For minimization operators, let represent the operator that minimizes the objective function, i.e., the spatial distance value. , , Parameter combinations; That is, the axial reference coordinates of the cold soot blowing boundary corresponding to the i-th heated surface tube; That is, the circumferential reference angle of the cold soot blowing boundary corresponding to the i-th heated surface tube; The range parameter is the distance between the centerline of the cold jet and the nearest point of contact. The solution process can employ the Levenberg-Marquardt nonlinear least squares optimization algorithm, or a combination of grid initial value search and local least squares optimization, to ensure a stable nearest point of contact even with complex tube-screen geometry. Finally, the range parameter corresponding to each heated surface tube is... and The pipe segment numbers and arrangement order are used to store the structured data as the reference data for subsequent calculation of the hot-state oblique phase offset. At the same time, the completed three-dimensional surface model of the heated pipe array is converted into a triangular facet model with UV mapping for subsequent defect three-dimensional location mapping and visualization rendering.

[0026] In one embodiment of the present invention, acquiring multi-channel detection response and scanning displacement, mapping the multi-channel detection response to the scanning displacement, and generating a tube wall unfolding response image includes: This embodiment uses a multi-channel eddy current detection response as a specific example of a multi-channel detection response for illustration. However, the multi-channel detection response described in this invention is not limited to eddy current detection response. As long as the detection device can output response values ​​corresponding to the detection time, scanning displacement, and circumferential channel, they can all be converted into a pipe wall unfolding response image according to the method of this embodiment. The response value can be eddy current amplitude, phase, combined response quantity, or other channel response quantities that can characterize the local state changes of the pipe wall. When generating the pipe wall unfolding response image, the selected response value is used as the image grayscale value or pixel intensity value to participate in subsequent image processing. The multi-channel non-destructive testing equipment used in this embodiment is a multi-channel eddy current testing system, equipped with an automatic scanner that travels along the axial direction of the heated surface tube. An incremental photoelectric follow-up encoder is installed on the scanner. The encoder and the multi-channel testing system achieve time synchronization through hardware trigger signals. The sampling frequencies of the two can be matched and set according to the detection scanning speed and resolution requirements. It is preferred to set them to the same sampling frequency to ensure that the timestamps of the detection response and the scanning displacement correspond one-to-one. The scanning displacement sequence continuously recorded by the servo encoder during the detection process, along with the detection time sequence synchronized with the displacement sequence, is acquired. The detection time is then precisely mapped to the axial position of the heated surface tube. Specifically, let the servo encoder at the detection time... The corresponding real-time scanning speed is Scanning speed The encoder pulse equivalent can be calculated by converting the number of pulses per unit time of the encoder with the preset encoder pulse equivalent. The encoder pulse equivalent can be pre-calibrated based on the circumference of the scanner wheel and the number of encoder lines. The axial position of the heated surface tube corresponds to the detection time t. The calculation formula is based on the time integral of the real-time scanning speed: , In the formula, the lower limit of integration 0 corresponds to the start time of the detection scan. This start time corresponds one-to-one with the scan start reference point of the tube end of the heated surface, ensuring that the axial coordinate origin is consistent with the axial coordinate origin of the aforementioned three-dimensional surface model. To detect the axial position of the probe along the pipe axis at time t; the axial position is calculated by integration, which can adapt to the actual working conditions of non-uniform scanning speed during on-site scanning, avoid the axial position deviation caused by uniform speed conversion, and ensure the accurate correspondence between the detection response and the axial position of the pipe wall. Next, the multi-channel detection response sequence synchronously acquired by the multi-channel detection system is obtained. In this embodiment, the detection probes are array probes uniformly arranged along the circumference of the heated surface tube, with each probe corresponding to an independent detection channel. The circumferential angle of each detection channel is calibrated one by one with the circumferential angle of the heated surface tube. Specifically, the center position of the tube screen facing the fire is taken as the 0° circumferential reference, and the direction of increasing the circumferential angle is counterclockwise along the positive direction of the tube axis. This circumferential angle reference is consistent with the circumferential angle reference of the aforementioned three-dimensional surface model. According to the circumferential arrangement spacing of the array probes, a unique corresponding circumferential angle is assigned to the m-th detection channel. , The value range is from 0° to 360°; for channels that are not continuously arranged in the circumferential direction, the response data of the intermediate angle can be supplemented by circumferential interpolation. The interpolation method can be linear interpolation or spline interpolation to ensure continuous coverage of the circumferential angle. Using the axial position of the heated surface tube as the first dimension coordinate of the image and the circumferential angle of the heated surface tube as the second dimension coordinate, and the corresponding multi-channel detection response value as the pixel grayscale value of the image, an initial tube wall unfolding response image is constructed. Specifically, discrete detection response sampling points are mapped into a two-dimensional digital image of a regular grid using bilinear interpolation. The axial pixel spacing and circumferential pixel spacing of the image can be adaptively adjusted according to the detection resolution requirements. The axial pixel spacing is preferably set to 0.5mm to 2mm, and the circumferential pixel spacing is preferably set to 0.5° to 2°, so that each pixel in the initial tube wall unfolding response image can uniquely correspond to an axial position and circumferential angle combination on the tube wall, realizing a one-to-one mapping between the detection response and the tube wall position. The above-mentioned axial pixel spacing and circumferential pixel spacing are preferred values ​​adapted to the detection resolution of this embodiment and are not intended to limit the invention. The axial pixel spacing is determined based on the minimum effective resolution of the scanning displacement sequence and the defect positioning accuracy requirements. The circumferential pixel spacing is determined based on the circumferential arrangement interval of the detection channel, the outer diameter of the heated surface tube, and the minimum effective resolution after circumferential interpolation. The axial pixel spacing should not be greater than the minimum axial anomaly length that the detection equipment can stably distinguish, and the circumferential arc length corresponding to the circumferential pixel spacing should not be greater than the minimum circumferential anomaly length that the detection equipment can stably distinguish. This selection method ensures that the pixel units in the tube wall unfolding response image match the actual spatial resolution capability of the detection equipment.

[0027] To reduce the impact of gain differences and background noise fluctuations between different detection channels on the response values, while preserving the relative amplitude characteristics of abnormal defect responses, robust normalization based on the median and median absolute deviation is performed on the response values ​​in the initial tube wall unfolding response image. For the i-th heated surface tube, the formula for calculating the normalized tube wall unfolding response image is as follows: , in: , In the formula, The axial position of the i-th heated surface tube and circumferential angle The normalized response value at the location; For the i-th heated surface tube, ... The original detection response value of each detection channel at time t; To detect the absolute deviation of the response value from the overall median benchmark, the absolute value can uniformly characterize the degree of abnormal deviation of the response value. Whether it is a positive or negative deviation, it can reflect the abnormal changes in the pipe wall condition. The median value is the median of all detected response values ​​of the i-th heated surface tube. Using the median value can avoid the interference of extreme defect response values ​​on the overall benchmark, ensure that the normalized benchmark is not affected by a small number of abnormal signals, and has strong robustness. The median absolute deviation of all test response values ​​of the i-th heated surface tube is used to characterize the background fluctuation level of the test response. The median absolute deviation is not sensitive to outliers, which can further improve the anti-interference ability of normalization processing and adapt to the actual working conditions of unstable background noise in field testing. To avoid extremely small positive numbers with a denominator of zero, its value can be adjusted according to the dynamic range of the detection equipment's response. Through this normalization process, the detection responses of different pipe sections and channels can be unified to the same dimensional range. At the same time, the difference between the defect abnormal response and the background noise is amplified, which facilitates the subsequent extraction of abnormal response regions. The resulting normalized pipe wall unfolding response image can be directly used for the subsequent extraction and analysis of candidate regions for abnormal responses.

[0028] In one embodiment of the present invention, extracting candidate regions of abnormal responses from the tube wall unfolding response image and calculating the weighted center of the candidate regions of abnormal responses includes: For the normalized unfolded response image of the i-th heated surface tube, threshold segmentation of abnormal regions is performed. The segmentation threshold can be adaptively adjusted according to the signal-to-noise ratio requirements of the on-site detection, preferably set to 3 to 5. Since the response values ​​of the normal tube wall background are concentrated around 0 after the aforementioned normalization processing based on the median and median absolute deviation, response values ​​exceeding 3 times the median absolute deviation can be considered statistically significant abnormal responses. Using this threshold, the normalized unfolded response image is binarized to obtain a set of abnormal pixels with response values ​​higher than the segmentation threshold. Subsequently, 8-neighborhood connected component analysis is performed on the binarized abnormal pixel set to extract locally connected regions with continuously distributed response values. Each independent connected region is a candidate region for abnormal response, denoted as . ,in, For the first The candidate regions for abnormal responses on the heated pipe surface are sequentially numbered. Simultaneously, a minimum area screening threshold can be set according to detection requirements to remove isolated noise connected regions with fewer than a preset threshold pixels. The minimum area screening threshold is preferably set to 3 to 10 pixels to avoid random noise being misidentified as candidate regions for abnormal responses, ensuring that the extracted candidate regions are all pipe wall abnormal responses with continuous distribution characteristics. The above segmentation threshold and minimum area screening threshold are not fixed restrictions; their selection is based on the background response distribution and image spatial resolution of the normalized pipe wall unfolded response image. The segmentation threshold is used to distinguish between background responses and local abnormal responses. Its determination principle is to ensure that the number of connected regions formed by random noise in the normal background area is lower than the allowable noise level for extracting candidate regions for abnormal responses, while maintaining the connectivity of continuously distributed local abnormal regions. The minimum area screening threshold is used to remove isolated noise connected regions. Its determination principle is: it is not less than the pixel scale formed by single-point random noise in the tube wall unfolding response image, and not greater than the pixel scale corresponding to the smallest resolvable abnormal area of ​​the detection device in the axial and circumferential directions. When the resolution of the detection device, the axial pixel spacing, the circumferential pixel spacing, or the outer diameter of the heated tube changes, the segmentation threshold and the minimum area screening threshold are re-determined according to the normalized background response distribution and the image spatial resolution.

[0029] For each extracted candidate region of anomaly response, its corresponding response weight is calculated using the following formula: , In the formula, Let be the response weight of the j-th abnormal response candidate region on the i-th heated surface tube; The axial position and circumferential angle coordinates of each pixel within the candidate region for abnormal responses; This is the normalized response value corresponding to the pixel. The summation operation is used to accumulate the total response amplitude of all pixels in the abnormal region. The sum of response values ​​is used as the response weight because this weight can comprehensively characterize the response intensity and coverage of the abnormal region. The larger the weight, the more significant the detection response of the abnormal region. At the same time, this weight can be used as a weighting coefficient when fitting the subsequent oblique band phase misalignment model, so that the abnormal region with a more significant response contributes more to the model fitting and improves the accuracy of model recognition. The area corresponding to each candidate region of an anomaly response is calculated using the following formula: , In the formula, The area of ​​the region corresponding to the candidate region for the abnormal response; This represents the total number of pixels contained in the candidate region for the abnormal response. The axial pixel spacing of the tube wall unfolding response image is consistent with the axial pixel spacing in the aforementioned image generation step. Let be the outer radius of the i-th heated surface tube; The circumferential pixel spacing of the tube wall unfolded response image; The circumferential arc length corresponding to a single pixel; this formula can convert the number of pixels in the image into the actual physical area of ​​the pipe wall surface. This area parameter can intuitively characterize the coverage of the defect and provide basic data for calculating the defect expansion rate during subsequent multi-cycle detection. In scenarios where physical size conversion is not required, the number of pixels can also be used as a representation of the area of ​​the candidate region for abnormal response. However, the same area representation method should be used in the same detection batch and the same cross-cycle comparison process. For each candidate region of anomaly response, its axial weighted center and circumferential weighted center are calculated based on the normalized response values ​​of each pixel within the region. The formula for calculating the axial weighted center is as follows: , The formula for calculating the circumferential weighted center is: , In the formula, , is the axial weighted center of the j-th abnormal response candidate region on the i-th heated surface tube, used to represent the core position of the abnormal response candidate region in the tube axis direction; , is the circumferential weighted center of the j-th abnormal response candidate region on the i-th heated surface tube, used to represent the core angular position of the abnormal response candidate region in the circumferential direction of the tube wall; This represents the j-th abnormal response candidate region on the i-th heated surface tube. Candidate regions for abnormal responses The coordinates of the tube wall position corresponding to any pixel point within the tube, where This indicates the axial position corresponding to the pixel. This indicates the circumferential angle corresponding to that pixel. Let the axial position s and circumferential angle of the i-th heated surface tube be... The normalized response value at the location is used to characterize the degree of anomaly of the detection response at that pipe wall location relative to the background response. The defect center is calculated by weighting the response value instead of directly using the geometric center of the region because there is a correlation between the detection response amplitude of the pipe wall defect and the size or degree of change in the local state of the defect. The higher the response amplitude, the closer the location is to the core area of ​​the abnormal response. Weighting the response value can effectively reduce the influence of low-amplitude noise at the edge of the region on the location of the defect center, and more accurately characterize the core position of the defect in the axial and circumferential directions of the pipe wall, ensuring the benchmark accuracy of the subsequent oblique band phase misalignment model fitting. The axial weighted center and circumferential weighted center corresponding to each candidate region of the abnormal response are used as candidate centers for the subsequent hot-state blowing boundary oblique band phase misalignment model identification. The response weight, region area and candidate center are simultaneously structured and stored to provide complete basic data for subsequent steps.

[0030] In one embodiment of the present invention, based on the axial and circumferential misalignment of the weighted center of adjacent pipe segments caused by the pipe segment sequence, the hot-state jetting boundary oblique zone phase misalignment model is identified, including: For the axial weighted center and circumferential weighted center of the j-th abnormal response candidate region on the i-th heated surface tube, convert them into the tube-screen unfolded coordinates composed of the axial position and the circumferential arc length position. The conversion formula is as follows: , In the formula, Let J be the tube-screen expansion coordinates of the j-th candidate region for abnormal response on the i-th heated surface tube; Let be the outer radius of the i-th heated surface tube; Let j be the axial weighted center of the candidate region for the abnormal response on the i-th heated surface tube. The circumferential weighted center of the j-th abnormal response candidate region on the i-th heated surface tube needs to be pre-converted to radians to ensure the dimensional accuracy of the circumferential arc length calculation. This is the circumferential arc length position corresponding to the candidate center; this transformation converts the circumferential angle into a length dimension that is consistent with the axial position, which can reduce the impact of differences in pipe diameters of different pipe segments on the quantification of circumferential position, make the positional deviations between different pipe segments have uniform comparability, and provide a standardized quantitative benchmark for subsequent distance error calculation.

[0031] Select adjacent pipe sections close to the sootblower's spray coverage boundary; specifically, based on the determined cold sootblowing boundary reference coordinates, select the heated surface pipe sections within the effective spray range coverage of the sootblower's design, and number the selected pipe sections sequentially along the sootblower's spray direction according to the arrangement layer sequence. Where n is the total number of adjacent pipe segments participating in the model fitting, the value of n can be preferably set to 3 to 20 based on the actual layout of the pipe screen and the distribution range of defects, thereby limiting the target range of the model fitting and avoiding interference from irrelevant pipe segments in non-flushing areas with the fitting results; the number of adjacent pipe segments participating in the model fitting is not limited to a fixed value, and its determination principle is: the selected pipe segments should be located within the effective spray range coverage of the sootblower, and at least contain effective abnormal response candidate areas that can determine the parameters of the oblique band phase misalignment model; when there are multiple spatially discontinuous abnormal response distributions within the coverage area of ​​the same sootblower, corresponding oblique band phase misalignment models are established for each segment of continuous abnormal response distribution. The interior point determination distance is determined based on the spatial resolution of the pipe wall unfolded response image, the pitch of adjacent pipe segments, and the positioning error of the abnormal response candidate areas. It should not be less than the maximum spatial size corresponding to a single pixel unit, and not greater than half of the mapping distance corresponding to the pitch of adjacent pipe segments, so as to avoid incorrectly including adjacent oblique bands or irrelevant abnormal areas in the same model.

[0032] For the selected adjacent pipe segments, a slant-zone phase-shifting model is established that linearly varies with the pipe segment sequence i. This model is used to characterize the continuous axial and circumferential offset of the injection boundary caused by the variation of the pipe segment sequence. The model expression is as follows: , In the formula, Let be the predicted tube-screen expansion coordinates of the oblique-band phase-shifting model on the i-th heated surface tube; The axial position of the inclined zone boundary on the i-th heated surface tube, as predicted by the model; The circumferential arc length of the inclined zone boundary on the i-th heated surface tube is predicted by the model. , These are the parameters for the starting position of the inclined zone, corresponding to the axial position and circumferential arc length position of the boundary of the inclined zone on the starting pipe segment where the sequence number i equals 1, respectively. The axial phase shift component of the oblique zone characterizes the axial positional shift of the oblique zone boundary as the sequence of the pipe section increases by one level. The circumferential phase-shifting component of the oblique band represents the circumferential arc length shift generated by the oblique band boundary when the pipe segment sequence increases by one level. This linear model can match the core characteristics of the oblique band phase-shifting that occurs when the sootblower jet continuously sweeps along the pipe row sequence, with the jet boundary gradually shifting axially and circumferentially on adjacent pipe segments. Compared with the nonlinear model, it has fewer parameters, stronger fitting stability, and can effectively avoid overfitting problems, making it suitable for actual working conditions with limited field detection data. A target function for model fitting is constructed. From multiple candidate regions of abnormal responses in each heated surface tube, the candidate center with the best match to the oblique-band model is selected. The optimization objective is to minimize the distance error from the candidate center to the oblique-band phase-shifting model after weighting by response weights. The initial parameters of the oblique-band phase-shifting model are then fitted. The response weights of the candidate centers participating in the fitting are normalized. The selected center in the i-th heated surface tube is... The formula for calculating the normalized response weights corresponding to each candidate center is: , In the formula, For the i-th heated surface tube, the first... Normalized response weights corresponding to each candidate center; For the i-th heated surface tube, the first... The response weights corresponding to each candidate center; To avoid extremely small positive numbers with a denominator of 0, normalization can map all weights to a comparable range of values, ensuring a balanced weight contribution across different pipe segments. Based on this, a weighted least squares optimization objective function is constructed, expressed as follows: , In the formula, The set of candidate center numbers selected for model fitting in each heated surface tube; The Euclidean norm is used to calculate the straight-line distance between the actual coordinates of the candidate center and the model-predicted coordinates. The distance error from the candidate center to the oblique band model is squared. By using normalized response weights to weight the distance error, the candidate center with higher response amplitude and stronger defect credibility can contribute more to the model fitting. This effectively reduces the interference of low-amplitude noise anomaly regions and pseudo-defect responses on the model fitting results and improves the matching degree between the oblique band phase-shifting model and the real hot-state jet boundary. This optimization problem can be solved using a combination of the Random Sample Consensus (RANSAC) algorithm and the Levenberg-Marquardt algorithm. Specifically, the pipe-screen unfolded coordinates of the candidate region of the abnormal response are first used as the sampling object. Each time, candidate center combinations that can determine the parameters of the oblique band phase reversal model are extracted from different pipe segments, and a temporary oblique band phase reversal model is calculated based on these candidate center combinations. Then, the distance error from the remaining candidate centers to the temporary oblique band phase reversal model is calculated, and candidate centers with distance errors less than the inlier determination distance are selected as inliers. After repeated sampling, the inlier set corresponding to the temporary oblique band phase reversal model with the largest number of inliers and the smallest weighted residual is selected as the effective fitting set. Then, using this effective fitting set as input, the Levenberg-Marquardt algorithm is used to perform least-squares optimization on the parameters of the oblique band phase reversal model to obtain the parameters of the oblique band phase reversal model. The inlier determination distance is determined based on the spatial resolution of the pipe wall unfolded response image, the pitch of adjacent pipe segments, and the localization error of the candidate region of the abnormal response, and is not limited to a fixed value. Finally, the initial parameters of the oblique band phase reversal model are obtained. , , , Complete the initial fitting of the model.

[0033] A phase-in-phase reference model is established for comparison and verification. This model corresponds to the conventionally used cold-state injection boundary reference. Its physical meaning is that the injection boundaries of adjacent pipe segments do not generate axial or circumferential phase-shifting components with the pipe segment sequence, meaning that the injection boundary reference positions of all pipe segments remain consistent. The expression of this phase-in-phase model adopts the same pipe-screen expansion coordinate system as the aforementioned oblique-band phase-shifting model to ensure the consistency of the reference for model comparison. The expression of the phase-in-phase model is as follows: , In the formula, Let the predicted tube-screen expansion coordinates of the in-phase model on the i-th heated surface tube; For the axial position parameters of the in-phase model; The circumferential arc length position parameter of the in-phase model; the model does not contain an offset term that varies with the pipe segment sequence i, that is, the axial phase fault component and the circumferential phase fault component are both 0, which can characterize the blowing boundary without oblique zone phase fault, and serve as a comparison benchmark for the oblique zone phase fault model. Using the same fitting dataset and weighting rules as the oblique-band phase-shifting model, the parameters of the in-phase model are fitted. The fitting objective function adopts the same weighted least squares form as described above, and its expression is: , In the formula, The normalized response weight is the candidate center selected in the i-th heated surface tube, which is consistent with the weight used when fitting the oblique zone phase-shifting model. The coordinates of the measured tube screen corresponding to the candidate center are unfolded; The Euclidean norm is used; the fitting rule consistent with the slant band model is adopted, which can reduce the interference of the difference in fitting conditions on the model comparison results and ensure the objectivity and fairness of the verification results; the fitting algorithm can be the same as the above-mentioned Levenberg-Marquardt algorithm, and finally the optimal parameters of the in-phase model are obtained. The residual sum of squares is calculated for both the oblique-band phase-shifted model and the in-phase model. The residual sum of squares quantifies the model's fitting accuracy to the measured candidate center data; a smaller value indicates a better fit. The formula for calculating the residual sum of squares for the oblique-band phase-shifted model is as follows: , The formula for calculating the sum of squared residuals in an in-phase model is: , In the two formulas above, This represents the sum of squared residuals for the oblique-band phase-shifting model; This represents the sum of squared residuals for the in-phase model. The predicted coordinates of the fitted oblique-band phase-shifting model in the i-th pipe segment; The predicted coordinates of the i-th pipe segment of the fitted in-phase model are given, and the remaining parameters are consistent with the parameter definitions in the fitting objective function. The residuals are weighted by normalized response weights, which can ensure that the contribution of high-confidence defect responses to the residual calculation is higher and avoid the interference of low-amplitude pseudo-defect responses on the fitting accuracy evaluation. Based on the sum of squared residuals, the number of model parameters, and the number of candidate centers participating in the fitting, the Akaike Information Criterion (AIC) values ​​for the two models are calculated respectively. The Akaike Information Criterion can objectively evaluate the overall performance of the model while balancing model fitting accuracy and model complexity, avoiding overfitting misjudgments caused by the larger number of parameters in the oblique-band phase-shifting model. The formula for calculating the AIC value of the oblique-band phase-shifting model is as follows: , The formula for calculating the AIC value of the in-phase model is: , In the two formulas above, The information criterion value for the oblique-band phase misalignment model; This represents the information criterion value for the in-phase model; In this embodiment, the number of independent parameters for the slant-band phase-shifting model is [number missing]. ,correspond , , , Four parameters to be fitted; In this embodiment, the number of independent parameters is [number missing]. ,correspond , Two parameters to be fitted; This represents the number of candidate centers actually involved in a single model fitting; in a single oblique-band phase-shifting model fitting, each heated surface tube involved in the fitting selects at most one candidate center; when each heated surface tube involved in the fitting is selected into a candidate center... This is equal to the number of heated surface tubes participating in the fitting; when some heated surface tubes fail to extract effective abnormal response candidate regions or their candidate centers are identified as outliers... This represents the actual number of candidate centers retained and participating in the fitting process. In this embodiment, a total of 12 heated surface tubes within the effective blowing coverage area of ​​the sootblower were selected into the fitting region, and each heated surface tube retained one candidate center. Therefore... =12; When comparing the oblique-band out-of-phase model and the in-phase model using information criteria, both use the same candidate center set and the same response weights to ensure that the model comparison benchmark is consistent; The first term in this formula is used to penalize the complexity of the model. The more parameters there are, the larger the penalty term is. The second term is used to quantify the fitting error of the model. The larger the fitting error, the larger the value of the second term is, which can achieve a balance between fitting accuracy and model complexity. In scenarios with a small sample size or where stronger constraints on model complexity are required, the Bayesian Information Criterion (BIC), which has the same input data as the AIC, can be used to compare the skewed out-of-phase model and the in-phase model. When using BIC, the skewed out-of-phase model and the in-phase model still use the same candidate center set, the same response weights, and the same residual sum of squares calculation method, only replacing the complexity penalty term of the information criterion with the complexity penalty term corresponding to BIC. Regardless of whether AIC or BIC is used, the model with the smaller information criterion value is the final model adopted.

[0034] The validity of the execution model and the final model determination are performed when the AIC value of the oblique-band phase-shifting model is less than that of the in-phase model. This indicates that, under the premise of balancing fitting accuracy and model complexity, the oblique-band phase-misalignment model has better overall fitting performance for measured defect data than the in-phase model. It can be determined that the target pipe section region has oblique-band phase-misalignment characteristics at the hot-state jetting boundary. In this case, the aforementioned fitted oblique-band phase-misalignment model is used as the final hot-state jetting boundary oblique-band phase-misalignment model for subsequent defect 3D localization and attribution correction. If this condition is not met... If the judgment condition is met, it indicates that there are no significant oblique phase misalignment features in the target area. The cold soot blowing boundary benchmark corresponding to the same phase model is used as the reference benchmark for defect location, so as to achieve objective identification of oblique phase misalignment features and adaptive selection of the model.

[0035] In one embodiment of the present invention, determining the three-dimensional location point, attribution distance, and normalized coordinates of the oblique band of the defect based on the hot-state blowing boundary oblique phase misalignment model and the cold-state soot blowing boundary reference coordinates includes: Based on the final determined hot-state jetting boundary band misalignment model, the misalignment offset of the hot-state band relative to the cold-state soot blowing boundary is calculated. For the i-th heated surface tube, the circumferential reference angle in the previously determined cold-state soot blowing boundary reference coordinates is first converted into the circumferential arc length, and then the misalignment offset is calculated. The calculation formula is as follows: , In the formula, This is the phase shift corresponding to the i-th heated surface tube; The axial phase shift is the deviation of the hot-state jet boundary from the cold-state design boundary. The circumferential arc length phase shift represents the circumferential positional deviation of the hot-state jetting boundary relative to the cold-state design boundary. The predicted axial position of the hot-state jetting boundary oblique zone phase-shifting model on the i-th heated surface tube; The position of the circumferential arc length predicted by the model; Let be the axial reference coordinate of the cold soot blowing boundary of the i-th heated surface tube; This refers to the circumferential reference arc length of the cold-state soot blowing boundary. The outer radius of the i-th heated surface tube is consistent with the tube diameter parameter during the construction of the three-dimensional surface model. The phase misalignment offset calculated by this formula can intuitively quantify the actual offset of the blowing boundary relative to the cold design value under hot conditions, and at the same time provide an offset benchmark for defect attribution correction. The phase misalignment offset is used to describe the offset of the equivalent blowing boundary during operation relative to the cold soot blowing boundary benchmark coordinate obtained by inverting the defect distribution detected during shutdown. It does not indicate that the tube screen still maintains the hot geometric position under shutdown conditions. Through this definition, the positional trend of blowing damage during operation can be correlated with the detection response data obtained during shutdown maintenance, thereby providing a unified coordinate benchmark for subsequent three-dimensional defect localization and oblique band attribution analysis.

[0036] Based on the axial and circumferential weighted centers of the candidate regions for abnormal responses, three-dimensional defect location points are generated in the three-dimensional surface model. For the j-th candidate region for abnormal response on the i-th heated surface tube, its axial and circumferential weighted centers are substituted into the parameterized equation of the three-dimensional surface of the tube wall to obtain the three-dimensional defect location points. The calculation formula is as follows: , Right now: , In the formula, Let J be the three-dimensional location point of the defect corresponding to the j-th abnormal response candidate region on the i-th heated surface tube; The mapping benchmark between the three-dimensional surface parameterization model of the i-th heated surface tube wall constructed above and the three-dimensional positioning point of the defect is unified; The three-dimensional coordinates of the pipe centerline corresponding to the axial weighted center of the defect; This is the principal normal unit vector of the tube centerline at this axial position; The subnormal unit vector of the pipe centerline at this axial position is the same as the boiler furnace general coordinate system used in the construction of the 3D model. The axial and circumferential weighted center is used as the mapping reference, rather than the geometric center of the abnormal area, because the weighted center uses the detection response amplitude as the weight, which can more accurately characterize the core position of the defect and avoid the influence of low amplitude noise at the edge of the area on the 3D positioning accuracy. Through this parameterized mapping, the abnormal area in the 2D pipe wall unfolded image can be uniquely mapped to the actual pipe wall position in the 3D space of the boiler furnace, realizing the 3D spatial positioning of the defect.

[0037] Calculate the attribution distance between the candidate region of the abnormal response and the phase-shifted model of the hot jet boundary band; for the j-th candidate region of the abnormal response on the i-th heated surface tube, first convert its candidate center into the tube screen unfolded coordinates, then calculate the Euclidean distance between it and the predicted coordinates of the hot jet boundary band model on the same tube segment to obtain the attribution distance. The calculation formula is as follows: , In the formula, is the affiliation distance between the j-th abnormal response candidate region on the i-th heated surface tube and the phase-shifted model of the hot-state jet boundary band. Let J be the tube-screen expansion coordinates of the j-th candidate center of the abnormal response on the i-th heated surface tube; The predicted tube-screen expansion coordinates on the i-th heated surface tube are given by the hot-state jetting boundary oblique zone phase-shifting model. The distance is the Euclidean norm; this attribution distance can directly quantify the spatial proximity between the core location of the defect and the hot blowing boundary. The smaller the distance, the higher the correlation between the defect and the pipe wall damage caused by the sootblower. This provides a quantitative basis for defect cause analysis and pipe segment attribution correction, and avoids misjudging the continuous oblique defect group caused by sootblowing as an isolated random defect.

[0038] Based on the axial and circumferential phase-shifting components of the hot-state jetting boundary oblique zone misalignment model, the oblique zone direction vector is formed, and the three-dimensional oblique zone direction vector is constructed. The expression is: , In the formula, the first component 1 corresponds to the unit increment of the pipe segment sequence, which represents the extension direction of the oblique zone along the pipe segment sequence arrangement. The axial phase-shifting component of the hot-state jetting boundary oblique zone misalignment model represents the axial offset of the oblique zone boundary when the pipe segment sequence increases by 1 level. The circumferential phase-shifting component of the model represents the circumferential arc length offset of the oblique zone boundary when the pipe segment sequence increases by one level. Both components are consistent with the model parameters obtained by the aforementioned fitting. This three-dimensional direction vector can completely represent the overall extension trend of the oblique zone in the three-dimensional pipe screen parameter space composed of pipe segment sequence, axial position, and circumferential arc length. It provides a unified directional reference for the subsequent projection calculation of the relative position of defects, ensuring the uniqueness and stability of the oblique zone's normalized coordinates.

[0039] The pipe segment sequence, axial weighted center, and circumferential arc length of the candidate abnormal response region are used to form the three-dimensional pipe screen parameter points; the expression for the three-dimensional pipe screen parameter points for the j-th candidate abnormal response region on the i-th heated surface pipe is: , In the formula, Let i be the three-dimensional tube screen parameter point corresponding to the j-th abnormal response candidate region on the i-th heated surface tube; i is the arrangement sequence number of the tube segment where the abnormal region is located, which is consistent with the sequence numbering rule when fitting the oblique band model. This is the axial weighting center of the anomalous region; This represents the circumferential arc length position corresponding to the abnormal area. This three-dimensional pipe screen parameter point combines the single-pipe position information of the defect with the overall layer sequence information of the pipe screen, which can completely characterize the spatial position of the defect in the entire pipe screen oblique zone distribution, rather than being limited to the relative position within a single pipe, providing a unified parameter benchmark for calculating the relative position of the oblique zone across pipe segments. Projecting the 3D tube screen parameter points onto the inclined band direction vector yields the inclined band normalized coordinates; for the j-th abnormal response candidate region on the i-th heated surface tube, the formula for calculating the inclined band normalized coordinates is: , in: , In the formula, Let be the slant normalized coordinate of the j-th abnormal response candidate region on the i-th heated surface tube; The predicted three-dimensional pipe screen parameter point at the starting pipe segment i=1 of the hot-state jetting boundary oblique zone phase-shifting model is used as the starting reference point for the oblique zone normalized coordinates. This is the position offset vector of the defect parameter point relative to the starting point of the slope; the superscript T is the vector transpose operator. The square of the magnitude of the oblique band direction vector; To avoid extremely small positive numbers with a denominator of 0, this formula uses vector projection to convert the absolute position of the defect in the three-dimensional tube screen space into a one-dimensional relative coordinate along the extension direction of the inclined band at the hot-state spray boundary. This normalized coordinate does not use a single tube segment number or a single axial absolute distance as a comparison benchmark, but rather characterizes the relative position of the defect within the inclined band distribution. It exhibits stability in multi-cycle detection and can serve as a unified alignment benchmark for cross-cycle defect tracking, effectively avoiding misjudging changes in tube screen attitude or differences in scanning position as defect expansion or migration. In scenarios requiring lateral comparison of inclined bands of different lengths, the normalized coordinate range corresponding to the entire length of the inclined band can be linearly mapped to the 0-1 interval.

[0040] In one embodiment of the present invention, based on the three-dimensional defect location point, the attribution distance, the oblique zone normalized coordinates, and the three-dimensional surface model of the heated surface tube bank, a three-dimensional visualized defect location result is generated, including: The predicted axial position and predicted circumferential angle of the hot-state jetting boundary band misalignment model on each pipe segment are mapped to the three-dimensional surface model of the heated surface pipe array constructed in the previous steps, generating the corresponding hot-state jetting boundary band annotation points for each pipe segment. For the i-th heated surface pipe, the calculation formula for the band annotation point is: , Right now: , In the formula, The three-dimensional annotation point of the hot-state jet boundary slope zone on the surface of the i-th heated surface tube; The parameterized model of the three-dimensional surface of the i-th heated pipe wall constructed above is consistent with the mapping benchmark of the three-dimensional positioning point of the defect. This ensures that the oblique label points fit the pipe wall surface and avoids spatial misalignment during the visualization process. The predicted axial position of the hot-state jetting boundary oblique zone phase-shifting model on the i-th heated surface tube; To correspond to the predicted circumferential angle, both are consistent with the parameters of the aforementioned fitted oblique phase fault model; To predict the three-dimensional coordinates of the tube centerline corresponding to the axial position; Let be the outer radius of the i-th heated surface tube; This is the principal normal unit vector of the tube centerline at this axial position; This is the secondary normal unit vector of the tube centerline at this axial position. All coordinate systems adopt the boiler furnace general three-dimensional coordinate system consistent with the above. According to the pipe segment arrangement sequence i from small to large, the marked points of the hot-state blowing boundary slope of adjacent pipe segments are connected sequentially to generate the hot-state blowing boundary slope. The connection method can be selected as a straight line connection or a curve fitting method according to the visualization requirements. The former has higher calculation efficiency, while the latter can make the slope appear to have a smoother and continuous transition effect. In order to more intuitively present the coverage range of the sootblower blowing boundary, the circumferential width of the hot-state blowing boundary slope can also be set according to the effective coverage range of the sootblower jet, and the single-line slope can be converted into a three-dimensional strip model that fits the pipe wall surface. The circumferential width is determined according to the sootblower nozzle structure, spray coverage range, pipe diameter and pipe screen pitch.

[0041] Based on the aforementioned generated 3D defect location points, the defect positions are displayed in the 3D surface model of the heated pipe bank. The displayed size of the defect can be determined based on the actual physical area of ​​the abnormal response candidate region, or it can be set according to the detection response amplitude. The correlation strength between the defect and the hot-state jet boundary slope is generated based on the previously calculated attribution distance. The formula for calculating the correlation strength is: , In the formula, The correlation strength between the j-th abnormal response candidate region on the i-th heated surface tube and the inclined band of the hot jet boundary; Let be the assigned distance between the j-th abnormal response candidate region on the i-th heated surface tube and the hot-blown boundary slope. The reason for using this mapping method is that it can convert the assigned distance into a continuous correlation strength between 0 and 1, and the smaller the assigned distance, the greater the correlation strength. When the assigned distance is 0, the correlation strength reaches a maximum of 1. As the assigned distance increases, the correlation strength gradually decreases and approaches 0. It can intuitively express the spatial proximity between the defect and the hot-blown boundary slope. If other monotonically decreasing mapping methods are used to calculate the correlation strength, it should be ensured that the mapping method satisfies the monotonic relationship that the smaller the assigned distance, the greater the correlation strength, and that all defects in the same inspection batch use the same mapping rule.

[0042] Color mapping rules can be used to visualize defects. For example, defects with a correlation strength of 0.8 to 1.0 can be rendered as red, those with a correlation strength of 0.5 to 0.8 as orange, those with a correlation strength of 0.3 to 0.5 as yellow, and those with a correlation strength below 0.3 as blue. This allows maintenance personnel to intuitively distinguish the degree of correlation between defects and damage caused by sootblowing. The above color grading is only used for visualization in this embodiment and is not a necessary limitation on the defect location results. The color grading boundary can be determined according to the distribution range of correlation strength within the same inspection batch, or according to the number of risk levels that maintenance personnel need to display, by dividing the data into equal intervals or quantiles. Regardless of the color grading method used, the numerical results of the defect three-dimensional location point, attribution distance, correlation strength, and oblique band normalized coordinates will not change due to the change in the color grading method. The output includes 3D visualized defect location results, including the defect's 3D location points, the hot-state jetting boundary slope, the phase shift, the attribution distance, the association strength, and the slope's normalized coordinates. The visualization results can be presented through an industrial 3D visualization engine or a power plant equipment 3D management platform, supporting switching, rotation, scaling, and sectioning operations for the entire pipe panel and individual pipes. It enables defect location and information retrieval, and simultaneously outputs a structured defect parameter list and a 3D model file. The structured defect parameter list can be stored in a common table format, containing pipe segment number, defect axial position, defect circumferential angle, defect 3D spatial coordinates, the phase shift of the hot-state slope relative to the cold-state boundary, the attribution distance between the defect and the slope, the association strength, and the slope's normalized coordinates. The 3D model file can be stored in a 3D model format capable of representing the pipe panel model, the hot-state jetting boundary slope, and defect location annotations, including the heated surface pipe panel 3D model, the hot-state jetting boundary slope model, and the defect location annotation model.

[0043] Based on the historical inspection data from three consecutive unit maintenance cycles, this embodiment simultaneously executes multi-cycle defect alignment and propagation rate analysis steps. The three inspection cycles correspond to the first, second, and third shutdown maintenance after the unit was put into operation, respectively. In each inspection cycle, the inspection and parameter calculation of the heating surface tube bank within the same sootblower coverage area were completed according to the aforementioned steps. The specific implementation process is as follows: First, a unified benchmark for multi-cycle defect alignment is established. For the k-th inspection cycle, In this embodiment, K=3, and the detection completion time for this cycle is denoted as . The cumulative number of days corresponding to the maintenance completion date is used for measurement, with the first maintenance completion date as the zero point to ensure a consistent benchmark for time difference calculation. Simultaneously, for the hot-state jetting boundary band misalignment model fitted for each inspection cycle, the normalized coordinates of all defects within that cycle are standardized, linearly mapping the normalized coordinate range corresponding to the entire length of the band to the interval between 0 and 1. The formula for calculating the normalized coordinates of the standardized band is: , In the formula, The normalized slant band coordinates are the coordinates of the m-th defect in the k-th detection cycle. The original normalized coordinate of the oblique band for the m-th defect in the k-th detection cycle is consistent with the definition of the oblique band normalized coordinate calculated above. The original normalized coordinates are the starting points of the hot-state jetting boundary slope zone in the kth detection cycle. The original normalized coordinates corresponding to the termination point of the slope band in this cycle; To avoid extremely small positive numbers with a denominator of 0, this standardization process is adopted because the coverage area of ​​the sootblower and the number of pipe segments involved in the fitting of the tube screen may differ in different detection cycles, resulting in different ranges of the original oblique band normalized coordinates. After standardization to the 0 to 1 range, the impact of the oblique band length difference in different cycles on the alignment results can be reduced, so that the oblique band normalized coordinates in different cycles have uniform comparability, providing a stable benchmark for defect alignment. Based on the standardized normalized coordinates of the inclined band, defects in different detection cycles are spatially aligned using a nearest neighbor bidirectional matching rule. Specifically: for the m-th defect in the k-th cycle, among all defects in the (k+1)-th cycle, the defect with the smallest absolute value of the difference in the standardized normalized coordinates of the inclined band is selected as a candidate matching defect; when the absolute value of the difference in the standardized normalized coordinates of two defects is less than or equal to the matching threshold, and they are nearest neighbors in bidirectional matching, the two defects are determined to be corresponding defects of the same sootblowing damage location in different detection cycles and are grouped into the same defect matching group; for defects that appear only in a single cycle, they are marked as new defects or repaired defects and are not included in the defect propagation rate calculation. The reason for adopting this matching rule is that the relative position of pipe wall damage defects caused by sootblowing along the inclined band of the hot blowing boundary has long-term stability and is not affected by the thermal expansion deformation of the pipe screen, the difference in the scanning starting point, or the fine adjustment of the sootblower installation posture. Compared with the traditional alignment method based on pipe segment number and axial absolute position, this method can effectively avoid defect mismatch caused by changes in pipe screen posture and scanning reference offset, and improve the accuracy of cross-cycle defect tracking. The matching threshold is not limited to a fixed value; its selection is based on the comprehensive results of the calculation error of the normalized coordinates of the oblique band, the spatial resolution of the detection equipment, and the modeling error of the tube screen in different detection cycles. The matching threshold should be greater than the range of oblique band normalized coordinate fluctuations generated by the same defect in repeated detections, and less than the minimum resolvable normalized distance between adjacent independent defects in the oblique band direction. When there are repeated detection samples, the matching threshold can be determined based on the deviation of the oblique band normalized coordinates of the same defect in the repeated detection samples; when there are no repeated detection samples, the matching threshold can be calculated based on the axial pixel spacing, circumferential pixel spacing, and the total length of the oblique band in the tube wall unfolding response image. In this embodiment, the matching threshold can be preferably set to 0.02 to 0.05 according to the detection resolution and the tube screen pitch, corresponding to a relative distance of 2% to 5% of the total length of the oblique band.

[0044] For the same defect matching group that has been matched, the area of ​​the candidate region for abnormal response and the detection time for each detection cycle within the group are obtained. The area of ​​the candidate region for abnormal response is the actual physical area of ​​the defect on the pipe wall surface calculated above, consistent with the area calculation rules in the previous steps, and can reflect the overall coverage of the defect. Subsequently, based on the difference in the area of ​​the candidate region for abnormal response and the difference in the detection time between adjacent detection cycles, the defect propagation rate of the defect between adjacent cycles is calculated using the following formula: , In the formula, The defect propagation rate in the k-th detection cycle relative to the (k-1)-th detection cycle; Let be the response area of ​​the defect in the k-th detection cycle; The response area of ​​the same matching defect in the (k-1)th detection cycle; The time interval between two adjacent detection cycles is used. The reason for using this formula to calculate the expansion rate is that it can quantify the average expansion speed of defects during unit operation. The time dimension uses the actual number of natural days of unit operation, which can reflect the development trend of defects during unit operation under load. To avoid interference from outliers, when the area difference between adjacent cycles is negative, it is determined that the defect was repaired by grinding or caused by detection error, and is marked as abnormal data, not included in the expansion rate calculation. When no effective operating cycle has elapsed between two detections, it is treated as a repeated detection; when an effective operating cycle has elapsed between two detections, it is treated as a cross-cycle detection and included in the defect expansion rate calculation. The time threshold is used to distinguish between repeated detections within the same maintenance cycle and cross-maintenance cycle detections. Its value is not limited to a fixed number of days, but is determined based on the unit maintenance records, detection batch number, and unit operating interval. In this embodiment, when the time interval is less than a preset threshold, it can be determined as a repeated detection within the same maintenance cycle and is not included in the expansion rate calculation. This preset threshold can be preferably set to 30 days according to the unit maintenance schedule. When the detection equipment can stably output other defect scale characterization quantities, the same defect scale characterization quantity as the output of the same detection equipment can also be used to calculate the rate of change, but multiple detection cycles of the same defect matching group should use the same scale characterization quantity.

[0045] The calculated defect propagation rate is incorporated into the 3D visualization defect location results. Specifically, in the structured defect parameter list, adjacent cycle propagation rate and cumulative average propagation rate fields are added for each defect, and stored synchronously with parameters such as defect 3D location points, normalized coordinates of the oblique band, and correlation strength. In the 3D visualization scene, color mapping rules can be used to render defects; for example, defects with higher propagation rates are displayed in a more prominent color, while defects with lower propagation rates or those that have been repaired are displayed in different colors. The above propagation rate color grading is only for display purposes and does not change the numerical results of defect 3D location points, normalized coordinates of the oblique band, correlation strength, and propagation rate. Based on the defect location and propagation data from multiple cycles, a time-series animation of defect evolution can also be generated, intuitively presenting the development process of blowdown damage defects with each detection cycle. The final output includes a three-dimensional visualization of defect location results containing the defect propagation rate. This is used to display the spatial location of defects on the same hot-state jetting boundary slope in the three-dimensional surface model of the heated pipe bank and the trend of their change with the detection cycle. This provides data display basis for maintenance personnel to view the spatial distribution of defects, confirm the corresponding pipe sections of defects, and observe the changes in defects.

[0046] like Figure 2 The figure shown is a structural diagram of a defect localization system based on three-dimensional visualization for detecting heated surface tubes, according to an embodiment of the present invention.

[0047] In this embodiment, the functions of each module / unit are as follows: The model baseline construction module is used to acquire the structural data of the heated surface tube bank and the cold installation data of the sootblower nozzles, construct a three-dimensional surface model of the heated surface tube bank, and determine the baseline coordinates of the cold sootblowing boundary. The tube wall unfolding imaging module is used to acquire multi-channel detection response and scanning displacement, and to generate tube wall unfolding response image by mapping the multi-channel detection response to the scanning displacement. An abnormal region extraction module is used to extract candidate regions of abnormal response from the unfolded response image of the pipe wall and to calculate the weighted center of the candidate regions of abnormal response. The oblique-band phase misalignment identification module is used to identify the oblique-band phase misalignment model of the hot-state jetting boundary based on the axial and circumferential misalignment of the weighted center of adjacent pipe segments as the pipe segment sequence occurs. The defect 3D location module is used to determine the defect 3D location point, the assigned distance, and the normalized coordinates of the oblique band based on the hot-state jetting boundary oblique phase misalignment model and the cold-state soot blowing boundary reference coordinates. The visualization module is used to generate three-dimensional visualized defect location results based on the three-dimensional defect location points, the assigned distance, the oblique zone normalized coordinates, and the three-dimensional surface model of the heated surface tube array.

[0048] like Figure 3As shown, a multi-channel detection assembly 2 is installed on the outer side of the heated surface tube 1. The multi-channel detection assembly 2 is mounted on a scanning bracket 3, which is equipped with a detection control unit 4 and a follow-up encoder 5. The scanning bracket 3 moves relative to the heated surface tube 1 along the scanning direction 6, so that each detection channel of the multi-channel detection assembly 2 sequentially collects the detection response at different positions around the heated surface tube 1 according to the channel number. The detection response is combined with the scanning displacement recorded by the follow-up encoder 5 to form a correspondence between the detection response and the axial position. The collected multi-channel detection response is expanded by coordinates to form a tube wall expansion response image 7. The pipe wall unfolding response image 7 uses axial position as the horizontal coordinate and circumferential angle as the vertical coordinate, and grayscale depth to represent the normalized response intensity. In the pipe wall unfolding response image 7, abnormal response candidate regions are extracted based on the local abnormal regions with continuous distribution of response values. The response values ​​in the abnormal response candidate regions are weighted to obtain the corresponding weighting centers 8 and 9. The weighting centers are used to characterize the core position of the corresponding abnormal response candidate region in axial position and circumferential angle, and serve as input data for hot-state jetting boundary oblique zone phase misalignment model identification and defect three-dimensional localization.

[0049] Electronic device 400 can be a desktop computer, laptop, handheld computer, cloud server, or other electronic device. Electronic device 400 may include, but is not limited to, processor 401 and memory 402. Those skilled in the art will understand that... Figure 4 This is merely an example of electronic device 400 and does not constitute a limitation on electronic device 400. It may include more or fewer components than shown, or combine certain components, or different components. For example, electronic device may also include input / output devices, network access devices, buses, etc.

[0050] Processor 401 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0051] The memory 402 can be an internal storage unit of the electronic device 400, such as a hard disk or RAM of the electronic device 400. The memory 402 can also be an external storage device of the electronic device 400, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the electronic device 400. Furthermore, the memory 402 can include both internal and external storage units of the electronic device 400. The memory 402 is used to store the computer program 403 and other programs and data required by the electronic device. The memory 402 can also be used to temporarily store data that has been output or will be output.

[0052] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A defect localization method based on three-dimensional visualization of heated surface tube inspection, characterized in that, Includes the following steps: S1. Obtain the structural data of the heated surface tube bank and the cold installation data of the soot blower nozzle, construct a three-dimensional surface model of the heated surface tube bank, and determine the reference coordinates of the cold soot blowing boundary. S2. Obtain the multi-channel detection response and scanning displacement, match the multi-channel detection response with the scanning displacement, and generate a pipe wall unfolding response image; S3. Extract candidate regions of abnormal responses from the expanded response image of the pipe wall and calculate the weighted center of the candidate regions of abnormal responses. S4. Based on the axial and circumferential misalignment of the weighted center of adjacent pipe segments with the pipe segment sequence, identify the phase misalignment model of the hot-state jetting boundary oblique zone. S5. Determine the three-dimensional location point, the assigned distance, and the normalized coordinates of the oblique zone of the hot-state jetting boundary and the reference coordinates of the cold-state soot blowing boundary based on the hot-state jetting boundary oblique zone misalignment model and the cold-state soot blowing boundary reference coordinates. S6. Based on the three-dimensional defect location points, the assigned distance, the oblique zone normalized coordinates, and the three-dimensional surface model of the heated surface tube array, generate three-dimensional visualized defect location results.

2. The method according to claim 1, characterized in that, Construct a three-dimensional surface model of the heated tube bank and determine the reference coordinates of the cold soot blowing boundary, including: The heat-receiving surface tube bank structure data includes tube segment number, tube diameter, tube centerline, pitch and arrangement sequence; the sootblower nozzle cold installation data includes nozzle outlet reference point and spray direction. A three-dimensional surface model of each heated tube is established based on the tube coreline, tube diameter, and arrangement sequence. Determine the cold injection centerline based on the nozzle exit reference point and the injection direction; Calculate the nearest interaction position between the cold jet centerline and the three-dimensional surface model of each heated surface tube; The axial coordinates and circumferential angles corresponding to the most recent action position are used as the reference coordinates for the cold soot blowing boundary.

3. The method according to claim 1, characterized in that, Acquire multi-channel detection responses and scanning displacements, correlate the multi-channel detection responses with the scanning displacements, and generate a tube wall unfolding response image, including: The scanning displacement sequence recorded by the servo encoder is obtained, and the detection time is mapped to the axial position of the heated surface tube according to the scanning displacement sequence. Acquire the multi-channel detection response sequence and map each detection channel to the circumferential angle of the heated surface tube; Using the axial position and circumferential angle of the heated surface tube as image coordinates, and the multi-channel detection response value as the image grayscale value, a tube wall unfolding response image is formed; The response values ​​in the tube wall unfolding response image are normalized based on the median and median absolute deviation to obtain the normalized tube wall unfolding response image.

4. The method according to claim 1, characterized in that, Extract candidate regions of abnormal responses from the expanded response image of the pipe wall, and calculate the weighted centers of the candidate regions of abnormal responses, including: Candidate regions for abnormal responses are obtained from local abnormal regions with continuous distribution of response values ​​in the tube wall unfolding response image; The response weights are obtained by summing the response values ​​within each candidate region of an anomaly response. Calculate the area of ​​the candidate region for each candidate region of an anomaly response. Based on the response values ​​of each image location within the candidate region of abnormal response, the axial weighted center and the circumferential weighted center are calculated, and the axial weighted center and the circumferential weighted center are used as candidate centers for identifying the phase-shifting model of the hot jet boundary band.

5. The method according to claim 4, characterized in that, Based on the axial and circumferential misalignment of the weighted centers of adjacent pipe segments with the pipe segment sequence, a hot-state jetting boundary oblique zone phase misalignment model is identified, including: The candidate center is converted into a tube-screen unfolding coordinate system consisting of axial position and circumferential arc length position; For adjacent pipe sections near the sootblower's spray coverage boundary, an oblique-band phase-shifting model is established that varies with the pipe section sequence. The oblique-band phase-shifting model includes axial phase-shifting components and circumferential phase-shifting components. Candidate centers are selected from the candidate regions of abnormal responses of each heated surface tube, and the initial parameters of the oblique-band phase-shifting model are fitted with the goal of minimizing the distance error from the candidate center to the oblique-band phase-shifting model after weighting by response weight. A phase-in-phase model is established, which indicates that the candidate centers of adjacent pipe segments do not generate axial phase-out components and circumferential phase-out components with the pipe segment sequence. Calculate the sum of squared residuals for the skew-band out-of-phase model and the in-phase model, respectively; The information criterion values ​​for the oblique-band out-of-phase model and the in-phase model are calculated based on the sum of squared residuals, the number of model parameters, and the number of candidate centers involved in the fitting, respectively. When the information criterion value of the oblique zone phase-shifting model is less than that of the in-phase model, the oblique zone phase-shifting model is used as the hot-state jet boundary oblique zone phase-shifting model.

6. The method according to claim 4, characterized in that, Based on the hot-state blowing boundary oblique zone phase misalignment model and the cold-state soot blowing boundary reference coordinates, the three-dimensional location point, assigned distance, and oblique zone normalized coordinates of the defect are determined, including: Based on the predicted axial position and predicted circumferential arc length position of the hot-state jetting boundary band misalignment model on the corresponding pipe section, and the reference coordinates of the cold-state soot blowing boundary, the misalignment offset of the hot-state jetting band relative to the cold-state soot blowing boundary is calculated. Based on the axial weighted center and circumferential weighted center of the candidate region of abnormal response, three-dimensional location points of defects are generated in the three-dimensional surface model of the heated surface tube bank. Calculate the attribution distance between the candidate regions of the abnormal response and the phase-shifted model of the thermal jet boundary band; The oblique zone direction vector is formed based on the axial and circumferential phase-shifting components of the hot-state jetting boundary oblique zone phase-shifting model. The pipe segment sequence, axial weighting center, and circumferential arc length position of the candidate region of abnormal response are used to form three-dimensional pipe screen parameter points; The three-dimensional tube screen parameter points are projected onto the oblique band direction vector to obtain the oblique band normalized coordinates.

7. The method according to claim 6, characterized in that, Based on the 3D defect location points, attribution distances, oblique zone normalized coordinates, and the 3D surface model of the heated tube bank, 3D visualized defect location results are generated, including: The predicted axial position and predicted circumferential angle of the hot-state jetting boundary band phase-shifting model on each pipe segment are mapped to the three-dimensional surface model of the heated pipe row to generate the hot-state jetting boundary band marking points. Connect the marked points of the hot-state jetting boundary slope band of adjacent pipe sections to generate the hot-state jetting boundary slope band; The defect location is displayed based on the three-dimensional positioning points of the defect, and the correlation strength between the defect and the hot-blown boundary slope is generated based on the attribution distance. The smaller the attribution distance, the greater the correlation strength. The output includes a 3D visualization of the defect location, including the defect's 3D location points, the hot-blown boundary band, the phase shift, the attribution distance, the correlation strength, and the normalized coordinates of the band.

8. The method according to claim 7, characterized in that, In the case of multiple inspection cycles, defects in different inspection cycles are aligned according to the normalized coordinates of the oblique band. Obtain the area and detection time of the candidate region for abnormal response corresponding to the normalized coordinates of the same slope; obtain the defect propagation rate based on the difference in area and detection time of the candidate region for abnormal response between adjacent detection cycles. The defect propagation rate is incorporated into the 3D visualization defect location results.

9. A defect localization system based on three-dimensional visualization of heated surface tube inspection, used to execute the defect localization method based on three-dimensional visualization of heated surface tube inspection as described in any one of claims 1-8, characterized in that, include: The model baseline construction module is used to acquire the structural data of the heated surface tube bank and the cold installation data of the sootblower nozzles, construct a three-dimensional surface model of the heated surface tube bank, and determine the baseline coordinates of the cold sootblowing boundary. The tube wall unfolding imaging module is used to acquire multi-channel detection response and scanning displacement, and to generate tube wall unfolding response image by mapping the multi-channel detection response to the scanning displacement. An abnormal region extraction module is used to extract candidate regions of abnormal response from the unfolded response image of the pipe wall and to calculate the weighted center of the candidate regions of abnormal response. The oblique-band phase misalignment identification module is used to identify the oblique-band phase misalignment model of the hot-state jetting boundary based on the axial and circumferential misalignment of the weighted center of adjacent pipe segments as the pipe segment sequence occurs. The defect 3D location module is used to determine the defect 3D location point, the assigned distance, and the normalized coordinates of the oblique band based on the hot-state jetting boundary oblique phase misalignment model and the cold-state soot blowing boundary reference coordinates. The visualization module is used to generate three-dimensional visualized defect location results based on the three-dimensional defect location points, the assigned distance, the oblique zone normalized coordinates, and the three-dimensional surface model of the heated surface tube array.

10. An electronic device, characterized in that, include: One or more processors; A storage unit for storing one or more programs that, when executed by one or more processors, enable the one or more processors to implement the method according to any one of claims 1 to 8.

Images