Air-cooled island finned tube bundle micro-deformation detection method and system

Abstract

The application discloses an air-cooled island finned tube bundle micro-deformation detection method and system, and belongs to the technical field of object deformation measurement. The method comprises the following steps: based on the spatial position of a support member and a platform structure in observation point cloud data, and combined with detection equipment position and posture information and observation ray distribution data, wedge blind area range data is constructed; based on the observation point cloud data and reference geometric model data, the actual center line of each finned tube is extracted and compared with the corresponding reference center line to obtain geometric feature change data of the finned tube bundle; based on the geometric feature change data, micro-deformation index data of the finned tube bundle is calculated; based on the observation ray distribution data and the wedge blind area range data, directional observable reliability is generated; based on the wedge blind area range data and the directional observable reliability, directional compensation is performed on the micro-deformation index data; the application can cover the external visible range of the finned tube bundle and the accessible range of the wedge-shaped limited area, and improve the reliability and timeliness of micro-deformation monitoring.

Description

Technical Field

[0001] This invention relates to the field of object deformation measurement technology, and more specifically, to a method and system for detecting micro-deformation of air-cooled island finned tube bundles. Background Technology

[0002] As a crucial heat exchange device in energy production facilities, the Type A overhead cooling island comprises finned tube bundles, supporting components, and platform structures, forming a spatially complex heat exchange unit. During long-term operation, the finned tube bundles are susceptible to micro-deformation due to thermal expansion and contraction, airflow impact, and load variations. Therefore, it is necessary to inspect for slight lateral bending of the tube centerline, mid-span deflection of the support, and minor offsets of the tube bundle relative to the support in order to conduct equipment maintenance and condition assessment.

[0003] Currently, the inspection of air-cooled island finned tube bundles mostly employs optical non-contact measurement schemes, such as laser scanning, structured light imaging, multi-view vision reconstruction, and point cloud registration—three-dimensional contour measurement techniques. These methods acquire surface topography data, image data, or point cloud data of the target area and compare the observation results with a reference geometric model to identify and measure minute deformations of the finned tube bundle. These methods essentially rely on the geometric relationship between the observation ray, the imaging line of sight, and the target topography. The measurement accuracy is closely related to the accessibility of the optical line of sight, the integrity of the three-dimensional contour, and the conditions for topography calculation.

[0004] However, in the actual scenario of a Type A overhead cooling island, the supporting components and platform structure continuously obstruct the finned tube bundles, limiting optical observation to a restricted viewpoint in some areas. Furthermore, convergent obstruction regions easily form along the depth of the tube array. For these convergent obstruction regions, existing technologies typically handle them using a unified point cloud registration and deformation calculation method, failing to effectively differentiate between uneven ray distribution, degraded projection geometry, and missing local 3D contours under obstruction conditions. Especially when the observation rays are nearly orthogonal to the principal direction of the target micro-deformation, the projection values ​​of the relevant displacement components in optical measurements are too small, easily affected by noise, registration errors, and local contour extraction errors, leading to underestimated, missed, or spatially discontinuous micro-deformation. These problems further affect the accuracy and comparability of the finned tube bundle micro-deformation results, hindering subsequent maintenance judgments. Therefore, how to obtain accurate measurement results of finned tube bundle micro-deformation based on non-contact optical measurement under complex obstruction conditions has become an urgent technical problem to be solved. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this application aims to provide a method and system for detecting micro-deformation of air-cooled island finned tube bundles. By combining the position and attitude information of the detection equipment, observation point cloud data, observation ray distribution data, and the reference geometric model data of the finned tube bundle, wedge-shaped blind zone range data is constructed, the actual centerline is extracted, and geometric feature change data is generated. Furthermore, directional compensation is performed on the micro-deformation index data based on the directional observability reliability. This solves the problems of insufficient line-of-sight accessibility, uneven distribution of observation rays, local missing three-dimensional contours, and underestimation or missed detection of the main direction displacement component of the target micro-deformation due to the degradation of projection geometry and optical geometry during optical non-contact measurement under complex obstruction conditions. This enables accurate measurement of micro-deformation of air-cooled island finned tube bundles and improves the reliability, continuity, and comparability of the detection results.

[0006] To achieve the above objectives, this application adopts the following technical solution:

[0007] Firstly, this application provides a method for detecting micro-deformation of air-cooled island finned tube bundles, including:

[0008] The position and attitude information of the detection equipment, the reference geometric model data of the finned tube bundle, the observation point cloud data and the observation ray distribution data are acquired in advance;

[0009] Based on the spatial location of the supporting components and platform structure in the observation point cloud data, and combined with the position and attitude information of the detection equipment and the observation ray distribution data, the range data of the wedge-shaped blind zone is constructed.

[0010] Based on the observation point cloud data and the benchmark geometric model data, the actual centerline of each finned tube is extracted and compared with the corresponding benchmark centerline to obtain the geometric feature change data of the finned tube bundle.

[0011] Calculate the micro-deformation index data of finned tube bundles based on geometric feature change data;

[0012] The observability of the direction can be generated based on the observed ray distribution data and the wedge blind zone range data.

[0013] Oriented compensation is performed on the micro-deformation index data based on the wedge blind zone range data and the observability reliability of the direction.

[0014] Preferably, the method for constructing wedge-shaped blind zone range data includes:

[0015] Extract the spatial locations of the supporting components and platform structure in the observed point cloud data to form surface fragment representations of the supporting components and platform structure.

[0016] By combining the position and attitude information of the detection equipment with the distribution data of the observed rays, a spatial set of observed rays is constructed. Based on the surface fragment representation of the supporting component and the surface fragment representation of the platform structure, the spatial intersection relationship of the spatial set of observed rays is determined, and the reachable area and the occlusion-restricted area are obtained.

[0017] Based on the reachable region and the occlusion-restricted region, the spatial set of observed rays is spatially divided to obtain candidate regions for wedge-shaped blind zones;

[0018] Spatial connectivity processing is performed on the candidate regions of the wedge-shaped blind zone to obtain the range data of the wedge-shaped blind zone.

[0019] By transforming the occlusion effects caused by supporting components and platform structures into structured wedge-shaped blind zone range data, a clear spatial constraint basis can be provided for subsequent generation of directional observability reliability and directional compensation, reducing measurement deviations caused by mixing occluded and unoccluded areas.

[0020] Preferably, the method for obtaining geometric feature variation data of finned tube bundles includes:

[0021] Based on the finned tube bundle arrangement relationship in the benchmark geometric model data, a reference position sequence is generated and a local reference direction is established in the reference position sequence;

[0022] Based on the reference position sequence, a spatial envelope range coaxial with the baseline centerline is constructed in the observed point cloud data. Points within the spatial envelope range are selected to form a candidate point set. A cross-sectional sampling zone is constructed based on the reference position sequence and the local reference direction. Points in the candidate point set that fall into the cross-sectional sampling zone are used as the cross-sectional point set sequence.

[0023] Extract the cross-sectional center point sequence based on the cross-sectional point set sequence, and generate the actual centerline based on the cross-sectional center point sequence;

[0024] The actual centerline is compared with the reference centerline based on the reference position sequence. The centerline offset information and the connection position offset information are recorded to obtain the geometric feature change data of the finned tube bundle.

[0025] By establishing a tube-by-tube, position-by-position correspondence between point cloud measurement results and a reference geometric model, optical 3D contour information can be transformed into calculable centerline offset results, improving the accuracy of geometric change identification and the clarity of result attribution.

[0026] Preferably, the method for calculating the micro-deformation index data of the finned tube bundle includes:

[0027] Calculate the lateral micro-bending of the pipe centerline based on geometric feature variation data;

[0028] Calculate mid-span deflection of the support based on geometric feature variation data;

[0029] The minute offset of the tube bundle relative to the support is calculated based on the geometric feature change data;

[0030] The micro-deformation index data of the finned tube bundle are calculated by taking into account the lateral slight bending of the centerline of the converging tube, the mid-span deflection of the support, and the slight offset of the tube bundle relative to the support.

[0031] By further transforming centerline offset information into micro-deformation index data with engineering significance, the detection results can be transformed from geometric difference descriptions into quantitative results that can be directly used for equipment condition assessment.

[0032] Preferably, the method for calculating the transverse micro-bend of the pipe centerline includes:

[0033] Read the centerline offset information and extract the offset component sequence in the lateral direction. Process the offset component sequence in the lateral direction based on the cross-end constraints provided by the connection position offset information, and extract the lateral micro-bending characterization.

[0034] The maximum value of the transverse microbending characteristic in multiple spans of the same finned tube is selected as the transverse microbending amount of the tube centerline.

[0035] By introducing cross-end constraints provided by connection position offset information, the interference caused by overall translation on the identification of lateral micro-bending can be reduced, making the lateral micro-bending amount of the tube centerline more concentrated in reflecting the bending shape change of the finned tube body.

[0036] Preferably, the method for generating directional observability confidence includes:

[0037] A set of spatial location indexes for finned tube bundles is established based on the wedge-shaped blind zone range data, and the blind zone attribution information is recorded;

[0038] Extract the set of observation ray directions corresponding to each location in the spatial location index set based on the observation point cloud data;

[0039] The classification results of the reliability of directional observability are determined based on the set of observed ray directions and the information on blind zone attribution.

[0040] The ranking results of the observable confidence level at each location direction are aggregated into the micro-deformation index data to generate the observable confidence level at each direction.

[0041] By linking the distribution of observed ray directions and the attribution of blind zones with the main direction of target micro-deformation, it is possible to distinguish the differences in optical observability at different spatial locations, thus avoiding the distortion of results caused by the use of uniform credibility processing for data under different observation conditions in existing technologies.

[0042] Preferably, the method for performing targeted compensation on micro-deformation index data includes:

[0043] Based on the wedge-shaped blind zone range data, the mid-span position index and the connection position index in the micro-deformation index data are mapped to obtain the blind zone attribution label;

[0044] Based on the blind zone attribution markers, the micro-deformation index data are grouped to obtain wedge-shaped blind zone grouping and transition region grouping;

[0045] Based on the directional observability confidence level, the compensation objects are determined by grouping the wedge blind zone and the transition region, and the main directional components of the target micro-deformation are limited to obtain the location interval and compensation components of the compensation objects.

[0046] Based on the geometric feature change data, the reference sequence in the cross-pipe direction and the reference sequence along the pipe length direction are extracted in the location interval of the compensation object, and the compensation path is constructed based on the reference sequence;

[0047] The compensation components of the compensation object's location range are updated according to the compensation path to obtain the micro-deformation index data after compensation.

[0048] By targeting low-confidence components only in the wedge-shaped blind zone, we can reduce the problems of underestimation, missed detection, and spatial discontinuity of micro-deformation caused by limited line of sight, small projection components, and missing local contours.

[0049] Preferably, a mutation retention threshold is set during the directional compensation process for micro-deformation index data, and structural mutation protection is implemented, specifically including:

[0050] Based on the acquisition conditions of observation point cloud data and the discrete characteristics of geometric feature change data, a mutation retention threshold is set for each change sequence.

[0051] The structural mutation protection coverage is determined based on the mutation retention threshold and directional observability confidence, combined with wedge blind zone range data.

[0052] In targeted compensation, compensation rules are switched for the compensation components of the location interval of the compensation object based on the coverage of structural mutation protection.

[0053] By adding structural abrupt change protection during the directional compensation process, we can avoid mistreating real local geometric changes as component compression caused by observational degradation, thereby improving the consistency between the compensation results and the actual structural state.

[0054] Preferably, the method for obtaining the lateral direction is as follows:

[0055] The lateral direction is established for each finned tube based on the baseline geometric model data.

[0056] Secondly, this application provides a micro-deformation detection system for air-cooled island finned tube bundles, including:

[0057] The data acquisition module pre-acquires the position and attitude information of the detection equipment, the reference geometric model data of the finned tube bundle, the observation point cloud data, and the observation ray distribution data;

[0058] The blind zone construction module constructs wedge-shaped blind zone range data based on the spatial position of supporting components and platform structure in the observation point cloud data, and combines the position and attitude information of the detection equipment with the observation ray distribution data.

[0059] The feature analysis module extracts the actual centerline of each finned tube based on the observed point cloud data and the benchmark geometric model data, and compares it with the corresponding benchmark centerline to obtain the geometric feature change data of the finned tube bundle.

[0060] The deformation analysis module calculates the micro-deformation index data of the finned tube bundle based on the geometric feature change data;

[0061] The trusted computing module generates directional observability trust based on observed ray distribution data and wedge-shaped blind zone range data.

[0062] The directional compensation module performs directional compensation on the micro-deformation index data based on the wedge blind zone range data and the observability reliability of the direction.

[0063] Compared with the prior art, the beneficial effects achieved by this application are as follows:

[0064] This application does not perform unified registration and deformation calculation on observation point cloud data. Instead, it addresses the problems of limited line-of-sight accessibility, uneven distribution of observation rays, local missing three-dimensional contours, and the susceptibility of the displacement components of the main direction of target micro-deformation to the degradation of projection geometry in optical non-contact measurement under complex occlusion conditions. It constructs a continuous processing chain that includes wedge blind zone identification, actual centerline extraction, geometric feature change analysis, micro-deformation index calculation, direction observability reliability generation, and orientation compensation.

[0065] By constructing wedge-shaped blind zone range data based on the spatial positions of supporting components and platform structures in the observation point cloud data, combined with the position and attitude information of the detection equipment and the observation ray distribution data, complex occlusion effects can be transformed into calculable spatial constraints. By extracting the actual centerline of each finned tube based on the observation point cloud data and the reference geometric model data and comparing it with the corresponding reference centerline, the geometric feature change data of the finned tube bundle can be obtained, transforming the point cloud, imaging line of sight, and 3D contour measurement results into traceable geometric change results. By calculating the micro-deformation index data of the finned tube bundle based on the geometric feature change data, and generating directional observability reliability based on the observation ray distribution data and the wedge-shaped blind zone range data, directional compensation can be further performed on the micro-deformation index data. This allows for differentiated processing of data under different optical observation conditions, thereby reducing the problems of underestimated micro-deformation, missed detection, and spatial discontinuity, and improving the detection accuracy, result reliability, and engineering applicability of air-cooled island finned tube bundles in optical non-contact measurement scenarios. Attached Figure Description

[0066] Figure 1 This is a structural diagram of the air-cooled island finned tube bundle micro-deformation detection system according to an embodiment of the present invention;

[0067] Figure 2 This is a schematic diagram of the micro-deformation detection method for air-cooled island finned tube bundles according to an embodiment of the present invention;

[0068] Figure 3 This is a flowchart illustrating the construction of wedge-shaped blind zone range data in this invention;

[0069] Figure 4 This is a flowchart illustrating the geometric feature variation data of the finned tube bundle obtained in this invention. Detailed Implementation

[0070] The technical solutions of the embodiments of the present invention will be described in detail, clearly, and completely below with reference to the accompanying drawings. It should be particularly noted that the specific embodiments described below are only for better illustrating and explaining the technical solutions of the present invention, and are intended to enable those skilled in the art to better understand and implement the present invention, and should not be construed as limiting the scope of protection of the present invention. Without departing from the spirit and substance of the present invention, those skilled in the art can modify, adjust, or make equivalent substitutions based on the content disclosed in the present invention, and these should all be considered within the scope of protection of the present invention.

[0071] Example 1

[0072] Please see Figure 1As shown, this embodiment discloses a micro-deformation detection system for air-cooled island finned tube bundles, including a data acquisition module, a blind zone construction module, a feature analysis module, a deformation analysis module, a reliable calculation module, and a directional compensation module. Each module is connected via wired or wireless means to achieve data transmission.

[0073] The data acquisition module pre-acquires the position and attitude information of the detection equipment, the reference geometric model data of the finned tube bundle, the observation point cloud data of the finned tube bundle, and the observation ray distribution data of the finned tube bundle.

[0074] The position and attitude information of the detection equipment is used to characterize the spatial position and attitude of the detection equipment at the moment of point cloud acquisition, and establishes a correspondence with the acquisition time of the finned tube bundle observation point cloud data, so that each frame of finned tube bundle observation point cloud data corresponds to a set of detection equipment position and attitude information. The detection equipment is used to acquire the finned tube bundle observation point cloud data. The detection equipment includes any one of laser scanner, structured light camera and multi-view camera. The detection equipment generates finned tube bundle observation ray distribution data at the same time as generating the finned tube bundle observation point cloud data. The finned tube bundle observation ray distribution data is used to characterize the observation ray direction of each measuring point in the finned tube bundle observation point cloud data and the correlation between each measuring point and the position and attitude information of the detection equipment, and is consistent with the finned tube bundle observation point cloud data according to the measuring point index, so that each measuring point in the finned tube bundle observation point cloud data has an observation ray direction record corresponding to the measuring point index and a detection equipment position and attitude information record corresponding to the acquisition time.

[0075] When acquiring point cloud data of finned tube bundles using a laser scanner, the direction of the observation ray is determined by the spatial connection between the laser scanner's emission center position and the corresponding measurement point's coordinates. The emission center position is determined by the position and attitude information of the detection equipment at the time of acquisition. The point coordinates are calculated from the laser scanning range and scanning angle and recorded in the same coordinate system, ensuring that the direction of the observation ray reflects the observation path direction from the detection equipment to the measurement point. When acquiring point cloud data of finned tube bundles using a structured light camera or multi-view camera, the camera's extrinsic parameters are first determined by the detection equipment's position and attitude information, and the camera's imaging model is determined by the camera calibration information. Then, the depth reconstruction process maps pixel observations to point coordinates and establishes a correspondence with the point cloud measurement point index. The direction of the observation ray is determined by the spatial connection between the camera's emission center position and the corresponding measurement point's coordinates. The spatial connection between the coordinates of the measurement points is determined, and the position of the camera optical center is determined by the position and attitude information of the detection equipment corresponding to the acquisition time, so that the direction of the observed ray can correspond to the point coordinates in the same coordinate expression. The ray distribution data of the finned tube bundle observation uses the measurement point index as the primary key to record the direction of the observed ray and records the acquisition time identifier corresponding to the measurement point index, thereby connecting with the acquisition time correspondence of the detection equipment position and attitude information. This ensures that any measurement point can be traced back to the detection equipment position and attitude information when the finned tube bundle observation ray distribution data was generated, and the corresponding observation ray direction can be obtained. By completing the establishment of the acquisition time correspondence and the measurement point index correspondence at the data source end, the finned tube bundle observation point cloud data carries traceable observation geometric information, which facilitates subsequent coordinate unification and geometric degradation identification.

[0076] The finned tube bundle reference geometric model data includes the original design geometric dimensions and arrangement of the air-cooled island finned tube bundle and its support structure, serving as a reference for micro-deformation judgment. Based on the finned tube bundle reference geometric model data, the data is analyzed to form finned tube bundle reference data. The analysis process includes: determining the geometric expression of the centerline of each finned tube, the position of the centerline endpoint, the direction of the centerline, and the ideal position of the support point in the finned tube bundle reference geometric model data; and recording the spatial position of the support components and the platform structure as constraint boundaries, so that the finned tube bundle reference reference data simultaneously includes the reference centerline information of the finned tube bundle and the geometric constraint information of the support connection in the same coordinate system.

[0077] Considering that the outer edge of the fins forms multiple contours in the point cloud data of finned tube bundle observations, the finned tube bundle reference data, while recording the reference centerline information, further establishes a reference position sequence for each finned tube along the centerline direction, and records a local reference direction consistent with the centerline direction in the reference position sequence. The reference position sequence and local reference direction are used to limit the cross-sectional sampling zone direction when extracting the centerline from the finned tube bundle observation point cloud data. The finned tube bundle reference data also records the spatial envelope range coaxial with the centerline. The spatial envelope range covers the point cloud distribution area that may appear on the outer surface of the finned tube and the outer contour of the fin. Within the spatial envelope range, it distinguishes the screening constraints between tube body candidate points and fin outer edge points, so that tube body candidate points can be preferentially used in the subsequent cross-sectional sampling zone and the influence of fin outer edge points on the centerline extraction can be suppressed. In the case of point cloud missing in the cross-sectional sampling zone, the finned tube bundle reference data records the continuity relationship of adjacent reference position sequences and the boundary constraints of the centerline endpoint positions, which are used to maintain the spatial continuity of the centerline in the missing section and keep it corresponding to the ideal position of the support point.

[0078] The aforementioned finned tube bundle reference data can provide a unified geometric reference for the registration and deviation calculation of subsequent observation data, and provide cross-sectional sampling zone, spatial envelope range and fin outer edge point suppression constraints in the centerline extraction stage, reducing the interference caused by the multi-layer profile of the fin outer contour on the actual centerline extraction, and avoiding systematic errors introduced by inconsistencies in reference between different data sources.

[0079] The finned tube bundle observation point cloud data is obtained by scanning the finned tube bundle area using detection equipment. The finned tube bundle observation point cloud data includes external point cloud data and supplementary point cloud data for the wedge-shaped occlusion-restricted area of ​​the finned tube bundle. The external point cloud data reflects the three-dimensional surface morphology and spatial position of adjacent supporting components and platform structures within the visible range of the finned tube bundle. The supplementary point cloud data for the wedge-shaped occlusion-restricted area reflects the local three-dimensional geometric information within the wedge-shaped occlusion-restricted area formed by the occlusion of supporting components and platform structures along the tube bundle depth direction. The wedge-shaped occlusion-restricted area refers to the poor observability area where the observation ray is nearly orthogonal to the main direction of the target micro-deformation, causing the displacement component in that direction to be easily affected by geometric degradation. The supplementary point cloud data for the wedge-shaped occlusion-restricted area is obtained by the detection equipment changing the observation angle under multi-station conditions and performing supplementary scanning along the reachable path of the wedge-shaped occlusion-restricted area to obtain the three-dimensional point cloud distribution of reachable measurement points within the wedge-shaped occlusion-restricted area and form a stitchable overlapping area with the external point cloud data of the finned tube bundle.

[0080] Preprocessing and preliminary registration are performed on the external point cloud data of the finned tube bundle. Preprocessing and preliminary registration use the finned tube bundle reference data as geometric constraints, the position and attitude information of the detection equipment as the coordinate transformation basis, and the observed ray distribution data of the finned tube bundle as the observation direction maintenance basis. The external point cloud data of the finned tube bundle is transformed to a coordinate system consistent with the finned tube bundle reference data, and the observed ray distribution data of the finned tube bundle is simultaneously transformed to the same coordinate system to maintain a consistent correspondence between the observed ray direction and the point cloud coordinates. After completing the coordinate unification, the external point cloud data of the finned tube bundle is identified... The surface points of the finned tube bundle are identified and background points unrelated to the finned tube bundle are removed. Then, the point cloud is partitioned according to the spatial position of each finned tube in the finned tube bundle reference data, so that each partition establishes a matching relationship with the corresponding finned tube. Based on the matching relationship, the external point cloud data of the finned tube bundle is globally aligned with the finned tube bundle reference data. Through the above processing, the external point cloud data of the finned tube bundle has tube-level index and unified coordinate reference before entering the subsequent fusion stage, which facilitates the splicing with the supplementary point cloud data of the wedge-shaped occlusion restricted area of ​​the finned tube bundle and reduces matching ambiguity during subsequent calculation.

[0081] After initial registration of the external point cloud data of the finned tube bundle, the wedge-shaped blind zone point cloud data of the finned tube bundle is fused to improve observation coverage. The fusion process includes: transforming the wedge-shaped blind zone point cloud data of the finned tube bundle to a coordinate system consistent with the reference data of the finned tube bundle based on the position and attitude information of the detection equipment; establishing a geometric correspondence between the wedge-shaped blind zone point cloud data and the external point cloud data of the finned tube bundle in the overlapping area near the boundary of the wedge-shaped blind zone; and performing stitching and alignment of the wedge-shaped blind zone point cloud data of the finned tube bundle under the constraints of the geometric correspondence, so that the wedge-shaped blind zone point cloud data of the finned tube bundle can form a continuous spatial coverage with the registered external point cloud data of the finned tube bundle; simultaneously, the ray distribution data of the finned tube bundle observation is fused along with the wedge-shaped blind zone point cloud data of the finned tube bundle according to the measurement point index. The data is then written into the fusion result to ensure that the fused point cloud still retains the observation ray direction information corresponding to each measurement point. After fusion, complete finned tube bundle observation point cloud data and complete observation ray distribution data are obtained. The complete finned tube bundle observation point cloud data includes supplementary observation information of the wedge-shaped blind zone region and has been aligned to a coordinate system consistent with the finned tube bundle reference data. The complete observation ray distribution data and the complete finned tube bundle observation point cloud data correspond one-to-one according to the measurement point index. Through the above processing, the geometric information of the wedge-shaped blind zone region can be entered into the subsequent solution link in a computable manner, and the observation ray direction, a key basis for identifying unilateral squint geometric degradation, can be retained, thus providing a data foundation for the subsequent observability analysis and correction of micro-deformation components.

[0082] The blind zone construction module constructs wedge-shaped blind zone range data based on the spatial position of supporting components and platform structure in the observation point cloud data, combined with the position and attitude information of the detection equipment and the distribution data of the observation rays.

[0083] Please see Figure 3 As shown, in order to obtain wedge-shaped blind zone range data that reflects the geometric characteristics of the A-type overhead cooling island obstruction, a calculable observation accessibility description is established based on the spatial positions of the supporting components and platform structure in the observation point cloud data, combined with the position and attitude information of the detection equipment and the observation ray distribution data. The specific implementation method can be carried out according to the following steps:

[0084] The process involves pre-acquiring the position and attitude information of the detection equipment, observation point cloud data, and observation ray distribution data, and establishing a temporal and measurement point correspondence among these three. The detection equipment position and attitude information is recorded according to the acquisition time of the observation point cloud data, ensuring that each frame of observation point cloud data corresponds to a set of detection equipment position and attitude information. The observation ray distribution data is recorded according to the measurement point index of the observation point cloud data, ensuring that each measurement point in the observation point cloud data corresponds to an observation ray direction. Furthermore, the observation ray direction is correlated with the detection equipment position and attitude information at the corresponding acquisition time, characterizing the spatial relationship between the observation ray emitted from the detection equipment and reaching the target surface at each measurement point. After establishing the correspondence, based on the detection equipment position and attitude information, the observation point cloud data and observation ray distribution data are unified under the same coordinate system, ensuring that subsequent descriptions of occlusion relationships can be performed under the same spatial reference. These steps enable the observation point cloud data to carry clear observation geometric information and possess traceable spatial references, providing a foundation for subsequent identification of supporting components, platform structures, and their occlusion effects within the same geometric framework.

[0085] Based on the observation point cloud data, the spatial positions of the supporting components and platform structure are extracted from the observation point cloud data, forming a structural shape representation that can be used to describe occlusion. During the extraction process, the relatively fixed arrangement and shape continuity of the supporting components and platform structure within the air-cooled island are utilized. Point sets matching the height range, connection direction, and plate-like or beam-like shape of the air-cooled island support system are selected as candidate structural point sets from the observation point cloud data. Then, combining the spatial extension characteristics of continuous surface regions in the observation point cloud data, the candidate structural point sets are organized into supporting component point sets and platform structure point sets. To provide operable geometric objects for subsequent occlusion judgment, the supporting component point sets and platform structure point sets are converted into continuous surface segment representations. These surface segment representations include the spatial coverage range of the structural surface and the outward-facing surface orientation information, while preserving the positional relationships of these surface segment representations under a unified coordinate system. These steps transform the supporting components and platform structure from discrete observation point cloud data into geometric boundaries that can be used to describe occlusion paths, thus enabling direct determination of whether the observation ray is occluded by the structure in subsequent steps.

[0086] A spatial set of observation rays is constructed by combining the position and attitude information of the detection equipment with the distribution data of the observation rays. The occlusion effect of the support components and platform structure is introduced into the spatial set of observation rays to obtain a single-frame reachability description. For each frame of observation point cloud data, the starting position and emission direction of the observation ray are determined according to the corresponding position and attitude information of the detection equipment. The direction of the observation ray in the distribution data of the observation ray is mapped to a unified coordinate expression to form a spatial set of observation rays covering the finned tube bundle. The surface segment expression of the support component and the surface segment expression of the platform structure are used as the occlusion boundary. A spatial intersection relationship determination rule is established between each observation ray in the spatial set of observation rays and the surface segment expression of the support component and the surface segment expression of the platform structure. The spatial intersection relationship determination rule is used to determine the set of intersection points of the observation ray in the propagation direction with the occlusion boundary under the unified coordinate expression, and based on this, the position of the nearest intersection point and the object to which the nearest intersection point belongs in the propagation direction of the observation ray are obtained.

[0087] To ensure clear intersection rules and priorities for determining spatial intersection relationships, the surface fragment representations of supporting components and platform structures are represented using triangular meshes or sets of regular faces, with a spatial index structure established for these meshes. The spatial intersection relationship between the observation ray set and the occlusion boundary is determined by finding the intersection between the observation ray and the triangular mesh, or between the observation ray and the set of regular faces. Specifically, during the intersection process, the set of intersection points obtained from the same observation ray is sorted according to the distance from the starting position of the observation ray to the intersection point. The intersection point with the smallest distance is determined as the nearest intersection point, and the corresponding surface fragment representation of the supporting component or the platform structure is identified as the nearest intersection point's assigned object. If both the supporting component surface fragment representation and the platform structure surface fragment representation satisfy the same condition for the nearest intersection point's assigned object, then the assignment is based on the pre-established priority of the occlusion boundary, defined as the support component surface fragment representation taking precedence over the platform structure surface fragment representation. To prioritize the determination of the occlusion boundary of the supporting components in the structural connection area and maintain consistency with the subsequent cross-segment semantics organized based on the connection position of the supporting components; based on the nearest intersection point position and the object to which the nearest intersection point belongs, it is determined whether the observed ray, when propagating from the starting position along the direction of the observed ray, spatially intersects with the surface segment expression of the supporting component or the surface segment expression of the platform structure before reaching the target area of ​​the finned tube bundle; when the nearest intersection point position is located outside the target area of ​​the finned tube bundle and the object to which the nearest intersection point belongs is the surface segment expression of the supporting component or the surface segment expression of the platform structure, the target spatial direction corresponding to the observed ray is marked as occluded and accessibility restricted; when there is no intersection point or the nearest intersection point position is located inside the target area of ​​the finned tube bundle, the target spatial direction corresponding to the observed ray is marked as accessible; based on the above markings, the single-frame observed point cloud data is spatially divided into accessible areas and occluded restricted areas, and the correspondence between the occluded restricted areas and the surface segment expressions of the supporting components and the surface segment expressions of the platform structure is preserved.

[0088] By combining the marker results of occluded accessibility in the single-frame accessibility description, spatial segments that present unilateral oblique coverage and convergence along the pipe row depth direction at most acquisition times are summarized as wedge-shaped blind zone candidate regions. This maps the spatial occlusion effects of support components and platform structures into a calculable wedge-shaped blind zone range description, providing a spatial index basis for the subsequent generation of directional observability confidence.

[0089] Based on the single-frame accessibility description, cross-frame fusion is performed to extract the occlusion-restricted morphology that gradually converges along the tube bundle depth direction and form an initial representation of the wedge-shaped blind zone range. Since the detection equipment may have different position and orientation information at different acquisition times, the observed ray distribution data also changes accordingly. During cross-frame fusion, the occlusion-restricted regions of each frame are first uniformly superimposed onto the same spatial reference. Then, using the spatial orientation of the finned tube bundle as the axis, the spatial set of observed rays is divided into several continuous spatial segments along the tube bundle depth direction. For each spatial segment, the coverage direction distribution of the spatial set of observed rays from different acquisition times is statistically analyzed to distinguish the coverage of observed rays from both sides of the finned tube bundle, and this is combined with the single-frame accessibility description... The marking results of occlusion accessibility limitation identify whether a spatial segment is in a state of long-term unilateral oblique view coverage. If a spatial segment has only unilateral oblique view accessible observation rays at most acquisition times, while the observation rays in the other direction are mainly blocked by support components or platform structures, then the spatial segment is classified into the wedge blind zone candidate region, and the spatial shape of the wedge blind zone candidate region is tracked along the pipe row depth direction to see whether it shows a layer-by-layer convergence. The above steps extend the occlusion effect from a single frame to the coverage characteristics across frames, so that the determination of the wedge blind zone not only reflects a certain instantaneous viewpoint, but also reflects the continuous occlusion structure formed by the combined action of support components, platform structures and detection paths, thus making it more suitable for limiting the scope of subsequent micro-deformation component unobservable risk.

[0090] Spatial connectivity processing is performed on candidate wedge-shaped blind zones to output wedge-shaped blind zone range data. This data includes the wedge-shaped blind zone boundary, the depth extension range of the wedge-shaped blind zone, and the transition region near the boundary. During spatial connectivity processing, candidate wedge-shaped blind zone regions are merged under a unified coordinate system, combining regions with continuous occlusion-restricted characteristics in adjacent spatial segments into a single wedge-shaped range. Scattered occlusion-restricted segments unconnected to the wedge-shaped range are separated to avoid the data being affected by local noise or occasional occlusion. Subsequently, the outer contour of the wedge-shaped range is extracted along the pipe row depth direction to form the wedge-shaped blind zone boundary, and the furthest extension position of the wedge-shaped range in the depth direction is used as the boundary. The depth extension range of the wedge-shaped blind zone is determined. Considering that the area near the boundary of the wedge-shaped blind zone is often in a transitional state between accessibility and occlusion limitation, in order to facilitate the subsequent detailed processing of unobservable risks, a spatial segment with intermittent unilateral squint coverage characteristics is reserved outside the boundary of the wedge-shaped blind zone as the transitional area near the boundary of the wedge-shaped blind zone. This allows the wedge-shaped blind zone range data to describe not only the blind zone itself, but also the degradation transition zone of the blind zone boundary. The above steps transform the occlusion limitation characteristics into structured wedge-shaped blind zone range data, enabling the wedge-shaped blind zone range data to be directly used to define the spatial scope of subsequent processing and to provide a clear spatial index basis for the generation of directional observability confidence, thereby reducing the expansion of the misprocessing range caused by unclear blind zone boundaries.

[0091] The feature analysis module extracts the actual centerline of each finned tube based on the observed point cloud data and the benchmark geometric model data, and compares it with the corresponding benchmark centerline to obtain the geometric feature change data of the finned tube bundle.

[0092] Please see Figure 4 As shown, in order to obtain geometric feature change data of the finned tube bundle that can be used to characterize the micro-deformation of the finned tube bundle, the actual centerline of each finned tube is extracted based on the observed point cloud data and the reference geometric model data, and compared with the corresponding reference centerline. The specific implementation method can be carried out according to the following steps:

[0093] The process involves loading baseline geometric model data and forming a baseline centerline index framework for each finned tube. The baseline geometric model data includes the original geometric dimensions and arrangement of the finned tube bundles, supporting components, and platform structure. After loading, a unique identifier is established for each finned tube according to the arrangement relationship of the finned tube bundles. Geometric constraint information, including the baseline centerline, endpoint positions, direction, and connection points with the supporting components, is extracted from the geometric representation of each finned tube. To provide clear spatial guidance for extracting the actual centerline from the observation point cloud data, a continuous sequence of reference positions is generated along the length of the baseline centerline. A local reference direction consistent with the direction of the baseline centerline is established at each reference position, ensuring that each reference position indicates both the spatial location of the finned tube and the cross-sectional sampling direction. These steps transform the baseline geometric model data into a baseline centerline index framework that can be called for each finned tube and each position, enabling subsequent processing of the observation point cloud data to be assigned and aligned at the tube-level granularity, reducing matching ambiguities caused by repetitive structures in the finned tube bundles.

[0094] Based on the baseline centerline indexing framework, a candidate point set for each finned tube is formed in the observation point cloud data, and the candidate point set is segmented and organized according to the reference position sequence. The observation point cloud data is obtained by scanning the finned tube bundle area with detection equipment. The point cloud includes points on the surface of the finned tube bundle, points on the surface of the supporting components, points on the surface of the platform structure, and possible environmental background points. To ensure that the actual centerline extraction is focused on the finned tube bundle body, a spatial envelope range coaxial with the baseline centerline is constructed around the baseline centerline of each finned tube. The spatial envelope range extends along the baseline centerline and covers the point cloud distribution area that may appear on the outer surface of the finned tube and the outer contour of the fin. Points falling within the spatial envelope range in the observation point cloud data are selected as the corresponding fins. The candidate point set of the tube is selected, and points falling within the spatial location range of the supporting components and platform structure are removed to reduce the mixing of points on the tube surface by the structural shading boundary. Then, the candidate point set is segmented and organized using the reference position sequence. At each reference position, a cross-sectional sampling zone perpendicular to the local reference direction is taken. Points in the candidate point set that fall into the cross-sectional sampling zone are assigned to the cross-sectional point set corresponding to each reference position, and the correspondence between the cross-sectional point set and the reference position sequence is retained. The above steps use the reference centerline as a priori guide to perform tube-level screening and segmentation of the observation point cloud data, so that each finned tube forms a structured cross-sectional point set sequence in the observation point cloud data, which is convenient for extracting the center point at the local cross-sectional level and splicing it into the actual centerline.

[0095] The cross-sectional center point sequence of each finned tube is extracted based on the cross-sectional point set sequence, and the actual centerline is generated from the cross-sectional center point sequence. Due to the finned structure on the outer surface of the finned tube bundle, the observed point cloud data may show a mixed distribution of multiple outer contour points at the cross-sectional level. To ensure that the actual centerline can reflect the true geometric position of the tube body, points that are more consistent with the continuity of the outer surface of the tube body and are distributed in a circumferential manner around the reference centerline are preferentially selected as tube body candidate points in each cross-sectional point set. At the same time, the point cloud density variation along the reference centerline direction is constrained to ensure that the selected tube body candidate points maintain spatial distribution continuity between adjacent reference positions. For the selected tube body candidate points, a set of corresponding opposing point pairs is constructed based on the positional relationship of the tube body candidate points relative to the reference centerline within the cross-sectional sampling zone. Each opposing point pair is located on both sides of the reference centerline on the cross-section, and the midpoint of the opposing point pair is used as the center of the cross-section. Candidate points are selected; when there are local missing points in the cross-section point set, the cross-section center point is determined by combining the direction of the candidate center points at adjacent reference positions with the direction of the reference center line, so that the sequence of cross-section center points remains coherent in space along the extension direction of the tube. Subsequently, the center points corresponding to each reference position are connected in the order of the reference position sequence to obtain the actual center line of each finned tube. At both ends of the actual center line, the geometric constraint information of the endpoint position of the reference center line and the connection position of the supporting component is combined to perform position consistency processing on the end center points, so that the end of the actual center line can maintain a corresponding relationship with the structural connection boundary. The above steps extract the center points at the cross-section level and splice them along the reference position sequence to generate the actual center line. The actual center line inherits the tube-level index provided by the reference geometric model data and reflects the real geometric distribution in the observed point cloud data, thereby providing an alignable center line expression for subsequent reference center line comparison.

[0096] Based on the comparison between the actual centerline of each finned tube and the corresponding reference centerline, geometric feature change data of the finned tube bundle is generated. During the comparison process, using the reference position sequence as a unified alignment index, the reference position point of the reference centerline and the center point of the actual centerline at the corresponding position are read at each reference position. The spatial deviation between the two is recorded as the centerline offset information at the reference position. Combined with the direction of the reference centerline, an expression method for the offset direction is established, so that the centerline offset information can distinguish the offset along the direction of the reference centerline and the offset perpendicular to the direction of the reference centerline. At the same time, the relative relationship between the end center point of the actual centerline and the connection position of the support component is read at the endpoint position of the reference centerline. The relationship change record is recorded as connection position offset information, so that the geometric feature change data of the finned tube bundle includes both the centerline offset distribution and the end connection position offset distribution. For each finned tube in the finned tube bundle, the geometric feature change data is organized according to the unique identifier of the finned tube and the reference position sequence index, forming a structured data set that can be directly used for subsequent micro-deformation index extraction. The above steps compare the actual centerline with the reference centerline position by position using a unified reference position sequence and form an offset distribution record, so that the geometric feature change data of the finned tube bundle can be spatially located to the specific finned tube and specific position, providing a continuous and traceable input basis for subsequent calculation of the micro-deformation index data of the finned tube bundle from the geometric feature change data.

[0097] The deformation analysis module calculates the micro-deformation index data of the finned tube bundle based on the geometric feature change data.

[0098] To transform the geometric feature variation data of finned tube bundles into quantitative results that can be used for maintenance decisions, a method is adopted to calculate the micro-deformation index data of finned tube bundles based on the geometric feature variation data. The centerline offset distribution and connection position offset information of each finned tube in the geometric feature variation data are normalized in direction, assigned to span, and aggregated into indexes to form micro-deformation index data such as the lateral micro-bending of the tube centerline, the mid-span deflection of the support, and the small offset of the tube bundle relative to the support. The specific implementation method can be carried out according to the following steps.

[0099] The geometric feature change data is prepared through index calculation to ensure that the offset information of each finned tube has consistent directional and span semantics. The geometric feature change data comes from the comparison results of observed point cloud data and baseline geometric model data, and is organized according to the unique identifier of the finned tube and the reference position sequence index. It includes centerline offset information at each reference position and connection position offset information at the endpoints of the baseline centerline. To avoid inconsistencies in the offset direction meaning due to differences in installation orientation among different finned tubes, the axial direction of each finned tube is established based on the baseline centerline orientation in the baseline geometric model data. Furthermore, under the same coordinate system, a transverse direction consistent with the tube array arrangement and a vertical direction consistent with the gravity direction are established, allowing the centerline offset information to be decomposed into... The components corresponding to the axial, lateral, and vertical directions are expressed. Simultaneously, based on the connection positions of the support components and finned tubes in the benchmark geometric model data, the support span boundary of each finned tube is determined. The reference position sequence is divided into several span position intervals according to the support span boundaries, and a correlation is established between each span position interval and the corresponding support component connection position. This allows for direct location of the mid-span and end positions when calculating the mid-span deflection of the support. Through the above processing, the offset distribution in the geometric feature change data is transformed from a simple spatial difference record into directional components and span assignments with engineering significance. This provides a unified semantic basis for the subsequent calculation of the lateral micro-bending of the tube centerline and the mid-span deflection of the support, and also facilitates maintaining consistency in the meaning of indicators across different testing batches.

[0100] The lateral micro-bending of the tube centerline is calculated based on geometric feature change data. This lateral micro-bending characterizes the degree of bending of the finned tube in the lateral direction, focusing on the displacement caused by morphological changes along the pipe rather than overall translation. Specifically, for each finned tube, the centerline offset information under the reference position sequence index is read, and the lateral offset component sequence is extracted to obtain the lateral offset component sequence. Considering that the acquisition of the observation point cloud data is affected by the position and attitude information of the detection equipment, the centerline offset information may simultaneously include the overall positional displacement of the entire finned tube and the morphological changes along the pipe bending. To ensure that the lateral micro-bending of the tube centerline mainly reflects the bending morphological changes, the cross-end constraint provided by the connection position offset information in the geometric feature change data is used to establish an alignment relationship between the lateral offset component sequence and the lateral offset component at the cross-end position within each support cross-segment position interval. This ensures that the lateral offset component sequence is consistent with the connection position offset information at the cross-end position. The components attributable to the overall offset between the ends are stripped from the lateral offset component sequence, making the remaining part more focused on reflecting the morphological offset within the span. Subsequently, within each support span position interval, the reference position with the largest lateral offset amplitude is extracted from the processed lateral offset component sequence as the lateral micro-bending characterization of the corresponding support span position interval. Then, the maximum value of the lateral micro-bending characterization is selected from multiple support span position intervals of the same finned tube as the lateral micro-bending amount of the tube centerline of the corresponding finned tube. If the same finned tube has multiple peak positions with similar lateral offset amplitudes in different spans, the corresponding reference position index is recorded simultaneously for subsequent positioning of the maintenance focus area. Through the above processing, the lateral micro-bending amount of the tube centerline can be directly generated from the geometric feature change data, and the overall offset influence is stripped away by the span-end constraint, making the index closer to the lateral bending itself. This facilitates maintaining the clarity of the index meaning when performing subsequent directional compensation for the offset compression caused by low sensitivity near the wedge blind zone boundary.

[0101] The mid-span deflection of the support is calculated based on geometric feature variation data. This mid-span deflection characterizes the degree of sag or arching of the finned tube relative to the end position in the vertical direction, and is consistent with the support conditions. Specifically, for each finned tube, based on the support span position intervals formed in the first step, the vertical offset component sequence of the centerline offset information within each support span position interval is read, and the vertical components of the connection position offset information at both ends of the support span are read as the vertical boundary of the span end. To avoid the mid-span deflection being affected by the overall vertical offset of the span end, a vertical reference relationship for the span is established using the vertical boundary of the span end, ensuring that the vertical offset component sequence is consistent with the vertical boundary of the span end at the span end position. The components that uniformly change between the span ends are separated from the vertical offset component sequence, making the vertical morphological offset within the span more prominent. Subsequently, the mid-span position is determined within each support span position interval. The reference position index corresponding to the mid-span position is determined by the distribution of the span length and the reference position sequence, ensuring the mid-span... The location represents the geometric midpoint of the span. The processed vertical offset component is read at the mid-span location, and compared with the vertical reference relationship formed by the vertical boundary at the span end to obtain the mid-span deflection characterization of the support span location interval. For cases where multiple local extrema appear in the vertical offset component sequence within the span, the vertical offset components within a certain range near the mid-span location are comprehensively expressed, ensuring that the mid-span deflection characterization reflects the overall shape of the mid-span region rather than a single-point anomaly. Finally, the corresponding support mid-span deflection is output for each of the multiple spans of the same finned tube, and the span identifier and mid-span location index are retained in the finned tube bundle micro-deformation index data. This facilitates the subsequent association of support mid-span deflection with the support component location for maintenance scheduling. Through the above processing, the support mid-span deflection is generated from the vertical offset component of the geometric feature change data under the span semantics, preserving the engineering meaning of the mid-span location while avoiding treating the overall vertical offset at the span end as deflection, thus making the index more aligned with the focus of support mid-span deflection.

[0102] The minute offset of the tube bundle relative to the support is calculated based on geometric feature change data. This minute offset characterizes the relative displacement of the finned tube bundle relative to the support member at the support connection position and directly corresponds to the connection reliability and stress state changes. Specifically, for each finned tube, the connection position offset information is read from the geometric feature change data. This information records the spatial change of the actual centerline end point relative to the reference connection position at the endpoint of the reference centerline. To ensure that the minute offset of the tube bundle relative to the support can distinguish the relative offset effects in different directions, the connection position offset information is decomposed into three components based on the axial, lateral, and vertical directions established in the first step, forming expressions for end lateral relative offset, end vertical relative offset, and end axial relative offset, respectively. Considering that a finned tube typically has two connection ends, the end components at both ends are recorded separately. When a single-value index is needed, the combined expression of the two end components is used as the tube bundle relative offset of that finned tube. The minute offset of the support is considered. To maintain consistency with the semantics of the support span, a correspondence is established between the minute offset of the tube bundle relative to the support at each end and the position interval of the adjacent support span. This allows for direct segment-based organization when analyzing the correlation between the mid-span deflection of the support and the minute offset of the tube bundle relative to the support. Finally, the lateral micro-bending of the tube centerline of each finned tube, the mid-span deflection of each support span, and the minute offset of the tube bundle relative to the support at both ends or in combination are aggregated into micro-deformation index data of the finned tube bundle. This data is then organized according to the unique identifier of the finned tube, the segment identifier, and the reference position index, forming data input that can be directly used for subsequent generation of directional observable reliability and directional compensation processing. Through the above processing, the minute offset of the tube bundle relative to the support can be directly obtained from the connection position offset information while maintaining clear directional semantics. This allows the micro-deformation index data to simultaneously cover the changes in tube body morphology along the pipe and the relative displacement changes at the connection ends, facilitating targeted processing of the component compression problem caused by the wedge blind zone in subsequent steps.

[0103] The trusted computing module generates directional observability trusted based on observed ray distribution data and wedge blind zone range data.

[0104] To enable subsequent directional compensation to distinguish between component compression caused by unilateral squint observation conditions within the wedge blind zone data and normal observation results outside the wedge blind zone data, a method is adopted to generate directional observability reliability based on observation ray distribution data and wedge blind zone data. This transforms the influence of observation geometry on the principal directional components of different micro-deformations into a traceable and locatable reliability description. The specific implementation can be carried out in the following steps:

[0105] An observability index for the spatial location of finned tube bundles is established based on wedge-shaped blind zone range data. The observed ray distribution data and the micro-deformation index data of the finned tube bundles are consistent at the spatial location level. The wedge-shaped blind zone range data includes the boundary of the wedge-shaped blind zone, the depth extension range of the wedge-shaped blind zone, and the transition area range near the boundary of the wedge-shaped blind zone. These ranges are expressed in a unified coordinate system. To ensure that the reliability of directional observability falls on each finned tube, each reference position sequence index, and each support connection position, the reference position sequence index used to organize centerline offset information is read from the geometric feature change data, and the cross-segment identifier, mid-span position index, and connection position index recorded in the micro-deformation index data are read to form a set of spatial location indexes for the finned tube bundles. Each position in the spatial location index set of the finned tube bundles is mapped to a wedge. In the blind zone range data, it is determined whether each location is within the depth extension range of the wedge blind zone, the transition area near the boundary of the wedge blind zone, or outside the wedge blind zone range data. The attribution result of each location is obtained, and the attribution result is written into the basic record of directional observability confidence as the blind zone attribution information of each location. Combining the point cloud measurement point set corresponding to each location in the observation point cloud data, the observation ray direction set and observation ray source information corresponding to the observation ray distribution data are extracted according to the measurement point index, so that each location has an observation ray direction set associated with itself. The above steps form a unified spatial index and blind zone attribution information before the directional observability confidence is generated. The observation ray distribution data and micro-deformation index data share the same location semantics, reducing the interpretation bias caused by the inconsistency between the observability description and the index calculation location.

[0106] Based on the baseline geometric model data and micro-deformation index data, the principal direction of the target micro-deformation is determined. The observability and reliability of the direction are described differently for different micro-deformation indices. The micro-deformation index data of the finned tube bundle includes the lateral micro-bending of the tube centerline, the mid-span deflection of the support, and the small offset of the tube bundle relative to the support. The displacement component directions corresponding to different indices are different. The baseline centerline orientation of each finned tube is read from the baseline geometric model data, and the axial, lateral, and vertical directions are established for each finned tube under a unified coordinate expression. The axial direction is consistent with the baseline centerline orientation, the lateral direction is consistent with the tube arrangement, and the vertical direction is consistent with the gravity direction. The principal direction of the target micro-deformation corresponding to the lateral micro-bending of the tube centerline is determined as the lateral direction. The direction is defined, and the location of action is limited to the set of reference position sequence indices within the support span position interval; the main direction of the target micro-deformation corresponding to the mid-span deflection of the support is determined as the vertical direction, and the location of action is limited to the mid-span position index and its adjacent reference position sequence index set within each support span position interval; the main direction of the target micro-deformation corresponding to the small offset of the tube bundle relative to the support is determined as a combination of the lateral, vertical and axial directions, and the location of action is limited to the set of connection position indices; the above steps align the main direction of the target micro-deformation with the physical meaning of the micro-deformation index data, and the observability and reliability of the direction provide a positioning basis for the index and components, supporting the targeted processing of component compression by subsequent directional compensation.

[0107] Based on the observed ray distribution data, an observation coverage morphology description is formed at each spatial location. Combined with the wedge blind zone range data, a grading result of the observability reliability of the direction is given. The grading result reflects the projection relationship between the set of observed ray directions and the main direction of the target micro-deformation, as well as the continuous occlusion characteristics caused by the blind zone attribution. For each location, the set of observed ray directions associated with that location is read, and the main direction of the target micro-deformation is used as the reference direction to form projection relationship description information and coverage morphology description information. The projection relationship description information is used to reflect the projection component level of the set of observed ray directions on the main direction of the target micro-deformation, and the coverage morphology description information is used to reflect the distribution pattern of the set of observed ray directions in terms of coverage on both sides and single-sided oblique coverage in space. The description of the cause of the observation coverage morphology obtained by cross-frame fusion is used as the source information of the continuous occlusion characteristics, so that the set of observed ray directions at the same spatial location at different acquisition times can be summarized under a unified coordinate expression.

[0108] The process of forming projection relationship description information includes establishing projection component records for each observation ray direction in the set of observation ray directions and the main direction of the target micro-deformation, and summarizing the projection component records into projection component distribution records according to spatial location index; the projection component distribution records are used to characterize the proportion of observation ray directions that can reflect the changes of the main direction component of the target micro-deformation at the spatial location and the distribution of projection component values, thereby distinguishing between the case where the projection component is concentrated in a smaller value and the case where the projection component covers multiple values ​​within the same location; the process of forming coverage morphology description information includes classifying the set of observation ray directions on both sides of the pipe row direction, and combining the wedge blind zone depth extension range and the transition area range near the boundary of the wedge blind zone in the wedge blind zone range data to form a two-sided coverage ratio record and a one-sided oblique coverage record.

[0109] The process of forming the coverage ratio record on both sides includes: reading the set of observation ray directions corresponding to the spatial location index based on a unified coordinate expression, and classifying the set of observation ray directions on both sides based on the tube array direction. The tube array direction is determined by the arrangement relationship of the finned tube bundle in the reference geometric model data. Observation ray directions that are in the same direction as the tube array direction are classified into the spatial direction on one side of the tube array direction, and observation ray directions that are opposite to the tube array direction are classified into the spatial direction on the other side of the tube array direction. The process of forming the coverage ratio record on both sides also includes: counting the number of observation ray directions on one side of the tube array direction and the number of observation ray directions on the other side of the tube array direction respectively, and using the total number of observation ray directions as a normalization benchmark to obtain the coverage ratio record on both sides. The record content includes the spatial location index, the coverage ratio of the spatial direction on one side of the tube array direction, the coverage ratio of the spatial direction on the other side of the tube array direction, and the corresponding acquisition time range. Thus, the coverage ratio record on both sides can reflect the proportion of observation ray directions from both sides of the spatial direction at the spatial location index level.

[0110] The formation process of a unilateral strabismus coverage record includes: based on multiple acquisition times within the cross-frame fusion range, reading the set of observation ray directions corresponding to the spatial position index at each acquisition time, and reusing the classification results of both sides within each acquisition time to generate a bilateral existence record. The bilateral existence record includes whether an observation ray direction exists on one side of the pipe-pile direction and whether an observation ray direction exists on the other side of the pipe-pile direction. The formation process of a unilateral strabismus coverage record also includes summarizing the bilateral existence records within the cross-frame fusion range to determine whether all acquisition times within the cross-frame fusion range satisfy the condition that an observation ray direction exists on one side of the pipe-pile direction and that an observation ray direction does not exist on the other side of the pipe-pile direction, or whether the cross-frame fusion range... All acquisition times within the timeframe satisfy the condition that there is an observation ray direction on the other side of the pipe array direction and no observation ray direction on one side of the pipe array direction. The spatial location index that meets the conditions is recorded as a single-sided squint coverage record. The single-sided squint coverage record includes a spatial location index, the spatial direction attribution of the single-sided squint coverage, the corresponding acquisition time range and the description of the reason for the observation coverage shape. The description of the reason for the observation coverage shape is generated by the blind zone attribution information of the depth extension range of the wedge blind zone and the transition area range near the boundary of the wedge blind zone, the coverage ratio record of both sides and the existence record of both sides. Thus, the single-sided squint coverage record can reflect whether the set of observation ray directions is continuously provided by the same spatial direction under cross-frame fusion conditions at the spatial location index level.

[0111] Directional observability confidence is used to classify and label the observability of the same spatial location under the main direction of the target's micro-deformation. The classification label of directional observability confidence includes three values: high confidence, medium confidence, and low confidence. The high confidence, medium confidence, and low confidence are generated by the directional observability confidence classification process based on the projection component distribution record, the two-sided coverage ratio record, the single-sided oblique coverage record, and the blind zone attribution information, and are written into the directional observability confidence classification result record.

[0112] The process of determining the projection relationship includes: reading the projection component value distribution from the projection component distribution record corresponding to the spatial location index, and deduplicating the projection component value distribution; the deduplication of the projection component value distribution is based on the actual projection component values ​​that appear in the projection component distribution record. If the projection component distribution record has at least two different projection component values, it is determined that the projection component distribution record covers multiple values. If the projection component distribution record has only one projection component value, it is determined that the projection component distribution record does not cover multiple values.

[0113] The process of forming the two-sided coverage determination result includes: reading the spatial coverage ratio on one side of the pipe row direction and the spatial coverage ratio on the other side of the pipe row direction from the two-sided coverage ratio record corresponding to the spatial location index, and determining whether the spatial coverage ratio on one side of the pipe row direction is greater than 0 and whether the spatial coverage ratio on the other side of the pipe row direction is greater than 0. When the spatial coverage ratio on one side of the pipe row direction is greater than 0 and the spatial coverage ratio on the other side of the pipe row direction is greater than 0, the two-sided coverage determination result is determined to be two-sided coverage. In other cases, the two-sided coverage determination result is determined not to meet the two-sided coverage requirement.

[0114] The process of forming the continuous occlusion determination result includes: reading the spatial direction of the unilateral strabismus coverage and the corresponding acquisition time range from the unilateral strabismus coverage record corresponding to the spatial location index; if there is a unilateral strabismus coverage record in the spatial location index, the continuous occlusion determination result is determined to be unilateral strabismus coverage; if there is no unilateral strabismus coverage record in the spatial location index, the continuous occlusion determination result is determined to be unilateral strabismus coverage.

[0115] The grading process for directional observability reliability includes: reading the projection relationship determination result, lateral coverage determination result, and persistent occlusion determination result under the target micro-deformation principal direction corresponding to the spatial location index, and generating a grading result for directional observability reliability by combining blind zone attribution information; specifically, when the projection relationship determination result shows that the projection component distribution record covers multiple values ​​and the lateral coverage determination result shows lateral coverage, the directional observability reliability is determined to be high reliability; when the projection relationship determination result shows that the projection component distribution record covers multiple values ​​and the lateral coverage determination result does not meet the lateral coverage requirement, or when the projection relationship determination result does not meet the requirement that the projection component distribution record covers multiple values ​​and the persistent occlusion determination result does not meet the requirement of unilateral oblique coverage, the directional observability reliability is determined to be medium reliability; ... When the projection component distribution record covers multiple values ​​and the continuous occlusion determination result is unilateral strabismus coverage, and the blind zone attribution information corresponds to the depth extension range of the wedge blind zone or the transition area range near the boundary of the wedge blind zone, the directional observability confidence is determined to be low confidence. The grading result of the directional observability confidence is saved to the directional observability confidence grading result record, and the description of the cause of the observed coverage morphology is written into the directional observability confidence grading result record. The description of the cause of the observed coverage morphology includes the summary results of the projection component distribution record, the summary results of the bilateral coverage ratio record, and the summary results of the unilateral strabismus coverage record. Thus, the directional compensation can determine the compensation domain and compensation component based on the description of the cause of the observed coverage morphology, and the selection of the compensation object is consistent with the spatial area defined by the wedge blind zone range data.

[0116] The hierarchical results of observable confidence at each location are aggregated into an index-level expression of the micro-deformation index data, while retaining traceable spatial index mapping relationships. Targeted compensation selects processing objects based on index and location. For the lateral micro-bending of the pipe centerline, the corresponding reference location sequence index set is used as input, and the hierarchical results of observable confidence at each location within the reference location sequence index set are aggregated to form the observable confidence at the direction of the lateral micro-bending of the pipe centerline, while retaining the distribution of observable confidence at the location level along the pipe. For the mid-span deflection of the support, the mid-span location index and its adjacent reference location sequence index set are used as input, and the hierarchical results of observable confidence at the corresponding location are aggregated to form the observable confidence at the direction of the mid-span deflection of the support. The reliability is measured, and the mapping relationship between the segment identifier and the mid-span position index to the reliability is retained. For the small offset of the tube bundle relative to the support, the directional observable reliability classification results in the lateral, vertical and axial directions at the connection position are aggregated as input to form the directional observable reliability of the small offset of the tube bundle relative to the support, and the records of the connection positions at both ends are retained. The above steps align the directional observable reliability and the micro-deformation index data in terms of data organization, forming an input basis that can be directly used for the next step of performing directional compensation on the micro-deformation index data based on the wedge blind zone range data and the directional observable reliability. The selection of the compensation range and the compensation object is consistent with the component compression mechanism.

[0117] The directional compensation module performs directional compensation on the micro-deformation index data based on the wedge blind zone range data and the observability reliability of the direction.

[0118] To specifically correct the compression of displacement components within the wedge-shaped blind zone data caused by unilateral oblique observation conditions, while avoiding unnecessary rewriting of normal observation results outside the wedge-shaped blind zone data, a directional compensation method is adopted for micro-deformation index data based on the wedge-shaped blind zone data and the reliability of directional observability. The compensation domain, compensation object, and compensation components are limited to the degradation region and degradation components jointly pointed to by the wedge-shaped blind zone data and the reliability of directional observability, forming traceable compensated micro-deformation index data. The specific implementation method can be described in the following steps:

[0119] Based on the wedge-shaped blind zone range data, the micro-deformation index data is spatially constrained and grouped by location, ensuring that subsequent directional compensation only covers the relevant areas of the wedge-shaped blind zone and maintains consistency with the spatial index of the geometric feature change data. The wedge-shaped blind zone range data includes the boundary of the wedge-shaped blind zone, the depth extension range of the wedge-shaped blind zone, and the transition area range near the boundary of the wedge-shaped blind zone. The micro-deformation index data is organized according to the unique identifier of the finned tube, the identifier of the support span, the mid-span position index, and the connection position index. During processing, the corresponding positions of each index in the micro-deformation index data are first mapped to the wedge-shaped blind zone range data to obtain the position interval of each finned tube in each support span. The blind zone is assigned a marker, and the connection position index and mid-span position index are mapped to the depth extension range of the wedge-shaped blind zone and the transition area range near the boundary of the wedge-shaped blind zone, respectively. Then, the micro-deformation index data is grouped according to the blind zone assignment marker, forming a wedge-shaped blind zone group, a transition area group, and a non-wedge-shaped blind zone group. The non-wedge-shaped blind zone group retains the original micro-deformation index data and does not enter the compensation process, while the wedge-shaped blind zone group and the transition area group enter the subsequent directional compensation process. At the same time, the spatial index of the micro-deformation index data in each group is retained, so that the subsequent compensation can return to the specific finned tube, the specific support span, and the specific position index.

[0120] Based on the directional observability confidence level, the micro-deformation index data is subject to object and component limitations, ensuring that directional compensation is only performed on micro-deformation indices with low directional observability confidence and the main directional component of the target micro-deformation. The directional observability confidence level has been aligned with the micro-deformation index data using the finned tube's unique identifier, reference position sequence index, mid-span position index, and connection position index. Furthermore, confidence descriptions consistent with the main directional direction of the target micro-deformation are provided for the lateral micro-bending of the tube centerline, the mid-span deflection of the support, and the small offset of the tube bundle relative to the support. During processing, within the wedge-shaped blind zone group and the transition region group, the micro-deformation index data and its corresponding directional observability confidence level are read item by item: for the lateral micro-bending of the tube centerline, the directional observability confidence level distribution of the reference position sequence index set contributing to the lateral micro-bending of the tube centerline within the support span position interval is read to determine the source position interval of the lateral directional component requiring compensation; for the mid-span deflection of the support, the span... The directional observable confidence distribution of the central position index and its adjacent reference position sequence index set is used to determine the source position interval of the vertical direction component that needs compensation. For the small offset of the tube bundle relative to the support, the directional observable confidence of the lateral, vertical and axial directions at the connection position index is read to determine the connection position direction component that needs compensation. Subsequently, the index items and position intervals with low directional observable confidence are marked as compensation objects, and the index items and position intervals with high directional observable confidence are marked as retention objects. Within the compensation objects, it is further limited to adjusting only the main direction component of the target micro deformation, and retaining the direction component that is consistent with the direction of the observation ray and can be stably reflected without being rewritten. Through the above processing, the directional compensation is changed from "smoothing the whole index" to "correcting the specific component of a specific region". The boundary of the compensation object is determined by the directional observable confidence, reducing the intervention of high confidence observation results.

[0121] Compensation reference information is constructed and a compensation path is formed based on geometric feature change data, enabling directional compensation to be carried out along the structural continuity and support constraint continuity of the finned tube bundle without relying on empirical inference. The geometric feature change data includes the centerline offset distribution and connection position offset information, and shares the unique identifier of the finned tube, reference position sequence index, and connection position index with the micro-deformation index data. During processing, two types of compensation reference information are established for each compensation object. The first type of compensation reference information comes from similar micro-deformation index data with high observable reliability in adjacent reachable regions. Specifically, finned tubes belonging to the same support span identifier or adjacent support span identifier are selected along the tube row direction outside the wedge-shaped blind zone boundary. The lateral micro-bending of the finned tube's centerline, the mid-span deflection of the support, or the small offset of the tube bundle relative to the support are read, and the centerline offset distribution or connection position offset information at the corresponding position in the geometric feature change data of the finned tube is read simultaneously to form a reference sequence in the tube-crossing direction. The second type of compensation reference information comes from the tube segment results with high observable reliability along the tube length direction of the same finned tube. Specifically, in the same... On a finned tube, a location interval is selected that is located in a non-wedge-shaped blind zone group or a transitional region group but has high reliability in directional observability. The lateral or vertical components of the centerline offset distribution in the geometric feature change data of the finned tube are read, and the structural relationship between the end position and the mid-span position is determined by combining the support span markers of the finned tube, forming a reference sequence along the tube length. Subsequently, the reference sequence in the tube length direction and the reference sequence in the tube length direction are used together to construct the compensation path: the compensation path takes the boundary of the wedge-shaped blind zone as the transition starting point, the position of the compensation object within the depth extension range of the wedge-shaped blind zone as the target position, and maintains the consistency trend of the end connection position offset information and the centerline offset distribution along the path under the constraint of the support span markers, so that the compensation result can maintain a continuous relationship with the adjacent reachable area and the observable section of the same finned tube in space. Through the above processing, directional compensation no longer depends solely on the single value of the micro-deformation index data, but uses the geometric feature change data as compensation reference information, so that the compensation path corresponds with the structural continuity and support constraint continuity of the finned tube bundle, thereby improving the physical interpretability of the compensation result.

[0122] Based on the compensation object, compensation components, and compensation reference information, directional compensation is completed, and the micro-deformation index data after compensation is output, enabling the output results to directly enter subsequent structural protection processing and maintain full traceability. During processing, component-level updates are performed for each compensation object while maintaining index-level consistency: For the lateral micro-bending of the pipe centerline, the reference position sequence index interval that contributes significantly to the lateral micro-bending of the pipe centerline is first located in the centerline offset distribution of geometric feature change data. Only the lateral component within the reference position sequence index interval is adjusted according to the compensation reference information, so that the adjusted lateral component is within the wedge shape. The lateral component trends near the blind zone boundary and adjacent reachable areas are connected, while maintaining a smooth transition relationship along the support span position interval within the depth extension range of the wedge-shaped blind zone. The updated lateral component is then re-converged into the lateral micro-bending amount of the pipe centerline and written back into the micro-deformation index data. For the support mid-span deflection, the compensation interval for the vertical component is first located in the mid-span position index and its adjacent reference position sequence index set. Only the vertical component within the compensation interval is adjusted according to the compensation reference information, so that the vertical component at the mid-span position index forms a consistent span semantic with the offset information of the span end connection position. The updated vertical components are then converged to form the mid-span deflection of the support and written back to the micro-deformation index data. For the small offset of the tube bundle relative to the support, the directional components that need to be compensated are first located at the connection position index. Only the directional components with low observable confidence are adjusted according to the compensation reference information, while the directional components with high observable confidence remain unchanged. This ensures that the updated results at the connection position are consistent with the distribution of the centerline offset along the same finned tube at the end position. The updated connection position is then written into the micro-deformation index data to form the compensated micro-deformation index data. The data organization retains the compensation object location range, support span identifier, connection position index, and the source type of the compensation reference information used. This allows the compensated micro-deformation index data to be directly called in subsequent steps of setting a mutation retention threshold and performing structural mutation protection during directional compensation. Through this processing, directional compensation limits the spatial range with wedge-shaped blind zone range data, limits the compensation object and compensation components with directional observable confidence, and provides compensation reference information with geometric feature change data. The output compensated micro-deformation index data takes into account both local correction and global consistency, and provides a traceable data foundation for subsequent structural protection.

[0123] In the process of performing directional compensation on micro-deformation index data based on wedge-shaped blind zone range data and directional observability reliability, in order to avoid misprocessing the actual local geometric feature changes in the finned tube bundle as component compression caused by observation degradation, a mutation retention threshold is set and structural mutation protection is implemented during the directional compensation process. The specific implementation method can be carried out according to the following steps:

[0124] A reference sample set for the mutation retention threshold is established based on directional observability confidence, forming a comparison sequence that reflects the geometric change gradient. The directional observability confidence is generated from the observed ray distribution data and wedge blind zone range data, and is already aligned with the micro-deformation index data at the levels of finned tube unique identifier, reference position sequence index, mid-span position index, and connection position index. To ensure the mutation retention threshold comes from more reliable observational geometry, index entries with high directional observability confidence are first selected from the micro-deformation index data. These index entries cover the reference position sequence index set corresponding to the lateral micro-bending of the tube centerline, the mid-span position index corresponding to the support mid-span deflection and its adjacent reference position sequence index set, and the connection position index set corresponding to the small offset of the tube bundle relative to the support. Subsequently, based on the centerline along the same index in the geometric feature change data... The process offset distribution and connection position offset information are used to construct a variation sequence for comparison: Inside the same finned tube, the adjacent differences of the lateral and vertical components are extracted according to the reference position sequence index order to form a variation sequence along the tube length; between adjacent finned tubes, the difference between the lateral micro-bending of the tube centerline and the mid-span deflection of the support under the same span marker is extracted according to the adjacent relationship of the tube row direction to form a variation sequence along the tube row direction; at the connection position index, the differences of the small offset of the tube bundle relative to the support in the lateral, vertical and axial directions are extracted to form an end connection variation sequence; the above steps screen high-confidence samples based on the observable confidence of the direction, and construct multi-directional variation sequences with geometric feature change data, so that the setting of the mutation retention threshold is based on more reliable spatial difference information, reducing the threshold drift caused by the degradation of wedge blind zone observation.

[0125] The logic for setting the mutation retention threshold is established by combining the acquisition conditions of the observed point cloud data and the discrete characteristics of the geometric feature change data, and the mutation retention threshold is solidified into a reusable judgment criterion. The observed point cloud data is obtained by scanning with a detection device. Different scanning distances, incident directions, and point cloud densities will lead to differences in the degree of dispersion of the point cloud surface. The geometric feature change data inherits this discrete characteristic in terms of the centerline offset distribution and connection position offset information. In order for the mutation retention threshold to distinguish between normal fluctuations caused by the discreteness of the observed point cloud data and local changes caused by structural state changes, the setting logic of the mutation retention threshold is based on the change sequence of high-confidence samples, and the point cloud density and point cloud dispersion of the observed point cloud data in the high-confidence region are used as the upper bound reference for fluctuation: in the change sequence along the pipe length direction, the change amplitude that appears in multiple consecutive reference position sequence indices is selected as the candidate feature to avoid isolated differences at a single reference position sequence index affecting the threshold; in the change sequence along the pipe length direction, the change amplitude that appears in multiple consecutive reference position sequence indices is selected as the candidate feature to avoid isolated differences at a single reference position sequence index affecting the threshold. In the sequence of changes in the tube arrangement direction, differences with consistent directionality on multiple adjacent finned tubes are selected as candidate features to avoid the influence of accidental differences caused by the absence of point clouds in individual finned tubes on the threshold. In the sequence of changes in the end connection, differences with consistent spatial orientation in the lateral, vertical, and axial directions at the same connection position index are selected as candidate features to avoid the influence of local discrete amplification of single-directional components on the threshold. Based on the distribution of the change amplitude of the candidate features, the mutation retention threshold is set to a threshold level that can cover common change amplitudes caused by the discreteness of the observed point cloud data and maintain separation for larger change amplitudes. This makes changes exceeding the mutation retention threshold more likely to reflect structural state changes rather than discrete fluctuations. The above steps bind the mutation retention threshold to the discreteness of the observed point cloud data and constrain the source of the threshold with the multi-directional change sequence of high-confidence samples, so that the mutation retention threshold has a clear setting basis and can maintain a consistent judgment meaning in different detection batches.

[0126] The triggering location for structural mutation protection is identified based on the mutation retention threshold and directional observability confidence level. The coverage area of ​​structural mutation protection is determined by combining wedge-shaped blind zone range data, enabling structural mutation protection to coordinate with the spatial domain of directional compensation. The triggering location for structural mutation protection preferentially comes from regions with high directional observability confidence level because the observed ray coverage in high-confidence regions is more complete, and the micro-deformation index data better reflects the true geometric differences. During processing, the variation sequences along the tube length direction, along the tube row direction, and at the end connection variation sequences are traversed within the high-confidence region. If adjacent differences exceed the mutation retention threshold, and the exceeding behavior is continuously distributed on adjacent finned tubes with the same continuous reference position sequence index or the same span identifier, then the corresponding reference position sequence index interval or the corresponding finned tube interval is marked as the structural mutation protection trigger interval. For differences at the connection position index, if at least two of the differences in the lateral, vertical, and axial directions exceed the mutation retention threshold, and the connection position index is at the support... If a component exhibits a consistent spatial orientation change in its vicinity, the connection location is indexed and marked as the structural mutation protection trigger location. Subsequently, the structural mutation protection trigger interval is spatially expanded using wedge-shaped blind zone range data: when the structural mutation protection trigger interval is adjacent to the boundary of the wedge-shaped blind zone, the structural mutation protection trigger interval is extended a certain distance along the transition area near the boundary of the wedge-shaped blind zone into the depth extension range of the wedge-shaped blind zone, ensuring that the structural mutation protection covers the neighborhood within the wedge-shaped blind zone range data that may be affected by directional compensation; when the structural mutation protection trigger interval is located in a non-wedge-shaped blind zone group, the structural mutation protection coverage area is kept within the depth extension range of the wedge-shaped blind zone, avoiding including areas outside the wedge-shaped blind zone range data in the compensation limit; the above steps use mutation retention thresholds to determine the trigger for changes in high-confidence areas and utilize wedge-shaped blind zone range data for boundary expansion, enabling the structural mutation protection to form a protection range around the actual change location while maintaining consistency with the directional compensation domain defined by the wedge-shaped blind zone range data.

[0127] In the directional compensation process, structural abrupt change protection constraints are introduced. After compensation, the micro-deformation index data retains its variation characteristics within the coverage area of ​​the structural abrupt change protection, while directional correction of component compression is completed outside the coverage area. Directional compensation adjusts the component levels of the micro-deformation index data based on the wedge-shaped blind zone range data and the directional observability confidence level, and provides compensation reference information using geometric feature change data. Directional compensation uses an interpolation extrapolation method based on the compensation reference information to perform component level updates. The interpolation extrapolation uses the reachable positions at both ends of the compensation object's location interval as boundary constraints, and uses similar micro-deformation index data with high directional observability confidence levels in adjacent reachable regions as anchor point constraints, with high directional observability confidence levels along the tube length of the same finned tube. In the pipe segment results, the reference position sequence index adjacent to the location interval of the compensation object is selected as an additional anchor point constraint, thereby forming a constrained compensation path within the transition area near the depth extension range of the wedge blind zone and the boundary of the wedge blind zone. The boundary constraint is used to limit the continuity of the values ​​of the location interval of the compensation object between the transition starting point and the target position. The anchor point constraint is used to limit the trend consistency between the update results within the location interval of the compensation object and the directional observable confidence of the adjacent reachable area. The connection position of the support member is written into the end constraint of the compensation path through the correspondence between the connection position offset information in the geometric feature change data and the centerline offset distribution along the path, so that the cross-end connection position offset information and the cross-segment semantics are consistent with the cross-segment centerline offset distribution inside the cross segment.

[0128] After introducing structural mutation protection, the intersection relationship with the coverage area of ​​structural mutation protection is determined in the compensation path of the compensation object. When the compensation path enters the coverage area of ​​structural mutation protection, the component level adjustment within the coverage area of ​​structural mutation protection is performed using a constraint tightening method. For the lateral micro-bending of the pipe centerline, when the lateral component within the coverage area of ​​structural mutation protection is updated by interpolation extrapolation, the direction of the difference in the adjacent reference position sequence index within the structural mutation protection trigger interval is kept consistent. The update amplitude of the structural mutation protection trigger interval is limited by the anchor point constraint outside the structural mutation protection trigger interval, so that the directional correction of the component compression corresponding to the update result does not change the relative difference of the structural mutation protection trigger interval. For the mid-span deflection of the support, when the vertical component within the coverage area of ​​structural mutation protection is updated by interpolation extrapolation, the segment semantic relationship between the vertical component near the mid-span position index and the offset information of the span connection position is kept. The update direction at the mid-span position index is limited by the anchor point constraint corresponding to the adjacent support span identifier, so that the change characteristics near the mid-span position index are retained in the update result. For the pipe bundle The small offset relative to the support is used to incorporate the combination of the lateral, vertical, and axial components at the connection position index of the structural mutation protection trigger location into the interpolation extrapolation update process as connection position constraints. Components with low confidence in the direction of observation and located within the wedge blind zone range are updated, while components with high confidence in the direction of observation retain their original values, thus preserving the multi-directional differences at the connection position index. Outside the coverage area of ​​the structural mutation protection, directional compensation performs interpolation extrapolation updates on the location interval of the compensation object according to boundary constraints and anchor point constraints, and maintains a continuous transition along the transition area near the boundary of the wedge blind zone to the depth extension range of the wedge blind zone. After compensation, the micro-deformation index data forms a continuous spatial organization structure as a whole. Through the above processing, the structural mutation protection triggered by the mutation retention threshold is embedded in the interpolation extrapolation update process of directional compensation in the form of boundary constraints, anchor point constraints, and connection position constraints. The micro-deformation index data after compensation is used to correct the component compression problem within the wedge blind zone range data, while preserving the variation characteristics within the coverage area of ​​the structural mutation protection.

[0129] Through the above scheme, the steps are correlated and have a constraint transmission relationship. The finned tube bundle observation point cloud data obtained from multi-source data acquisition, the position and attitude information of the detection equipment, and the finned tube bundle observation ray distribution data form a traceable data link under a unified coordinate expression. Combined with the wedge blind zone range data and the observability reliability of the direction, directional compensation is performed on the micro-deformation index data. In the directional compensation process, a mutation retention threshold is set and structural mutation protection is implemented. This forms a directional correction path for the component compression problem caused by the unilateral oblique observation conditions corresponding to the wedge blind zone, reducing the interpretation bias caused by the spatial discontinuity of the micro-deformation index data near the boundary of the wedge blind zone. The micro-deformation detection results of the finned tube bundle obtained in this way can cover the visible range outside the finned tube bundle and the reachable range of the wedge-shaped occlusion restricted area. The micro-deformation index data such as the lateral micro-bending of the tube centerline, the mid-span deflection of the support, and the small offset of the tube bundle relative to the support are output to improve the reliability and timeliness of micro-deformation monitoring of the finned tube bundle of the A-type overhead cooling island and provide data basis for maintenance decisions.

[0130] Example 2

[0131] Please see Figure 2 As shown, this embodiment provides a method for detecting micro-deformation of air-cooled island finned tube bundles, including:

[0132] The position and attitude information of the detection equipment, the reference geometric model data of the finned tube bundle, the observation point cloud data and the observation ray distribution data are acquired in advance;

[0133] Based on the spatial location of the supporting components and platform structure in the observation point cloud data, and combined with the position and attitude information of the detection equipment and the observation ray distribution data, the range data of the wedge-shaped blind zone is constructed.

[0134] Based on the observation point cloud data and the benchmark geometric model data, the actual centerline of each finned tube is extracted and compared with the corresponding benchmark centerline to obtain the geometric feature change data of the finned tube bundle.

[0135] Calculate the micro-deformation index data of finned tube bundles based on geometric feature change data;

[0136] The observability of the direction can be generated based on the observed ray distribution data and the wedge blind zone range data.

[0137] Oriented compensation is performed on the micro-deformation index data based on the wedge blind zone range data and the observability reliability of the direction.

[0138] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of protection of the claims.

Claims

1. A method for detecting micro-deformation of air-cooled island finned tube bundles, characterized in that, include: The position and attitude information of the detection equipment, the reference geometric model data of the finned tube bundle, the observation point cloud data and the observation ray distribution data are acquired in advance; Based on the spatial location of the supporting components and platform structure in the observation point cloud data, and combined with the position and attitude information of the detection equipment and the observation ray distribution data, the range data of the wedge-shaped blind zone is constructed. Based on the observation point cloud data and the benchmark geometric model data, the actual centerline of each finned tube is extracted and compared with the corresponding benchmark centerline to obtain the geometric feature change data of the finned tube bundle. Calculate the micro-deformation index data of finned tube bundles based on geometric feature change data; The observability of the direction can be generated based on the observed ray distribution data and the wedge blind zone range data. Oriented compensation is performed on the micro-deformation index data based on the wedge blind zone range data and the observability reliability of the direction.

2. The method for detecting micro-deformation of air-cooled island finned tube bundles according to claim 1, characterized in that, Methods for constructing wedge-shaped blind zone range data include: Extract the spatial locations of the supporting components and platform structure in the observed point cloud data to form surface fragment representations of the supporting components and platform structure. By combining the position and attitude information of the detection equipment with the distribution data of the observed rays, a spatial set of observed rays is constructed. Based on the surface fragment representation of the supporting component and the surface fragment representation of the platform structure, the spatial intersection relationship of the spatial set of observed rays is determined, and the reachable area and the occlusion-restricted area are obtained. Based on the reachable region and the occlusion-restricted region, the spatial set of observed rays is spatially divided to obtain candidate regions for wedge-shaped blind zones; Spatial connectivity processing is performed on the candidate regions of the wedge-shaped blind zone to obtain the range data of the wedge-shaped blind zone.

3. The method for detecting micro-deformation of air-cooled island finned tube bundles according to claim 1, characterized in that, Methods for obtaining geometric feature variation data of finned tube bundles include: Based on the finned tube bundle arrangement relationship in the benchmark geometric model data, a reference position sequence is generated and a local reference direction is established in the reference position sequence; Based on the reference position sequence, a spatial envelope range coaxial with the baseline centerline is constructed in the observed point cloud data. Points within the spatial envelope range are selected to form a candidate point set. A cross-sectional sampling zone is constructed based on the reference position sequence and the local reference direction. Points in the candidate point set that fall into the cross-sectional sampling zone are used as the cross-sectional point set sequence. Extract the cross-sectional center point sequence based on the cross-sectional point set sequence, and generate the actual centerline based on the cross-sectional center point sequence; The actual centerline is compared with the reference centerline based on the reference position sequence. The centerline offset information and the connection position offset information are recorded to obtain the geometric feature change data of the finned tube bundle.

4. The method for detecting micro-deformation of air-cooled island finned tube bundles according to claim 3, characterized in that, Methods for calculating the micro-deformation index data of finned tube bundles include: Calculate the lateral micro-bending of the pipe centerline based on geometric feature variation data; Calculate mid-span deflection of the support based on geometric feature variation data; The minute offset of the tube bundle relative to the support is calculated based on the geometric feature change data; The micro-deformation index data of the finned tube bundle are calculated by taking into account the lateral slight bending of the centerline of the converging tube, the mid-span deflection of the support, and the slight offset of the tube bundle relative to the support.

5. The method for detecting micro-deformation of air-cooled island finned tube bundles according to claim 4, characterized in that, Methods for calculating the lateral slight bend of the pipe centerline include: Read the centerline offset information and extract the offset component sequence in the lateral direction. Process the offset component sequence in the lateral direction based on the cross-end constraints provided by the connection position offset information, and extract the lateral micro-bending characterization. The maximum value of the transverse microbending characteristic in multiple spans of the same finned tube is selected as the transverse microbending amount of the tube centerline.

6. The method for detecting micro-deformation of air-cooled island finned tube bundles according to claim 1, characterized in that, Methods for generating directional observability confidence include: A set of spatial location indexes for finned tube bundles is established based on the wedge-shaped blind zone range data, and the blind zone attribution information is recorded; Extract the set of observation ray directions corresponding to each location in the spatial location index set based on the observation point cloud data; The classification results of the reliability of directional observability are determined based on the set of observed ray directions and the information on blind zone attribution. The ranking results of the observable confidence level at each location direction are aggregated into the micro-deformation index data to generate the observable confidence level at each direction.

7. The method for detecting micro-deformation of air-cooled island finned tube bundles according to claim 1, characterized in that, Methods for performing targeted compensation on micro-deformation index data include: Based on the wedge-shaped blind zone range data, the mid-span position index and the connection position index in the micro-deformation index data are mapped to obtain the blind zone attribution label; Based on the blind zone attribution markers, the micro-deformation index data are grouped to obtain wedge-shaped blind zone grouping and transition region grouping; Based on the directional observability confidence level, the compensation objects are determined by grouping the wedge blind zone and the transition region, and the main directional components of the target micro-deformation are limited to obtain the location interval and compensation components of the compensation objects. Based on the geometric feature change data, the reference sequence in the cross-pipe direction and the reference sequence along the pipe length direction are extracted in the location interval of the compensation object, and the compensation path is constructed based on the reference sequence; The compensation components of the compensation object's location range are updated according to the compensation path to obtain the micro-deformation index data after compensation.

8. The method for detecting micro-deformation of air-cooled island finned tube bundles according to claim 7, characterized in that, During the directional compensation process for micro-deformation index data, a mutation retention threshold is set, and structural mutation protection is implemented, specifically including: Based on the acquisition conditions of observation point cloud data and the discrete characteristics of geometric feature change data, a mutation retention threshold is set for each change sequence. The structural mutation protection coverage is determined based on the mutation retention threshold and directional observability confidence, combined with wedge blind zone range data. In targeted compensation, compensation rules are switched for the compensation components of the location interval of the compensation object based on the coverage of structural mutation protection.

9. The method for detecting micro-deformation of air-cooled island finned tube bundles according to claim 5, characterized in that, The method for obtaining the lateral direction is as follows: The lateral direction is established for each finned tube based on the baseline geometric model data.

10. A micro-deformation detection system for air-cooled island finned tube bundles, used to implement the micro-deformation detection method for air-cooled island finned tube bundles as described in any one of claims 1-9, characterized in that, include: The data acquisition module pre-acquires the position and attitude information of the detection equipment, the reference geometric model data of the finned tube bundle, the observation point cloud data, and the observation ray distribution data; The blind zone construction module constructs wedge-shaped blind zone range data based on the spatial position of supporting components and platform structure in the observation point cloud data, and combines the position and attitude information of the detection equipment with the observation ray distribution data. The feature analysis module extracts the actual centerline of each finned tube based on the observed point cloud data and the benchmark geometric model data, and compares it with the corresponding benchmark centerline to obtain the geometric feature change data of the finned tube bundle. The deformation analysis module calculates the micro-deformation index data of the finned tube bundle based on the geometric feature change data; The trusted computing module generates directional observability trust based on observed ray distribution data and wedge blind zone range data. The directional compensation module performs directional compensation on the micro-deformation index data based on the wedge blind zone range data and the observability reliability of the direction.

Images