Method for detecting butt joint of overhead conductor steel strand

Abstract

The application belongs to the technical field of overhead line butt joint detection, and specifically discloses an overhead conductor steel strand butt joint detection method, which comprises the following steps: constructing and correcting a digital twin model of a to-be-detected clamp based on CAD drawings and three-dimensional scanning data of the to-be-detected clamp; extracting interface geometric parameters and identifying the actual shape thereof; generating an initial grid detection path based on a shape matching grid node distribution strategy; collecting a first ultrasonic data set according to the detection path and identifying an abnormal area; dynamically adjusting the grid density according to the abnormal index of the abnormal area, generating an updated detection path, and collecting a second ultrasonic data set; and merging the first and second ultrasonic data sets to form a final detection data set. Through model correction and adaptive path planning, the application overcomes the shortcomings of a fixed point distribution mode, realizes a transition from extensive screening to accurate focusing, and significantly improves the defect identification accuracy and evaluation reliability.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of overhead line butt joint detection, and particularly relates to a method for detecting overhead conductor steel strand after butt joint. BACKGROUND

[0002] As the core load-bearing component of a power transmission line, the butt joint quality of the overhead conductor steel strand directly determines the safe and stable operation of the power grid. However, due to the complexity of high-altitude operation, hidden defects such as underpressure, overpressure, and loose strands are prone to occur, and therefore post-butt joint detection is required.

[0003] Prior art 1 is an ultrasonic detection method for a strain clamp and the pressure bonding quality of a steel-cored aluminum strand disclosed in Chinese patent application No. 201911265625.X. The method uses the same test conditions as actual detection to obtain a reference signal by establishing a model, detects the test piece to obtain an ultrasonic echo signal time sequence, compares the measured signal with the reference signal, and obtains the pressure bonding quality result.

[0004] Prior art 2 is a power transmission line clamp pressure joint quality ultrasonic detection and analysis system disclosed in Chinese patent application No. 202211469989.1. The system collects wall thickness data of the steel core and steel anchor, the notch and aluminum sleeve, and the aluminum strand and aluminum sleeve pressure bonding position through multiple ultrasonic data acquisition modules, combines a pressure interval screening module and a wall thickness interval integration module, and comprehensively analyzes and judges whether the strain clamp pressure bonding is qualified by a central analysis system.

[0005] Obviously, prior art 1 and prior art 2 both use non-destructive testing technology, but neither of them focuses on the optimization of front-end data acquisition. They usually use a fixed point monitoring mode, which cannot dynamically increase, reduce, or move the detection points for focused fine detection according to the preliminary collected signal quality or suspected defect areas, resulting in missing some data and thus insufficient accuracy of butt joint quality inspection and evaluation.

[0006] Secondly, the fixed point monitoring mode usually only relies on ideal CAD drawing models. However, the insertion depth and coaxiality of the steel core of the actual pressure bonded clamp must have machining and installation errors, which cannot truly reflect the actual state of the steel strand butt joint, and the detection path planning based on it has a large deviation compared with the actual situation.

[0007] At the same time, the current monitoring arrangement cannot be targeted for different shaped areas, nor can it be self-adaptive to complex spatial geometric features, and thus cannot generate detection paths self-adaptively for complex spatial geometric features, resulting in insufficient detection coverage of complex shaped areas. SUMMARY

[0008] In view of this, in order to solve the above problems, an overhead conductor steel strand butt joint detection method is proposed.

[0009] The purpose of the application can be achieved by the following technical solutions: the application provides an overhead conductor steel strand butt joint detection method, which comprises the following steps: constructing a digital twin model based on the CAD drawing of the to-be-inspected clamp, and extracting the geometric parameters of all compression joint interfaces and connection interfaces from the model.

[0010] According to the geometric parameters, the geometric feature set of each interface shape pre-stored in the inspection information library is called to identify the actual geometric shape of each interface.

[0011] Based on the identified actual geometric shape, a corresponding grid node distribution strategy is matched for each interface to generate an initial grid detection path covering all interfaces.

[0012] According to the initial grid detection path, a first ultrasonic data set is collected, and based on the first ultrasonic data set, an abnormal area is identified, and the spatial coordinates of the abnormal area are recorded.

[0013] Based on the spatial coordinates of the identified abnormal area and the first ultrasonic data set, the abnormal index of the abnormal area is counted, the grid node density of the abnormal area is adjusted based on the abnormal index, and an updated detection path is generated.

[0014] According to the updated detection path, a second ultrasonic data set is collected, and combined with the first ultrasonic data set, a butt joint detection data set is formed.

[0015] Compared with the prior art, the application has the following advantages: (1) The application combines spatial correction of the digital twin model with adaptive grid distribution strategy of the interface shape to realize dynamic optimization of the detection path, and overcomes the problem of detection blind area caused by geometric deviation and shape complexity in the fixed point distribution mode.

[0016] (2) The application constructs a digital twin model of the to-be-inspected clamp and implements spatial correction based on installation deviation parameters, so that the model is accurately aligned with the three-dimensional scanning data in space, effectively eliminating the geometric deviation caused by the manufacturing and installation process, and truly reflecting the actual butt joint state of the steel strand, and providing a real and reliable geometric basis for accurate planning of the subsequent detection path.

[0017] (3) The application automatically matches the corresponding grid node distribution strategy for different shaped interfaces based on the identification of the actual geometric shape of each interface to generate an initial grid detection path, solves the problem that the fixed point distribution mode cannot adapt to complex spatial geometric features, realizes targeted monitoring and layout of regions with different shapes, greatly improves the effectiveness of path planning, and reduces the risk of subsequent defect omission.

[0018] (4) This invention identifies suspected defect areas by analyzing the first ultrasonic dataset and dynamically adjusts the grid node density of the area based on its spatial coordinates and ultrasonic signal characteristics, thereby achieving precise focused detection of abnormal areas and significantly improving the detection rate and reliability of hidden defects. It overcomes the limitations of insufficient hidden defect detection capability under the fixed-point mode, effectively enhances the detection accuracy of potential quality risk areas, and further improves the accuracy and reliability of subsequent docking quality inspection and evaluation. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the overall implementation process of the present invention.

[0021] Figure 2 This is a schematic diagram of the actual geometric shape recognition process of the interface of the present invention.

[0022] Figure 3 This is a schematic diagram of the interface grid node distribution strategy matching process of the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Please see Figure 1 As shown, this invention provides a method for detecting the connection of overhead conductor steel strands after splicing, demonstrating the complete process of steps S1 to S6. Specifically, it includes constructing a digital twin model, extracting geometric parameters, identifying the interface shape, generating an initial detection path, acquiring a first ultrasonic dataset and identifying abnormal regions, calculating the anomaly index and adjusting the grid density, generating an updated detection path, and acquiring a second ultrasonic dataset to form a splicing detection dataset.

[0025] In one specific embodiment, the specific implementation process of the method includes: S1, constructing a digital twin model based on the CAD drawing of the clamp to be inspected, and extracting the geometric parameters of all crimping interfaces and connection interfaces from the model.

[0026] Considering that the clamp mainly faces the risks of structural loosening and electrical overheating in long-term operation, and the states of the three interfaces of the compression interface of the steel core and the steel anchor, the compression interface of the aluminum stranded wire and the aluminum sleeve, and the connection interface of the drainage plate and the wire directly correspond to the main mechanical tension, the main electrical current passage, and the key current diversion node, the failure of the three interfaces will directly lead to serious consequences such as wire breakage and overheating.

[0027] In addition, considering that there is a deviation between the actual spatial posture and the ideal design model of the overhead conductor clamp in the actual installation process due to the limitations of construction technology and site conditions, if the detection path is planned directly based on the original CAD model, the positioning of the ultrasonic probe and the real interface to be detected will be spatially misaligned, a detection blind area will be generated, and the evaluation accuracy will be affected.

[0028] Therefore, the present application marks the specific functional interfaces in the digital twin model in advance, so that the subsequent detection can be directly focused on the key parts that are most prone to failure and determine the service life of the product, realizing the leap from rough overall scanning to accurate diagnosis, and ensuring the effectiveness of the detection strategy and the pertinence of the risk assessment. At the same time, the theoretical model is rotated and translated through a three-dimensional scanning and spatial transformation algorithm, so that it is accurately aligned with the actual installation state of the clamp, thereby ensuring the consistency of the digital twin model and the actual spatial posture of the clamp.

[0029] In one specific embodiment, the specific implementation process of the construction of the digital twin model includes: S11, importing the CAD drawing of the clamp to be detected into a digital twin modeling software platform, automatically analyzing and identifying the key components such as the steel core, the steel anchor, the aluminum sleeve, and the drainage plate, and constructing a three-dimensional geometric model based on the geometric parameters designed for the components.

[0030] S12, identifying the compression region of the steel core and the steel anchor cylinder, the compression region of the aluminum stranded wire wrapped by the sleeve, and the contact region of the drainage plate and the wire connected by bolts or compression from the three-dimensional geometric model, and marking them as the compression interface of the steel core and the steel anchor, the compression interface of the aluminum stranded wire and the aluminum sleeve, and the connection interface of the drainage plate and the wire, respectively, thereby obtaining the three key functional interfaces of the compression interface of the steel core and the steel anchor, the compression interface of the aluminum stranded wire and the aluminum sleeve, and the connection interface of the drainage plate and the wire.

[0031] S13, using a three-dimensional scanner to obtain the surface point cloud data of the clamp on site, establishing a global coordinate system based on the point cloud data, taking the center of the steel anchor end face as the origin (0, 0, 0), establishing the Z-axis along the steel anchor axis, establishing the X-axis perpendicular to the steel anchor axis and passing through the center of the keyway, determining the Y-axis according to the right-hand rule, extracting the direction vectors of the steel core axis and the steel anchor axis in the global coordinate system, calculating the rotation axis vector through vector cross multiplication of the direction vectors of the steel core axis and the steel anchor axis, and calculating the included angle cosine value through vector dot multiplication of the two direction vectors. Finally, the actual coaxiality deviation angle between the steel core and the steel anchor axis is calculated by the arccosine function , Meanwhile, the point cloud data of the region where the cross section of the exposed strand corresponding end is located is extracted from the point cloud data, denoted as the strand end surface point cloud, the strand end surface point cloud is fitted by the least square method to obtain a fitting plane, the normal vector of the fitting plane is extracted, the Z coordinate of the intersection point of the normal vector and the Z axis is taken as the actual insertion depth of the strand, and the difference between the actual insertion depth of the strand and the design insertion depth marked in the CAD drawing is taken as the insertion depth deviation value of the strand. Subsequently, the normal vector and the curvature value of each point in the point cloud are calculated, the point cloud region whose curvature value exceeds the set curvature threshold value is denoted as the indentation region, for each indentation region, the center point thereof is extracted and denoted as the indentation center point, and a plurality of indentation center points are identified along the axial direction of the crimping region, the actual center axis of the crimping region is obtained by linear fitting, the projection point of the actual center axis in the Z axis direction of the global coordinate system is calculated, the Z coordinate of the projection point is extracted as the actual center position of the crimping, and the distance between the actual center position of the crimping and the design center position marked in the CAD drawing is calculated as the crimping position offset.

[0032] S14, three principal directions of the point cloud are determined by the principal component analysis method, and a local coordinate system is established with the centroid of the point cloud as the origin. Subsequently, the singular value decomposition algorithm is used to calculate the rotation matrix R between the two coordinate systems, and the translation vector T is calculated by the least square method. The spatial transformation matrix M is established based on the rotation matrix and the translation vector, M = [R | T], and all vertex coordinates of the three-dimensional geometric model are multiplied by the spatial transformation matrix M. The boundary representation data in the model is simultaneously transformed, including: the same coordinate transformation is applied to each NURBS surface or mesh patch and each boundary curve pair in the three-dimensional geometric model, while the average distance error between the model surface and the point cloud data is required to be less than the accuracy tolerance of the ultrasonic detection, so as to align the model and the point cloud data in the spatial position.

[0033] S15, the boundary representation data of the steel core is extracted from the model after the overall alignment, the boundary representation data includes the surface patch, the boundary curve and the vertex, and the three-dimensional model of the steel core assembly is established based on the boundary representation data. Then, based on the calculated installation deviation parameters, the corresponding spatial transformation is performed on the assembly: according to the strand insertion depth deviation value , the translation vector , is constructed, the translation transformation is performed on all vertex coordinates of the steel core assembly based on the translation vector , and the spatial position of the vertex of the steel core assembly is updated. Subsequently, based on the coaxiality deviation angle , the rotation matrix , , is a unit matrix, is a cross product matrix of the rotation axis vector, and then a rotation transformation is performed on the translated steel core assembly with the midpoint of the steel core axis as the rotation center, so as to update the spatial pose and the direction vector of the steel core assembly.

[0034] The embodiment of the present application ensures that the detection path generated in the subsequent step can accurately correspond to the real spatial coordinates of the steel core, the aluminum sleeve and other components through the two-stage correction of overall alignment and local adjustment, avoids problems such as poor coupling of the ultrasonic probe and the curved surface, signal misalignment and even missing scanning caused by path deviation, and simultaneously provides a reliable spatial reference for subsequent ultrasonic detection.

[0035] S16, re-integrate the corrected steel core assembly three-dimensional model into the three-dimensional geometric model, and output the spatially corrected three-dimensional geometric model.

[0036] The embodiment of the present application can accurately align the model and the three-dimensional scanning data in space by constructing the digital twin model of the wire clamp to be detected and implementing spatial correction based on the installation deviation parameters, effectively eliminates the geometric deviation caused by the machining and installation process, can truly reflect the actual butt joint state of the steel strand, and provides a real and reliable geometric basis for accurate planning of the subsequent detection path.

[0037] It should be noted that before the steel core assembly three-dimensional model is re-integrated into the three-dimensional geometric model, verification of the assembly rationality of the corrected steel core assembly independent three-dimensional model is further included, and only when the assembly rationality is verified, the corrected steel core assembly independent three-dimensional model is re-integrated into the three-dimensional geometric model, otherwise, the correction step is re-executed.

[0038] As a preferred example, the specific verification implementation process of the assembly rationality includes: calculating the minimum Euclidean distance between the outer surface of the steel core and the inner hole surface of the steel anchor, and if the minimum Euclidean distance is greater than 0, the contact interference verification is passed.

[0039] The conformal contact surface of the steel core and the steel anchor is extracted, the surface area of the actual contact area is calculated, and it is judged whether the design requirement is met, and if it is met, the contact area verification is passed.

[0040] The actual length of the crimping area is measured along the steel anchor axis direction, the spatial included angle between the steel core and the steel anchor axis is calculated, and if the actual length and the spatial included angle both meet the design tolerance requirement, the size verification is passed.

[0041] If the contact interference, the contact area and the size are all verified, it is judged that the assembly rationality of the corrected steel core assembly independent three-dimensional model is verified, otherwise, it is judged that the assembly is not reasonable.

[0042] If the assembly is unreasonable, the correction step is re-executed, and if it still fails after multiple corrections, it is determined that the line clamp has serious installation defects and the detection is terminated. In one specific embodiment, in order to ensure detection efficiency, the number of re-corrections does not exceed three times.

[0043] It should be noted that the design requirement is the minimum effective contact area specified in the technical specification of the type of strain clamp, and the design tolerance is derived from the assembly process file of the type of steel anchor and steel core.

[0044] On the digital twin model that has completed spatial correction, in order to facilitate the generation of subsequent detection paths, for each interface, an automatic geometric parameter extraction is performed: Since the compression interface of the steel core and the steel anchor is commonly cylindrical, for the compression interface of the steel core and the steel anchor, the extracted geometric parameters include but are not limited to the diameters and center coordinates of the interface at each equidistant cross section in the axial direction.

[0045] Since the compression interface of the aluminum strand and the aluminum sleeve is commonly spiral-shaped, for the compression interface of the aluminum strand and the aluminum sleeve, the extracted geometric parameters include but are not limited to the spatial vectors between adjacent points in the interface and the angle change between adjacent spatial vectors.

[0046] Since the connection interface of the drainage plate and the conductor is mostly planar, for the connection interface of the drainage plate and the conductor, the extracted geometric parameters include but are not limited to the surface normal vector of the connection contact area and the distance from each point in the point cloud to the fitted plane of the connection contact area.

[0047] Among them, the fitted plane is fitted by the Principal Component Analysis (PCA) algorithm, and the Principal Component Analysis (PCA) algorithm is a common technical means, which will not be described here.

[0048] S2, according to the geometric parameters, call the geometric feature set of each interface shape pre-stored in the inspection information library, and identify the actual geometric shape of each interface.

[0049] To support the subsequent interface shape recognition step, the present application preferably uses a pre-constructed inspection information library. Understandably, the inspection information library stores the geometric feature set of each standard interface shape and its setting conditions.

[0050] In one specific embodiment, the inspection information library can be constructed by the following steps: First, collect industry design drawings and historical similar overhead conductor steel strand butt joint qualified case data. The interface is divided into three categories: cylinder, spiral and plane, corresponding to the steel core-steel anchor, aluminum strand-aluminum sleeve, and drainage plate-conductor.

[0051] Next, for each type of interface, extract several parameters from its butt joint qualified case data to form a geometric feature set.

[0052] The present application preferably extracts the diameters and the coordinates of the center of the axial equidistant cross sections of the cylindrical-shaped compression interface as the extracted geometric parameters.

[0053] For the spiral shape, the three-dimensional coordinates of the continuous spatial points in the spiral-shaped compression interface are extracted, and the spatial vectors between adjacent points are calculated, and the central axis of the interface is extracted, the direction vector of the central axis is taken as the rotation axis, the spatial vectors between adjacent points are projected onto the plane perpendicular to the rotation axis to obtain the corresponding projection vectors, the dot product and the cross product of the two projection vectors are calculated, the angle change between the adjacent spatial vectors is calculated by the atan2 function, and the spatial vectors between adjacent points and the angle change between adjacent spatial vectors are taken as the extracted geometric parameters.

[0054] The atan2 function is an existing calculation function and will not be described again.

[0055] For the planar shape, the surface normal vector of the connection contact area where the planar-shaped connection interface is located and the distance of each point in the point cloud to the fitting plane of the connection contact area are extracted, and the surface normal vector and the distance of each point to the fitting plane of the connection contact area are taken as the extracted geometric parameters.

[0056] For each interface to which each shape belongs, the geometric parameters extracted from all qualified connection cases are generated into a geometric feature set.

[0057] Then, the setting conditions of the geometric parameters are determined: 1) the cylindrical shape setting condition: select the steel core and steel anchor connection cases that have passed the detection from the butt joint qualified case data, and take them as qualified samples.

[0058] Based on the qualified samples, the diameter measurement values of the axial equidistant cross sections are counted, the average value and the standard deviation of each cross section are calculated, ±2 times the standard deviation of the average value is taken as the diameter qualified range, and the coordinates of the center of the axial equidistant cross sections are counted, the radial offset of the center of each cross section relative to the theoretical axis is calculated, and the 95% percentile value is taken as the maximum allowed center offset.

[0059] 2) The setting condition of the spiral shape: select the qualified wire connection cases from the butt joint qualified case data as qualified wire samples.

[0060] Based on the qualified wire samples, the angle change of the adjacent spatial vectors is counted, the average value and the standard deviation are calculated, and the qualified range of the angle change is set in the same way as the setting method of the diameter qualified range, and based on the qualified wire samples, the lengths of the spatial vectors are counted, the standard deviation and the average value of the lengths are calculated, the ratio of the standard deviation and the average value is taken as the coefficient of variation, and the average value of the coefficients of variation corresponding to all qualified wire samples is taken as the maximum allowed coefficient of variation.

[0061] 3) Plane shape setting condition: select the detected qualified drainage plate connection case from the interface qualification case data as the qualified connection sample.

[0062] Based on the qualified connection sample, the angle deviation of the surface normal vector and the theoretical normal vector is calculated, the 95% quantile value is taken as the maximum allowed deviation angle, and the distance of each point in the point cloud to the fitting plane is calculated based on the qualified sample, and a distance sequence is constructed.

[0063] The range value and the coefficient of variation of the distance sequence are calculated, and the average of the range value and the coefficient of variation of the distance sequence corresponding to all qualified samples is taken as the maximum allowed range value and the maximum allowed coefficient of variation, respectively.

[0064] Finally, the three interface shape categories, their corresponding geometric feature sets, and the setting conditions of each geometric feature parameter are mapped and stored in the database to form the final inspection information library.

[0065] It should be noted that by calling this library, the present application realizes the rapid identification of the geometric shape of each interface in the digital twin model, which provides a key prerequisite for subsequent generation of precise and adaptive detection paths.

[0066] In actual overhead conductor steel strand butt joint engineering, due to manufacturing tolerances, installation errors and long-term running wear, the actual geometric shape of the clamp interface often differs from the ideal standard shape, and a single yes / no judgment logic cannot adapt to this complexity. Secondly, the generation quality of the ultrasonic detection path directly depends on the accuracy of shape recognition, and different interface shapes have different effects on the reflection and propagation characteristics of ultrasonic waves. Based on this, the present application designs a hierarchical identification strategy from complete matching to approximate matching, which not only ensures the identification accuracy under standard conditions, but also has the ability to adapt to complex conditions.

[0067] Further, as a preferred example, please refer to Figure 2 As shown, the actual geometric shape of each interface is identified by comparing the interface geometric parameters extracted from the digital twin model with the geometric feature set of each standard interface shape in the inspection information library. If the geometric parameters of a certain interface match the geometric feature set of a certain interface shape, further compare these matched geometric parameters with the corresponding setting conditions. If all geometric parameters meet the setting conditions of the standard, the interface shape is taken as the final actual geometric shape of the interface.

[0068] For example, the sequence of the angle change of the adjacent spatial vectors of a certain spiral interface is The average value is And the standard deviation is All are in the eligible range of the set condition. The coefficient of variation of each space vector module length is 0.08, which is less than the maximum allowable coefficient of variation 0.12, so the interface is determined to be a completely matched spiral shape.

[0069] When all the geometric parameters do not completely meet the set condition, the geometric parameters of the interface are matched and compared with the geometric parameters in the corresponding geometric feature set of each interface shape associated with the interface, the ratio of the number of matched geometric parameters to the total number of geometric parameters of the interface is taken as the reference matching degree, for the parameters that do not meet the set condition, the relative deviation value from the maximum allowable value in the set condition is calculated, and the minimum-maximum normalization processing is performed to obtain the normalized deviation. Take the average value of all normalized deviations as the reference normalized deviation, and input the reference normalized deviation as an input variable into the exponential decay function , to obtain a matching degree correction factor, , the input variable, , the matching degree correction factor, , a natural constant, wherein the larger the input variable, the greater the deviation degree, and the more dissimilar the corresponding shape, that is, the smaller the correction factor.

[0070] The product of the reference matching degree and the matching degree correction factor is output as the final matching degree, so as to obtain the matching degree of the interface and each interface shape associated with the interface. The interface shape with the highest matching degree is taken as the final actual geometric shape of the interface, and an approximate mark is made, and the matching degree of the interface and the corresponding matching interface shape is recorded.

[0071] The present application expands the judgment of interface shape from the traditional binary judgment to a quantitative evaluation mode containing complete matching and approximate matching, which not only ensures the optimal generation of the detection path under the standard shape, but also dynamically adjusts the grid density according to the matching degree to adapt to the shape deviation in actual engineering, thereby effectively improving the accuracy of defect recognition under complex working conditions.

[0072] S3, based on the recognized actual geometric shape, matching the corresponding grid node distribution strategy for each interface, generating an initial grid detection path covering all interfaces.

[0073] The sound wave propagation characteristics of different geometric shapes are different, and it is necessary to design a detection node distribution mode matched therewith. Considering the current overhead conductor steel strand detection setting experience, the cylindrical interface adopts matrix distribution in the axial and circumferential directions to ensure complete coverage of the continuous curved surface. The spiral interface adopts path planning along the spiral trend to maintain geometric consistency with the strand structure. The planar interface adopts uniform grid distribution to achieve efficient full-area scanning.

[0074] For the approximately matched interface, in order to further guarantee the coverage of detection, the application adjusts the detection density dynamically through the matching degree, increases the detection points in the area where the shape deviation may exist, and realizes the balance between detection accuracy and efficiency.

[0075] In one specific embodiment, referring to Figure 3 The matching process of the grid node distribution strategy is implemented in the following manner: if the interface is identified as a cylindrical shape, axial grids are divided at fixed intervals along the axis of the cylinder, circumferential grids are divided at fixed angles along the circumference, four corner points and a center point of each grid cell are taken as detection nodes, and a matrix type detection path around the surface of the cylinder is constructed.

[0076] If the interface is identified as a spiral shape, the aluminum strand is divided into several spiral turns along the axis direction according to the pitch length, the grid center point is taken as the detection node for the plane where each spiral turn is located, and the detection nodes of each spiral turn are generated in spiral order to generate a spiral detection path.

[0077] Example: for an aluminum strand with a pitch of 25mm, divided into 2 complete spiral turns within a range of 50mm, 12 detection nodes are divided at intervals for each turn, a total of 24 detection points are generated along the spiral direction.

[0078] If the interface is identified as a planar shape, the contact area is divided according to the rectangular grid, the grid size is dynamically adjusted according to the size of the interface, the center point of each grid is taken as the detection node, and a planar detection path is formed.

[0079] If the interface has an approximate mark, the number of detection nodes arranged for each interface is counted and divided by the corresponding interface surface area to obtain the detection density, denoted as , and the recorded matching degree is denoted as . The value range is between 0 and 1.

[0080] The updated grid density , , and the detection path is updated based on the updated grid density.

[0081] Through this shape adaptive grid node distribution strategy, the application realizes the targeted matching of the detection path and the geometric characteristics of the interface, ensuring the comprehensiveness and effectiveness of ultrasonic detection. For standard shape interfaces, the node distribution scheme corresponding to the shape is adopted to improve the efficiency while ensuring the detection quality. For approximately shaped interfaces with deviations, the detection density is dynamically adjusted through the matching degree, improving the defect detection capability under complex working conditions, and providing reliable technical support for the butt joint quality evaluation of overhead conductors.

[0082] ​Secondly, the application also overcomes the limitation of insufficient detection ability of hidden defects in the fixed point mode, effectively enhances the detection accuracy of potential quality risk areas, and further improves the accuracy and reliability of subsequent quality inspection and evaluation.

[0083] S4, collect a first ultrasonic data set according to the initial meshing detection path, identify an abnormal area based on the first ultrasonic data set, and record the spatial coordinates of the abnormal area.

[0084] The ultrasonic signal will produce specific time domain and frequency domain characteristic changes when encountering material defects, based on which, the application extracts time domain characteristic parameters and frequency domain characteristic parameters from the first ultrasonic data set, and identifies the abnormal area.

[0085] In one specific embodiment, the abnormal area identification process is implemented in the following steps: filtering and denoising each detection node signal in the first ultrasonic data set, extracting time domain characteristic parameters including but not limited to initial amplitude, primary bottom wave amplitude and signal attenuation rate, and extracting frequency domain characteristic parameters including but not limited to center frequency, frequency bandwidth and spectral peak.

[0086] For each detection node, it is judged whether all the following conditions are met: the ultrasonic signal collected by the detection node is complete.

[0087] The signal-to-noise ratio of the ultrasonic signal collected by the detection node is higher than a preset first threshold.

[0088] The initial amplitude or primary bottom wave amplitude of the detection node is within a preset qualified range.

[0089] If all the conditions are met, the detection node is recorded as an effective node.

[0090] It should be noted that the first threshold and the qualified range are determined by analyzing a large number of historical ultrasonic data of qualified line clamp butt joint detection cases.

[0091] It should also be noted that the filtering and denoising both use current filtering and denoising technology, which is not limited here.

[0092] Specifically, collect historical butt joint qualified case data of the same type of overhead conductor steel strand, extract the ultrasonic data in the cases, count the signal-to-noise ratios of all signals from the ultrasonic data, take the mean value of the signal-to-noise ratios corresponding to all signals as the reference signal-to-noise ratio of the corresponding butt joint qualified case, sort all the reference signal-to-noise ratios of the butt joint qualified cases in ascending order, and take the value corresponding to the 5% quantile in the sorting as the first threshold.

[0093] The initial wave amplitude or the first bottom wave amplitude of the ultrasonic signal corresponding to all historical butt joint detection cases of the wire clamp is counted, and initial wave amplitude sequence and first bottom wave amplitude sequence are constructed, and the 5% quantile of the amplitude value sequence is taken as the lower limit of the qualified range, and the 95% quantile is taken as the upper limit of the qualified range.

[0094] For all valid detection nodes on the same interface, the feature parameter sample set is formed by collecting each feature parameter, the first quartile (Q1) ) and the third quartile (Q3) ) of each feature parameter sample set are calculated, and the feature parameter reference range of the interface is established as Q1 , is a constant greater than 0, and k is taken as a reference for normal distribution experience rule, and is used to distinguish the main data and most potential abnormalities, and as a preferred example, the value can be 1.5.

[0095] It should be noted that the specific value of k can be dynamically selected according to the specific industrial detection scene, for example, for the overhead conductor steel strand butt joint detection scene with high precision requirement, the value of k can be reduced to 1, so as to reduce the reference range of the feature parameter, and a more strict judgment strategy is executed to improve the identification precision of the abnormal area, and for the overhead conductor steel strand butt joint detection scene with low precision requirement, the value of k can be enlarged to 2, so as to improve the identification sensitivity of the abnormal area.

[0096] The detection nodes whose feature parameters exceed the corresponding feature parameter reference range are recorded as abnormal detection nodes, a spatial clustering algorithm (such as DBSCAN) based on Euclidean distance is used to cluster the abnormal nodes to form a connected suspected defect area, which is recorded as an abnormal area, and the spatial boundary coordinates of the area are recorded.

[0097] It should be noted that, in order to avoid the error combination of different defects which are far apart in space and independent of each other, and to ensure that the abnormal points belonging to the same physical defect but possibly slightly divided by the initial grid are correctly aggregated together, the clustering distance threshold can be set to be slightly larger than the grid spacing, and as a preferred example, the clustering distance threshold is set to 1.5 times the grid spacing.

[0098] The clustering distance threshold can be defined by the detection personnel according to the actual detection precision requirement, and the given value by the present application is only a preferred value example, not a fixed value.

[0099] The embodiment of the present application can effectively distinguish normal areas from abnormal areas by establishing a characteristic parameter benchmark range based on effective detection node statistics. Meanwhile, considering the continuity of defects, a spatial clustering algorithm is used to aggregate discrete abnormal detection nodes into connected abnormal areas, which can provide accurate spatial targets for subsequent fine detection. At the same time, it fundamentally avoids the risk of misjudging invalid or abnormal signals as defects, and further prevents defects from being missed due to signal quality problems, providing a reliable data basis for subsequent accurate identification of real material defects.

[0100] S5, based on the spatial coordinates of the identified abnormal area and the first ultrasonic data set, statistics the abnormal index of the abnormal area, adjusts the grid node density of the abnormal area based on the abnormal index, and generates an updated detection path.

[0101] Considering that the abnormal area found in the initial gridding detection may have potential defects, but its severity and characteristics are not clear. If the previous detection path is continued, it is easy to cause the key area to be unable to focus.

[0102] Based on this, the present application adjusts the grid node density of the abnormal area by statistics the abnormal index of the abnormal area, so as to preferentially allocate limited detection resources to the most suspicious area, that is, by increasing detection nodes in high-risk areas and maintaining basic coverage in low-risk areas, the evaluation accuracy of potential defects is greatly improved on the premise of increasing total detection time, so as to complete the optimization from preliminary screening to accurate diagnosis.

[0103] In one specific embodiment, the abnormal index of the abnormal area is implemented according to the following steps: based on the spatial coordinates of the identified abnormal area, the ultrasonic signals of all detection nodes in the area are extracted from the first ultrasonic data set.

[0104] The average signal attenuation rate of all detection nodes in the abnormal area is calculated, the signal attenuation rate is the ratio of the first bottom wave amplitude to the initial wave amplitude, and the amplitude coefficient of variation corresponding to the initial wave amplitude of all detection nodes in the calculation area is calculated, and at the same time, the proportion of abnormal nodes to the total number of detection nodes is calculated to obtain the abnormal node proportion.

[0105] The average signal attenuation rate is subjected to minimum-maximum normalization processing to obtain a signal attenuation rate normalization value, and the linear weighted sum of the signal attenuation rate normalization value, the amplitude coefficient of variation and the abnormal node proportion is taken as the abnormal index of the abnormal area.

[0106] It should be noted that the average signal attenuation rate reflects the severity of the overall attenuation of ultrasonic energy caused by defects, and when encountering volume defects such as pores and inclusions, the scattering of the acoustic wave will cause the attenuation rate to increase significantly, therefore, the average signal attenuation rate weight is set to be the largest, the wave amplitude coefficient of variation reflects the inconsistency or dispersion degree of the signals in the region, which can effectively capture the microscopic heterogeneity of the defects, the weight is second, the proportion of abnormal nodes represents the size of the defect range from the spatial distribution, although it can effectively identify the macroscopic influence range of the defect, but since it depends on the results of the preliminary determination, therefore, the weight is set to be the smallest, and the weights of the three satisfy the sum of 1, as a preferred example, the weights of the signal attenuation rate, the wave amplitude coefficient of variation and the proportion of abnormal nodes can be respectively 0.5, 0.3 and 0.2.

[0107] In another specific embodiment, the adjustment of the grid node density of the abnormal region is implemented by the following steps: taking the geometric center of the suspected defect region as the center, a three-dimensional minimum circumscribed rectangular body is constructed based on the spatial boundary coordinates thereof. The rectangular body is expanded outward by a safety boundary distance to form a final local encryption area.

[0108] The current grid density of the local encryption area is denoted as In the local encryption area, the grid node density is adjusted to , The abnormal index is denoted as

[0109] Since the grid density changes suddenly from the encryption area to the normal area, it will cause the ultrasonic probe to need to frequently stop and start, which not only reduces the detection efficiency, but also introduces noise due to mechanical vibration, and even damages the equipment. Therefore, a transition zone with a preset width is provided inside and outside the boundary of the local encryption area, in the transition zone, the grid node density is linearly decreased from to according to a preset decreasing gradient.

[0110] The grid node density outside the local encryption area is maintained unchanged.

[0111] It should be noted that the safety boundary distance and the transition zone width can be determined according to the detection accuracy requirement and the equipment performance, and are usually 1-2 times the current grid spacing.

[0112] It should be further noted that the decreasing gradient is determined according to the equipment requirements of the overhead wire, and can be defined by the detection personnel, or can take the existing transition decreasing empirical value.

[0113] Through the embodiment, the grid encryption degree is ensured to be positively correlated with the severity of the potential defect. The transition zone avoids the sudden change of the grid density, and ensures the stability of the ultrasonic probe movement and the continuity of the data acquisition.

[0114] Secondly, the optimized detection grid finally generated by the embodiment of the present application can keep the overall detection efficiency while carrying out higher-density investigation on the high-risk area, thereby greatly improving the detection probability of micro-defects and critical defects, providing a more reliable data basis for the safety evaluation of the overhead conductor, and further realizing the precise focused detection of the abnormal area, significantly improving the detection rate and evaluation reliability of hidden defects. The limitations of insufficient detection ability of hidden defects under the fixed point mode are overcome, the detection accuracy of potential quality risk areas is effectively enhanced, and the accuracy and reliability of subsequent joint quality inspection and evaluation are further improved.

[0115] S6, collect a second ultrasonic data set according to the updated detection path, and form a joint detection data set combining the first ultrasonic data set, the joint detection data set being formed by merging the first ultrasonic data set and the second ultrasonic data set, and the detection nodes of the overlapping area being based on the second ultrasonic data set to ensure data uniqueness.

[0116] It should be noted that the first data set provides complete reference information based on the initial gridded detection path, and the second data set provides higher-density scanning of the abnormal area. By merging the two, the detection background of the overall structure can be retained, and more detailed detection data of the abnormal area can be obtained, which not only effectively improves the accuracy of defect identification and eliminates accidental errors that may exist in single scanning, but also provides a more comprehensive and reliable data basis for subsequent quantitative evaluation, defect classification, etc.

[0117] The above is merely an example and description of the concept of the present application, and those skilled in the art can make various modifications or supplements to the described specific embodiments or use similar ways to replace them, as long as they do not deviate from the concept of the present application or exceed the scope defined by the present application, which shall belong to the protection scope of the present application.

Claims

1. A method for testing the connection of steel strands in an overhead conductor, characterized in that, The method includes: A digital twin model is constructed based on the CAD drawings of the clamp to be inspected, and the geometric parameters of all crimping interfaces and connection interfaces are extracted from the model. The construction of digital twin models includes: Import the CAD drawing of the clamp to be inspected; Based on the CAD drawings, a three-dimensional geometric model is constructed, including the crimping interface between the steel core and the steel anchor, the crimping interface between the aluminum stranded wire and the aluminum sleeve, and the connection interface between the diversion plate and the conductor. The point cloud data of the surface of the clamp to be inspected is obtained by a 3D scanner. Based on the point cloud data, the coaxial deviation angle between the steel core and the steel anchor, the deviation value of the stranded wire insertion depth and the offset of the crimping position are calculated and recorded as installation deviation parameters. Based on the installation deviation parameters, the three-dimensional geometric model is spatially corrected, and the spatially corrected three-dimensional geometric model is output. Based on the geometric parameters, the geometric feature set of each interface shape is called from the inspection information database to identify the actual geometric shape of each interface. Based on the identified actual geometry, a corresponding grid node distribution strategy is matched for each interface to generate an initial gridded detection path covering all interfaces; The first ultrasound dataset is collected according to the initial gridded detection path. Based on the first ultrasound dataset, abnormal areas are identified and the spatial coordinates of the abnormal areas are recorded. Based on the spatial coordinates of the identified abnormal regions and the first ultrasound dataset, the abnormality index of the abnormal regions is calculated, the grid node density of the abnormal regions is adjusted based on the abnormality index, and an updated detection path is generated. Among them, the abnormality index of the statistically abnormal area includes: From the first ultrasound dataset, the average signal attenuation rate and amplitude variation coefficient of the ultrasound signals corresponding to all detection nodes in the abnormal region are statistically analyzed, and the ratio of abnormal nodes to the total number of detection nodes is recorded as the abnormal node ratio. The average signal attenuation rate is normalized to obtain a normalized signal attenuation rate value. The weighted sum of the normalized signal attenuation rate value, the amplitude variation coefficient, and the proportion of abnormal nodes is used as the anomaly index of the abnormal region. The second ultrasound dataset is collected according to the updated detection path and combined with the first ultrasound dataset to form a docking detection dataset. The docking detection dataset is formed by merging the first ultrasound dataset and the second ultrasound dataset.

2. The method for detecting the connection of overhead conductor steel strands as described in claim 1, characterized in that: The spatial correction process for the three-dimensional geometric model includes: Establish a global coordinate system based on CAD drawings and a local coordinate system based on 3D scanned point cloud data; Calculate the spatial transformation matrix between the global coordinate system and the local coordinate system; Based on the spatial transformation matrix, a global rotation and translation transformation is performed on the three-dimensional geometric model to align the model with the point cloud data in spatial position. Identify and extract steel core components from the aligned 3D geometric model; Based on the stranded wire insertion depth deviation value, an axial translation transformation is performed on the steel core assembly to correct the actual insertion depth of the steel core in the steel anchor; Based on the coaxial deviation angle between the steel core and the steel anchor, a rotation transformation is performed on the steel core assembly to correct the directional deviation between the steel core axis and the steel anchor axis; The corrected steel core assembly is reintegrated into the three-dimensional geometric model, and the spatially corrected three-dimensional geometric model is output.

3. The method for detecting the connection of overhead conductor steel strands as described in claim 1, characterized in that: The interface shapes include cylindrical, spiral, and planar shapes, wherein identifying the actual geometric shape of each interface includes: The geometric parameters extracted from each interface are matched and compared with the geometric feature set of each interface shape. If the geometric parameters of a certain interface match the geometric feature set of a certain interface shape, establish the association between the interface and the interface shape, and compare the matching geometric parameters with the set conditions of the corresponding geometric parameters in the matching geometric feature set. If all geometric parameters meet their calibration settings, the interface shape will be taken as the final actual geometric shape of the interface. Otherwise, calculate the matching degree between the interface and the shapes of each associated interface, take the interface shape with the highest matching degree as the final actual geometric shape of the interface, make an approximate mark, and record the matching degree between the interface and the corresponding matching interface shape. This allows us to identify the actual geometry of each interface.

4. The method for detecting the connection of overhead conductor steel strands as described in claim 3, characterized in that: The statistical process for the matching degree includes: The geometric parameters of the interface are matched and compared with the geometric parameters in the geometric feature set corresponding to the shapes of the associated interfaces. The ratio of the number of matched geometric parameters to the total number of geometric parameters of the interface is used as the benchmark matching degree. For parameters that do not meet the set conditions, calculate their relative deviation from the set conditions and normalize them to obtain the normalized deviation. The average value of all normalized deviations is taken as the reference normalized deviation. The reference normalized deviation is then input into the exponential decay function to obtain the matching degree correction factor. The product of the baseline matching degree and the matching degree correction factor is output as the final matching degree, thereby obtaining the matching degree of the interface with the shapes of each associated interface.

5. The method for detecting the connection of overhead conductor steel strands as described in claim 3, characterized in that: The strategy for matching the corresponding grid node distribution for each interface includes: If the interface is identified as a cylinder, it is divided into grids along the circumference and axis of the cylinder, and the corner points and center points of the grid are taken as detection nodes to construct a matrix detection path around the surface of the cylinder. If the interface is identified as a spiral, a grid is divided along the circumferential plane of each spiral loop. The center point of each grid is taken as the detection node of the corresponding spiral loop. The detection nodes of each spiral loop are then used to generate a spiral detection path according to the spiral sequence. If the interface is identified as a planar shape, the planar shape is divided into grids, and the center point of the grid is taken as the detection node to form a planar detection path; If there are approximate markers on the interface, count the number of detection nodes placed on each interface and divide by the corresponding interface surface area to obtain the detection density, denoted as . And record the matching degree as ; Calculate the updated mesh density , The detection path is then updated based on the updated grid density.

6. The method for detecting the connection of overhead conductor steel strands as described in claim 1, characterized in that: The abnormal region identification includes: The ultrasound signal of each detection node in the first ultrasound dataset is filtered and denoised. Based on the filtered and denoised ultrasonic signal, the time-domain and frequency-domain feature parameters of each detection node are extracted, and the effective detection nodes are identified. For each extracted feature parameter, the feature parameters corresponding to all valid detection nodes on the same interface are collected to form a parameter sample set; Calculate the first and third quartiles of the parameter sample set, and establish the baseline range of the characteristic parameters of the interface; Detection nodes whose feature parameters exceed the baseline range of the corresponding feature parameters are recorded as abnormal detection nodes. Abnormal nodes are spatially clustered to form a connected abnormal region, and the spatial boundary coordinates of the region are recorded.

7. The method for detecting the connection of overhead conductor steel strands as described in claim 6, characterized in that: The identified valid detection nodes include: For each detection node, determine whether all of the following conditions are met: The ultrasound signals acquired by this detection node are complete; The signal-to-noise ratio of the ultrasonic signal acquired by this detection node is higher than a preset first threshold. The initial amplitude or primary bottom wave amplitude of the detection node is within the preset acceptable range; If all conditions are met, the detection node is recorded as a valid node.

8. The method for detecting the connection of overhead conductor steel strands as described in claim 1, characterized in that: The adjustment of the grid node density in the abnormal region includes: Using the geometric center of the abnormal region as a reference, a minimum bounding rectangle is constructed based on its spatial boundary coordinates; Extend the rectangle outward by a preset security boundary distance, and mark the extended space as a local encrypted area; The current grid density of this local encryption region is denoted as Within the local encryption zone, the grid node density is adjusted to , Indicates an abnormality index; A transition zone is set inside and outside the boundary of the local encryption zone. Within the transition zone, the grid node density decreases from... linearly decreasing to ; Maintain the grid node density outside the local encrypted zone unchanged.

Images