A flexible magnetic flux leakage inspection method for wire rope of a sling anchorage zone

Abstract

The present application relates to the technical field of ferromagnetic cable body nondestructive testing, and discloses a flexible magnetic flux leakage inspection method for steel wires in a sling anchorage zone. The method establishes a steel wire bundle topology model and divides candidate defect units, and selects an excitation coding set according to the expected magnetic response of different candidate defect units; a flexible partition magnetization sensing component wrapped in the anchorage zone is used to adjust the fitting state and excitation parameters according to the lift-off distance, fitting pressure and local curvature, perform forward and reverse coding excitation, and collect double-layer three-axis magnetic field data; the defect sensitive response is obtained through lift-off correction, coordinate conversion and baseline difference, the defect position, cross-section loss and positioning uncertainty are inverted in combination with the coding magnetic response matrix and topology constraints; when the positioning uncertainty exceeds the threshold, a fine scanning action is selected according to the information gain and retested. The method can improve the positioning and quantitative detection capability of steel wire defects in the sling anchorage zone.

Description

Technical Field

[0001] This invention relates to the field of nondestructive testing technology for ferromagnetic cables, and in particular to a flexible magnetic flux leakage inspection method for steel wires in the anchorage zone of suspension cables. Background Technology

[0002] Lifting slings are typically composed of multiple high-strength steel wires, with their ends connected to the main structure via anchoring structures. The anchoring zone bears the responsibility of transferring the sling load to the main structure. During long-term service, they may be affected by factors such as load changes, environmental corrosion, and local structural constraints, making the steel wires prone to breakage, corrosion, wear, slippage, and abnormal stress. Therefore, effective inspection of the steel wires in the anchoring zone is a crucial step in ensuring the safe operation of the slings.

[0003] Magnetic flux leakage (MFL) testing is a non-destructive testing method used for defect detection in ferromagnetic materials. It involves magnetizing the object under test and collecting the changes in the magnetic field caused by defects to determine if any abnormalities exist. This method can be used to inspect ferromagnetic cables such as wire ropes and slings.

[0004] However, compared to the free section of the cable, the structural morphology and magnetic field environment of the anchorage zone are more complex. The steel wires within the anchorage zone may exhibit expansion, contraction, dispersion, bending, or localized torsion, making it difficult for the detection device to maintain a stable fit at different locations. This results in variations in magnetization and signal acquisition conditions at each detection point. Furthermore, the distance between the detection component and the steel wire may change with local surface morphology, causing fluctuations in the acquired magnetic field signal.

[0005] In addition, the anchorage zone typically includes anchor cups, anchor plates, sleeves, or other ferromagnetic components. These components affect the magnetic flux distribution near the steel wire and create a strong background magnetic field. Changes in sling load and ambient temperature can also cause changes in the magnetic properties of the steel wire, causing the defect response to superimpose with the background changes, increasing the difficulty of anomaly identification.

[0006] Because slings consist of multiple steel wires arranged in different circumferential and radial layers, the magnetic field responses generated by the wires at different locations may couple with each other. Existing detection results often struggle to accurately distinguish the wire location and radial layer corresponding to anomalies, and also fail to maintain consistency in detection results when detection conditions change. If the initial detection results are insufficient to determine the defect state, the information obtained from repeated detections may still be limited.

[0007] Therefore, how to adapt to the complex structural form and magnetic field environment of the anchorage zone of the sling, reduce the influence of factors such as fit, detection distance, load and temperature on the detection results, and improve the ability to identify, locate and quantify steel wire defects inside the anchorage zone remains a technical problem that needs to be solved in this field. Summary of the Invention

[0008] To address the shortcomings of existing technologies, the present invention aims to provide a flexible magnetic flux leakage inspection method for steel wires in the anchorage zone of slings, which reduces the influence of complex structural morphology, lift-off changes, load temperature and background magnetic field on the detection results, and improves the ability to identify, locate and quantitatively detect defects in different steel wire sub-bundles and radial levels.

[0009] To achieve the above objectives, the present invention provides the following technical solution: a flexible magnetic flux leakage inspection method for steel wires in the anchorage zone of a sling, comprising: Step S1: Obtain the wire arrangement, radial hierarchy, axial direction and anchor structure, and establish a wire bundle topology model containing the adjacent relationships of the wires; divide the wire sub-bundles according to the adjacent relationships of the wires, and divide the candidate defect units according to the axial interval, wire sub-bundles and radial hierarchy. Step S2: The flexible partitioned magnetization sensing component, which includes a flexible bearing body, a circumferential excitation sector, an axial detection partition, a partitioned pushing mechanism, and a double-layer triaxial magnetic sensitive array, is wrapped around the anchoring area; the candidate excitation code is composed of the positive, negative, or stopped states of the circumferential excitation sector. Step S3: Obtain the expected magnetic response of the candidate defect unit under the candidate excitation coding, select the excitation coding set according to the response difference of different candidate defect units, and establish the coding magnetic response matrix of the corresponding candidate defect unit. Step S4: Obtain the lift-off distance, bonding pressure and local curvature of each axial detection zone, and adjust the bonding state and excitation parameters accordingly; when the lift-off distance and bonding pressure meet the acquisition conditions, drive the circumferential excitation sector according to the excitation code set, and at least one code performs forward and reverse excitation, and simultaneously acquire the double-layer triaxial magnetic field, cable force and temperature. Step S5: Perform lifting correction based on the double-layer triaxial magnetic field and lifting distance; reconstruct the local coordinate system of the steel wire according to the local curvature and the attitude of the carrier; update the preset positive and negative baseline responses based on cable force and temperature; and subtract the corrected positive and negative excitation responses from the corresponding baselines to obtain the defect-sensitive response. Step S6: The defect-sensitive responses are combined into a detection response vector. The defect location, cross-sectional loss, and positioning uncertainty are inverted based on the encoded magnetic response matrix and topological constraints. When the positioning uncertainty is greater than the preset threshold, the information gain is determined according to the uncertainty reduction and cost of the bonding or encoding adjustment. The maximum value is selected for retesting and inversion until it is no greater than the preset threshold or the upper limit of the number of times is reached. The result is then output.

[0010] Furthermore, the anchorage area is continuously divided into multiple axial intervals along the axial direction of the sling, and each axial interval covers the anchorage area. The boundaries of adjacent axial intervals are connected and their interiors do not overlap. Within each radial level of each axial interval, all the wires in that radial level are divided into multiple wire sub-bundles according to the wire adjacency relationship. Each wire sub-bundle includes at least one wire that is spatially adjacent to the other. The same wire belongs to only one wire sub-bundle within the same axial interval and the same radial level. A combination of an axial interval, a wire bundle, and a radial level is considered as a candidate defect unit.

[0011] Furthermore, each candidate excitation code includes multiple coding elements that correspond one-to-one with multiple circumferential excitation sectors, and each coding element represents the forward excitation state, reverse excitation state, or stop excitation state of the corresponding circumferential excitation sector. The dual-layer triaxial magnetic array includes multiple triaxial magnetic sensors, and each magnetic field component collected by each triaxial magnetic sensor on the three measurement axes is used as a magnetic channel. Each magnetic sensing channel is uniquely determined by the axial detection position, circumferential detection position, magnetic array level, and measurement axis of the corresponding triaxial magnetic sensing sensor; The combination of a candidate excitation code, an excitation intensity, and a magnetic sensing channel is taken as a response term. The expected defect-sensitive responses obtained by differentiating each candidate defect unit from the forward and reverse baselines are arranged in the order of candidate excitation code, excitation intensity, axial detection position, circumferential detection position, magnetic sensing array level, and measurement axis to form a response column vector. The difference between 1 and the absolute value of the normalized inner product of any two response column vectors is determined as the response resolution. The minimum value among the response resolutions is determined as the minimum response resolution of the candidate excitation code set. The candidate excitation code set with the largest minimum response resolution is selected as the excitation code set, and the coded magnetic response matrix is ​​established according to the corresponding response column vector; When there are multiple parallel candidate excitation code sets, the one with the fewest candidate excitation codes is selected.

[0012] Furthermore, the flexible support body includes multiple rigid sections, and adjacent rigid sections are connected by flexible connecting strips or rotating connectors; Each of the circumferential excitation sectors includes a flexible magnetic yoke, an excitation coil, a first magnetic pole, and a second magnetic pole, wherein the first magnetic pole and the second magnetic pole are spaced apart along the axial direction of the sling; When the excitation coil is energized, the magnetic flux passes sequentially through the flexible magnetic yoke, the first magnetic pole, the steel wire located between the first magnetic pole and the second magnetic pole, and the second magnetic pole, and returns to the flexible magnetic yoke to form a partially closed magnetic circuit. Non-magnetic isolation gaps are set between the flexible magnetic yokes of adjacent circumferential excitation sectors, and each excitation coil is driven by an independent bidirectional current drive branch.

[0013] Furthermore, multiple axial detection zones are arranged continuously along the axial direction of the sling and cover the anchorage area, with the boundaries of adjacent axial detection zones being connected and their interiors not overlapping. Each axial detection zone is equipped with a lift-off distance detection unit, a bonding pressure detection unit, a curvature detection unit, and an independent pushing unit; Based on the mapping relationship of the bonding excitation parameters obtained from the calibration of the defect-free steel wire bundle, the pushing amount, excitation current compensation amount and allowable acquisition flag are determined by the lifting distance, bonding pressure and local curvature of the current axial detection zone. The corresponding independent pushing unit is controlled according to the pushing amount to adjust the bonding state, and the excitation current amplitude of each circumferential excitation sector in the excitation state is adjusted according to the excitation current compensation amount while keeping the excitation direction specified by the current excitation code unchanged. When the allowed acquisition flag is set to prohibited acquisition, the triaxial magnetic field data of the corresponding axial detection zone is prohibited from being used to generate a defect-sensitive response.

[0014] Furthermore, the dual-layer triaxial magnetic array includes a first triaxial magnetic array and a second triaxial magnetic array spaced apart along the detection normal. Both triaxial magnetic arrays include multiple triaxial magnetic sensors, and the triaxial magnetic sensors with the same axial position and circumferential position are corresponding magnetic sensors to each other. Each magnetic field component acquired by each triaxial magnetic sensor on the three measurement axes is used as a magnetic sensing channel. For the same detection position, excitation code, excitation direction and excitation intensity, the local attenuation coefficient is determined based on the ratio of the magnetic field component amplitudes collected by the corresponding magnetic channels in the first and second three-axis magnetic arrays, and the difference between the actual lift-off distance of the corresponding magnetic channel in the second and first three-axis magnetic arrays and the actual lift-off distance of the corresponding magnetic channel in the first three-axis magnetic array. The magnetic field component collected by the first three-axis magnetic array is then corrected to a preset reference lift-off distance based on the local attenuation coefficient. When the actual lift-off distance or local attenuation coefficient exceeds the corresponding calibration range, the corresponding magnetic sensitive channel is marked as a low-confidence magnetic sensitive channel, and its corresponding defect-sensitive response channel weight is reduced during defect inversion.

[0015] Further, in step S5, the local extension direction of the steel wire at the detection position is determined according to the local curvature and the posture of the flexible support body, and a local coordinate system of the steel wire composed of local tangential, local normal and local subnormal is established; The expected magnetic response and the triaxial magnetic field response after lift-off correction are respectively transformed into the local coordinate system of the steel wire at the corresponding detection position; A positive baseline response and a negative baseline response are pre-established based on at least one of the following: defect-free wire bundle calibration data, initial inspection data of the sling, and magnetic field simulation data. The positive baseline response and the negative baseline response are updated according to the cable force and the temperature, respectively. The positive difference is obtained by subtracting the positive excitation response from the updated positive baseline response, and the negative difference is obtained by subtracting the negative excitation response from the updated negative baseline response. Half of the difference obtained by subtracting the negative difference from the positive difference is taken as the defect-sensitive response.

[0016] Furthermore, the dual-layer triaxial magnetic array includes multiple triaxial magnetic sensors. Each magnetic field component collected by each triaxial magnetic sensor on the three measurement axes is used as a magnetic channel. Each magnetic channel is uniquely determined by the axial detection position, circumferential detection position, magnetic array level, and measurement axis of the corresponding triaxial magnetic sensor. A combination of an excitation code, an excitation intensity, and a magnetic channel is used as a response item. The expected magnetic response in the encoded magnetic response matrix and the defect-sensitive response in the detection response vector are made to use the same lift-off correction method, wire local coordinate definition and forward and reverse baseline difference method, and the response items are arranged in the order of excitation code, excitation intensity, axial detection position, circumferential detection position, magnetic array level and measurement axis. The channel reliability is determined based on the lift-off distance, the change in bonding pressure within a preset time, the saturation state of the triaxial magnetic sensor, the signal noise, and the difference between repeated sampling results of the same response item, and a channel reliability weighting matrix is ​​established. A topological constraint relationship is established based on the adjacency relationship of steel wires and the radial hierarchy. The defect state is then inverted based on the encoded magnetic response matrix, the detection response vector, the channel confidence weighting matrix, and the topological constraint relationship. The positioning uncertainty is determined based on the distribution range of candidate defect units with defect state values ​​not less than a preset defect state threshold in the axial interval, wire bundle, and radial level.

[0017] Further, in step S6, when the positioning uncertainty is greater than a preset threshold, multiple candidate fine scanning actions are established. The candidate fine scanning actions include at least one of coding adjustment action and bonding adjustment action. The coding adjustment action includes adding excitation coding for driving the local circumferential excitation sector. The bonding adjustment action includes adjusting the bonding pressure of the axial detection partition corresponding to the current defect inversion result. For the aforementioned coding adjustment action, the expected new magnetic response is obtained based on finite element calculations of the magnetic field or standard wire bundle calibration data, and a candidate coding magnetic response matrix is ​​established based on the expected new magnetic response. For the aforementioned bonding adjustment action, the current encoded magnetic response matrix is ​​used as a candidate encoded magnetic response matrix, and the expected channel confidence level is determined based on the expected signal noise, lift-off distance, and bonding pressure. The expected positioning uncertainty is determined based on the candidate coding magnetic response matrix corresponding to each candidate fine scanning action, the current defect inversion result, and the expected channel confidence. The expected information gain is determined based on the reduction of the expected positioning uncertainty, the execution time, and the energy consumption. Perform the candidate fine scan action with the highest expected information gain and retest it; when the candidate fine scan action is the encoding adjustment action, update the encoding magnetic response matrix according to the corresponding expected magnetic response; update the detection response vector according to the defect sensitive response obtained from the retest, update the channel confidence weighting matrix according to the retest status, and perform defect inversion again.

[0018] Furthermore, when the response resolution of any two candidate defective units under the excitation code set is lower than a preset resolution threshold, at least one of the following is determined as a new candidate excitation code: a single-sector excitation code that drives only one circumferential excitation sector, an adjacent double-sector excitation code that drives two adjacent circumferential excitation sectors, and an opposing double-sector excitation code that drives two circumferential excitation sectors that are opposite each other along the circumference of the sling. This new candidate excitation code set is then added to the candidate excitation code set. Based on the candidate excitation code sets formed after adding the new candidate excitation codes, the minimum response resolution of each candidate excitation code set is determined, and the candidate excitation code set with the largest minimum response resolution is selected as the updated excitation code set. At least two excitation intensities are applied under each excitation code included in the updated excitation code set. When switching excitation codes, the excitation current of each circumferential excitation sector is reduced to zero, and the next excitation code is executed after the magnetic field reaches the preset stability condition.

[0019] The beneficial effects of this invention are: 1. By independently encoding and controlling the circumferential excitation sector, the magnetic flux distribution inside the wire bundle is changed, thereby improving the distinguishability between different wire sub-bundles and radial level defect responses.

[0020] 2. Select the excitation code set based on the expected magnetic response of the candidate defect unit, so that the excitation mode matches the arrangement of the sling wires and the magnetic circuit of the anchor, thereby improving the targeting of defect location.

[0021] 3. Use dual-layer triaxial magnetic field data for lift-off correction and local coordinate transformation to reduce the impact of curvature, fit status and lift-off distance changes on the consistency of detection response.

[0022] 4. Update the forward and reverse baseline responses based on cable force and temperature, and obtain the defect-sensitive response through forward and reverse differential to reduce interference caused by environmental changes, sensor bias and anchor background magnetic field.

[0023] 5. By combining the encoded magnetic response matrix and the topological constraints of the wire bundle to invert the defect state, the physical rationality of the defect location, radial level, and equivalent metal cross-section loss results is improved.

[0024] 6. Select the retesting method based on the positioning uncertainty and the information gain of the candidate fine scanning action to avoid aimless repeated detection and improve inspection efficiency and detection reliability. Attached Figure Description

[0025] Figure 1 This is a flowchart of the steps of the flexible magnetic flux leakage inspection method for the steel wire in the anchorage zone of the suspension cable in this invention. Detailed Implementation

[0026] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0027] In the following embodiments, the defect state value of the candidate defect cell uniformly represents the equivalent metal cross-sectional area loss ratio of the candidate defect cell, and the value ranges from 0 to 1. A defect state value of 0 indicates that no equivalent metal cross-sectional area loss was detected, a defect state value of 0.05 indicates that the equivalent metal cross-sectional area loss is 5%, and a defect state value of 1 indicates that the effective steel wire cross-sectional area within the candidate defect cell, which is the object of inversion, is completely lost.

[0028] Each column in the encoded magnetic response matrix corresponds to a candidate defect element and represents the sensitivity of the magnetic flux density change of each response term when the candidate defect element produces a unit equivalent metal cross-section loss ratio. When obtaining the expected magnetic response using a standard defect with a known loss ratio, the change in magnetic flux density produced by the standard defect is divided by the corresponding equivalent metal cross-section loss ratio to obtain the unit response sensitivity.

[0029] When the equivalent metal cross-section loss ratio is no greater than 20%, and the relative error between the finite element response or standard sample response and the linearly estimated response is no greater than 10%, linear inversion is performed using unit response sensitivity. When the equivalent metal cross-section loss ratio is greater than 20%, or the relative error of the linear estimation is greater than 10%, piecewise response data corresponding to 5%, 10%, 20%, 40%, and 60% equivalent metal cross-section losses are established respectively, and interpolation is performed based on the response data corresponding to two adjacent loss ratios.

[0030] The default defect state threshold is uniformly set to 0.05. Candidate defect elements with a defect state value of not less than 0.05 are used as valid defect candidate elements to participate in the determination of defect location, equivalent metal cross-section loss, and location uncertainty.

[0031] Example 1, as Figure 1 As shown, this embodiment provides a flexible magnetic flux leakage inspection method for steel wires in the anchorage zone of a sling, which reduces the influence of changes in the structural morphology of the anchorage zone, lift-off distance, cable force, temperature, and background magnetic field of the anchor on the detection results, and improves the ability to identify, locate, and quantitatively detect defects in different steel wire bundles and radial levels.

[0032] This embodiment uses a parallel wire sling consisting of 91 galvanized high-strength steel wires with a diameter of 7mm as the inspection object. The ends of the sling are fixed by anchor cups and anchor plates, and the axial length of the anchorage area to be inspected is 420mm. Near the anchor cup, the wire bundle gradually transitions from an approximately cylindrical section to an expanded diameter section, and some wires exhibit axial bending and circumferential displacement within the expanded diameter section.

[0033] The inspection process includes establishing a topology model of the wire bundle, dividing candidate defect units, selecting excitation code sets, flexible bonding control, coded excitation, acquiring a double-layer triaxial magnetic field, correcting the detection response, inverting topological constraints and defects, and fine scanning with information gain.

[0034] Flexible partitioned magnetization sensing component: The flexible zone magnetization sensing component is installed outside the cable anchorage area and includes a flexible bearing body, 8 circumferential excitation sectors, 6 axial detection zones, a first triaxial magnetic array, a second triaxial magnetic array, a lift-off distance detection array, a bonding pressure detection array, a curvature detection array, a zoned pushing mechanism, an excitation drive circuit, and a data acquisition controller.

[0035] The flexible support structure consists of eight rigid sections arranged circumferentially along the suspenders, with adjacent rigid sections connected by a 1.2mm thick fiber-reinforced silicone rubber connecting strip. The flexible connecting strip allows adjacent rigid sections to rotate circumferentially and move radially, enabling the flexible support structure to adapt to the expansion, bending, and local surface undulations in the anchorage zone.

[0036] Each rigid section corresponds to a circumferential excitation sector. Each circumferential excitation sector includes a flexible magnetic yoke, an excitation coil, a first magnetic pole, and a second magnetic pole. The first and second magnetic poles are spaced apart along the axial direction of the suspension cable, with an axial spacing of 60 mm. The flexible magnetic yoke is formed by stacking multiple layers of soft magnetic alloy strips to balance magnetic conductivity and bending adaptability.

[0037] The excitation coil is wound in the middle of the flexible magnetic yoke and connected to an independent bidirectional current drive branch. The excitation current adjustment range is -3A to 3A, and the current control accuracy is not less than 1% of the full scale. Each circumferential excitation sector can independently be in a forward excitation state, a reverse excitation state, or a stopped excitation state.

[0038] When the excitation coil is energized, the magnetic flux passes sequentially through the flexible magnetic yoke, the first magnetic pole, the steel wire located between the first and second magnetic poles, the second magnetic pole, and returns to the flexible magnetic yoke, forming a partially closed magnetic circuit.

[0039] A non-magnetic isolation gap with a width of 2mm is set between the flexible magnetic yokes of adjacent circumferential excitation sectors. The isolation material is glass fiber reinforced engineering plastic, which is used to reduce the uncontrolled bypass magnetic flux between adjacent circumferential excitation sectors.

[0040] Six axial detection zones are arranged continuously along the sling axis and cover a 420mm long anchorage area. The axial length of each axial detection zone is 70mm. The boundaries of adjacent axial detection zones are contiguous and do not overlap internally.

[0041] The first and second triaxial magnetic sensing arrays are positioned within the detection window between the first and second magnetic poles. The first triaxial magnetic sensing array is positioned close to the wire bundle, with its sensitive surface 1.5 mm from the inner surface of the flexible support body. The second triaxial magnetic sensing array is positioned outside the first triaxial magnetic sensing array along the detection normal, with a normal distance of 2.5 mm between the two triaxial magnetic sensing arrays.

[0042] Both triaxial magnetic arrays employ triaxial tunneling magnetoresistive sensors with a range of not less than ±10mT and a bandwidth of not less than 20kHz. Each circumferential excitation sector is equipped with one triaxial tunneling magnetoresistive sensor in each axial detection zone. Each layer of the triaxial magnetic array includes 48 triaxial tunneling magnetoresistive sensors, forming a total of 288 magnetic channels across the two layers.

[0043] The data acquisition controller includes an STM32H743 microcontroller, an Artix-7 field-programmable gate array (FPGA), and a 16-bit synchronous analog-to-digital converter (ADC). The FPGA is used to generate excitation timing, perform multi-channel synchronous sampling, and timestamp alignment; the microcontroller is used to execute fit control, excitation coding scheduling, response correction, defect inversion, and fine-scan decision-making. The sampling frequency of the three-axis magnetic field data is set to 10kHz.

[0044] The zoned pushing mechanism employs six independent air chambers, each corresponding to an axial detection zone. Each air chamber is connected to an electro-proportional pressure regulating valve, with a working pressure range of 10 kPa to 60 kPa.

[0045] Each axial detection zone is equipped with at least one contact pressure detection unit, one lift-off distance detection unit, and one curvature detection unit. The contact pressure detection unit uses a thin-film pressure sensor with a range of 0 kPa to 100 kPa; the lift-off distance detection unit uses a miniature eddy current displacement sensor with a range of 0.5 mm to 8 mm; and the curvature detection unit uses a miniature inertial measurement unit and an angle sensor located at the connection position of adjacent rigid sections.

[0046] The actual cable force is provided by the existing cable force monitoring system. The wire temperature is collected by a platinum resistance temperature sensor installed inside the flexible load-bearing body, with a temperature measurement range of -20℃ to 80℃ and a measurement error of no more than 0.5℃.

[0047] Acquisition threshold and stability conditions: The lift-off distance calibration range of the first triaxial magnetic array relative to the surface of the steel wire bundle is 0.8 mm to 3.5 mm, and the lift-off distance calibration range of the second triaxial magnetic array relative to the surface of the steel wire bundle is 3.3 mm to 6.0 mm.

[0048] The allowable bonding pressure ranges from 18 kPa to 35 kPa, and the allowable local curvature is no greater than 0.012 mm. -1 .

[0049] When the actual lift-off distance of the first triaxial magnetic array in any axial detection zone is greater than 3.5 mm, or the bonding pressure is less than 18 kPa, the allow acquisition flag of that axial detection zone is set to prohibit acquisition, and the corresponding triaxial magnetic field data is not used to generate defect-sensitive responses.

[0050] When the lift-off distance, bonding pressure, and local curvature are all within the allowable range, and the change in lift-off distance is no greater than 0.05 mm and the change in bonding pressure is no greater than 1 kPa within 500 ms, the allowable acquisition flag is set to allow acquisition.

[0051] When the excitation current enters the range of ±1% of the target excitation current and remains continuously for 20ms, the excitation current is considered to have reached a steady state. After reaching a steady state, triaxial magnetic field data is continuously acquired for 60ms.

[0052] Before switching the excitation encoding, the excitation current of all circumferential excitation sectors is reduced to zero. The magnetic field is considered to have reached the preset stability condition when the total excitation current is no greater than 1% of the rated excitation current, and the change in magnetic induction intensity within the detection window is no greater than 0.05 mT within 15 ms. The preset resolution threshold is set to 0.18, the positioning uncertainty threshold is set to 0.20, and the maximum number of fine scans is set to 3.

[0053] Step S1: Establish the topological model of the wire bundle and divide it into candidate defect elements: Obtain the sling design drawings, wire arrangement records, anchor assembly data, and structural dimensions of the anchorage zone. This includes the number of wires, wire diameter, circumferential arrangement of wires, radial layering of wires, axial direction of wires, anchor cup structure, anchor plate structure, relative positions of wires and anchors, and adjacent relationships between wires.

[0054] Based on the above information, a three-dimensional spatial arrangement model of 91 steel wires was established. The three-dimensional spatial arrangement model records the axial path, circumferential position, radial position, and relationship of each steel wire within the anchorage zone, thereby forming a topological model of the steel wire bundle.

[0055] Based on the spatial position of the steel wires in the circumferential direction, the 91 steel wires are divided into 8 steel wire bundles; based on the distance of the steel wires from the outer surface of the steel wire bundles, the steel wires are divided into 3 radial layers: outer layer, middle layer and inner layer.

[0056] Within each radial level, at least one wire that is spatially adjacent and has a similar magnetic response at the current magnetization and measurement resolution is grouped into a wire sub-bundle. The same wire belongs to only one wire sub-bundle within the same axial interval and the same radial level.

[0057] The 420mm long anchorage zone is continuously divided into 21 axial sections along the sling axis, each axial section being 20mm long. The boundaries of adjacent axial sections are contiguous and do not overlap internally, with the 21 axial sections collectively covering the anchorage zone to be inspected.

[0058] A candidate defect element is a combination of an axial interval, a wire bundle, and a radial level. Each axial interval contains 24 candidate defect elements, and the entire anchorage zone contains 504 candidate defect elements.

[0059] The topology model of the steel wire bundle was established using an engineering workstation. The external contour of the anchorage zone and the installation attitude of the flexible partition magnetization sensing component were verified using a structured light scanner.

[0060] Step S2: Install the flexible partitioned magnetization sensing component and establish candidate excitation codes: The flexible partitioned magnetization sensing component is wrapped around the outside of the anchorage area, so that the eight circumferential excitation sectors are arranged along the circumference of the sling, and the six axial detection zones are arranged along the axial direction of the sling.

[0061] During installation, the pressure of each independent air chamber is first set to 10 kPa. Then, the air chamber pressure is gradually increased according to the outer surface contour of the anchorage area to make the flexible bearing conform to the surface of the anchorage area. After installation, the attitude and lift-off distance of each axial detection zone are checked by the curvature detection unit and the lift-off distance detection unit.

[0062] Each candidate excitation code consists of 8 coding elements, which correspond one-to-one with 8 circumferential excitation sectors. Each coding element represents the forward excitation state, reverse excitation state, or stopped excitation state of the corresponding circumferential excitation sector.

[0063] Candidate excitation codes include at least the following: full-sector unidirectional excitation code, odd-even sector alternating excitation code, front-and-back grouped reverse excitation code, opposite double-sector excitation code, adjacent double-sector excitation code, and single-sector excitation code.

[0064] Step S3: Establish the encoded magnetic response matrix and select the excitation encoding set: The encoded magnetic response matrix is ​​established through nonlinear static magnetic field finite element calculation, standard wire bundle calibration data, or a combination of both.

[0065] Based on the sling design data, a three-dimensional geometric model was established for the steel wire, anchor cup, anchor plate, flexible magnetic yoke, first magnetic pole, second magnetic pole, and excitation coil. Each steel wire was modeled separately according to its actual diameter, circumferential position, radial hierarchy, and axial direction.

[0066] The magnetization curves of the steel wire, anchor cup, anchor plate, and flexible magnetic yoke are used for their respective materials. The magnetization curves include at least multiple magnetic field strength nodes and the corresponding magnetic induction intensity at each node.

[0067] The excitation coil is constructed according to the actual number of turns, winding area, coil cross-section, and current direction. Forward and reverse excitation are achieved by changing the direction of the excitation current.

[0068] An air calculation zone is established around the wire bundle and anchorage. This air calculation zone extends along the sling axis to both sides of the detection window by at least three times the axial length of the detection window, and radially by at least twice the outer diameter of the wire bundle. The outer boundary of the air calculation zone is either a magnetically insulated boundary or an infinitely far boundary.

[0069] Candidate defect elements are modeled by setting wire breaks, reducing the effective cross-sectional area of ​​the wire, setting local corrosion depressions, or simultaneously reducing the effective cross-sectional area of ​​multiple adjacent wires.

[0070] The maximum grid size is set to 0.5 mm for candidate defect areas, magnetic pole edges, wire contact areas, and near the sampling position of the triaxial magnetic sensor; the maximum grid size is set to 1.5 mm for other wire areas; and a gradient grid with a maximum size not exceeding 5 mm is used for air areas far from the detection window.

[0071] If, after two consecutive mesh refinements, the expected magnetic response amplitude of all effective magnetic channels changes by no more than 1%, the finite element mesh is deemed to meet the convergence requirement.

[0072] The expected magnetic response of each triaxial magnetometer is calculated using the average value of the magnetic flux density components within the sensor's sensitive volume.

[0073] For each candidate defect unit, the original expected magnetic response is obtained under different candidate excitation codes, different excitation intensities, different excitation directions, different axial detection positions, different circumferential detection positions, different magnetic array levels, and different measurement axes. The original expected magnetic response in the forward direction and the original expected magnetic response in the reverse direction under the same excitation code, the same excitation intensity, and the same magnetic channel are processed using the same forward and reverse baseline difference method as in step S5 to obtain the expected defect sensitive response; the excitation direction is used to form the expected defect sensitive response and is not used as an independent arrangement dimension of the response terms.

[0074] Each magnetic sensing channel is uniquely determined by the axial detection position, circumferential detection position, magnetic array level, and measurement axis of the corresponding triaxial magnetic sensor. A combination of a candidate excitation code, an excitation intensity, and a magnetic sensing channel is considered as a single response term. The expected defect-sensitive responses of the same candidate defect unit are arranged in the order of candidate excitation code, excitation intensity, axial detection position, circumferential detection position, magnetic array level, and measurement axis, forming a response column vector.

[0075] For any two candidate defect units, the difference between the absolute value of the normalized inner product of the two response column vectors and 1 is determined as the response resolvability. For each candidate excitation code set, the response resolvability of all pairwise combinations of candidate defect units is calculated, and the minimum value is determined as the minimum response resolvability.

[0076] The candidate excitation code set with the largest minimum response distinguishability is selected from multiple candidate excitation code sets as the excitation code set to be used in practice.

[0077] When multiple candidate excitation code sets exist, the candidate excitation code set with the fewest excitation codes is selected first; when the number of excitation codes is the same, the candidate excitation code sets with shorter expected detection time and lower expected energy consumption are selected in turn.

[0078] When the number of candidate excitation codes does not exceed 12, a combined exhaustive search method is used; when the number of candidate excitation codes exceeds 12, a greedy search method that gradually increases the number of excitation codes is used.

[0079] Stop adding excitation codes when the minimum response resolution reaches 0.18, or when adding excitation codes increases the minimum response resolution by less than 0.01.

[0080] Each column of the encoded magnetic response matrix corresponds to a candidate defect unit, and each row corresponds to a response term determined by the excitation code, excitation intensity, and magnetic sensing channel. Each magnetic sensing channel is uniquely determined by the axial detection position, circumferential detection position, magnetic array level, and measurement axis. The rows are arranged in the order of excitation code, excitation intensity, axial detection position, circumferential detection position, magnetic array level, and measurement axis.

[0081] Before the formal inspection, channel gain correction is performed using a defect-free reference area or a standard magnetic calibration piece. When the channel gain correction factor is within the range of 0.5 to 2.0, the corresponding row of the encoded magnetic response matrix is ​​adjusted according to the correction factor; if it exceeds this range, the corresponding magnetic sensitive channel is marked as an abnormal magnetic sensitive channel.

[0082] Fitting the excitation parameter mapping relationship: The bonding excitation parameter mapping relationship takes the lift-off distance, bonding pressure, and local curvature as inputs, and the pushing amount, excitation current compensation amount, and allowable acquisition flag as outputs.

[0083] The lift-off distance calibration nodes are set to 0.8mm, 1.5mm, 2.5mm, and 3.5mm; the bonding pressure calibration nodes are set to 18kPa, 25kPa, and 35kPa; and the local curvature calibration nodes are set to 0 and 0.004mm. -1 0.008mm -1 and 0.012mm -1 .

[0084] Each combination of lift-off distance calibration nodes, contact pressure calibration nodes, and local curvature calibration nodes records the pushing amount, excitation current compensation amount, and allowable acquisition flag, forming a three-dimensional parameter lookup table.

[0085] When the current state coincides with the calibration node, the corresponding parameters are read directly; when the current state is between adjacent calibration nodes, trilinear interpolation is used to obtain the push amount and excitation current compensation amount; when the current state exceeds the calibration range, no extrapolation is performed, and the allow acquisition flag is set to prohibit acquisition.

[0086] The coupling coefficient between adjacent circumferential excitation sectors is obtained by driving a single circumferential excitation sector. The ratio of the change in magnetic flux density in the detection area at the center of the adjacent sector to the change in magnetic flux density in the detection area at the center of the current sector is used as the coupling coefficient between adjacent sectors.

[0087] No compensation is performed when the coupling coefficient between adjacent sectors is no greater than 0.05; when it is greater than 0.05 but no greater than 0.10, the excitation current increases by 3%; when it is greater than 0.10 but no greater than 0.20, the excitation current increases by 6%; and when it is greater than 0.20, the excitation current increases by 10%. The increment of excitation current compensation shall not exceed 12% of the target excitation current.

[0088] Step S4: Adjust the bonding state and perform coded excitation: The data acquisition controller reads the lifting distance, bonding pressure, and local curvature of each axial detection zone, and determines the pushing amount, excitation current compensation amount, and allowable acquisition flag based on the bonding excitation parameter mapping relationship.

[0089] When the lifting distance is greater than 3.5 mm or the bonding pressure is less than 18 kPa, increase the corresponding air chamber pressure; when the bonding pressure is greater than 35 kPa, decrease the corresponding air chamber pressure.

[0090] The data acquisition controller adjusts the independent air chambers of the corresponding axial detection zone according to the pushing amount, and adjusts the excitation current amplitude of the circumferential excitation sector in the excitation state according to the excitation current compensation amount while keeping the excitation direction specified by the current excitation code unchanged.

[0091] Each excitation code uses three excitation intensities: 0.8A, 1.6A, and 2.4A, corresponding to weak magnetization, medium magnetization, and near-saturation magnetization, respectively. For each excitation code and each excitation intensity included in the coded magnetic response matrix, forward excitation and overall reverse excitation are performed respectively. During overall reverse excitation, the current direction of each non-zero coded element is reversed simultaneously, the stop state remains unchanged, and the paired forward and reverse responses are used to generate the defect-sensitive response in formula (3).

[0092] After the excitation state reaches a stable condition, the first and second triaxial magnetic sensing arrays synchronously acquire triaxial magnetic field data at a sampling frequency of 10kHz, simultaneously obtaining actual cable force, wire temperature, lift-off distance, bonding pressure, local curvature, and the attitude of the flexible support body. The synchronization error between the two triaxial magnetic sensing arrays is no greater than one sampling period.

[0093] Step S5, detect response correction: For the same detection position, the same excitation code, the same excitation direction, the same excitation intensity, and the same measurement axis, the magnetic induction intensity components of the corresponding magnetic channels in the first and second triaxial magnetic arrays are obtained respectively.

[0094] Based on the amplitude ratio of the corresponding magnetic induction intensity components of the two triaxial magnetic arrays and the actual lift-off distance difference between the two triaxial magnetic arrays, the local attenuation coefficient of the corresponding measurement axis is determined, and the magnetic induction intensity component acquired by the first triaxial magnetic array is corrected to a reference lift-off distance of 1.5 mm.

[0095] When the amplitude of the magnetic flux density component is less than three times the root mean square value of the corresponding noise, this component is not used to calculate the local attenuation coefficient. In the event of the above situation, the corrected magnetic flux density component and the defect-sensitive response of the corresponding excitation direction and magnetic sensing channel are set to zero, and the confidence level of the corresponding channel is set to zero so that it does not participate in defect inversion.

[0096] If the actual lift-off distance of the first or second three-axis magnetic array exceeds the calibration range, or the local attenuation coefficient exceeds the corresponding calibration range, the corresponding magnetic channel will be marked as a low-confidence magnetic channel.

[0097] The local extension direction of the steel wire at the detection location is determined based on the local curvature and the posture of the flexible load-bearing body, and a local coordinate system of the steel wire consisting of local tangential, local normal, and local subnormal is established.

[0098] The local tangential direction is set along the local extension direction of the steel wire; the local normal direction points from the center of the steel wire bundle to the detection position; the local sub-normal direction is orthogonal to both the local tangential direction and the local normal direction.

[0099] The triaxial magnetic flux density response, after lift-off correction, is transformed to the wire local coordinate system. The expected magnetic response in the encoded magnetic response matrix is ​​defined using the same wire local coordinate system.

[0100] Establish the positive baseline response and the negative baseline response based on at least one of the following: calibration data of defect-free wire bundles, initial inspection data of slings, and magnetic field simulation data.

[0101] The baseline response is stored separately according to excitation code, excitation intensity, magnetic sensing channel, and excitation direction. The stored dimension is used to distinguish between the forward and reverse original responses. After the forward and reverse baselines are differentially divided, the excitation direction is not used as an independent arrangement dimension for the response terms. During inspection, the forward and reverse baseline responses are updated based on the actual cable force and actual temperature.

[0102] The positive excitation response is subtracted from the updated positive baseline response to form a positive difference value; the reverse excitation response is subtracted from the updated reverse baseline response to form a reverse difference value; half of the positive difference value minus the reverse difference value is taken as the defect-sensitive response.

[0103] Channel reliability and data scaling: Each response item has its own lift confidence, bonding pressure confidence, sensor saturation confidence, signal-to-noise ratio confidence, and resampling confidence. The final channel confidence is the minimum value among the five confidence items.

[0104] When both the lift-off distance and the local attenuation coefficient are within the calibration range, the lift-off confidence level is set to 1; when it is near the calibration boundary, it is set to 0.7; and when it is outside the calibration range, it is set to 0.2.

[0105] The bonding pressure reliability is set to 1 when the change in bonding pressure within 500ms is no greater than 1kPa; 0.7 when the change is greater than 1kPa but no greater than 2kPa; and 0.4 when the change is greater than 2kPa.

[0106] When the measured value of magnetic induction intensity reaches 90% of the sensor's range, the sensor saturation confidence level is set to 0.2; otherwise, it is set to 1.

[0107] When the signal-to-noise ratio (SNR) is not less than 12dB, the SNR confidence level is set to 1; when it is less than 12dB but not less than 8dB, it is set to 0.7; when it is less than 8dB, it is set to 0.4.

[0108] When the amplitude difference between two repeated sampling results of the same response item is no greater than 8%, the confidence level of repeated sampling is set to 1; when the difference is greater than 8% but no greater than 15%, it is set to 0.7; when the difference is greater than 15%, it is set to 0.2.

[0109] The channel confidence weighting matrix is ​​a diagonal matrix, with each diagonal element being the square root of the corresponding final channel confidence.

[0110] Based on the range of each magnetic sensing channel and the noise standard deviation under defect-free conditions, each row of the encoded magnetic response matrix and the corresponding element of the detection response vector are processed using the same scale to convert both into dimensionless data.

[0111] Step S6, Topological constraint defect inversion: Each candidate defective unit serves as a node in the topology graph.

[0112] Establish circumferential topological edges between candidate defect units located in the same axial interval and the same radial level and circumferentially adjacent; establish radial topological edges between candidate defect units located in the same axial interval and the same wire bundle and radially adjacent; establish axial topological edges between candidate defect units located in adjacent axial intervals and wire bundles with the same radial level.

[0113] The weight of the circumferential topological edge is determined based on the magnetic coupling coefficient between the wire sub-bundles; the weight of the radial topological edge is determined based on the magnetic coupling coefficient between adjacent radial levels; and the weight of the axial topological edge is determined based on the reciprocal of the center distance between two axial intervals.

[0114] The weights of different types of topological edges are normalized to the range of 0 to 1. The topological constraint matrix adopts a weighted incidence matrix, with each topological edge corresponding to one row and each candidate defective unit corresponding to one column.

[0115] Defect inversion takes the encoded magnetic response matrix, detection response vector, channel confidence weighting matrix and topological constraint matrix as inputs, while reducing detection response residuals, restricting unfounded scattered defects, and restricting unfounded drastic state changes between candidate defect units with topological adjacency.

[0116] Candidate defect state values ​​are limited to the range of 0 to 1. The sparsity constraint coefficient is set to 0.015, and the topology constraint coefficient is set to 0.08.

[0117] An accelerated proximal gradient algorithm with interval projection is used to solve for the defect state, with a maximum number of iterations set to 500. The solution is considered convergent when the relative change in the objective function between two consecutive iterations is no greater than one part per million, or the relative change in the defect state vector between two consecutive iterations is no greater than one part per ten thousand.

[0118] Positioning uncertainty: The normalized position weight is obtained by dividing the defect state value of each effective defect candidate unit by the sum of the defect state values ​​of all effective defect candidate units.

[0119] The axial uncertainty is calculated by weighting the standard deviation based on the center position of the axial interval where each effective defect candidate unit is located, and then normalized by dividing by the total length of the anchorage zone.

[0120] The circumferential uncertainty is calculated based on the center angles of each wire sub-bundle. Each center angle is converted into a direction vector on the unit circle, and a weighted composite vector is formed according to the normalized position weights. The circumferential uncertainty is obtained by subtracting the length of the weighted composite vector from 1.

[0121] The radial uncertainty is calculated by weighting the standard deviation based on the radial level number and then normalized by dividing by the maximum possible standard deviation given the current number of radial levels.

[0122] The positioning uncertainty is formed by a weighted average of axial, circumferential, and radial uncertainties, with the axial uncertainty having a weight of 0.5, the circumferential uncertainty having a weight of 0.3, and the radial uncertainty having a weight of 0.2. When the positioning uncertainty is no greater than 0.20, the positioning result is deemed to meet the requirements; when the positioning uncertainty is greater than 0.20, the candidate fine scanning action evaluation is triggered.

[0123] Based on the currently acquired detection response vector and its inverted defect state vector, the current positioning uncertainty is obtained according to the determination methods of axial uncertainty, circumferential uncertainty, and radial uncertainty, and their corresponding weights of 0.5, 0.3, and 0.2. The current positioning uncertainty is used to determine whether to initiate the evaluation of candidate fine scanning actions.

[0124] If the current positioning uncertainty is no greater than 0.20, the defect positioning result is determined to meet the preset requirements; if the current positioning uncertainty is greater than 0.20, the candidate fine scanning action evaluation is initiated.

[0125] The predicted positioning uncertainty corresponding to the candidate fine cleaning action is determined using the same candidate defect element selection rules, axial uncertainty calculation rules, circumferential uncertainty calculation rules, radial uncertainty calculation rules, and weighting method as the current positioning uncertainty. During the evaluation of candidate fine cleaning actions, the variance, trace of the covariance matrix, or other statistical quantities of the defect state vector are not used to directly replace the positioning uncertainty.

[0126] Information gain fine scan: Candidate fine scanning actions include adding single-sector excitation coding, adding adjacent double-sector excitation coding, adding opposing double-sector excitation coding, increasing intermediate excitation intensity, increasing forward and reverse excitation, increasing sampling frequency, reducing axial movement speed, and adjusting the bonding pressure of the target axial detection zone.

[0127] For candidate fine sweeping actions that change the excitation code or excitation intensity, the expected new magnetic response is obtained based on the finite element calculation of the magnetic field or the standard wire bundle calibration data, and a candidate code magnetic response matrix is ​​established.

[0128] For candidate fine scan actions that increase the number of samplings or adjust the bonding pressure, the current encoded magnetic response matrix is ​​used, and the expected channel confidence is determined based on the expected signal noise, expected lift-off distance, expected bonding pressure, and expected stability.

[0129] When the number of samples is increased, the noise standard deviation is expected to decrease by the square root of the ratio of the original number of samples to the increased number of samples.

[0130] After adjusting the bonding pressure, if the expected lift-off distance is still within the lift-off correction calibration range, the candidate coding magnetic response matrix remains unchanged, and only the expected channel confidence is updated; if the expected lift-off distance exceeds the calibration range or the local curvature enters a new calibration interval, the relevant rows of the candidate coding magnetic response matrix are updated according to the corresponding calibration response.

[0131] For the current inspection status and each candidate fine cleaning action, the expected positioning uncertainty before fine cleaning and the expected positioning uncertainty after performing the corresponding candidate fine cleaning action are determined respectively. The expected positioning uncertainty before fine cleaning and the expected positioning uncertainty after fine cleaning are both determined according to the weighted result of axial uncertainty, circumferential uncertainty and radial uncertainty.

[0132] The difference between the expected location uncertainty before and after fine cleaning is taken as the reduction in expected location uncertainty. A positive reduction in expected location uncertainty indicates that the candidate fine cleaning action is expected to improve the concentration of defect location; a zero or negative reduction in expected location uncertainty indicates that the candidate fine cleaning action cannot reduce the location uncertainty.

[0133] The expected information gain of the candidate fine scanning action is determined by the expected reduction in positioning uncertainty, the expected execution time, and the expected energy consumption. The specific process for determining the expected positioning uncertainty follows the process in Example 3, which generates candidate defect state samples from the approximate defect state covariance matrix and propagates them to the positioning uncertainty.

[0134] The time cost factor is set to 0.02s. -1 The energy consumption cost coefficient is set to 0.01 kJ. -1 .

[0135] The candidate fine scan action with the highest expected information gain is selected for fine scan. When the difference in expected information gain between two candidate fine scan actions is less than 0.01, the candidate fine scan action with shorter execution time is selected first; when the execution time is the same, the candidate fine scan action with lower energy consumption is selected.

[0136] When the excitation code or excitation intensity is changed, the coded magnetic response matrix is ​​updated according to the corresponding expected magnetic response; the detection response vector is updated according to the defect-sensitive response obtained from the retest; the channel confidence weighting matrix is ​​updated according to the retest status, and then defect inversion is performed again.

[0137] The coarse scan moving speed is set to 5 mm / s, and the fine scan moving speed is set to 1.5 mm / s. Fine scanning is stopped when the positioning uncertainty is no greater than 0.20; if the positioning uncertainty is still greater than 0.20 after 3 fine scans, a prompt is output indicating that other non-destructive testing methods need to be used for verification.

[0138] Working principle and technical effects: The forward excitation, reverse excitation, and de-excitation states of different circumferential excitation sectors can change the magnetic flux path inside the wire bundle, enabling different wire sub-bundles and different radial levels to form distinguishable coded magnetic responses under multiple excitation codes.

[0139] When a steel wire breaks, corrodes, or loses its equivalent metallic cross-section, the effective magnetic permeability of the defect region decreases, and some magnetic flux escapes from the inside of the wire, forming a leakage magnetic field. Selecting the excitation code set based on response distinguishability can improve the differentiation between candidate defect units with similar responses.

[0140] The flexible support structure and independent air chamber can adapt to the expansion, bending, and local surface undulations of the anchorage zone. The lift-off distance, contact pressure, and local curvature are used together to adjust the pushing force and excitation current amplitude to improve the consistency of magnetization conditions at different detection positions.

[0141] The dual-layer triaxial magnetic array uses the corresponding magnetic induction intensity components at two detection normal positions to determine the local attenuation relationship and corrects the field response to a unified reference lift-off distance. Local coordinate transformation of the steel wire is used to reduce directional errors caused by changes in the attitude of the flexible support and the direction of the steel wire.

[0142] The forward and reverse baseline responses are updated based on cable force and temperature. The forward and reverse differential can reduce the effects of sensor bias, temperature drift, cable force variation and anchor background magnetic field.

[0143] Channel credibility weighting reduces the impact of low-quality magnetic channels on defect inversion. Topological constraints introduce the axial, circumferential, and radial adjacency relationships between wires into defect state inversion, reducing isolated spurious defects that do not conform to the actual arrangement of the wires.

[0144] The location uncertainty is always represented by a weighted sum of axial, circumferential, and radial uncertainties. The approximate defect state covariance matrix is ​​used only to predict the discrete range of the defect state under candidate fine-scanning actions, and is converted into the predicted location uncertainty through candidate defect state samples, so that the predicted information gain directly reflects the degree of reduction in the predicted location uncertainty.

[0145] Example 2 further illustrates the implementation scheme based on the overall inspection process of Example 1, employing a rotating connection structure, a miniature electric pushing method, a triaxial Hall sensor, and different steel wire bundle division rules. The processing steps in Example 1 that have not been significantly altered continue to apply to this example.

[0146] This embodiment uses a parallel wire sling composed of 127 galvanized high-strength steel wires with a diameter of 5.2 mm as an example. The ends of the sling are fixed by anchor cups, anchor plates, and grouting medium, and the axial length of the anchoring zone is 360 mm. Before entering the anchor cup, the wire bundle gradually transitions from a cylindrical section to an expanded diameter section.

[0147] Flexible partitioned magnetization sensing component: The flexible support structure is divided into 10 rigid sections along the circumference of the sling, and adjacent rigid sections are connected by miniature rotating connectors. The miniature rotating connectors allow relative rotation between adjacent rigid sections in the circumferential and radial directions of the sling.

[0148] Flexible connecting belts can also replace micro-rotating connectors. Flexible connecting belts can be made of fiber-reinforced silicone rubber belts, polyurethane elastic belts, or flexible composite material belts.

[0149] Each rigid section corresponds to a circumferential excitation sector. Each circumferential excitation sector includes a flexible magnetic yoke, an excitation coil, a first magnetic pole, and a second magnetic pole. The first and second magnetic poles are arranged at intervals along the axial direction of the suspension cable, with an axial distance of 54 mm.

[0150] A 3mm non-magnetic isolation gap is set between the flexible magnetic yokes of adjacent circumferential excitation sectors. The excitation drive circuit includes 10 independent bidirectional current drive branches, each driving one of the 10 excitation coils.

[0151] The flexible partitioned magnetization sensing component is divided into 5 axial detection zones along the axial direction of the suspension cable, with each axial detection zone having an axial length of 72mm.

[0152] Both the first and second triaxial magnetic arrays employ triaxial Hall sensors with a range of ±20mT and a bandwidth of not less than 15kHz. The detection normal distance between the two triaxial magnetic arrays is 3mm.

[0153] The calibrated range of the lift-off distance between the first three-axis magnetic sensitive array and the steel wire surface is 0.6mm to 3.8mm, and the nominal lift-off distance is 1.2mm.

[0154] The data acquisition controller uses an industrial control computer and an 18-bit multi-channel synchronous acquisition card, with a sampling frequency of 8kHz. The synchronization error of the two triaxial magnetic arrays is no greater than 125μs.

[0155] Zoned fit control: Each axial detection zone is equipped with two contact pressure sensors, two lift-off distance sensors, and one curvature detection unit.

[0156] The bonding pressure sensor uses a flexible thin-film pressure sensor with a range of 0 kPa to 150 kPa; the lifting distance sensor uses a laser triangular displacement sensor with a range of 0.3 mm to 10 mm; and the curvature detection unit uses an angle sensor set between adjacent rigid sections.

[0157] The zoned pushing mechanism employs five sets of miniature electric actuators, each corresponding to one axial detection zone. The effective stroke of the miniature electric actuator is 8mm, the rated thrust is 120N, and the displacement control resolution is 0.02mm.

[0158] The lift-off distance ranges are set to 0.6mm to 1.5mm, 1.5mm to 2.5mm, 2.5mm to 3.2mm, and 3.2mm to 3.8mm. The bonding pressure ranges are set to 20kPa to 30kPa, 30kPa to 45kPa, and 45kPa to 55kPa. The local curvature range is set to no more than 0.004mm. -1 Greater than 0.004mm -1 And not greater than 0.008mm -1 and greater than 0.008mm -1 And not greater than 0.015mm -1 .

[0159] The minimum pressure threshold is set to 20 kPa, and the maximum allowable lift-off distance is set to 3.8 mm.

[0160] When the bonding pressure is below 20 kPa or the lifting distance is greater than 3.8 mm, the data acquisition flag is set to prohibit data acquisition, and the data acquisition controller first adjusts the corresponding miniature electric push rod.

[0161] When the lift-off distance, bonding pressure, and local curvature are all within the allowable range, and the change in lift-off distance is no greater than 0.08 mm and the change in bonding pressure is no greater than 1.5 kPa within 400 ms, the allowable acquisition flag is set to allow acquisition.

[0162] Topological model of wire bundle and candidate defect elements: The wire bundle topology model records the number of wires, wire diameter, circumferential arrangement of wires, radial hierarchy of wires, axial orientation of wires, relative position between wires and anchor cups, relationship between adjacent wires, sub-bundle division results of wires, and magnetic coupling coefficient between sub-bundles of wires.

[0163] Within the same axial range and the same radial level, the adjacency relationship of steel wires is established based on the center-to-center distance. Two steel wires are considered spatially adjacent when the center-to-center distance is no greater than 1.15 times the diameter of the steel wire.

[0164] Spatially adjacent wires with a magnetic coupling coefficient difference of no more than 0.08 are grouped into the same wire sub-bundle. The sub-bundle division is completed using a connected component method, with each connected component constituting a wire sub-bundle.

[0165] The anchorage zone is continuously divided into 12 axial intervals along the sling axis, each interval being 30mm in length. The steel wires are divided into four radial layers according to their distance from the outer surface of the wire bundle: outer layer, sub-outer layer, middle layer, and inner layer.

[0166] Ten wire sub-bundles are formed within each radial level of each axial interval. The combination of an axial interval, a wire sub-bundle, and a radial level constitutes a candidate defect unit. Each axial interval includes 40 candidate defect units, and the entire anchorage zone includes 480 candidate defect units.

[0167] In this embodiment, a reference response is established using an 8% equivalent metal cross-sectional loss within the steel wire bundle, and an encoded magnetic response matrix is ​​formed according to the unit response sensitivity conversion rule.

[0168] Minimum response resolution: Before calculating the resolvability of the response, the expected magnetic response is dimensionlessly processed according to the channel scale value corresponding to each magnetic sensing channel. The channel scale value of each response item is denoted as The unit is mT, and the larger of the noise standard deviation of the response term under defect-free conditions and the reference response amplitude of the standard magnetic calibration component is taken. The expected magnetic response is divided by the corresponding channel scale value to form a dimensionless response column vector.

[0169] The minimum response resolvability of the candidate excitation code set is determined by the following formula: ; In the formula, Represents the set of candidate excitation codes; This represents the minimum response distinguishability of the candidate excitation code set; and Indicates the sequence number of two different candidate defect units; and These represent the response column vectors of the corresponding candidate defect units after channel scaling; superscript Indicates transpose; This represents the L2 norm.

[0170] In formula (1), each response column vector is a dimensionless vector, and the minimum response resolution is a dimensionless quantity, ranging from 0 to 1. The L2 norm of any response column vector is less than 1. When the minimum response resolution of the current candidate excitation code set is determined to be 0, the candidate excitation code set is excluded from the formal excitation code set.

[0171] When the number of candidate excitation codes does not exceed 12, a combined exhaustive search method is used; when the number exceeds 12, a greedy search method that gradually increases the number of excitation codes is used.

[0172] In this embodiment, the preset resolution threshold is set to 0.16. When the minimum response resolution reaches 0.16, or when the increase in minimum response resolution due to the addition of excitation code is less than 0.01, the addition of excitation code is stopped.

[0173] Magnetic field model, coupling compensation, and coded excitation: Each of the 127 steel wires was modeled according to its actual spatial centerline. The maximum mesh size for the candidate defect region, magnetic pole edge, and corresponding positions of the double-layer triaxial magnetic sensitive array was set to 0.4 mm.

[0174] The expected magnetic response of the corresponding sensor in the two triaxial magnetic arrays is calculated using the average magnetic flux density component within the sensitive region. Forward and reverse excitation are solved separately.

[0175] The coupling coefficient between adjacent circumferential excitation sectors is obtained by driving a single circumferential excitation sector. Clockwise and counterclockwise adjacent coupling coefficients are established for each circumferential excitation sector.

[0176] The excitation compensation is limited to 3% to 12% of the target excitation current. If any magnetic sensing channel is expected to reach more than 85% of the sensor's range after compensation, the excitation compensation should be reduced.

[0177] Each excitation code uses three excitation intensities: 0.6A, 1.3A, and 2.2A, corresponding to weak magnetization, medium magnetization, and near-saturation magnetization, respectively.

[0178] At least for the entire sector unidirectional excitation code, perform forward excitation and overall reverse excitation respectively.

[0179] When switching between different excitation codes, all excitation currents are reduced to zero. If the change in magnetic flux density within the detection window is no greater than 0.05 mT within 15 ms, the next excitation code is executed.

[0180] When it is necessary to reduce the impact of residual magnetism history, a demagnetization sequence is executed with excitation current amplitudes of 2A, 1.5A, 1A, 0.6A, 0.3A and 0A, and the directions change alternately.

[0181] Local attenuation coefficient and lift-off correction: The excitation current is considered to have reached a stable condition when the deviation between the excitation current and the target excitation current is no greater than 1.5% and remains stable for 18 ms. After the stable condition is met, data is continuously collected for 50 ms.

[0182] For the same detection location, excitation code, excitation intensity, and measurement axis, index according to the excitation direction. Lift-off corrections are applied to both forward and reverse excitation data. For each excitation direction, the local attenuation coefficient is determined using the following formula, and the magnetic flux density component acquired by the first triaxial magnetic array in that excitation direction is corrected to the reference lift-off distance: ; In the formula, Indicates the detection location sequence number; Indicates the excitation code sequence number; Indicates the excitation intensity sequence number; Indicates the measurement axis number of the triaxial magnetic sensor; Indicates the excitation direction index. , This indicates positive excitation, which is executed according to each coding element in the candidate excitation code; This indicates reverse excitation, where the direction of the excitation current for all non-zero coded elements is simultaneously reversed, while the stop state remains unchanged. This represents the local attenuation coefficient, in mm. -1 ; This represents the magnetic flux density component acquired by the first three-axis magnetic array, in mT. This represents the magnetic flux density component acquired by the second and third axis magnetic arrays, in mT. This represents the magnetic flux density component after correction to the reference lift-off distance, in mT. This represents the magnetic flux density bias, measured in mT. and These represent the actual lift-off distances of the first and second triaxial magnetic arrays, respectively, in mm; This indicates the baseline lift-off distance, in mm.

[0183] This embodiment will Set as mT sets the reference lift-off distance to 1.2mm. For = + and For each excitation direction in the equation, the actual lift-off distance of the second and third-axis magnetic sensing arrays should be greater than that of the first and third-axis magnetic sensing arrays. If the difference between the two is less than 1 mm or not positive in any excitation direction, the local attenuation coefficient for that excitation direction is not calculated, and the corresponding magnetic sensing channel in that direction is marked as a low-confidence magnetic sensing channel. The local attenuation coefficient calibration range is 0.04 mm. -1 Up to 1.25mm -1 When the local attenuation coefficient in any excitation direction is negative or exceeds the calibration range, the channel confidence of the corresponding magnetic sensitive channel in that excitation direction is set to 0.25.

[0184] In each excitation direction, the first and second triaxial magnetic sensing arrays are synchronously sampled within the same stable acquisition window, and the effective sampled values ​​of the same measurement axis are calculated by arithmetic mean. The average values ​​are used as the near-layer magnetic induction intensity component and the far-layer magnetic induction intensity component in formula (2), respectively. The actual lift-off distance obtained synchronously with the stable acquisition window is used as the near-layer actual lift-off distance and the far-layer actual lift-off distance in formula (2), respectively. When calculating the local attenuation coefficient, only the absolute value of the magnetic induction intensity component is taken to characterize the attenuation of the amplitude with the lift-off distance. When calculating the corrected magnetic induction intensity component, the sign of the near-layer magnetic induction intensity component is retained to maintain the direction information of the magnetic field component.

[0185] The first and second triaxial magnetic arrays are corrected to the same reference lift-off distance using formula (2), and the correction results of the two layers are respectively entered into formula (3) to form the corresponding final response terms according to the magnetic array level.

[0186] After obtaining the positive and negative corrected magnetic induction intensity components according to formula (2), they are transformed to the same local coordinate system of the steel wire, and subtracted from the corresponding positive and negative baseline responses, respectively. After completing the positive and negative baseline differences, they are combined into a defect-sensitive response according to formula (3). Therefore, the excitation direction index in formula (2) The data indexes belonging to the original response and lift-off correction stages are not used as independent arrangement dimensions for the final response items.

[0187] Working principle and technical effects: This embodiment determines the steel wire sub-bundles by considering the center distance of the steel wires, differences in magnetic coupling, and connectivity, thus providing clear geometric and magnetic response basis for the division of the steel wire sub-bundles.

[0188] The rotating connection structure can adapt to the large circumferential angle changes in the expanded diameter section, and the miniature electric actuator can adjust the pushing amount according to the axial detection zones.

[0189] Candidate excitation codes are selected based on the minimum response resolution to match the excitation code with the arrangement of the 127 steel wires and the magnetic circuit of the anchor.

[0190] Local closed magnetic circuits, non-magnetic isolation gaps, and coupling compensation between adjacent circumferential excitation sectors can improve the controllability of different circumferential excitation states.

[0191] The dual-layer triaxial magnetic array uses the difference in magnetic induction intensity at two detection normal positions to determine the local attenuation coefficient, and corrects the magnetic induction intensity component collected by the first triaxial magnetic array to a unified reference lift-off distance, thereby improving the comparability of detection responses under different bonding states.

[0192] Example 3: Based on the flexible partitioned magnetization sensing component, candidate defect unit division and coded magnetic response matrix construction method in Example 1, this example further illustrates the local coordinate reconstruction of the steel wire, forward and reverse baseline updates, channel confidence weighting, topological constraint defect inversion and information gain fine scanning.

[0193] The defect state definition, topology constraint matrix construction rules, channel credibility criteria, and location uncertainty determination methods disclosed in Example 1 continue to apply to this example.

[0194] Response items and data arrangement: Each magnetic flux density component acquired by each triaxial magnetic sensor on the three measurement axes is used as a magnetic channel. Each magnetic channel is uniquely determined by the axial detection position, circumferential detection position, magnetic array level, and measurement axis of the corresponding triaxial magnetic sensor. A combination of an excitation code, an excitation intensity, and a magnetic channel is used as a response term. The coded magnetic response matrix and the detection response vector are arranged in the order of excitation code, excitation intensity, axial detection position, circumferential detection position, magnetic array level, and measurement axis. The excitation direction index in formula (2) is used to distinguish between the positive and negative original responses and to perform lift-off corrections for the two excitation directions respectively. After the forward and reverse baseline difference is completed using formula (3), the data from the two excitation directions are combined into a single defect-sensitive response. The excitation direction is not used as an independent arrangement dimension for the final response term. The expected magnetic response in the encoded magnetic response matrix and the defect-sensitive response in the detection response vector adopt the same lift-off correction method, wire local coordinate definition, forward and reverse baseline difference method, channel gain correction method, and response term arrangement order. When any response term lacks valid field data, the corresponding element in the detection response vector is set to zero, and the corresponding channel confidence level is set to zero.

[0195] Component-based lift-off correction and wire local coordinate reconstruction: The component-based lift-off correction uses formula (2) from Example 2, and calculates the local attenuation coefficient and the corrected magnetic flux density component under both forward and reverse excitation conditions according to the excitation direction index. In this example, the reference lift-off distance is set to 1.5 mm. The lift-off distance calibration range for the first triaxial magnetic array is 0.8 mm to 3.5 mm, and the lift-off distance calibration range for the second triaxial magnetic array is 3.3 mm to 6.0 mm. Validity is determined for η=+ and η=- respectively: the local attenuation coefficient calibration range for the local tangential magnetic flux density component is 0.05 mm. -1 Up to 1.20mm -1 The calibration range for the local attenuation coefficient corresponding to the local normal magnetic induction intensity component is 0.08 mm. -1 Up to 1.50mm -1 The local attenuation coefficient corresponding to the local secondary normal magnetic induction intensity component is calibrated within a range of 0.04 mm. -1 Up to 1.00mm -1 .

[0196] When the actual lift-off distance or local attenuation coefficient exceeds the corresponding calibration range, the corresponding magnetic sensing channel is marked as a low-confidence magnetic sensing channel. A cubic spline curve is established based on the centerline of the wire bundle surface obtained from structured light scanning, with the direction of the first derivative of the spline curve at the detection position as the initial local tangent. When structured light scanning data is unavailable, the local extension direction of the wire is continuously determined based on the curvature detection array and the attitude of the flexible support, and the cumulative direction error is limited using the design centerline of the suspension cable. The radial vector from the detection position to the centerline of the wire bundle is projected into a plane perpendicular to the local tangent, and the projected unit vector is used as the local normal. The local sub-normal is determined jointly by the local tangent and the local normal.

[0197] When the local tangential directions of adjacent detection positions are opposite, the local tangential and local subnormal directions of the subsequent detection position are simultaneously reversed to ensure continuity of the wire's local coordinate system along the cable axis. A third-order direction cosine matrix is ​​established based on the local tangential, local normal, and local subnormal directions. The triaxial magnetic flux density, after lift-off correction, is then projected onto the wire's local coordinate system using this direction cosine matrix. The direction cosine matrix is ​​a dimensionless orthogonal matrix, and the magnetic flux density vectors before and after the transformation are both in mT units.

[0198] Forward baseline response and reverse baseline response: The forward and reverse baseline responses are stored separately according to the excitation code, excitation intensity, detection position, and magnetic sensing channel, and indexed by the excitation direction. To distinguish them, among them = + corresponds to the positive baseline response, = - Corresponding to the reverse baseline response. The original field response under the same excitation direction index is first corrected by formula (2), and then the difference is made with the baseline response under the same excitation direction index; after the forward and reverse baseline difference is completed, the excitation direction is not used as an independent arrangement dimension of the response term. The baseline response is obtained through defect-free standard steel wire bundles, initial inspection data of slings, data of anchorage structures of the same type, or magnetic field finite element models.

[0199] When multiple baseline data exist simultaneously, the initial inspection data of the sling should be used first; for inspection locations where the initial inspection data is missing, the data of the defect-free standard steel wire bundle should be used; when there is still no effective reference data near the anchor cup, magnetic field finite element simulation data should be used to supplement it.

[0200] The cable force calibration nodes are set to 20%, 40%, 60%, 80%, and 100% of the rated cable force; the temperature calibration nodes are set to -10℃, 10℃, 30℃, 50℃, and 60℃.

[0201] When the actual cable force and actual temperature are between adjacent calibration nodes, bilinear interpolation is used to obtain the forward and reverse baseline responses. When the actual cable force or actual temperature exceeds the calibration range, no extrapolation is performed; instead, the boundary calibration value closest to the current state is used, and the channel confidence of the corresponding magnetic sensing channel is reduced to 0.4.

[0202] After obtaining the positive and negative corrected magnetic induction intensity components from formula (2), transform them to the same local coordinate system of the steel wire, and subtract the updated baseline response under the same excitation direction, the defect-sensitive response is determined according to the following formula: ; In the formula, Indicates the detection location sequence number; Indicates the excitation code sequence number; Indicates the excitation intensity sequence number; This represents the defect-sensitive response vector, with units of mT; This represents the triaxial magnetic induction intensity vector under positive excitation conditions, corrected by the reference lift-off distance and transformed to the local coordinate system of the steel wire, with units of mT; This represents the triaxial magnetic flux density vector after the same treatment under reverse excitation conditions, with units of mT; and These represent the use of the same baseline lift-off distance, the same local coordinate definition for the steel wire, and the same channel gain correction method, respectively, and are based on the actual cable force. and actual temperature Updated forward and reverse baseline responses.

[0203] The four vectors in formula (3) are all three-dimensional magnetic induction intensity vectors, and have the same data arrangement order and dimensions.

[0204] The slowly varying components that change in the same direction under both forward and reverse excitation mainly reflect temperature drift, sensor bias, cable force changes, or anchor background changes; the components that reverse with the magnetization direction and have local abrupt changes in space serve as the main input for defect inversion.

[0205] Channel credibility weighted matrix: The channel reliability is determined based on the lift-off distance corresponding to each response item, the change in bonding pressure within 500ms, the saturation state of the triaxial magnetic sensor, the difference between signal noise and repeated sampling results.

[0206] When the lift-off distance and local attenuation coefficient are within the calibration range, the change in bonding pressure is no greater than 1 kPa, the magnetic induction intensity measurement value does not reach 90% of the sensor range, the signal-to-noise ratio is no less than 12 dB, and the difference in repeated sampling is no greater than 8%, the channel confidence level is set to 1.

[0207] When only one condition is close to the allowable boundary, the channel confidence level is set to 0.7; when one of the following occurs, namely, lift-off distance fluctuation, bonding pressure fluctuation, or signal-to-noise ratio below 12dB, the channel confidence level is set to 0.4; when the actual lift-off distance or local attenuation coefficient exceeds the calibration range, the sensor is saturated, or the difference in repeated sampling is greater than 15%, the channel confidence level is set to 0; when there is no valid sampling data, the channel confidence level is set to 0.

[0208] When multiple anomalies occur simultaneously, the lowest confidence level is used as the final channel confidence level.

[0209] Topological constraint defect inversion: For the Each response item sets a positive channel scale value. The unit is mT. The channel scale value is the larger of the noise standard deviation of the response term and the reference response amplitude of the standard magnetic calibrator.

[0210] The first in the detection response vector Divide each element by This forms a dimensionless detection response vector; the encoded magnetic response matrix is ​​then used to generate the first... Divide all elements of the row by This forms a dimensionless encoded magnetic response matrix.

[0211] The defect state is inverted according to the following formula: ; In the formula, This represents the defect state vector obtained through inversion; This represents the candidate defect state vector to be solved; The nth element in the equation represents the equivalent metal cross-sectional loss ratio of the nth candidate defect element; This represents the dimensionless detection response vector after channel-scale processing. This represents the dimensionless encoded magnetic response matrix after processing at the same channel scale. This represents the channel credibility weighting matrix; Represents the normalized topological constraint matrix; This represents the sparsity constraint coefficient, which is set to 0.015. This represents the topological constraint coefficient, taken as 0.08. Describes the norm, This represents the L2 norm.

[0212] The channel credibility weighting matrix is ​​a diagonal matrix, the first... Take the diagonal element as the first The square root of the channel credibility of each response item.

[0213] The encoded magnetic response matrix is ​​not column-normalized; therefore, the elements in the defect state vector directly represent the equivalent metal cross-sectional loss ratio of the corresponding candidate defect element.

[0214] Formula (4) is solved using an accelerated proximal gradient algorithm with interval projection, with a maximum number of iterations set to 500. The relative change in the objective function is no greater than [value missing]. Or the relative change in the defect state vector is no greater than Stop iterating when the time comes.

[0215] Location uncertainty and set of active candidate defect elements: Candidate defect cells with a defect state value of not less than 0.05 are considered valid candidate defect cells. If there are no candidate defect cells with a defect state value of not less than 0.05 in the current inversion results, the output will show no defects that have reached the preset threshold, the positioning uncertainty will not be calculated, and the candidate fine scanning action evaluation will not be initiated.

[0216] Based on the defect state vector obtained from the current inversion, and following the processing method disclosed in Example 1, the axial uncertainty, circumferential uncertainty, and radial uncertainty are determined respectively, and the current positioning uncertainty is formed with weights of 0.5, 0.3, and 0.2.

[0217] If the current positioning uncertainty is no greater than 0.20, the evaluation of candidate fine scanning actions will not be initiated; if the current positioning uncertainty is greater than 0.20, candidate fine scanning actions will be established and the expected positioning uncertainty corresponding to each candidate fine scanning action will be determined.

[0218] The active candidate defect element set consists of candidate defect elements with a current defect state value of not less than 0.05, and candidate defect elements that have axial, circumferential, or radial topological adjacency with them. Different candidate fine-scanning actions use the same active candidate defect element set and the same candidate defect element arrangement order.

[0219] The current positioning uncertainty and the expected positioning uncertainty corresponding to the candidate fine sweeping action both represent the weighted result of axial uncertainty, circumferential uncertainty and radial uncertainty, and they have the same definition, value range and physical meaning.

[0220] Excitation coding extension: If the response resolution between any two candidate defect units under the current excitation code set is less than 0.18, at least one of the following: single-sector excitation code, adjacent double-sector excitation code, and opposed double-sector excitation code is added to the candidate excitation code set. After adding the new candidate excitation code, the minimum response resolution of the candidate excitation code set is recalculated, and a new excitation code set is selected.

[0221] New excitation codes should use at least two excitation intensities of 0.8A and 2.4A respectively; when further differentiation of nonlinear magnetization response is required, an intermediate excitation intensity of 1.6A should be added. When switching excitation codes, the excitation current of each circumferential excitation sector should be reduced to zero. The next excitation code should be executed when the total excitation current is no greater than 1% of the rated excitation current, and the change in magnetic induction intensity within the detection window is no greater than 0.05mT within 15ms.

[0222] Approximate defect state covariance matrix and predicted location uncertainty: Based on the current inspection status and each candidate fine-scanning action, under the same set of active candidate defect units and the same order of arrangement of candidate defect units, the approximate defect status covariance matrix is ​​determined according to the following formula: ; In the formula, This represents the approximate defect state covariance matrix corresponding to the h-th candidate fine-scanning action; when h=0, This represents the approximate defect state covariance matrix before fine scanning; when h≥1, This represents the approximate defect state covariance matrix expected after executing the h-th candidate fine-scan action; Represents the set of active candidate defective units; This represents the column submatrix of the dimensionless candidate encoded magnetic response matrix corresponding to the h-th candidate fine scanning action on the set of active candidate defect elements; This represents the channel confidence weighting matrix corresponding to the h-th candidate fine scan action; This represents the column submatrix of the topological constraint matrix on the set of active candidate defective elements; Denotes the diagonal stability coefficient, taken as... ; The identity matrix represents the number of active candidate defective units, with the order equal to the number of such units.

[0223] Formula (5) is a local quadratic approximation of the objective function of Formula (4) within the positive value region of the active candidate defect unit. The sparse constraint term does not generate a second-order term within the positive value region of the active candidate defect unit, and therefore is not written into the approximate information matrix.

[0224] The approximate defect state covariance matrix obtained by formula (5) is used to describe the expected dispersion of the defect state value of the active candidate defect unit under the current inspection state or candidate fine cleaning action. The approximate defect state covariance matrix is ​​intermediate data for determining the expected positioning uncertainty. It is not directly used as the positioning uncertainty, nor is the matrix trace directly used to replace the positioning uncertainty.

[0225] For each h, the defect state value of the currently inverted active candidate defect unit is used as the defect state mean, and the corresponding approximate defect state covariance matrix is ​​used as the mean. As the discrete range of defect states, 1000 sets of candidate defect state samples are generated.

[0226] Candidate defect status samples are generated using a multivariate normal sampling method. To ensure the same expected positioning uncertainty under the same input conditions, the pseudo-random number seed is determined based on the inspection task number and the candidate fine-scanning action sequence number h.

[0227] After each set of candidate defect state samples is generated, defect state values ​​less than 0 are corrected to 0, defect state values ​​greater than 1 are corrected to 1, and the defect state values ​​of candidate defect units outside the active candidate defect unit set are set to 0.

[0228] For each group of candidate defect state samples, candidate defect units with defect state values ​​not less than 0.05 are determined as valid candidate defect units for that group of samples. When valid candidate defect units exist, the axial uncertainty, circumferential uncertainty, and radial uncertainty of that group of samples are calculated according to the processing method disclosed in Example 1, and the positioning uncertainty of that group of samples is formed according to weights of 0.5, 0.3, and 0.2.

[0229] If there are no candidate defect units with a defect state value of not less than 0.05 in a set of candidate defect state samples, the positioning uncertainty of the set of samples is set to 1, indicating that the set of samples cannot form a valid positioning result that meets the defect state threshold, and positioning fine scanning is not started.

[0230] The arithmetic mean of the positioning uncertainties corresponding to 1000 sets of candidate defect state samples is used to obtain the expected positioning uncertainty of the corresponding inspection state or candidate fine sweeping action.

[0231] When h=0, according to The determined predicted positioning uncertainty is denoted as the predicted positioning uncertainty before fine scanning. When h≥1, according to The determined predicted positioning uncertainty is denoted as the predicted positioning uncertainty after performing the h-th candidate fine scan action. .

[0232] Predicted positioning uncertainty before fine scanning And the estimated positioning uncertainty after fine scanning The same axial uncertainty, circumferential uncertainty, and radial uncertainty definitions and weighting methods as in Example 1 are used, and the difference between the two represents the expected reduction in positioning uncertainty.

[0233] Expected information gain: Candidate fine scanning actions include adding single-sector excitation coding, adding adjacent double-sector excitation coding, adding opposing double-sector excitation coding, increasing intermediate excitation intensity, increasing forward and reverse excitation, reducing axial movement speed, increasing sampling frequency, and adjusting the bonding pressure of the target axial detection zone.

[0234] When the candidate fine scanning action is to reduce the axial moving speed, the sampling frequency remains unchanged, and the new axial sampling interval is determined according to the ratio of moving speed to sampling frequency. The candidate encoded magnetic response matrix is ​​expanded according to the expected magnetic response of the newly added axial detection position, and the detection response vector is expanded and the channel confidence weighting matrix is ​​updated after retesting.

[0235] When the excitation code or excitation intensity is changed, the expected new magnetic response is obtained based on the finite element calculation of the magnetic field or the standard steel wire bundle calibration data, and a candidate code magnetic response matrix is ​​established.

[0236] When increasing the number of samplings or adjusting the bonding pressure, the current encoded magnetic response matrix is ​​used, and the expected channel reliability is determined based on the expected signal noise, expected lift-off distance, and expected bonding pressure.

[0237] The expected information gain of the h-th candidate fine scan action is determined by the following formula: ; In the formula, Let represent the expected information gain of the h-th candidate fine scan action, which is a dimensionless quantity; This represents the estimated positioning uncertainty before fine scanning, and is a dimensionless quantity. This represents the expected positioning uncertainty after performing the h-th candidate fine scanning action, and is a dimensionless quantity. This represents the expected reduction in positioning uncertainty corresponding to the h-th candidate fine scanning action.

[0238] This indicates the estimated execution time, in seconds. This indicates the projected energy consumption, expressed in kJ. This represents the time cost coefficient, taken as 0.02s. -1 ; This represents the energy consumption cost coefficient, taken as 0.01 kJ. -1 .

[0239] Since the expected reduction in positioning uncertainty, time cost, and energy cost are all dimensionless, the dimensions of each item in formula (6) are consistent.

[0240] when When the expected reduction in positioning uncertainty is positive, it indicates that the corresponding candidate fine-scanning action is expected to improve the concentration of defect positioning; when If the expected reduction in positioning uncertainty is zero or negative, it means that the corresponding candidate fine scanning action cannot reduce the positioning uncertainty, or it is expected to further disperse the positioning results.

[0241] When circumferential uncertainty is the primary source of uncertainty, priority is given to evaluating single-sector excitation coding and adjacent double-sector excitation coding; when radial uncertainty is the primary source of uncertainty, priority is given to evaluating opposed double-sector excitation coding and different excitation intensities; when axial uncertainty is the primary source of uncertainty, priority is given to evaluating reducing axial movement speed and increasing axial sampling frequency; when low-confidence magnetic sensing channels are mainly caused by changes in lift-off distance or bonding pressure, priority is given to evaluating bonding pressure adjustment actions. When the expected information gain of all candidate fine scanning actions is not greater than zero, the fine scanning is stopped and a prompt indicating that other detection methods need to be used for verification is output.

[0242] Select the candidate fine scan action with the highest expected information gain. When the difference in expected information gain between two candidate fine scan actions is less than 0.01, prioritize the candidate fine scan action with the shorter expected execution time; when the expected execution times are the same, select the candidate fine scan action with the lower expected energy consumption.

[0243] After executing the candidate fine scan action that changes the excitation code or excitation intensity, the coded magnetic response matrix is ​​updated according to the newly added expected magnetic response, the detection response vector is updated according to the defect sensitive response obtained from the retest, the channel confidence weighting matrix is ​​updated according to the retest status, and the defect status inversion, current positioning uncertainty calculation and subsequent candidate fine scan action evaluation are performed again.

[0244] The maximum number of fine scans is set to 3. Fine scanning stops when the positioning uncertainty is no greater than 0.20; if the positioning uncertainty is still greater than 0.20 after 3 fine scans, a prompt will be output indicating that other non-destructive testing methods need to be used for verification.

[0245] Working principle and technical effects: The first and second triaxial magnetic arrays simultaneously acquire triaxial magnetic flux density data at two detection normal positions. The ratio of magnetic flux density amplitudes of the corresponding magnetic channels and the difference in actual lift-off distance are used to determine the local attenuation coefficient, enabling the field response to be converted to a unified reference lift-off distance.

[0246] The local coordinate system of the steel wire unifies the triaxial magnetic induction intensity at different detection positions into local tangential, local normal, and local subnormal components, reducing the directional error caused by changes in the attitude of the flexible load-bearing body and the direction of the steel wire.

[0247] Actual cable force and actual temperature are used to update the forward and reverse baseline responses, respectively. The forward and reverse baseline differential can preserve the local defect response that reverses with the magnetization direction and reduce slowly varying responses such as temperature drift, sensor bias, and anchor background magnetic field.

[0248] The encoded magnetic response matrix and the detected response vector adopt the same lift-off correction, coordinate transformation, baseline difference and response term arrangement, so that the defect inversion is based on a unified physical response definition.

[0249] The channel confidence weighting matrix reduces the impact of low-quality magnetic sensitive channels on the inversion results, while the topological constraint matrix suppresses isolated spurious defects that do not conform to the actual arrangement of the wire bundle.

[0250] When the existing excitation coding cannot adequately distinguish adjacent candidate defect units, new magnetic flux paths are formed by adding single-sector, adjacent double-sector, or opposing double-sector excitation coding.

[0251] The approximate defect state covariance matrix is ​​used to predict the statistical discrete range of candidate defect states under different candidate fine scanning actions, and is propagated into axial, circumferential, and radial positioning uncertainties through candidate defect state samples. The expected information gain is directly adopted as the difference between the expected positioning uncertainty before and after fine scanning, so that the fine scanning evaluation object is consistent with the definition of positioning uncertainty.

[0252] Information gain fine scanning selects the corresponding excitation, sampling, or bonding adjustment actions based on the main sources of uncertainty in the axial, circumferential, or radial directions, avoiding aimless repetitive scanning under the same parameter conditions.

[0253] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A flexible magnetic flux leakage inspection method for steel wires in the anchorage zone of a suspension cable, characterized in that, include: Step S1: Obtain the wire arrangement, radial hierarchy, axial direction and anchor structure, and establish a wire bundle topology model containing the adjacent relationships of the wires; divide the wire sub-bundles according to the adjacent relationships of the wires, and divide the candidate defect units according to the axial interval, wire sub-bundles and radial hierarchy. Step S2: The flexible partitioned magnetization sensing component, which includes a flexible bearing body, a circumferential excitation sector, an axial detection partition, a partitioned pushing mechanism, and a double-layer triaxial magnetic sensitive array, is wrapped around the anchoring area. The candidate excitation code consists of the forward, reverse, or stop states of the circumferential excitation sector; Step S3: Obtain the expected magnetic response of the candidate defect unit under the candidate excitation coding, select the excitation coding set according to the response difference of different candidate defect units, and establish the coding magnetic response matrix of the corresponding candidate defect unit. Step S4: Obtain the lift-off distance, bonding pressure and local curvature of each axial detection zone, and adjust the bonding state and excitation parameters accordingly; when the lift-off distance and bonding pressure meet the acquisition conditions, drive the circumferential excitation sector according to the excitation code set, and at least one code performs forward and reverse excitation, and simultaneously acquire the double-layer triaxial magnetic field, cable force and temperature. Step S5: Perform lifting correction based on the double-layer triaxial magnetic field and lifting distance; reconstruct the local coordinate system of the steel wire according to the local curvature and the attitude of the carrier; update the preset positive and negative baseline responses based on cable force and temperature; and subtract the corrected positive and negative excitation responses from the corresponding baselines to obtain the defect-sensitive response. Step S6: The defect-sensitive responses are combined into a detection response vector. The defect location, cross-sectional loss, and positioning uncertainty are inverted based on the encoded magnetic response matrix and topological constraints. When the positioning uncertainty is greater than the preset threshold, the information gain is determined according to the uncertainty reduction and cost of the bonding or encoding adjustment. The maximum value is selected for retesting and inversion until it is no greater than the preset threshold or the upper limit of the number of times is reached. The result is then output.

2. The flexible magnetic flux leakage inspection method for the steel wire in the anchorage zone of the suspension cable according to claim 1, characterized in that, In step S1, the anchorage area is continuously divided into multiple axial intervals along the axial direction of the sling. Each axial interval covers the anchorage area, and the boundaries of adjacent axial intervals are connected and do not overlap. Within each radial level of each axial interval, all the wires in that radial level are divided into multiple wire sub-bundles according to the wire adjacency relationship. Each wire sub-bundle includes at least one wire that is spatially adjacent to the other. The same wire belongs to only one wire sub-bundle within the same axial interval and the same radial level. A combination of an axial interval, a wire bundle, and a radial level is considered as a candidate defect unit.

3. The flexible magnetic flux leakage inspection method for the steel wire in the anchorage zone of the suspension cable according to claim 2, characterized in that, Each candidate excitation code includes multiple coding elements that correspond one-to-one with multiple circumferential excitation sectors. Each coding element represents the forward excitation state, reverse excitation state, or stop excitation state of the corresponding circumferential excitation sector. The dual-layer triaxial magnetic array includes multiple triaxial magnetic sensors, and each magnetic field component collected by each triaxial magnetic sensor on the three measurement axes is used as a magnetic channel. Each magnetic sensing channel is uniquely determined by the axial detection position, circumferential detection position, magnetic array level, and measurement axis of the corresponding triaxial magnetic sensing sensor; The combination of a candidate excitation code, an excitation intensity, and a magnetic sensing channel is taken as a response term. The expected defect-sensitive responses obtained by differentiating each candidate defect unit from the forward and reverse baselines are arranged in the order of candidate excitation code, excitation intensity, axial detection position, circumferential detection position, magnetic sensing array level, and measurement axis to form a response column vector. The difference between 1 and the absolute value of the normalized inner product of any two response column vectors is determined as the response resolution. The minimum value among the response resolutions is determined as the minimum response resolution of the candidate excitation code set. The candidate excitation code set with the largest minimum response resolution is selected as the excitation code set, and the coded magnetic response matrix is ​​established according to the corresponding response column vector; When there are multiple parallel candidate excitation code sets, the one with the fewest candidate excitation codes is selected.

4. The flexible magnetic flux leakage inspection method for the steel wire in the anchorage zone of the suspension cable according to claim 1, characterized in that, The flexible support body includes multiple rigid sections, and adjacent rigid sections are connected by flexible connecting strips or rotating connectors. Each of the circumferential excitation sectors includes a flexible magnetic yoke, an excitation coil, a first magnetic pole, and a second magnetic pole, wherein the first magnetic pole and the second magnetic pole are spaced apart along the axial direction of the sling; When the excitation coil is energized, the magnetic flux passes sequentially through the flexible magnetic yoke, the first magnetic pole, the steel wire located between the first magnetic pole and the second magnetic pole, and the second magnetic pole, and returns to the flexible magnetic yoke to form a partially closed magnetic circuit. Non-magnetic isolation gaps are set between the flexible magnetic yokes of adjacent circumferential excitation sectors, and each excitation coil is driven by an independent bidirectional current drive branch.

5. The flexible magnetic flux leakage inspection method for the steel wire in the anchorage zone of the suspension cable according to claim 4, characterized in that, Multiple axial detection zones are arranged continuously along the axial direction of the sling and cover the anchorage area, with the boundaries of adjacent axial detection zones connected and their interiors not overlapping. Each axial detection zone is equipped with a lift-off distance detection unit, a bonding pressure detection unit, a curvature detection unit, and an independent pushing unit; Based on the mapping relationship of the bonding excitation parameters obtained from the calibration of the defect-free steel wire bundle, the pushing amount, excitation current compensation amount and allowable acquisition flag are determined by the lifting distance, bonding pressure and local curvature of the current axial detection zone. The corresponding independent pushing unit is controlled according to the pushing amount to adjust the bonding state, and the excitation current amplitude of each circumferential excitation sector in the excitation state is adjusted according to the excitation current compensation amount while keeping the excitation direction specified by the current excitation code unchanged. When the allowed acquisition flag is set to prohibited acquisition, the triaxial magnetic field data of the corresponding axial detection zone is prohibited from being used to generate a defect-sensitive response.

6. The flexible magnetic flux leakage inspection method for the steel wire in the anchorage zone of the suspension cable according to claim 1, characterized in that, The dual-layer triaxial magnetic array includes a first triaxial magnetic array and a second triaxial magnetic array spaced apart along the detection normal. Both triaxial magnetic arrays include multiple triaxial magnetic sensors, and the triaxial magnetic sensors with the same axial position and circumferential position are corresponding magnetic sensors to each other. Each magnetic field component acquired by each triaxial magnetic sensor on the three measurement axes is used as a magnetic sensing channel. For the same detection position, excitation code, excitation direction and excitation intensity, the local attenuation coefficient is determined based on the ratio of the magnetic field component amplitudes collected by the corresponding magnetic channels in the first and second three-axis magnetic arrays, and the difference between the actual lift-off distance of the corresponding magnetic channel in the second and first three-axis magnetic arrays and the actual lift-off distance of the corresponding magnetic channel in the first three-axis magnetic array. The magnetic field component collected by the first three-axis magnetic array is then corrected to a preset reference lift-off distance based on the local attenuation coefficient. When the actual lift-off distance or local attenuation coefficient exceeds the corresponding calibration range, the corresponding magnetic sensitive channel is marked as a low-confidence magnetic sensitive channel, and its corresponding defect-sensitive response channel weight is reduced during defect inversion.

7. The flexible magnetic flux leakage inspection method for the steel wire in the anchorage zone of the suspension cable according to claim 6, characterized in that, In step S5, the local extension direction of the steel wire at the detection position is determined according to the local curvature and the posture of the flexible support body, and a local coordinate system of the steel wire consisting of local tangential, local normal and local subnormal is established; The expected magnetic response and the triaxial magnetic field response after lift-off correction are respectively transformed into the local coordinate system of the steel wire at the corresponding detection position; A positive baseline response and a negative baseline response are pre-established based on at least one of the following: defect-free wire bundle calibration data, initial inspection data of the sling, and magnetic field simulation data. The positive baseline response and the negative baseline response are updated according to the cable force and the temperature, respectively. The positive difference is obtained by subtracting the positive excitation response from the updated positive baseline response, and the negative difference is obtained by subtracting the negative excitation response from the updated negative baseline response. Half of the difference obtained by subtracting the negative difference from the positive difference is taken as the defect-sensitive response.

8. The flexible magnetic flux leakage inspection method for the steel wire in the anchorage zone of the suspension cable according to claim 1, characterized in that, The dual-layer triaxial magnetic array includes multiple triaxial magnetic sensors. Each magnetic field component collected by each triaxial magnetic sensor on the three measurement axes is used as a magnetic channel. Each magnetic channel is uniquely determined by the axial detection position, circumferential detection position, magnetic array level and measurement axis of the corresponding triaxial magnetic sensor. A combination of an excitation code, an excitation intensity and a magnetic channel is used as a response item. The expected magnetic response in the encoded magnetic response matrix and the defect-sensitive response in the detection response vector are made to use the same lift-off correction method, wire local coordinate definition and forward and reverse baseline difference method, and the response items are arranged in the order of excitation code, excitation intensity, axial detection position, circumferential detection position, magnetic array level and measurement axis. The channel reliability is determined based on the lift-off distance, the change in bonding pressure within a preset time, the saturation state of the triaxial magnetic sensor, the signal noise, and the difference between repeated sampling results of the same response item, and a channel reliability weighting matrix is ​​established. A topological constraint relationship is established based on the adjacency relationship of steel wires and the radial hierarchy. The defect state is then inverted based on the encoded magnetic response matrix, the detection response vector, the channel confidence weighting matrix, and the topological constraint relationship. The positioning uncertainty is determined based on the distribution range of candidate defect units with defect state values ​​not less than a preset defect state threshold in the axial interval, wire bundle, and radial level.

9. The flexible magnetic flux leakage inspection method for the steel wire in the anchorage zone of the suspension cable according to claim 1, characterized in that, In step S6, when the positioning uncertainty is greater than a preset threshold, multiple candidate fine scanning actions are established. The candidate fine scanning actions include at least one of coding adjustment action and bonding adjustment action. The coding adjustment action includes adding excitation coding for driving the local circumferential excitation sector. The bonding adjustment action includes adjusting the bonding pressure of the axial detection partition corresponding to the current defect inversion result. For the aforementioned coding adjustment action, the expected new magnetic response is obtained based on finite element calculations of the magnetic field or standard wire bundle calibration data, and a candidate coding magnetic response matrix is ​​established based on the expected new magnetic response. For the aforementioned bonding adjustment action, the current encoded magnetic response matrix is ​​used as a candidate encoded magnetic response matrix, and the expected channel confidence level is determined based on the expected signal noise, lift-off distance, and bonding pressure. The expected positioning uncertainty is determined based on the candidate coding magnetic response matrix corresponding to each candidate fine scanning action, the current defect inversion result, and the expected channel confidence. The expected information gain is determined based on the reduction of the expected positioning uncertainty, the execution time, and the energy consumption. Perform the candidate fine scan action with the highest expected information gain and retest it; when the candidate fine scan action is the encoding adjustment action, update the encoding magnetic response matrix according to the corresponding expected magnetic response; update the detection response vector according to the defect sensitive response obtained from the retest, update the channel confidence weighting matrix according to the retest status, and perform defect inversion again.

10. The flexible magnetic flux leakage inspection method for the steel wire in the anchorage zone of the suspension cable according to claim 3, characterized in that, When the resolution of the expected magnetic response of any two candidate defect units under the excitation code set is lower than the preset resolution threshold, at least one of the following is determined as a new candidate excitation code and added to the candidate excitation code set: single-sector excitation code that drives only one circumferential excitation sector, adjacent double-sector excitation code that drives two adjacent circumferential excitation sectors, and opposing double-sector excitation code that drives two circumferential excitation sectors that are arranged opposite each other along the circumference of the sling. Based on the candidate excitation code sets formed after adding the new candidate excitation codes, the minimum response resolution of each candidate excitation code set is determined, and the candidate excitation code set with the largest minimum response resolution is selected as the updated excitation code set. At least two excitation intensities are applied under each excitation code included in the updated excitation code set. When switching excitation codes, the excitation current of each circumferential excitation sector is reduced to zero, and the next excitation code is executed after the magnetic field reaches the preset stability condition.

Images