Cable eccentricity detection method

Abstract

The application discloses a cable eccentricity detection method and relates to the technical field of power distribution network detection, and comprises the following steps: collecting multi-point magnetic field component data of a target cable under an energized state; performing data processing on the measurement results, calculating the magnetic field ratio between each pair of opposing sensors; mapping and calibrating the magnetic field ratio to generate an isovalue space trajectory; determining the position of a unique intersection point based on the isovalue space trajectory and a comprehensive scoring mechanism to obtain the spatial offset coordinates of the target core wire relative to the center of the magnetic concentrating ring; and calculating the eccentricity value of the target cable based on the spatial offset coordinates. The application constructs a two-dimensional space trajectory intersection model based on the isovalue space trajectory, determines the position of a unique intersection point from multiple intersection point candidates by using a comprehensive scoring mechanism, and obtains the spatial offset coordinates of the target core wire, so that high-precision measurement of the cable core wire offset is realized, and the problems of large magnetic field interference, low measurement precision and difficulty in real-time positioning in the traditional method are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power distribution network detection, and more particularly to a cable eccentricity detection method. BACKGROUND

[0002] The eccentricity of the core of a high-voltage cable can seriously affect the reliability of its insulation. At present, standards require that the cable be actually measured before it is purchased and installed to ensure that the cable eccentricity does not exceed a threshold value. However, the cable in the running state often changes due to factors such as changes in environmental temperature, wind action, icing suspension, and load fluctuations. This change is sometimes reversible but can also be irreversible. Actual measurement data shows that when the cable is running at 10% eccentricity for 3 years, its breakdown failure rate is 4 times higher than that of the standard cable. Therefore, it is necessary to monitor the eccentricity index of the cable in the running state in real time, and timely maintenance and maintenance of the power transmission line safety and reliability to improve the safety of the power transmission system.

[0003] The above disclosed technical solutions have at least the following technical problems: first, the traditional X-ray scanning method and mechanical pressure detection method require separate equipment and power supply, and the cost of long-term continuous monitoring is high; second, the equipment is not convenient to install on the suspended power transmission line, and its weight may exceed the load bearing capacity of the cable; finally, the above methods are difficult to support live detection, or are easily disturbed by environmental factors, and it is also difficult to apply to real-time monitoring of cable eccentricity on the running line. SUMMARY

[0004] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a cable eccentricity detection method, which constructs a two-dimensional space trajectory intersection model based on an equal ratio space trajectory, and determines the position of a unique intersection point from multiple intersection point candidates by using a comprehensive scoring mechanism, to obtain the spatial offset coordinates of the target core, so as to realize high-precision measurement of the cable core offset, and solve the problems of large magnetic field interference, low measurement precision and difficulty in real-time positioning in the traditional method.

[0005] To achieve the above object, the present application provides the following technical scheme:

[0006] The cable eccentricity detection method comprises the following steps: collecting multi-point magnetic field component data of a target cable in a powered state; performing data processing on the measurement results to calculate the magnetic field ratio between each pair of opposite sensors; mapping and calibrating the magnetic field ratio to generate an equal ratio space trajectory; determining the position of a unique intersection point based on the equal ratio space trajectory and a comprehensive scoring mechanism to obtain the spatial offset coordinates of the target core relative to the center of the magnetic concentration ring; and calculating the eccentricity value of the target cable based on the spatial offset coordinates.

[0007] In a preferred embodiment, the acquisition target cable collects multi-point magnetic field component data in a powered state, specifically: selecting a core wire non-offset test cable for testing; collecting magnetic field component signals on the circumferential side of the magnetic concentrating ring in the powered state of the cable; replacing the test cable with different core wire offset distances, repeating the above steps, and obtaining magnetic field data under different core wire offset states to form a magnetic field component sample set.

[0008] In a preferred embodiment, the magnetic field component signals are obtained by a TMR magnetic field sensor arranged outside the magnetic concentrating ring; the magnetic concentrating ring is composed of two symmetrically spliced half-rings, and a plurality of U-shaped grooves for focusing the magnetic field of the core wire and weakening external electromagnetic interference are uniformly arranged at the edges of the half-rings; the magnetic field sensor is arranged at the U-shaped outer opening of the magnetic concentrating ring.

[0009] In a preferred embodiment, the acquisition target cable collects multi-point magnetic field component data in a powered state, and further includes setting an opposite cooperative purity as a composite trigger criterion, specifically: dividing the diametrically opposite position sensors into two groups; calculating the ratio of the absolute value of the two-channel readings of each group of sensors to the sum of the absolute values of the two-channel readings to obtain a first opposite difference degree and a second opposite difference degree; performing exponential decay function transformation on the first opposite difference degree and the second opposite difference degree to obtain two dimensionless suppression quantities, and multiplying the two dimensionless suppression quantities to obtain the opposite cooperative purity; when either the first opposite difference degree or the second opposite difference degree exceeds a preset proportion threshold, and the opposite cooperative purity is lower than a preset purity threshold, the subsequent processing steps are triggered; otherwise, continuous signal acquisition and caching are maintained.

[0010] In a preferred embodiment, the magnetic field ratio is mapped and calibrated by establishing a ratio-position mapping calibration library, and the ratio-position mapping calibration library is specifically constructed as follows:

[0011] Each group of magnetic field ratios is one-to-one corresponding to the corresponding known eccentric coordinates to form a coordinate sample set; based on the sample set, a nonlinear mapping model between the magnetic field ratio and the position is established; the fitting residual of the nonlinear mapping model at each sample point is calculated, and the residual samples are fed back to the model parameter update layer after Gaussian weighting processing for model correction; the corrected fitting model parameters, the weight matrix, and the residual compensation coefficient are stored as the ratio-position mapping calibration library.

[0012] In a preferred embodiment, the step of establishing a nonlinear mapping model between the magnetic field ratio and the position based on the sample set comprises: performing multi-dimensional coding on the magnetic field ratio vector in the sample set to form a multi-dimensional input feature matrix; establishing a hybrid model combining partition adaptive polynomial fitting and lightweight neural network regression based on the multi-dimensional input feature matrix and the corresponding spatial offset coordinate output; introducing dynamic weight decay and error correlation coefficient constraints in the model training process, determining the model structure parameters through cross-validation, and minimizing the total sample fitting residual sum of squares to obtain an initial fitting model of the magnetic field ratio-space position.

[0013] In a preferred embodiment, the step of generating an equal-ratio space trajectory comprises: extracting a target ratio vector based on the relationship between each feature ratio vector and its corresponding spatial coordinate in the ratio-position mapping calibration library, and calculating its similarity distribution function in the mapping library; calculating the equal set of the similarity distribution function in the entire calibration space, and defining each equal curve as an equal-ratio space trajectory corresponding to the measured ratio; performing curvature continuity constraint and boundary smoothing interpolation on the equal-ratio space trajectory to obtain the final equal-ratio space trajectory.

[0014] In a preferred embodiment, the step of determining the unique intersection position based on the equal-ratio space trajectory and the comprehensive scoring mechanism comprises constructing a two-dimensional space trajectory intersection model, and the specific steps are: receiving two trajectories and unifying them to the same cross-sectional coordinate reference; discretizing each trajectory into a plurality of continuous line segments, and storing the endpoint coordinates, tangent vector, and curvature estimation window index of each line segment; constructing a trajectory narrowband buffer based on the allowable error band set by the working condition; performing intersection detection on the narrowband buffers of the two trajectories to obtain the intersection region and calculate the geometric center as the intersection candidate, and recording the corresponding tangent direction, local curvature, and same-value residual at each intersection candidate.

[0015] In a preferred embodiment, the step of determining the unique intersection position based on the equal-ratio space trajectory and the comprehensive scoring mechanism obtains the spatial offset coordinate of the target core relative to the center of the magnetic concentrating ring, and the specific steps are: calculating the tangent angle sine of the two trajectories at each intersection candidate as the angle component, and estimating the cubic root of the product of the local curvatures of the two trajectories as the curvature component; multiplying the angle component and the curvature component to obtain the trajectory intersection sharpness; calculating the normalized residual of the mapping values of the two pairs of opposite ratios and the real-time ratio, and multiplying them after exponential decay to obtain the convergence potential; taking the product of the trajectory intersection sharpness and the convergence potential as the comprehensive score, and selecting the one with the highest comprehensive score as the unique intersection under the constraints of feasible radius and time sequence proximity; taking the coordinates of the unique intersection as the instantaneous spatial offset coordinate of the core in the cross section.

[0016] In a preferred embodiment, the eccentricity value of the target cable is calculated based on the spatial offset coordinates, specifically: the eccentricity value of the target cable is calculated based on the spatial offset coordinates; the straight line distance from the core position to the origin is calculated with the cross-section geometric center as the origin; and the straight line distance is converted into a dimensionless eccentricity value according to the cable cross-section reference radius.

[0017] The technical effects and advantages of the cable eccentricity detection method of the present application are as follows:

[0018] 1. The present application establishes a nonlinear mapping model of magnetic field ratio-space offset, comprehensively utilizes multiple pairs of sensor ratio inputs, and introduces dynamic weight decay and channel correlation constraints in model training, thereby maintaining physical consistency, avoiding multiple solutions and false solution problems, and ensuring that the output spatial offset coordinates are continuous, monotonic and interpretable. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 The flowchart of the cable eccentricity detection method of the present application is shown;

[0020] Figure 2 The application scenario diagram of the measuring device of an exemplary embodiment of the present application is shown;

[0021] Figure 3 The structure diagram of the measuring device of an exemplary embodiment of the present application is shown;

[0022] Figure 4 The front perspective structure diagram of the measuring device of an exemplary embodiment of the present application is shown;

[0023] Figure 5 The perspective structure diagram of the measuring device of an exemplary embodiment of the present application in the open state is shown;

[0024] Figure 6 The structure diagram of the magnetic concentrating ring of the measuring device of an exemplary embodiment of the present application is shown;

[0025] In the figure: 1 cable detection device; 11, shell; 12, magnetic concentrating ring; 13, support assembly; 14, TMR magnetic field sensor; 2, micro power supply; 3, wireless communication device; 6, upper computer; 7, signal relay device; 8, test cable; 9, power frequency alternating current generating device DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0027] As Figure 2 shown, the device used in the cable eccentricity detection method of the embodiment includes:

[0028] Two parallelly arranged power frequency alternating current generating devices 9, the test cable 8 is installed between the two power frequency alternating current generating devices 9, and the cable detection device is installed on the cable 8.

[0029] As Figures 3-5 shown, the cable detection device 1 of the embodiment includes a protective shell 11, a miniature power supply 2 and a wireless communication device 3 are installed on the inner edge of the protective shell 11, and a magnetic concentrating ring 12 is installed at the center of the protective shell 1, the protective shell 11 and the magnetic concentrating ring 12 are both connected by a buckle type connection structure, the center structure size of the shell body is adapted to the outer diameter of the cable 8 to be measured, and the entire detection device 1 can be stably clamped on the cable 8.

[0030] The connected test device is signal connected with the upper computer 6, and the upper computer 6 is signal connected with the signal relay device 7. Among them, the shell 11 has an opening near the wireless communication device 3, which facilitates the sending of wireless signals to the signal relay device 7;

[0031] The magnetic concentrating ring 12 installed in the protective shell 11 is in any direction, and there are four U-shaped clamping grooves on the magnetic concentrating ring 12, and a supporting assembly 13 is installed at each of the four clamping grooves; the supporting assembly 13 can be stably installed in the four U-shaped outer opening structures of the magnetic concentrating ring, and is not easy to fall off and slide;

[0032] Among them, there are four U-shaped clamping grooves, and there are TMR magnetic field sensors 14 distributed at the four U-shaped clamping grooves, the TMR magnetic field sensors 14 are signal connected with the magnetic concentrating ring 12, and after installation, the measurement axis of the TMR magnetic field sensors 14 is perpendicular to the plane of the magnetic concentrating ring and parallel to the to-be-measured section of the cable;

[0033] Among them, the four TMR magnetic field sensors 14 are electrically connected with the miniature power supply 2, so that the TMR magnetic field sensors 14 can normally carry out measurement work.

[0034] In addition, the test cable 8 is a set composed of a plurality of cables with different states, the outer diameters of the cables in the set are the same as the diameters of the core wires, and at least includes a reference cable without offset of the core wire relative to the center of the cable cross section, and a plurality of cables with a predetermined offset of the core wire relative to the center of the cable cross section, the predetermined offset is gradually increased from the minimum set offset to the maximum set offset at equal intervals, and the reference cable and each offset cable together constitute a standard cable group covering the range from no offset to the target maximum offset; in a specific embodiment, the offset distance of the core wire relative to the center of the cable cross section should include no offset, offset 1mm, offset 2mm, offset 3mm, offset 4mm, offset 5mm, a total of six kinds of cables.

[0035] In actual assembly, the TMR magnetic field sensor 14 can be mounted on the four support assemblies 13 so that the measurement axis of the TMR magnetic field sensor 14 is perpendicular to the plane of the magnetic concentrating ring, and then the support assemblies 13 are fixed on the centers of the four symmetrically and equidistantly distributed U-shaped outer opening structures on the outer side of the magnetic concentrating ring 12. Finally, the magnetic concentrating ring 12, the micro power supply 2 and the wireless communication device 3 are fixedly installed in the shell 11.

[0036] In testing, the power supply is turned on, the test cable 8 is connected to the power frequency alternating current generating device 9 to form a test current path structure, the power supply of the power frequency alternating current generating device 9 is turned on, and the assembled measuring device is firmly clamped on the test cable 8 under the simulated live working condition.

[0037] At this time, the power frequency alternating current flows through the test cable 8, a power frequency magnetic field is excited around, the magnetic concentrating ring 12 confines the part of the magnetic field to be excited outward, and in turn forms a strengthened uniform magnetic field at the four U-shaped openings. At this time, the instantaneous strength of the strengthened uniform magnetic field is consistent, and has a good linear correspondence with the instantaneous value of the current passing through the test cable. The four TMR magnetic field sensors 14 measure the uniform magnetic field in the U-shaped outer opening area, the measurement signals are transmitted to the wireless communication device 3 through the signal lines, the wireless communication device 3 converts the wired signals into wireless signals and transmits them to the signal relay device 7, and then to the upper computer 6.

[0038] As shown in Figure 1 , in combination with the above-mentioned testing device, the cable eccentricity detection method of the present application is given, which comprises the following steps:

[0039] S1, collecting multi-point magnetic field component data of the target cable under the energized state;

[0040] The magnetic field component signals are obtained by the TMR magnetic field sensors arranged outside the magnetic concentrating ring;

[0041] The magnetic concentrating ring is composed of two symmetrically spliced semicircular rings, and a plurality of U-shaped grooves for focusing the magnetic field of the core wire and weakening external electromagnetic interference are uniformly arranged at the edges of the semicircular rings.

[0042] The magnetic field sensors are arranged at the U-shaped outer openings of the magnetic concentrating ring.

[0043] The micro power supply 2 is a rechargeable battery, which independently supplies power to the measuring device to avoid magnetic field measurement fluctuations caused by external power supply interference.

[0044] Taking any magnetic field sensor as the coordinate origin and the connecting line of the two magnetic field sensors as the horizontal and vertical coordinate axes, a coordinate system is established, as shown in Figure 6 At this time, the magnetic induction intensity excited by the core wire at the magnetic field sensor 1 is:

[0045] (1)

[0046] At this time, the magnetic field sensor 1 can measure the magnetic induction intensity in its axial direction as:

[0047] (2)

[0048] And the magnetic field sensor 2 can measure the magnetic induction intensity as similar. Therefore, the ratio of the measurement results of the magnetic field sensors 1 and 2 is:

[0049] (3)

[0050] S6: The measurement data shows that the measurement value of the magnetic field sensor 1 is times of the measurement value of the magnetic field sensor 2, then every time there is a certain value, and a certain can be obtained according to formula (3). At this time, for the line connecting the magnetic field sensor 1 and the magnetic field sensor 2, the cable core wire must be on two straight lines which are and angle with the line. Therefore, the cable core wire at this time should be at the intersection of the two straight lines.

[0051] Let from 0 to 90, the cable core wire positions corresponding to all different angles under the explicit value can be obtained. The collection of these core wire positions is embodied as two axisymmetric curves.

[0052] The measurement data shows that the measurement value of the magnetic field sensor 3 is times of the measurement value of the magnetic field sensor 4, then every time there is a certain value, and a certain can be obtained according to formula (3). At this time, for the line connecting the magnetic field sensor 3 and the magnetic field sensor 4, the cable core wire must be on two straight lines which are and angle with the line. Therefore, the cable core wire at this time should be at the intersection of the two straight lines.

[0053] Let from 0 to 90, the cable core wire positions corresponding to all different angles under the explicit value can be obtained. The collection of these core wire positions is embodied as two axisymmetric curves.

[0054] All the data are integrated, and there is only one core wire position meeting all the conditions, and the intersection position of the curve is obtained. Since the core wire offset is always continuous and develops from small to large, the point with the smallest difference from the adjacent data is selected in the stored data as the position (coordinates) of the core wire at this time, and the offset distance and offset angle of the core wire at this time relative to the cable center are output by the upper computer.

[0055] The application further provides that a test cable with no core wire offset is selected and connected to a power frequency alternating current generating device, a real-time current sensor for a power distribution network transmission line is clamped on the test cable before the power supply is turned on, a miniature power supply and a wireless communication device in the sensor are turned on, a signal relay device is turned on, and a communication upper computer receives measurement data.

[0056] The power frequency current generating device is turned on, the output current is adjusted, and after the test cable with no core wire offset passes through the current, a magnetic field is excited in the surrounding space, under the action of the magnetic concentrating ring, there are uniform and strengthened magnetic fields at the outer openings of the four U-shaped rings that meet the test requirements, the four TMR magnetic field sensors corresponding to the openings measure the strengthened magnetic fields, transmit the measurement signals to the wireless communication device, wirelessly transmit the measurement data to the upper computer, display four groups of real-time current values measured, and turn off the power frequency current generating device after the measurement is completed.

[0057] The test cable with no core wire offset is replaced by a test cable with different core wire offset distances, the real-time current sensor for the power distribution network transmission line is clamped on the test cable, the miniature power supply and the wireless communication device in the sensor are turned on, and the communication upper computer receives measurement data.

[0058] The measurement data of the four TMR magnetic field sensors and the differences thereof are counted, when the cable has no offset, the measurement results of the four TMR magnetic field sensors are the same, when the wire passes through two TMR magnetic field sensors with a diameter of the cable cross section, and the difference between the results of the two TMR magnetic field sensors exceeds the error range of a single TMR magnetic field sensor, it is determined that the cable core wire has a significant offset, and the measured magnetic field strengths of the four TMR magnetic field sensors are taken as inputs.

[0059] The application further provides that the opposite cooperative purity is set as a composite trigger criterion for entering S2, and the opposite cooperative purity is obtained by:

[0060] The first pair of sensors and the second pair of sensors located at diametrically opposite positions are grouped respectively, the ratio of the absolute value of the reading difference to the sum of the absolute values of the readings of the two channels of the first pair of sensors is calculated to obtain the first pair of difference degrees, and the second pair of difference degrees is obtained in the same way; specifically, two sensors located at diametrically opposite positions are divided into a first pair, and the other two sensors located at diametrically opposite positions are divided into a second pair; in each sampling period, the ratio of the absolute value of the reading difference to the sum of the absolute values of the readings of the two channels of the first pair is calculated to obtain the first pair of difference degrees, and the second pair of difference degrees is obtained in the same way;

[0061] The first pair of difference degrees and the second pair of difference degrees are respectively subjected to exponential decay function transformation and then multiplied to obtain dimensionless pair coordination purity; specifically, the first pair of difference degrees and the second pair of difference degrees are respectively subjected to exponential decay function transformation, and the transformation mode is to input the negative value of the difference degree into the exponential function to obtain two dimensionless suppression quantities; the two dimensionless suppression quantities are multiplied to obtain the dimensionless pair coordination purity. The closer the pair coordination purity is to one, the more consistent the two pairs are; any increase in the difference degree rapidly reduces the coordination purity;

[0062] When any pair difference degree exceeds the preset proportion and the pair coordination purity is lower than the preset purity threshold, it is determined to enter S2, otherwise continuous acquisition and caching are maintained; specifically, the preset proportion threshold of the pair difference degree and the preset purity threshold of the coordination purity are set; when the following conditions occur, it is determined to enter step two: any pair difference degree exceeds the preset proportion, and the pair coordination purity is lower than the preset purity threshold; when the above conditions are not met, continuous acquisition and caching are maintained, and S2 is not entered; the next sampling period is reviewed and determined.

[0063] S2, data processing is performed on the measurement results to calculate the magnetic field ratio between each pair of opposing sensors;

[0064] The magnetic field ratio is used to represent the spatial offset directional characteristics of the core relative to the center of the magnetic concentrating ring.

[0065] The real-time measured magnetic field strength values of the two TMR magnetic field sensors located at diametrically opposite positions in the cross section are obtained respectively, and the ratio thereof is calculated;

[0066] The ratio calculation is sequentially performed on all pairs of opposing sensor groups to form a set of multiple pairs of magnetic field ratios.

[0067] S3, mapping calibration is performed on the magnetic field ratio to generate an equiratio spatial trajectory;

[0068] In this embodiment, the mapping calibration of the magnetic field ratio is performed by establishing a ratio-position mapping calibration library, and the specific construction steps of the ratio-position mapping calibration library are as follows:

[0069] Each group of magnetic field ratio is one-to-one corresponding to the corresponding known eccentric coordinates, forming a magnetic field ratio-space offset coordinate sample set; wherein each sample pair represents a group of magnetic field response patterns that can be used for training, for reflecting the nonlinear relationship between magnetic field ratio and spatial offset;

[0070] Based on the magnetic field ratio-space offset coordinate sample set, a nonlinear mapping model between the magnetic field ratio and the position is established;

[0071] Based on the nonlinear mapping model, the fitting residual of each sample is calculated, and the residual sample is fed back to the model parameter update layer after Gaussian weighting processing;

[0072] The corrected fitting model parameters, weight matrix and residual compensation coefficient are solidified and stored to form a ratio-position mapping calibration library;

[0073] The calibration library includes: magnetic field ratio input dimension, spatial offset output dimension and error correction factor, which is used to quickly calculate the core wire offset position according to the ratio input in subsequent real-time detection.

[0074] In this embodiment, the nonlinear mapping model between the magnetic field ratio and the position is established based on the magnetic field ratio-space offset coordinate sample set, specifically:

[0075] The magnetic field ratio vector in the sample set is divided into dimension coding to structure the response difference of different opposite sensors, forming a multi-dimensional input feature matrix;

[0076] The dimension coding includes principal component analysis and correlation reduction processing of the magnetic field ratio sequence to avoid fitting deviation caused by feature redundancy;

[0077] Based on the multi-dimensional input feature matrix and the corresponding spatial offset coordinate output, an initial fitting structure is established by combining a partition adaptive polynomial fitting and a lightweight neural network regression hybrid model;

[0078] Among them, the partition adaptive polynomial is used to approximate the approximate linear variation law of the magnetic field ratio in the low offset interval, and the neural network regression module is used to describe the nonlinear distortion characteristics in the high offset interval, and the two are fused through an error balancing factor.

[0079] In the model training process, dynamic weight decay and error correlation coefficient constraints are introduced to ensure that the weight distribution of the multi-sensor ratio input in different offset directions remains physically consistent;

[0080] This constraint mechanism constrains the gradient update direction of each sensor input channel, so that the model output is continuous and monotonic in space, avoiding the "multiple solutions" or "false extreme points" of traditional single regression models.

[0081] The polynomial order, the number of neural network layers and the fusion factor weight are determined by cross-validation to minimize the total sample fitting residual sum of squares, to obtain an initial fitting model of magnetic field ratio-space position;

[0082] The initial fitting model forms a two-dimensional spatial coordinate prediction value at the output end, providing a basic mapping framework for subsequent residual feedback correction.

[0083] The hybrid model combining the partition adaptive polynomial fitting and the lightweight neural network regression is specifically:

[0084]

[0085]

[0086]

[0087] wherein, is the predicted position output, is a model fusion coefficient, representing the weight ratio of linear model and nonlinear model (set according to historical data), is a polynomial regression function, is a neural network regression function, is a polynomial order, is a polynomial coefficient (obtained by least squares method), is a multi-dimensional input feature matrix, , is a network weight matrix, corresponding to the weights of input layer and hidden layer, hidden layer and output layer respectively, , is a preset bias term, is an activation function, is an output layer mapping function.

[0088] The dynamic weight decay and error correlation coefficient constraint are introduced, specifically:

[0089]

[0090]

[0091] wherein, is a total loss function, is an actual spatial position label, , are weight decay coefficients respectively, used to control the model complexity and the correlation penalty coefficient, used to control the upper limit of the correlation between sensor channels, is a gradient with respect to the weight of the i-th layer, is a channel ratio correlation coefficient, is the ratio response coefficient of the i-th pair of sensors, is the forward measured magnetic field intensity of the i-th pair of magnetic sensors, is the reverse measured magnetic field intensity of the i-th pair of magnetic sensors.

[0092] The polynomial order, the number of neural network layers, and the fusion factor weight are determined by cross-validation to minimize the total sample fitting residual sum of squares, specifically:

[0093]

[0094]

[0095] wherein, is the residual sum of squares objective function, is the number of samples, , , are the polynomial order, the number of neural network layers, and the fusion weight, respectively, is the actual spatial position label, is the model prediction output, is the final mapping function, , , is the optimal parameter combination.

[0096] In the embodiment, the generation of the equal ratio value space trajectory is specifically:

[0097] Based on the relationship between each feature ratio value vector and its corresponding spatial coordinates in the ratio-position mapping calibration library, the target ratio value vector is extracted, and its similarity distribution function in the mapping library is calculated;

[0098] The equal value set of the similarity distribution function is calculated in the entire calibration space, and each equal value curve is defined as the equal ratio value space trajectory corresponding to the measured ratio value;

[0099] Curvature continuity constraint and boundary smoothing interpolation are performed on the equal ratio value space trajectory to eliminate trajectory breaks caused by local noise, and the final equal ratio value space trajectory is obtained.

[0100] The similarity distribution function is specifically:

[0101]

[0102] wherein, is the similarity of the position (x, y), is the target ratio value vector, is the standard ratio value vector corresponding to the position (x, y) in the calibration library, Similarity decay coefficient, used to adjust the spatial distribution smoothness (set according to historical experience).

[0103] S4, determining the unique intersection position based on the equal ratio spatial trajectory and the comprehensive scoring mechanism, to obtain the spatial offset coordinates of the target core relative to the center of the magnetic ring;

[0104] In this embodiment, the determination of the unique intersection position based on the equal ratio spatial trajectory and the comprehensive scoring mechanism includes constructing a two-dimensional spatial trajectory intersection model, specifically:

[0105] Receiving two core equal ratio position trajectories and unifying them to the same section coordinate reference; Specifically, performing coordinate reference unification on the two trajectories to ensure consistent origin and orientation definition;

[0106] Discretize each trajectory into continuous segments and construct a narrow band buffer zone, calculate the intersection area of the two buffer zones, form an intersection candidate according to the geometric center of the intersection area and record the tangent direction, local curvature and same value residual at the candidate; Specifically, discretize each trajectory into several continuous segments in order of adjacent points, store the endpoint coordinates, tangent vector and curvature estimation window index of each segment; Set the trajectory narrow band according to the allowable error band of the trajectory under the corresponding working condition, and the narrow band width is set as a set multiple of the allowable error band, which is used to cover the discretization and measurement disturbance; Perform intersection detection on the narrow bands of the two trajectories to obtain several intersection areas; For each intersection area, calculate the geometric center of the area as the "intersection candidate", and record the tangent direction, local curvature and same value residual of the two trajectories at this point.

[0107] The unique intersection position is determined in the multiple intersection candidate set through the comprehensive scoring mechanism, and the spatial offset coordinates of the target core relative to the center of the magnetic ring are obtained, specifically:

[0108] Get the tangent angle of the two trajectories at the intersection candidate, and take the sine of the tangent angle as the angle component; Specifically, take the tangent direction in the neighborhood of the two trajectories at the intersection candidate respectively, and calculate the tangent angle; Apply the sine transform to the tangent angle, and normalize the result to a dimensionless quantity between zero and one as the angle component; The closer the angle is to a right angle, the closer the angle component is to one;

[0109] Calculate the local curvature of the two trajectories at the candidate, and take the cube root of the product of the two as the curvature component; Specifically, estimate the local curvature of the two trajectories in the neighborhood of the candidate respectively, take the cube root of the product of the two and normalize it to zero to one as the curvature component; The more obvious the curvature is, the more straight the two trajectories are, and the greater the curvature component is;

[0110] Multiplying the angular component by the curvature component yields the trajectory intersection sharpness, which is used to characterize geometric solvability. Specifically, the higher the value, the sharper the intersection and the better the geometric conditions.

[0111] At the candidate coordinates, the deviations between the mapped values ​​and real-time ratios of the first and second pairs of opposing ratios are calculated, and normalized according to the allowable error band to form dimensionless residuals. The two residuals are multiplied after exponential decay with power-law suppression to obtain the consistency potential, which is used to characterize the consistent support of the two pairs of constraints for the candidate point. Specifically, at the candidate coordinates, the deviations between the mapped values ​​and real-time ratios of the first and second pairs of opposing ratios are calculated, and normalized according to the allowable error band of the corresponding working condition to obtain two dimensionless residuals. The two dimensionless residuals are multiplied after exponential decay with power-law suppression to obtain the consistency potential. The smaller the residual, the closer the consistency potential is to one.

[0112] The product of trajectory intersection sharpness and convergence potential is used as the comprehensive score. From the candidate set satisfying the feasible radius and temporal proximity constraints, the candidate with the highest comprehensive score is selected as the unique intersection point. Specifically, a feasible radius constraint is applied to all candidates, and candidates located outside the feasible radius are eliminated. If a previous time step position exists, candidates with smaller Euclidean distances to the previous time step position are preferentially retained (temporal proximity constraint). A comprehensive score is calculated for each candidate, which is the product of trajectory intersection sharpness and convergence potential. From the candidate set satisfying the constraints, the candidate with the largest comprehensive score is selected as the unique intersection point.

[0113] The unique intersection point is determined as the instantaneous position of the core wire within the cross section, and the rectangular coordinates and polar coordinates of this position are obtained; specifically, the unique intersection point is taken as the instantaneous position of the core wire within the cross section, and two representations, rectangular coordinates and polar coordinates, are given.

[0114] The straight-line distance from the instantaneous position to the geometric center is used as the offset distance, and the angle between the geometric center and the instantaneous position is used as the offset azimuth angle, with the agreed coordinate reference axis as the zero azimuth. Specifically, the straight-line distance from the instantaneous position to the origin is calculated using the geometric center as the reference, and the angle between the geometric center and the instantaneous position is calculated using the reference azimuth axis as the zero azimuth, and the positive direction and the angle counting direction are specified.

[0115] The offset distance is converted into dimensionless eccentricity based on the reference radius of the cable cross-section; specifically, the offset distance is converted into dimensionless eccentricity by ratio of the reference radius to the reference radius of the cable cross-section.

[0116] S5, calculate the eccentricity value of the target cable based on the spatial offset coordinates.

[0117] In the embodiment, the eccentricity value of the target cable is calculated based on the spatial offset coordinates, specifically:

[0118] The eccentricity value of the target cable is calculated based on the spatial offset coordinates;

[0119] The straight-line distance from the core position to the origin is calculated with the cross-sectional geometric center as the origin;

[0120] The straight-line distance is converted to a dimensionless eccentricity value according to the cable cross-sectional reference radius.

[0121] The above formulas are all dimensionless numerical calculations, and the formula is obtained by software simulation of a large amount of data to obtain the most real situation, and the preset parameters in the formula are set by the person skilled in the art according to the actual situation.

[0122] The above embodiments can be realized wholly or partially by software, hardware, firmware or any combination thereof. When realized by software, the above embodiments can be realized in the form of a computer program product, wholly or partially.

[0123] Those skilled in the art can realize that the modules and algorithm steps of the examples described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0124] In addition, the functional modules in each embodiment of the present application can be integrated in one processing module, or each module can exist physically alone, or two or more modules can be integrated in one module.

[0125] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any skilled person in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0126] Finally: the above is only the preferred embodiment of the present application, and is not used to limit the present application, any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application, should be included in the protection scope of the present application.

Claims

1. A method of cable eccentricity detection, characterized by, The method comprises the following steps: Collecting multi-point magnetic field component data of the target cable in a powered state; Processing the measurement data to calculate the magnetic field ratio between each pair of opposing sensors; Mapping and calibrating the magnetic field ratio to generate an equal-ratio spatial trajectory; Determining the unique intersection position based on the equal-ratio spatial trajectory and a comprehensive scoring mechanism to obtain the spatial offset coordinates of the target core wire relative to the center of the magnetic ring; the determination of the unique intersection position comprises constructing a two-dimensional spatial trajectory intersection model, and the specific steps are as follows: receiving two trajectories and unifying them to the same cross-sectional coordinate reference; Discretizing each trajectory into a plurality of continuous line segments, and storing the endpoint coordinates, tangent vector, and curvature estimation window index of each line segment; setting an allowable error band based on the working condition, and constructing a trajectory narrow band buffer; detecting the intersection of the narrow band buffers of the two trajectories to obtain the intersection region and calculate the geometric center as the intersection candidate, and recording the corresponding tangent direction, local curvature, and same value residual at each intersection candidate; The determination of the spatial offset coordinates of the target core wire relative to the center of the magnetic ring comprises: calculating the sine of the tangent angle of the two trajectories at each intersection candidate as the angle component, and estimating the cubic root of the product of the local curvatures of the two trajectories as the curvature component; multiplying the angle component and the curvature component to obtain the trajectory intersection sharpness; calculating the normalized residual of the mapping value and the real-time ratio of the two pairs of opposing ratios, and multiplying the power exponential decay to obtain the convergence potential; taking the product of the trajectory intersection sharpness and the convergence potential as the comprehensive score, and selecting the one with the highest comprehensive score as the unique intersection under the constraints of the feasible radius and time sequence proximity; Taking the coordinates of the unique intersection as the instantaneous spatial offset coordinates of the core wire in the cross section; Based on the spatial offset coordinates, the eccentricity value of the target cable is calculated.

2. The cable eccentricity detection method according to claim 1, characterized in that, The collection of multi-point magnetic field component data of the target cable in a powered state comprises: Selecting a core wire non-offset test cable for testing; Under the powered state of the cable, collecting the magnetic field component signals on the side of the magnetic ring; Replace the test cable with different core wire offset distances, repeat the above steps, and obtain the magnetic field data under different core wire offset states to form a magnetic field component sample set.

3. The cable eccentricity detection method of claim 2, wherein The magnetic field component signals are obtained by TMR magnetic field sensors arranged outside the magnetic ring; The magnetic ring is composed of two symmetrically spliced half-rings, and a plurality of U-shaped grooves for focusing the magnetic field of the core wire and weakening external electromagnetic interference are uniformly arranged at the edges of the half-rings; The magnetic field sensors are arranged at the U-shaped outer opening of the magnetic ring.

4. The cable eccentricity detection method according to claim 3, characterized in that, The collection of multi-point magnetic field component data of the target cable in a powered state further comprises setting the opposing cooperative purity as a composite trigger criterion, which comprises: Dividing the diametrically opposite position sensors into two groups; Calculating the ratio of the absolute values of the two-channel readings of each group of sensors and the sum of the absolute values of the two-channel readings to obtain the first and second opposing difference degrees; Performing exponential decay function transformation on the first and second opposing difference degrees to obtain two dimensionless suppression quantities, and multiplying the two dimensionless suppression quantities to obtain the opposing cooperative purity; When any one of the first or second opposite difference exceeds the preset proportion threshold, and the opposite synergy purity is lower than the preset purity threshold, a subsequent processing step is triggered; otherwise, continuous signal acquisition and caching are maintained.

5. The cable eccentricity detection method of claim 4, wherein, The magnetic field ratio is mapped and calibrated by establishing a ratio-position mapping calibration library, and the specific construction steps of the ratio-position mapping calibration library are as follows: Each group of magnetic field ratios is one-to-one corresponding to the corresponding known eccentric coordinates to form a coordinate sample set; Based on the sample set, a nonlinear mapping model between the magnetic field ratio and the position is established; The fitting residual of the nonlinear mapping model at each sample point is calculated, and the residual sample is feedback to the model parameter updating layer after Gaussian weighting processing for model correction; The corrected fitting model parameters, weight matrix and residual compensation coefficient are stored as a ratio-position mapping calibration library.

6. The cable eccentricity detection method of claim 5, wherein, Based on the sample set, a nonlinear mapping model between the magnetic field ratio and the position is established, specifically as follows: The magnetic field ratio vector in the sample set is divided into multi-dimensional coding to form a multi-dimensional input feature matrix; Based on the multi-dimensional input feature matrix and the corresponding spatial offset coordinates output, a hybrid model combining partition adaptive polynomial fitting and lightweight neural network regression is established; In the model training process, dynamic weight decay and error correlation coefficient constraints are introduced, the model structure parameters are determined through cross-validation, and the magnetic field ratio-space position initial fitting model is obtained by minimizing the total sample fitting residual sum of squares.

7. The cable eccentricity detection method of claim 6, wherein, The step of generating the equal-ratio space trajectory includes: Based on the relationship between each feature ratio vector and its corresponding spatial coordinate in the ratio-position mapping calibration library, the target ratio vector is extracted, and the similarity distribution function thereof in the mapping library is calculated; The equal set of the similarity distribution function is calculated in the whole calibration space, and each equal value curve is defined as the equal-ratio space trajectory corresponding to the measured ratio; The curvature continuity constraint and boundary smoothing interpolation are performed on the equal-ratio space trajectory to obtain the final equal-ratio space trajectory.

8. The cable eccentricity detection method of claim 7, wherein, The specific steps of calculating the eccentricity value of the target cable based on the spatial offset coordinates are as follows: The eccentricity value of the target cable is calculated based on the spatial offset coordinates; Taking the cross-sectional geometric center as the origin, the straight line distance from the core position to the origin is calculated; According to the cable cross-section reference radius, the straight line distance is converted into a dimensionless eccentricity value.

Images