A multi-sensor information fusion method for high-voltage cable insulation status monitoring

Abstract

The invention discloses a multi-sensor information fusion method for monitoring the insulation state of a high-voltage cable, and relates to the field of information technology. The invention obtains magnetic flux distribution data at different spatial positions in a grounding box according to a three-axis orthogonal magnetic flux sensor, analyzes the magnetic flux distribution characteristics, and obtains a spatial sensitivity distribution spectrum; obtains a magnetic field signal generated by a cable fault discharge source, and uses an inversion method based on a magnetic field gradient tensor to decompose and calculate the magnetic field gradient tensor, locates the defect position of the cable insulation, and obtains the three-dimensional spatial coordinate information of the defect position; according to the results of the spatial distribution evaluation of the cable insulation state, the arrangement density of the fluxgate sensor is increased in the area with high incidence of insulation faults, the layout of the tetrahedron configuration sensor array is optimized, and a comprehensive evaluation index system is constructed. The invention realizes the accurate spatial distribution evaluation of the cable insulation state, provides data support for state monitoring and life prediction, and improves the reliability of cable operation.

Description

Technical Field

[0001] The invention relates to the field of information technology, and in particular to a multi-sensor information fusion method for monitoring the insulation status of a high-voltage cable. Background Art

[0002] When deploying a tetrahedron configuration sensor array in a cable grounding box, the spatial arrangement of the three-axis orthogonal magnetic flux sensor will directly affect the distribution characteristics of the magnetic flux. Under different arrangements, the spatial sensitivity distribution map of the tetrahedron configuration sensor array will also produce significant differences. This results in the positioning accuracy and reliability being restricted by the arrangement when using the magnetic field gradient tensor inversion algorithm to locate insulation defects. In addition, the discharge characteristics of the cable fault discharge source are complex and changeable, and the discharge signal usually shows non-stationarity in the time domain and frequency domain, which further increases the design difficulty of the three-dimensional positioning algorithm. At the same time, due to the small internal space of the grounding box and the complex electromagnetic environment, the arrangement of the tetrahedron configuration sensor array is also restricted by site conditions, making it difficult to achieve an ideal arrangement plan.

[0003] Solving this problem is the key to establishing a spatial distribution evaluation system for cable insulation status, which is of great significance for achieving early warning and accurate diagnosis of cable faults.

[0004] The present invention solves the above-mentioned technical shortcomings of the prior art by:

[0005] The magnetic flux distribution characteristics of the tetrahedron configuration sensor array in the grounding box under different spatial arrangements are explored. Combined with the spatial sensitivity distribution map of the tetrahedron configuration sensor array, a three-dimensional positioning algorithm for the cable fault discharge source is constructed. The insulation defects are located through magnetic field gradient tensor inversion, and then a spatial distribution evaluation system for the cable insulation status is established. Summary of the invention

[0006] In order to solve the above-mentioned technical problems, the present invention provides a multi-sensor information fusion method for monitoring the insulation status of a high-voltage cable.

[0007] The technical solution of the present invention is implemented as follows: a multi-sensor information fusion method for monitoring the insulation status of a high-voltage cable comprises:

[0008] S1. Obtain magnetic flux distribution data at different spatial positions in the grounding box through a three-axis orthogonal magnetic flux sensor, analyze the magnetic flux distribution characteristics, and obtain a spatial sensitivity distribution map of the tetrahedron configuration sensor array;

[0009] S2. Obtain the magnetic field signal generated by the cable fault discharge source, locate the defect position of the cable insulation, and obtain the three-dimensional spatial coordinate information of the defect position;

[0010] S3, obtaining the spatial layout information and material characteristic parameters of the cable, and combining the three-dimensional spatial coordinate information of the defect position to construct a spatial distribution evaluation model of the cable insulation state, and obtaining the spatial distribution characteristics of the cable insulation performance through finite element simulation analysis;

[0011] S4. Classify and predict the spatial distribution characteristics of the cable insulation status, identify potential high-incidence areas of insulation faults, perform fine-grained analysis on the magnetic field signals in high-incidence areas of insulation faults, and obtain the spatial distribution evaluation results of the cable insulation status;

[0012] S5. According to the insulation fault high-incidence areas shown by the cable insulation status spatial distribution evaluation results, increase the arrangement density of fluxgate sensors in the insulation fault high-incidence areas;

[0013] S6. Combine the spatial distribution evaluation results of the cable insulation status with the cable operating parameters and environmental factors to construct a comprehensive evaluation index system. Through weighted fusion of multiple indicators, a quantitative score of the cable insulation health status is obtained to provide data support for cable status monitoring and life prediction.

[0014] Furthermore, the step S1 acquires magnetic flux distribution data at different spatial positions in the grounding box through a three-axis orthogonal magnetic flux sensor, analyzes the magnetic flux distribution characteristics, and obtains a spatial sensitivity distribution map of the tetrahedron configuration sensor array.

[0015] Furthermore, the step S2 obtains the magnetic field signal generated by the cable fault discharge source, and uses an inversion method based on the magnetic field gradient tensor to decompose and calculate the magnetic field gradient tensor to locate the defect position of the cable insulation and obtain the three-dimensional spatial coordinate information of the defect position.

[0016] Furthermore, the step S3 obtains the spatial layout information and material characteristic parameters of the cable, and combines the three-dimensional spatial coordinate information of the defect position to construct a spatial distribution evaluation model of the cable insulation status, and obtains the spatial distribution characteristics of the cable insulation performance through finite element simulation analysis.

[0017] Furthermore, the step S4 classifies and predicts the insulation status and spatial distribution characteristics of the cable, identifies potential high-incidence areas of insulation faults, performs fine-grained analysis on the magnetic field signals in the high-incidence areas of insulation faults, and obtains the spatial distribution evaluation results of the cable insulation status.

[0018] Furthermore, the step S5 increases the arrangement density of fluxgate sensors in the insulation fault high-incidence area according to the insulation fault high-incidence area shown by the cable insulation state spatial distribution evaluation result.

[0019] Furthermore, step S6 combines the spatial distribution evaluation results of the cable insulation status with the operating parameters and environmental factors of the cable to construct a comprehensive evaluation index system, and obtains a quantitative score of the cable insulation health status through weighted fusion of multiple indicators, providing data support for cable status monitoring and life prediction. Beneficial Effects

[0020] The present invention obtains the magnetic flux distribution data in the grounding box through a tetrahedron configuration sensor array, locates the cable insulation defect position by using a magnetic field gradient tensor inversion algorithm, constructs an insulation state spatial distribution evaluation model in combination with the cable spatial layout and material parameters, obtains the insulation performance spatial distribution characteristics through finite element simulation analysis, classifies and predicts the insulation state spatial distribution characteristics, identifies potential high-incidence areas for faults and performs fine-grained analysis, and increases the density of fluxgate sensors in high-incidence areas and reduces the density in low-incidence areas according to the evaluation results, balances the area of ​​high-risk areas with monitoring requirements through mapping analysis, optimizes the layout of the tetrahedron configuration sensor array, and finally combines the evaluation results with operating parameters and environmental factors to construct a comprehensive evaluation index system to obtain a quantitative score for the insulation health status. The present invention realizes accurate spatial distribution evaluation of the cable insulation status, provides data support for status monitoring and life prediction, and improves the reliability of cable operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 The process of a multi-sensor information fusion method for monitoring the insulation status of a high-voltage cable in an embodiment of the present invention is as follows Figure 1 ;

[0022] Figure 2 The process of a multi-sensor information fusion method for monitoring the insulation status of a high-voltage cable in an embodiment of the present invention is as follows Figure 2 ;

[0023] Figure 3 The process of a multi-sensor information fusion method for monitoring the insulation status of a high-voltage cable in an embodiment of the present invention is as follows Figure 3 . DETAILED DESCRIPTION

[0024] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0025] The preferred implementation methods of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these implementation methods are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0026] It should be noted that when an element is considered to be "connected" to another element, it can be directly connected to the other element, or connected to the other element through an intermediate element. In addition, the "connection" in the following embodiments should be understood as "electrical connection", "communication connection", etc. if there is transmission of electrical signals or data between the connected objects.

[0027] When used herein, the singular forms "a", "an", and "said / the" may also include plural forms, unless the context clearly indicates otherwise. It should also be understood that the terms "include / comprise" or "have" and the like specify the presence of stated features, wholes, steps, operations, components, parts, or combinations thereof, but do not exclude the possibility of the presence or addition of one or more other features, wholes, steps, operations, components, parts, or combinations thereof. At the same time, the term "and / or" used in this specification includes any and all combinations of the relevant listed items.

[0028] See also Figure 1-Figure 3 As shown, a multi-sensor information fusion method for high-voltage cable insulation status monitoring includes:

[0029] S1. Obtain magnetic flux distribution data at different spatial positions in the grounding box through a three-axis orthogonal magnetic flux sensor, analyze the magnetic flux distribution characteristics, and obtain a spatial sensitivity distribution map of the tetrahedron configuration sensor array;

[0030] S2. Obtain the magnetic field signal generated by the cable fault discharge source, locate the defect position of the cable insulation, and obtain the three-dimensional spatial coordinate information of the defect position;

[0031] S3, obtaining the spatial layout information and material characteristic parameters of the cable, and combining the three-dimensional spatial coordinate information of the defect position to construct a spatial distribution evaluation model of the cable insulation state, and obtaining the spatial distribution characteristics of the cable insulation performance through finite element simulation analysis;

[0032] S4. Classify and predict the spatial distribution characteristics of the cable insulation status, identify potential high-incidence areas of insulation faults, perform fine-grained analysis on the magnetic field signals in high-incidence areas of insulation faults, and obtain the spatial distribution evaluation results of the cable insulation status;

[0033] S5. According to the insulation fault high-incidence areas shown by the cable insulation status spatial distribution evaluation results, increase the arrangement density of fluxgate sensors in the insulation fault high-incidence areas;

[0034] S6. Combine the spatial distribution evaluation results of the cable insulation status with the cable operating parameters and environmental factors to construct a comprehensive evaluation index system. Through weighted fusion of multiple indicators, a quantitative score of the cable insulation health status is obtained to provide data support for cable status monitoring and life prediction.

[0035] Furthermore, the step S1 acquires magnetic flux distribution data at different spatial positions in the grounding box through a three-axis orthogonal magnetic flux sensor, analyzes the magnetic flux distribution characteristics, and obtains a spatial sensitivity distribution map of the tetrahedron configuration sensor array.

[0036] Furthermore, a tetrahedron configuration sensor array group is arranged for a number of monitoring areas divided inside the grounding box, and the tetrahedron configuration sensor array group is configured with a three-axis orthogonal magnetic flux sensor on each surface to obtain magnetic flux spatial distribution calibration data;

[0037] According to the magnetic flux spatial distribution calibration data, a radial basis function interpolation fitting curve is used to calculate the magnetic flux density value of the area where the three-axis orthogonal magnetic flux sensor is not deployed, the radial basis function interpolation radius is a preset percentage of the side length of the monitoring area, and the complete magnetic flux spatial distribution data is obtained from the magnetic flux density value;

[0038] The complete magnetic flux spatial distribution data is subjected to spectrum decomposition by using fast Fourier transform, and the magnetic flux vector divergence is calculated from the spectrum decomposition to obtain the magnetic field source distribution characteristics;

[0039] According to the magnetic field source distribution characteristics, wavelet transform is used to perform multi-scale decomposition on the magnetic flux data, and the temporal and spatial variation characteristics of the magnetic field are obtained based on the multi-scale decomposition data;

[0040] The spline interpolation method is used to perform spatial interpolation calculation on the temporal and spatial variation characteristics of the magnetic field, wherein a Gaussian smoothing kernel function is used to process boundary data, perform spatial interpolation calculation, and generate a magnetic flux isosurface map.

[0041] Specifically, a three-dimensional cubic grid structure with a side length of 0.5 meters is constructed inside the grounding box, and eight monitoring areas are divided from it. A tetrahedron configuration sensor array group is arranged for each monitoring area. The tetrahedron configuration sensor array group is configured with a three-axis orthogonal magnetic flux sensor on each face. The sampling frequency of the three-axis orthogonal magnetic flux sensor is 1000 Hz, and the magnetic flux spatial distribution calibration data is obtained from the tetrahedron configuration sensor array group;

[0042] A radial basis function interpolation fitting curve is established according to the calibration data, wherein the radial basis function interpolation radius is 0.5 times the side length of the monitoring area, and the radial basis function weight coefficient is inversely proportional to the square of the distance. A supplementary calculation is performed on the magnetic flux density value of the area where the three-axis orthogonal magnetic flux sensor is not arranged in the three-dimensional grid, and complete magnetic flux spatial distribution data is obtained from the supplementary calculation;

[0043] Performing spectrum decomposition on the complete magnetic flux spatial distribution data by using fast Fourier transform, calculating the magnetic flux vector divergence in the frequency domain, obtaining the magnetic field source distribution characteristics from the divergence data, performing multi-scale decomposition on the magnetic flux data by using wavelet transform, and obtaining the temporal and spatial variation characteristics of the magnetic field from the multi-scale decomposed data;

[0044] Using cubic spline interpolation to perform spatial interpolation calculation on the temporal and spatial variation characteristics of the magnetic field, the cubic spline interpolation uses a Gaussian smoothing kernel function to process boundary data, generating a magnetic flux isosurface map according to the spatial interpolation result, and obtaining spatial sensitivity distribution data of the tetrahedral configuration sensor array at each monitoring point from the isosurface map, wherein the spatial sensitivity distribution data includes a sensitivity coefficient and a monitoring area coverage radius;

[0045] A three-dimensional cube grid structure with a side length of 0.5 meters is provided with 64 vertex nodes in total, and a tetrahedron configuration sensor array is arranged at each vertex node. The tetrahedron configuration sensor array is composed of 4 three-axis orthogonal magnetic flux sensors forming a triangular pyramid structure. The length of the pyramid base is 0.2 meters and the height is 0.15 meters. The three-axis orthogonal magnetic flux sensors are evenly arranged at an angle of 120 degrees on each face. The sampling frequency of the three-axis orthogonal magnetic flux sensor is 1000 Hz. Each three-axis orthogonal magnetic flux sensor obtains the magnetic flux values ​​in the three directions of x, y, and z at each sampling moment;

[0046] In the radial basis function interpolation calculation, the Gaussian kernel function is selected as the basis function, the interpolation radius is 0.25 meters, the weight coefficient adopts the quadratic inverse function, the weight coefficient is 0.8 at a distance of 0.1 meters, and the weight coefficient is 0.4 at a distance of 0.2 meters. The interpolation grid density is 4 times the original grid density. After interpolation, the magnetic flux values ​​at 256 grid nodes are obtained;

[0047] The complete magnetic flux data is subjected to 1024-point fast Fourier transform to obtain the amplitude and phase information in the spectral components. The magnetic flux data is decomposed into four layers using the db4 wavelet basis to obtain the time domain features at different scales, from which the magnetic flux change rate and spatial gradient information are extracted.

[0048] The cubic spline function is used to generate the magnetic flux isosurface. The function node spacing is 0.05 meters, the standard deviation of the smoothing kernel function is 0.1 meters, and 16 isosurface layers are generated. The interval between each layer is 0.2μT. The sensitivity coefficient is calculated based on the curvature and gradient of the isosurface. The coverage radius of the monitoring area varies in the range of 0.1 meters to 0.4 meters.

[0049] Furthermore, step S2 obtains the magnetic field signal generated by the cable fault discharge source, and uses an inversion method based on the magnetic field gradient tensor to decompose and calculate the magnetic field gradient tensor to locate the defect position of the cable insulation and obtain the three-dimensional spatial coordinate information of the defect position.

[0050] Furthermore, the magnetic field signal timing data is obtained from the fluxgate sensor arranged at the vertex position of the cube;

[0051] The time series data is decomposed and denoised using a wavelet basis function, and the magnetic field intensity and direction values ​​are extracted from the denoised signal;

[0052] A magnetic field gradient tensor matrix is ​​constructed according to the magnetic field strength and direction values, a singular value decomposition is performed on the magnetic field gradient tensor matrix, and the main axis direction of the magnetic field source is obtained by retaining eigenvalues ​​and eigenvectors with energy accounting for more than a preset percentage;

[0053] Using the steepest descent method to perform an inversion operation on the main axis direction;

[0054] According to the quantitative relationship, an equipotential surface distribution of grid density is constructed in three-dimensional space, and the spatial coordinate value of the discharge source position is obtained through Monte Carlo random sampling verification. If the standard deviation of the spatial coordinate value is less than a threshold, the position is determined to be a cable insulation defect point.

[0055] Specifically, eight fluxgate sensors are arranged at the vertices of a cube with a side length of 0.5 meters. The sampling frequency of the fluxgate sensors is 2000 Hz and the measurement range is 0-100 Tesla. The magnetic field signal time series data is obtained from the fluxgate sensors. The db4 wavelet basis function is used to perform 5-layer decomposition and denoising on the time series data. The magnetic field intensity and direction values ​​of the discharge amplitude in the range of 1-10 millivolts are extracted through the denoised signal.

[0056] A magnetic field gradient tensor matrix is ​​constructed according to the magnetic field strength and direction values, wherein the magnetic field gradient tensor matrix includes magnetic field change rates in three spatial directions, singular value decomposition is performed on the magnetic field gradient tensor matrix, eigenvalues ​​and eigenvectors with an energy share of more than 95% are retained, and the principal axis direction of the magnetic field source in space is obtained through the eigenvalues ​​and eigenvectors;

[0057] An inversion operation is performed on the main axis direction by using the steepest descent method, wherein the inversion operation objective function includes a magnetic field intensity attenuation term and a direction deviation term, a multipole expansion coefficient of a magnetic field source is obtained from the inversion operation, and a quantitative relationship between the attenuation of the magnetic field source intensity and the quadratic attenuation of the distance is established by using the multipole expansion coefficient;

[0058] According to the quantitative relationship, an equipotential surface distribution with a grid density of 0.01 meters is constructed in three-dimensional space, and 1000 Monte Carlo random sampling verifications are performed on the equipotential surface distribution. The spatial coordinate value of the discharge source position is obtained through the random sampling verification. If the standard deviation of the spatial coordinate value is less than 0.01 meters, the position is determined to be a cable insulation defect point;

[0059] The fluxgate sensors of model FGM3D / 100 are arranged at the eight vertices of a cube with a side length of 0.5 meters. The sensitivity of the fluxgate sensor is 0.1 nanotesla, the linearity error is less than 0.0001%, the sampling frequency is set to 2000 Hz, and the measurement range is 0-100 Tesla. The original magnetic field signal is decomposed by db4 wavelet 5 layers to obtain the fundamental component with a frequency band of 0-500 Hz and the high-frequency component of 500-1000 Hz. The discharge signal in the range of 1-10 millivolts is extracted after wavelet threshold denoising. The magnetic field intensity value shows obvious dipole distribution characteristics in the xy plane.

[0060] The constructed magnetic field gradient tensor matrix contains two main eigenvalues. The first eigenvalue 0.85 represents the radial gradient component, and the second eigenvalue 0.12 represents the tangential gradient component. The energy sum of the two eigenvalues ​​accounts for 97% of the total energy.

[0061] In the inversion calculation, the weight of the magnetic field intensity term is 0.7, the weight of the direction deviation term is 0.3, and the steepest descent method with a step size of 0.01 is used for iterative optimization. The obtained multipole expansion coefficients have a dipole term of 85%, a quadrupole term of 12%, and an octupole term of 3%. The inversion error is less than 1%.

[0062] An equipotential surface grid was constructed in a spatial range of 0.5 m × 0.5 m × 0.5 m with a grid density of 0.01 m. The spatial coordinates of the discharge source were obtained after 1000 Monte Carlo samplings. The sampling results showed that the standard deviation in the x-direction was 0.008 m, the standard deviation in the y-direction was 0.007 m, and the standard deviation in the z-direction was 0.009 m. The standard deviations in the three directions were all less than the preset threshold of 0.01 m, and the positioning results converged to the cable insulation defect point.

[0063] Furthermore, the step S3 obtains the spatial layout information and material characteristic parameters of the cable, and combines the three-dimensional spatial coordinate information of the defect position to construct a spatial distribution evaluation model of the cable insulation status, and obtains the spatial distribution characteristics of the cable insulation performance through finite element simulation analysis.

[0064] Furthermore, a laser scanner is used to obtain three-dimensional point cloud data of the outer surface of the cable, and the spatial trajectory point coordinates, bending radius and torsion angle of the cable are extracted from the three-dimensional point cloud data;

[0065] Generate a spatial geometric surface equation of the cable according to the bending radius and the torsion angle, and construct a physical structure of the cable through the spatial geometric surface equation;

[0066] Dividing the solid structure by a hexahedral grid, generating basic grid units and encrypted grid units at defect locations from the hexahedral grid;

[0067] A thermodynamic-electric three-field coupled constitutive equation is established according to the basic grid unit and the encrypted grid unit, and the dielectric constant, thermal conductivity and elastic modulus values ​​of the insulating layer are obtained from the material database;

[0068] Performing numerical iterative calculation on the thermodynamic and electric three-field coupled constitutive equation, and if the relative error is less than a preset threshold, obtaining the spatial distribution characteristic value of the cable insulation performance from the numerical iterative calculation;

[0069] It is determined whether the dielectric breakdown field strength at the insulation defect exceeds a preset field strength threshold value according to the spatial distribution characteristic value.

[0070] Specifically, a laser scanner with a resolution of 0.1 mm is used to perform three-dimensional modeling on the outer surface of the cable within a scanning range of 2 meters, and the coordinates of the spatial trajectory points of the cable are extracted based on the three-dimensional modeling data, and the bending radius and torsion angle values ​​of the cable are obtained from the spatial trajectory points. The spatial geometric surface equation of the cable is generated according to the bending radius and torsion angle, and the three-dimensional solid structure of the cable is constructed through the surface equation;

[0071] According to the three-dimensional solid structure, the cable insulation layer is divided into hexahedral grids, the basic grid unit size is set to 5 mm, and an encrypted grid of 0.5 mm size is used at the insulation defect position. The dielectric constant of the insulation layer in the temperature range of 20-80 degrees Celsius is 2.3-2.8, the thermal conductivity is 0.2-0.3 watts per meter Kelvin, and the elastic modulus is 2.8-3.2 GPa. The thermoelectric three-field coupling constitutive equation is established through the material parameters;

[0072] An electrostatic field solver is used to apply a 1000 volt potential difference boundary condition to the cable insulation layer, the electric field distribution of the insulation layer is calculated under the potential difference, the temperature distribution of the insulation layer is calculated for a current carrying capacity of 500 amperes, the spatial distribution function of the dielectric constant and the elastic modulus is obtained according to the temperature distribution, and the thermal stress and electric field intensity distribution are calculated from the spatial distribution function;

[0073] A multi-field coupling simulation grid is constructed according to the distribution of thermal stress and electric field intensity, and a numerical iterative calculation is performed on the coupling simulation grid. If the relative error is less than 0.1%, a distribution characteristic value of the cable insulation performance in three-dimensional space is obtained from the iterative calculation, and the characteristic value is used to determine whether the dielectric breakdown field strength at the insulation defect exceeds 30 kilovolts per millimeter;

[0074] The LS300 laser scanner was used to perform 3D modeling of the cable. The laser scanning resolution was set to 0.1 mm, and the scanning range covered a 2-meter-long cable segment. The point cloud data density obtained by scanning was 1,000 points per square centimeter. The cable surface grid was obtained through the point cloud registration algorithm, with 500,000 grid units. The extracted spatial trajectory points showed that the minimum bending radius of the cable was 0.8 meters, and the maximum torsion angle was 15 degrees. The root mean square error of the geometric equation obtained based on surface fitting was less than 0.2 mm.

[0075] The cable insulation layer is meshed, with an initial mesh unit size of 5 mm, and a total of 80,000 hexahedral units are divided. A 0.5 mm mesh is used within a range of 0.1 m around the insulation defect, and the number of cells in the encrypted area is 20,000. The dielectric constant of the insulation layer at 25 degrees Celsius is read from the material database as 2.4, the thermal conductivity is 0.25 watts per meter Kelvin, and the elastic modulus is 3.0 GPa;

[0076] When a current of 500 amperes passes through the cable, the temperature field calculation shows that the maximum temperature of the insulation layer reaches 65 degrees Celsius, the dielectric constant at the corresponding position decreases to 2.2, the thermal conductivity increases to 0.28 watts per meter Kelvin, and the elastic modulus decreases to 2.8 GPa. The thermal stress calculation results show that the stress concentration factor at the defect is 2.3, and the electric field strength reaches 28 kilovolts per millimeter;

[0077] A convergence threshold of 0.001 was used in the multi-field coupling calculation. After 86 iterations, the relative error dropped to 0.08%. The calculation results showed that the breakdown field strength at the defect was 32 kilovolts per millimeter, exceeding the preset threshold of 30 kilovolts per millimeter.

[0078] Furthermore, the step S4 classifies and predicts the spatial distribution characteristics of the cable insulation status, identifies potential high-incidence areas of insulation faults, performs fine-grained analysis on the magnetic field signals in the high-incidence areas of insulation faults, and obtains the spatial distribution evaluation results of the cable insulation status.

[0079] Furthermore, a gradient boosting tree classifier is used to classify the spatial distribution characteristics of electric field strength, temperature distribution, mechanical stress, dielectric constant, insulation thickness, current carrying capacity, operating time, ambient humidity, material aging, and historical fault frequency, and a risk level threshold is set according to the classification result, and the spatial range data of the fault area is obtained from the risk level threshold;

[0080] Divide the spatial range data of the fault area into monitoring sub-intervals in the cable length direction, and arrange fluxgate sensors in the monitoring sub-intervals to obtain magnetic field signal time series data;

[0081] Decomposing the signal using a wavelet basis function according to the magnetic field signal time series data, and correcting the amplitude of the decomposed signal using a temperature compensation coefficient;

[0082] Extract the time-frequency domain feature parameters such as mean, variance, peak value, and spectrum entropy from the corrected data, construct a feature vector matrix for the feature parameters, calculate the matching degree between the feature vector and the fault template using the cosine similarity criterion, and obtain the fault probability distribution function from the matching degree;

[0083] The insulation performance degradation index is calculated based on the fault probability distribution function. The sampling frequency of the three-axis fluxgate sensor is increased in the region where the degradation index exceeds a preset threshold. The high-frequency sampling data is randomly verified using the Monte Carlo method, and the cable insulation status evaluation value is obtained from the verified results.

[0084] Specifically, a classifier based on a gradient boosting tree is used to classify data on ten spatial distribution characteristics, including electric field strength, temperature distribution, mechanical stress, dielectric constant, insulation thickness, current carrying capacity, operating time, ambient humidity, material aging, and historical fault frequency. Four risk level thresholds of 0.3, 0.5, 0.7, and 0.9 are set for the classification results. The spatial range data of high-incidence areas of faults are obtained from the classification results, and the spatial position of the fault area is calibrated by the spatial range data.

[0085] According to the fault area calibration data, 50 monitoring sub-intervals are divided every 2 meters in the cable length direction, and a three-axis fluxgate sensor with a sampling frequency of 1000 Hz is arranged for each sub-interval, and the magnetic field signal time series data is obtained from the three-axis fluxgate sensor, and the magnetic field signal is decomposed into 5 layers using the db4 wavelet basis, and the magnetic field signal amplitude is corrected by the temperature compensation coefficient;

[0086] Extracting four types of time-frequency domain feature parameters, namely, mean, variance, peak, and spectrum entropy, from the corrected magnetic field signal, constructing a 16-dimensional feature vector matrix for the feature parameters, acquiring a fault feature template from a historical fault database, calculating the matching degree between the feature vector and the fault template using a cosine similarity criterion, and obtaining a fault probability distribution function from the matching degree;

[0087] The insulation performance degradation index is calculated according to the fault probability distribution function, the sampling frequency of the three-axis fluxgate sensor is increased to 10 kHz in the area where the degradation index exceeds 0.8, and the high-frequency sampling data is randomly verified 1000 times using Monte Carlo sampling. If the standard deviation of the verification result is less than 0.1, the spatial distribution evaluation value of the cable insulation state is obtained from the verification result;

[0088] In a 100-meter-long cable section, the gradient boosting tree classifier uses a parameter configuration of 20 minimum leaf nodes, a learning rate of 0.1, and a tree depth of 6. Among the ten input features, the electric field strength is distributed in the range of 10-50 kilovolts per millimeter, the temperature is distributed in the range of 20-80 degrees Celsius, the mechanical stress is distributed in the range of 0.5-2.5 megapascals, and the dielectric constant is distributed in the range of 2.0-3.0. The features are input into the classifier after being standardized. The classification results show that the fault risk values ​​of the three sections of the cable 15-25 meters, 48-52 meters, and 73-78 meters exceed 0.7;

[0089] In the sections where the risk value exceeds 0.7, three-axis fluxgate sensors are arranged at intervals of 2 meters. The three-axis fluxgate sensor model is FGM3D / 1000, with a sensitivity of 0.1 nanotesla, a sampling frequency of 1000 Hz, and a temperature compensation coefficient that changes linearly from 1.00 to 0.85 in the range of 20-80 degrees Celsius;

[0090] The acquired magnetic field signal was decomposed by db4 wavelet 5 layers, and the energy ratio in the high-frequency 3 layers reached 85%. The feature extraction showed that the mean was 125 nanotesla, the variance was 15 nanotesla, the peak was 380 nanotesla, and the spectrum entropy was 1.85.

[0091] The cosine similarity between the feature and 500 groups of typical fault features in the historical fault database was calculated. The three groups with the highest similarity values ​​were 0.92, 0.89, and 0.87, respectively, and the corresponding fault probabilities were 0.85, 0.82, and 0.79, respectively;

[0092] The degradation index detected in the 48-52 meter section reached 0.83. The sampling frequency of the three-axis fluxgate sensor in this section was increased to 10 kHz. After 1000 Monte Carlo random sampling verifications, the calculated standard deviation was 0.08. The verification results confirmed that the insulation status assessment value of this section was 0.86.

[0093] Furthermore, in step S5, according to the insulation fault prone areas shown by the cable insulation status spatial distribution evaluation result, the arrangement density of fluxgate sensors is increased in the insulation fault prone areas, and for the insulation fault prone areas, the arrangement density of fluxgate sensors is reduced.

[0094] Furthermore, the fault risk values ​​are obtained, and the fault risk values ​​are spatially grouped using a K-means clustering algorithm to obtain spatial range data of high-risk areas, medium-risk areas, and low-risk areas;

[0095] Gridding the high-incidence area according to the spatial range data, setting a fluxgate sensor according to the gridding result, and obtaining layout coordinate points from the fluxgate sensor at a first spacing threshold in the high-incidence area, a second spacing threshold in the medium-risk area, and a third spacing threshold in the low-risk area;

[0096] Setting the fluxgate sensor sampling parameters for the layout coordinate points to obtain the data collection process for high-risk areas, medium-risk areas, and low-risk areas;

[0097] Calculate the fluxgate sensor signal strength attenuation curve according to the data acquisition process; if the signal strength attenuates with the square of the distance and exceeds a threshold, add a fluxgate sensor node in the area; if the crosstalk signal strength between adjacent fluxgate sensors exceeds a preset percentage of the main signal strength, adjust the fluxgate sensor node position;

[0098] Acquire collected data for the fluxgate sensor node positions, calculate spatial coverage from the collected data, and determine monitoring grid quality based on the spatial coverage.

[0099] Specifically, according to the results of the spatial distribution assessment of the cable insulation status, the K-means clustering algorithm with 3 cluster centers is used to spatially group the fault risk values. The areas with fault risk values ​​exceeding 0.8 are divided into high-risk areas, the areas with fault risk values ​​between 0.3 and 0.8 are divided into medium-risk areas, and the areas with risk values ​​below 0.3 are divided into low-risk areas. The spatial range data are obtained from the three areas, and the density adjustment planning of the monitoring area is carried out through the spatial range data.

[0100] In addition, according to the spatial sensitivity distribution map of the magnetic field tetrahedron configuration sensor array, the arrangement density of fluxgate sensors in different areas can be dynamically optimized and adjusted to match the sensitivity distribution of the tetrahedron configuration sensor array with the probability distribution of insulation faults. For example, according to the evaluation results of the spatial distribution of cable insulation status, in the three sections of 15-25 meters, 48-52 meters, and 73-78 meters where insulation faults are most frequent, the arrangement density of fluxgate sensors is increased to 10 per meter, so that the number of fluxgate sensors per unit area is increased by 5 times compared with the original arrangement plan, and the spatial resolution of magnetic field signal acquisition is increased from 2 meters to 1 meter.

[0101] In other sections with low insulation fault incidence, the density of fluxgate sensors was reduced to one per 5 meters. The number of fluxgate sensors per unit area was reduced by 60% compared with the original layout plan. The total number of fluxgate sensors was optimized from 50 to 28, and the layout cost was reduced by 44%.

[0102] By optimizing the density of fluxgate sensors in different areas three times, the sensitivity of the tetrahedron sensor array in the three high-risk sections of 15-25 meters, 48-52 meters, and 73-78 meters reached 95%, which accurately matched the fault probability distribution.

[0103] Delaunay triangulation is used to grid the high-incidence area, and a fluxgate sensor with a sensitivity of 0.1 nanotesla and a measurement range of 0-100 Tesla is set according to the grid division result. The fluxgate sensor spacing in the high-incidence area is set to 0.5 meters, the fluxgate sensor spacing in the medium-risk area is set to 2 meters, and the fluxgate sensor spacing in the low-risk area is set to 5 meters. The layout coordinate points are obtained from the fluxgate sensor spacing, and the spatial distribution map of the monitoring points is constructed through the layout coordinate points;

[0104] According to the monitoring point distribution map, the data acquisition parameters are configured. For high-incidence areas, the sampling frequency of the fluxgate sensor is set to 2000 Hz, and the signal bandwidth is 1000 Hz. For medium-risk areas, the sampling frequency is set to 500 Hz, and the signal bandwidth is 250 Hz. For low-risk areas, the sampling frequency is set to 200 Hz, and the signal bandwidth is 100 Hz. The data acquisition process is obtained from the sampling parameter configuration, and a monitoring data time series library is established through the data acquisition process;

[0105] Calculate the fluxgate sensor signal strength attenuation curve according to the monitoring data time series library, add fluxgate sensor nodes in the area where the signal strength decays by more than 70% with the square of the distance, and if the crosstalk signal strength between adjacent fluxgate sensors exceeds 5% of the main signal strength, adjust the fluxgate sensor node position, obtain collected data from the fluxgate sensor node position, calculate the spatial coverage rate for the collected data, and judge the monitoring grid quality according to the spatial coverage rate;

[0106] In the 100-meter-long cable section, the K-means clustering algorithm uses the fault risk value distribution data to perform three-level spatial grouping. After 200 iterations, the cluster centers converge to three values: 0.25, 0.55, and 0.85. The spatial division results show that 15-25 meters and 48-52 meters are high-risk areas, 32-45 meters and 60-70 meters are medium-risk areas, and the rest of the sections are low-risk areas.

[0107] In the high-risk area, a Delaunay triangulated grid with a side length of 0.5 meters is used, and a total of 22 FGM3D / 1000 fluxgate sensors are deployed, with a sensitivity of 0.1 nanotesla and a measurement range of 0-100 Tesla. In the medium-risk area, 12 fluxgate sensors are deployed at a spacing of 2 meters, and in the low-risk area, 8 fluxgate sensors are deployed at a spacing of 5 meters.

[0108] The fluxgate sensor in the high-risk area is configured with a sampling frequency of 2000 Hz. The sampling data shows that the signal bandwidth is mainly distributed in the range of 800-1000 Hz. The signal bandwidth obtained by 500 Hz sampling in the medium-risk area is concentrated in the range of 200-250 Hz. The signal bandwidth of 200 Hz sampling in the low-risk area is in the range of 80-100 Hz.

[0109] The signal strength measurement of the fluxgate sensor shows that it is 100% at a distance of 1 meter, attenuated to 25% at 2 meters, attenuated to 11% at 3 meters, and attenuated to 6% at 4 meters. The crosstalk test of adjacent fluxgate sensors shows that the crosstalk strength is 3.8% at a distance of 0.5 meters, 0.9% at a distance of 2 meters, and 0.2% at a distance of 5 meters. The spatial coverage calculation results show that the coverage rate in high-risk areas reaches 95%, the coverage rate in medium-risk areas is 85%, and the coverage rate in low-risk areas is 75%;

[0110] Obtain the spatial coordinate distribution of areas with high and low insulation fault incidence, as well as the current layout of the tetrahedron configuration sensor array, map and analyze the two, weigh the size of the high-risk area with the required monitoring coverage and accuracy, increase or decrease according to actual conditions, so that the magnetic field tetrahedron configuration sensor array focuses on areas with high insulation fault risks, and derive an adjustment plan for the position of the fluxgate sensor;

[0111] Establish a risk level map in a rectangular coordinate system based on the insulation fault spatial distribution data, and calculate the regional risk value distribution for the risk level map; obtain the cumulative probability of faults within the detection radius of the fluxgate sensor from the risk value distribution, divide the high-incidence area and the low-incidence area according to the cumulative probability, and construct a monitoring area spatial coordinate database through regional division;

[0112] The spatial coordinate database is used to perform Moran spatial autocorrelation calculation, and the monitoring area is gridded according to the autocorrelation calculation result;

[0113] Calculate the correlation coefficient between adjacent units according to the grid division, and determine the fluxgate sensor interval in the high-incidence area and the fluxgate sensor interval in the low-incidence area from the correlation coefficient;

[0114] The Kriging interpolation algorithm is used to perform spatial optimization on the fluxgate sensor intervals, the effective coverage radius of a single fluxgate sensor is calculated based on the spatial optimization result, and a tetrahedron configuration sensor array arrangement scheme is obtained from the coverage radius.

[0115] Specifically, a risk level map in a rectangular coordinate system is established based on the spatial distribution data of insulation faults. A kernel density estimation algorithm with a bandwidth of 2 meters is used to calculate the risk value distribution of each region for the map. The cumulative probability of faults is calculated within the detection radius of a fluxgate sensor with a sensitivity of 0.1 nanotesla. Areas with a risk level greater than 0.8 are obtained from the cumulative probability as high-incidence areas, and areas with a risk level less than 0.3 are obtained as low-incidence areas. A spatial coordinate database of the monitoring area is constructed through the regional division.

[0116] The Moran spatial autocorrelation algorithm is used to divide the current fluxgate sensor arrangement position into a grid with a side length of 0.5 meters. The correlation coefficient between adjacent cells is calculated for the grid cells. The fluxgate sensor interval is set to 0.5 meters in the area where the correlation coefficient is greater than 0.8, corresponding to a coverage index of 0.95. The fluxgate sensor interval is set to 5 meters in the area where the correlation coefficient is less than 0.3, corresponding to a coverage index of 0.75. The minimum number of fluxgate sensors is calculated through a signal strength attenuation model, and the optimized arrangement number that meets the interference threshold of 5% is obtained from the number;

[0117] A monitoring grid is constructed according to the optimized arrangement quantity of the fluxgate sensors, a dual-objective fitness function of coverage and cost is established for the monitoring grid, a genetic algorithm is used to perform spatial optimization on the arrangement positions of the fluxgate sensors, a cost weight of 1.0 per square meter is taken in a high-incidence area, and a cost weight of 0.2 per square meter is taken in a low-incidence area, and an arrangement scheme of 32 fluxgate sensors in a high-incidence area and 8 fluxgate sensors in a low-incidence area is obtained from the weight calculation;

[0118] The Kriging interpolation algorithm is used to divide the monitoring area space of the fluxgate sensor arrangement scheme, and the effective coverage radius of a single fluxgate sensor at a 95% confidence level is calculated for the monitoring area. A single fluxgate sensor is set to cover an area of ​​0.8 square meters in a high-incidence area, and the signal crosstalk of adjacent fluxgate sensors is less than 3%. A single fluxgate sensor is set to cover an area of ​​8 square meters in a low-incidence area, and the signal crosstalk of adjacent fluxgate sensors is less than 8%. The tetrahedral configuration sensor array adjustment scheme is obtained from the coverage constraints;

[0119] In the rectangular coordinate system of the 100-meter-long cable section, the fault risk distribution is calculated using the kernel density estimation with a bandwidth of 2 meters. The cumulative probability of the fluxgate sensor with a sensitivity of 0.1 nanotesla within a detection radius of 1 meter shows that the risk level of the 15-25 meter section and the 48-52 meter section exceeds 0.85, the risk level of the 32-45 meter section and the 60-70 meter section is between 0.35-0.65, and the risk level of the remaining sections is less than 0.28;

[0120] The monitoring area was divided into 0.5-meter-long grid cells. The Moran spatial autocorrelation calculation results showed that the correlation coefficient of adjacent cells in the high-risk area was 0.92, 0.65 in the medium-risk area, and 0.25 in the low-risk area;

[0121] The signal strength attenuation model calculation shows that the crosstalk between adjacent fluxgate sensors is 3.2% at a 0.5-meter spacing, which meets the 5% threshold requirement. At a 5-meter spacing, the crosstalk drops to 0.8%;

[0122] The genetic algorithm uses the parameter configuration of population size 100, iteration number 500, crossover probability 0.8, and mutation probability 0.1. The convergence results of the dual-objective fitness function show that under the constraint of a total cost of 1.2 million yuan, the coverage rate reaches 97.5% when 32 fluxgate sensors are deployed in high-incidence areas, and the coverage rate reaches 78.5% when 8 fluxgate sensors are deployed in low-incidence areas.

[0123] The Kriging interpolation results show that at a 95% confidence level, the effective coverage radius of a single fluxgate sensor in a high-incidence area is 0.6 meters, the actual coverage area is 0.78 square meters, and the crosstalk between adjacent fluxgate sensors is 2.8%. The effective coverage radius in a low-incidence area is 1.8 meters, the actual coverage area is 7.85 square meters, and the crosstalk between adjacent fluxgate sensors is 6.5%.

[0124] Furthermore, step S6 combines the spatial distribution evaluation results of the cable insulation status with the operating parameters and environmental factors of the cable to construct a comprehensive evaluation index system, and obtains a quantitative score of the cable insulation health status through weighted fusion of multiple indicators, providing data support for cable status monitoring and life prediction.

[0125] Furthermore, principal component analysis is used to extract features of cable operating temperature, current carrying capacity, ambient humidity, and mechanical stress parameters, and a correlation coefficient matrix is ​​constructed from the features, and a corrected operating status evaluation index is obtained through the correlation coefficient matrix;

[0126] Establishing a hierarchical analysis structure according to the operating status evaluation index, constructing a discriminant matrix for the hierarchical analysis structure, and using the discriminant matrix to calculate the operating parameter score;

[0127] Using fuzzy comprehensive evaluation to perform fusion operation on the spatial distribution characteristics of insulation status, establishing a comment set for the fusion operation, obtaining a membership function value from the comment set, and calculating the status score through the membership function value;

[0128] Constructing a quantitative scoring formula according to the status score, setting a performance indicator weighting coefficient in the quantitative scoring formula, and calculating a comprehensive cable health score according to the weighting coefficient;

[0129] If the comprehensive score is higher than a preset threshold, it is determined to be in a healthy state; if the comprehensive score is lower than a preset threshold, it is determined to be in a damaged state, and the predicted value of the remaining life of the cable is obtained from the state determination result.

[0130] Specifically, principal component analysis is used to extract features of four parameters, namely, cable operating temperature, current carrying capacity, ambient humidity, and mechanical stress. A correlation coefficient matrix is ​​constructed for the features. Eigenvalues ​​with cumulative variance contribution rates greater than 85% are selected from the correlation coefficient matrix. The principal component scores are calculated from the eigenvalues. The principal component scores are compared with the standard values ​​in the historical operation database to obtain the revised operation status evaluation index.

[0131] A hierarchical analysis structure is established according to the operating status evaluation index, a weight coefficient of 0.3 is set for the operating temperature, a weight coefficient of 0.25 is set for the current carrying capacity, a weight coefficient of 0.15 is set for the ambient humidity, and a weight coefficient of 0.3 is set for the mechanical stress, a discriminant matrix is ​​constructed from the weight coefficients, and the consistency ratio is calculated through the discriminant matrix. If the consistency ratio is less than 0.1, the operating parameter score is obtained from the discriminant matrix;

[0132] Fuzzy comprehensive evaluation is used to perform fusion operation on the spatial distribution characteristics of the insulation state, and a five-level comment set of excellent, good, qualified, poor, and dangerous is established for the fusion operation. A factor set including six indicators of electric field strength, temperature distribution, mechanical stress, dielectric loss, insulation resistance, and discharge frequency is set, and a membership function value is obtained from the factor set. The state score in the range of 0-100 is calculated by the membership function value;

[0133] A quantitative scoring formula is constructed according to the state score, and weighted coefficients of the insulation performance index 0.4, the operating parameter index 0.3, and the environmental factor index 0.3 are set in the quantitative scoring formula. The cable health comprehensive score is calculated according to the weighted coefficients. If the comprehensive score is higher than 85 points, it is judged as a healthy state, 60-85 points are judged as a sub-healthy state, and lower than 60 points are judged as a damaged state. The remaining life prediction value of the cable is obtained from the state judgment result;

[0134] The principal component analysis calculation of the 100-meter-long cable section shows that the first principal component eigenvalue is 2.86, and the variance contribution rate is 52%, which corresponds to the combined influence of operating temperature and mechanical stress. The second principal component eigenvalue is 1.92, and the variance contribution rate is 35%, reflecting the coupling effect of current carrying capacity and ambient humidity. The cumulative variance contribution rate reaches 87%. After comparing with 1,000 sets of operating records in the historical database, the correction coefficients of temperature deviation -5% and stress deviation +8% are calculated.

[0135] The analytic hierarchy process showed that the maximum eigenvalue of the discriminant matrix was 4.15, the consistency index was 0.05, the consistency ratio was 0.056, which met the threshold requirement of 0.1, and the comprehensive score of the operating parameters was 82 points;

[0136] In the fuzzy evaluation process, the membership of the six indicators of electric field strength, temperature distribution, mechanical stress, dielectric loss, insulation resistance, and discharge frequency were 0.82, 0.75, 0.68, 0.71, 0.77, and 0.65, respectively, corresponding to the five-level comment set of excellent, good, qualified, poor, and dangerous, and the calculated status score was 78 points;

[0137] The weighted calculation of the quantitative score shows that the insulation performance score is 76 points, the operating parameter score is 82 points, the environmental factor score is 85 points, and the comprehensive score is 80.5 points. The calculation formula for the comprehensive score is: Comprehensive score = Insulation performance score * Insulation performance weighting coefficient + Operating parameter score * Operating parameter weighting coefficient + Environmental factor score * Environmental factor weighting coefficient, that is: Comprehensive score = 76 * 0.4 + 82 * 0.3 + 85 * 0.3 = 30.4 + 24.6 + 25.5 = 80.5, which belongs to the sub-health status range. The remaining life prediction value calculated based on Weibull distribution is 8.5 years, and the prediction confidence is 92%.

[0138] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0139] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and rules of the present invention shall be included in the protection scope of the present invention.

Claims

1. A multi-sensor information fusion method for high-voltage cable insulation status monitoring, characterized in that: S1. Obtain magnetic flux distribution data at different spatial positions in the grounding box through a three-axis orthogonal magnetic flux sensor, analyze the magnetic flux distribution characteristics, and obtain a spatial sensitivity distribution map of the tetrahedron configuration sensor array; S2, obtaining the magnetic field signal generated by the cable fault discharge source, locating the defect position of the cable insulation, and obtaining the three-dimensional spatial coordinate information of the defect position; S3. Acquire the spatial layout information and material characteristic parameters of the cable, and build a spatial distribution evaluation model of the cable insulation status in combination with the three-dimensional spatial coordinate information of the cable insulation defects. Through finite element simulation analysis, obtain the spatial distribution characteristics of the cable insulation performance. S4. Classify and predict the spatial distribution characteristics of the cable insulation status, identify potential high-incidence areas of insulation faults, perform fine-grained analysis on the magnetic field signals in high-incidence areas of insulation faults, and obtain the spatial distribution evaluation results of the cable insulation status; S5. According to the insulation fault high-incidence areas shown by the cable insulation state spatial distribution evaluation results, the arrangement density of the fluxgate sensors is increased in the insulation fault high-incidence areas, and the arrangement density of the fluxgate sensors is dynamically optimized and adjusted according to the spatial sensitivity distribution map of the magnetic field tetrahedron configuration sensor array; S6. Combine the spatial distribution evaluation results of the cable insulation status with the cable operating parameters and environmental factors to construct a comprehensive evaluation index system. Through weighted fusion of indicators, a quantitative score of the cable insulation health status is obtained.

2. A multi-sensor information fusion method for high-voltage cable insulation status monitoring according to claim 1, characterized in that: The step S1 arranges a tetrahedron configuration for several monitoring areas divided inside the grounding box, and adopts a tetrahedron configuration sensor array group. The tetrahedron configuration sensor array group configures a three-axis orthogonal magnetic flux sensor on each surface to obtain magnetic flux spatial distribution calibration data. After acquisition, a radial basis function interpolation fitting curve is used to calculate the magnetic flux density value of the area where the three-axis orthogonal magnetic flux sensor is not arranged, and the complete magnetic flux spatial distribution data is obtained from the magnetic flux density value and a fast Fourier transform is used to perform spectrum decomposition. In the decomposition, the flux vector divergence is calculated to obtain the magnetic field source distribution characteristics and a wavelet transform is used to perform multi-scale decomposition to obtain the magnetic field space-time variation characteristics. The spatial interpolation calculation of the magnetic field space-time variation characteristics is performed by cubic spline interpolation. The cubic spline interpolation uses a Gaussian smoothing kernel function to process boundary data, and a magnetic flux isosurface spectrum is generated according to the spatial interpolation result, and the spatial sensitivity distribution data of the sensor array at each monitoring point is obtained from the isosurface spectrum.

3. The multi-sensor information fusion method for monitoring the insulation status of a high-voltage cable according to claim 1 is characterized in that: In the step S2, the magnetic field signal time series data is obtained by a fluxgate sensor, and after the acquisition, the wavelet basis function is used to decompose and denoise the signal, and then the magnetic field intensity and direction values ​​are extracted from the denoised signal, and the magnetic field gradient tensor matrix is ​​constructed by using the magnetic field intensity and direction values. The magnetic field gradient tensor matrix is ​​subjected to singular value decomposition, and the main axis direction of the magnetic field source is obtained by retaining the eigenvalues ​​and eigenvectors with energy accounting for more than a preset percentage; An inversion operation is performed on the main axis direction by using the steepest descent method, wherein the objective function of the inversion operation includes a magnetic field intensity attenuation term and a direction deviation term, a multipole expansion coefficient of a magnetic field source is obtained from the inversion operation, and a quantitative relationship between the attenuation of the magnetic field source intensity and the quadratic attenuation of the distance is established by using the multipole expansion coefficient; According to the quantitative relationship, an equipotential surface distribution of grid density is constructed in three-dimensional space, and the spatial coordinate value of the discharge source position is obtained through Monte Carlo random sampling verification. If the standard deviation of the spatial coordinate value is less than a threshold, the position is determined to be a cable insulation defect point.

4. The multi-sensor information fusion method for high-voltage cable insulation status monitoring according to claim 1 is characterized in that: In step S4, a gradient boosting tree classifier is used to classify the spatial distribution characteristics of electric field strength, temperature distribution, mechanical stress, dielectric constant, insulation thickness, current carrying capacity, operating time, ambient humidity, material aging and historical fault frequency; Setting a risk level threshold according to the classification result, and obtaining spatial range data of the fault area from the risk level threshold; Divide the spatial range data of the fault area into monitoring sub-intervals in the cable length direction, and arrange fluxgate sensors in the monitoring sub-intervals to obtain magnetic field signal time series data; Decomposing the signal using a wavelet basis function according to the magnetic field signal time series data, and correcting the amplitude of the decomposed signal using a temperature compensation coefficient; Extracting mean, variance, peak value and spectrum entropy from the corrected data to obtain time-frequency domain feature parameters, and constructing a feature vector matrix for the feature parameters; The cosine similarity criterion is used to calculate the matching degree between the feature vector and the fault template, and the fault probability distribution function is obtained from the matching degree; The insulation performance degradation index is calculated according to the fault probability distribution function, the sensor sampling frequency is increased in the area where the degradation index exceeds the preset threshold, the high-frequency sampling data is randomly verified using the Monte Carlo method, and the cable insulation status evaluation value is obtained from the spatial distribution evaluation verification result of the cable insulation status.

5. The multi-sensor information fusion method for high-voltage cable insulation status monitoring according to claim 1 is characterized in that: The step S5 obtains the fault risk value, and uses the K-means clustering algorithm to spatially group the fault risk value to obtain the spatial range data of the high-risk area, the medium-risk area and the low-risk area; Gridding the high-incidence area according to the spatial range data, setting fluxgate sensors according to the gridding results, setting the fluxgate sensor spacing in the high-incidence area to 0.5 meters, setting the fluxgate sensor spacing in the medium-risk area to 2 meters, and setting the fluxgate sensor spacing in the low-risk area to 5 meters, obtaining the layout coordinate points from the fluxgate sensor spacing, and constructing a spatial distribution map of the monitoring points through the layout coordinate points; Configure data sampling parameters according to the monitoring point distribution map, obtain a data acquisition process from the sampling parameter configuration, and establish a monitoring data time series library through the data acquisition process; Calculate the fluxgate sensor signal strength attenuation curve according to the monitoring data time series library, if the signal strength attenuates with the square of the distance and exceeds a threshold, add a fluxgate sensor node in the area, if the crosstalk signal strength between adjacent fluxgate sensors exceeds a preset percentage of the main signal strength, adjust the fluxgate sensor node position; Acquire collected data for the fluxgate sensor node positions, calculate spatial coverage from the collected data, and determine monitoring grid quality based on the spatial coverage.

Images