Transformer internal anomaly detection method, system and device and storage medium

Abstract

The invention relates to the technical field of transformer detection, in particular to a transformer internal anomaly detection method, system and device and a storage medium. In order to solve the technical problem of low transformer anomaly detection accuracy in the prior art, the method comprises the following steps: firstly, obtaining one-circle ultrasonic data of a transformer, and carrying out filtering and alignment processing to obtain a plurality of pieces of transformer ultrasonic alignment data; converting the ultrasonic alignment data of the transformer into a point cloud, and performing registration and splicing processing to obtain a three-dimensional point cloud model of the transformer; and finally, based on the three-dimensional point cloud model of the transformer, accurately and quickly obtaining parameters reflecting the shape and position characteristics of the windings in the transformer, and quickly evaluating the deformation degree of each winding after comparing the parameters with corresponding standard parameters. And when the deformation degree of the windings is not large, the accumulated stress condition of all the windings in the transformer is calculated according to the diameters and radians of the windings, the internal safety of the transformer is comprehensively analyzed, and the accuracy and stability of internal anomaly detection of the transformer are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of transformer detection, and in particular to a method, system, device and storage medium for detecting internal abnormality of a transformer. Background Art

[0002] The deformation of the internal winding of the transformer will cause electric field distortion, leading to discharge and affecting the safety of the power grid. Therefore, accurate assessment of transformer safety is the basis for reliable operation of the power grid. At present, existing technologies often use oil chromatography, partial discharge, infrared, ultraviolet and other technical methods to detect transformer anomalies. However, the above methods are all process detection after discharge or creepage occurs. The detection timing is relatively delayed and lacks effective early warning signals and time. At the same time, the above methods are easily affected by the operating environment, detection range and sensitivity, resulting in poor accuracy and stability of transformer anomaly detection results. Summary of the invention

[0003] The object of the present invention is to provide a method, system, device and storage medium for detecting internal abnormality of a transformer.

[0004] The technical solution of the present invention is as follows: A method for detecting anomalies inside a transformer includes the following operations: S1. Arrange a number of ultrasonic sensors around the transformer to obtain ultrasonic data in real time to obtain a number of transformer ultrasonic data; perform alignment processing on the several transformer ultrasonic data after filtering processing respectively to obtain a number of transformer ultrasonic alignment data; S2. Converting several transformer ultrasonic alignment data into point cloud data respectively to obtain several transformer point cloud data; performing registration and splicing processing on several transformer point cloud data to obtain a three-dimensional point cloud model of the transformer; S3. From the three-dimensional point cloud model of the transformer, obtain the diameter, height, radian and turn-to-turn distance of the winding, calculate the differences with the standard diameter, standard height, standard radian and standard turn-to-turn distance respectively, and obtain the diameter change, height change, radian change and turn-to-turn distance change of the winding; if the diameter change, height change, radian change, or turn-to-turn distance change of the winding are respectively greater than their respective change thresholds, there is an abnormality inside the transformer; if the diameter change, height change, radian change, and turn-to-turn distance change of the winding are all not greater than their respective change thresholds, obtain the internal safety value of the transformer based on the diameter and height of the winding; if the internal safety value of the transformer is less than the safety value threshold, there is an abnormality inside the transformer.

[0005] The operation of converting the current transformer ultrasonic alignment data into point cloud data in S2 is specifically as follows: based on the position of each point in the current transformer ultrasonic alignment data, the coordinates of each point in the current transformer ultrasonic alignment data are obtained to obtain the local coordinates of each point; based on the distance between the ultrasonic sensor and the center of the transformer corresponding to the current transformer ultrasonic alignment data, the local coordinates of each point are converted to a coordinate system with the center of the transformer as the origin to obtain the global coordinates of each point, thereby forming the current transformer point cloud data.

[0006] The specific operation of registering the current transformer point cloud data with the next set of transformer point cloud data in S2 is as follows: in the next set of transformer point cloud data, obtain the corresponding point cloud of the feature point cloud of the current transformer point cloud data to obtain several feature point cloud pairs; based on the several feature point cloud pairs, construct several linear equations to form a linear equation group; the linear equation group is processed by the least squares method to obtain a rotation matrix and a translation vector; based on the rotation matrix and the translation vector, perform spatial transformation on the next set of transformer point cloud data to obtain the next set of transformer point cloud initial registration data; the next set of transformer point cloud registration data and the current transformer point cloud data are processed by the nearest point iteration method to obtain the current transformer point cloud registration data and the next set of transformer point cloud registration data.

[0007] The method for obtaining the characteristic point cloud of the current transformer point cloud data is specifically as follows: obtaining the neighborhood covariance matrix of each point cloud in the current transformer point cloud data; and taking the corresponding point cloud whose eigenvalue ratio of the neighborhood covariance matrix is ​​greater than the eigenvalue ratio threshold as the characteristic point cloud.

[0008] The internal safety value of the transformer in S3 is obtained by the following calculation formula: , , , is the internal safety value of the transformer, is the coefficient, For the i The maximum allowable stress of a winding, For the i The mechanical stress of a winding, For the i The electromagnetic stress of the winding, is the total number of windings, For the i The elastic modulus of the winding, For the i The curvature of the winding, For the i The diameter of the winding, For the iThe arc of the winding, For the i The current density of the winding, For the i The magnetic field strength of the winding.

[0009] The alignment processing operation in S1 can be achieved by performing spatial transformation on each transformer ultrasonic filtering data, and converting each transformer ultrasonic filtering data from its own local coordinate system to the global coordinate system.

[0010] S2 also includes downsampling several transformer point cloud data respectively to obtain several transformer point cloud downsampled data for performing subsequent alignment operations; the downsampling operation is specifically: dividing the current transformer point cloud data into several point cloud units, retaining the centroid point cloud in each point cloud unit, and obtaining the current transformer point cloud downsampled data; the centroid point cloud is the average value of all point clouds in the corresponding point cloud unit.

[0011] A transformer internal abnormality detection system, used to implement the above transformer internal abnormality detection method, comprising: The transformer ultrasonic alignment data generation module is used to arrange a number of ultrasonic sensors around the transformer to obtain ultrasonic data in real time and obtain a number of transformer ultrasonic data; the several transformer ultrasonic data are respectively filtered and aligned to obtain a number of transformer ultrasonic alignment data; The transformer 3D point cloud model generation module is used to convert several transformer ultrasonic alignment data into point cloud data respectively to obtain several transformer point cloud data; perform registration and splicing processing on several transformer point cloud data to obtain the transformer 3D point cloud model; The transformer has an abnormality judgment module inside, which is used to obtain the diameter, height, radian and turn-to-turn distance of the winding from the three-dimensional point cloud model of the transformer, and calculate the differences with the standard diameter, standard height, standard radian and standard turn-to-turn distance respectively, to obtain the diameter change, height change, radian change and turn-to-turn distance change of the winding; if the diameter change, height change, radian change, or turn-to-turn distance change of the winding are respectively greater than the respective change thresholds, then the transformer has an abnormality inside; if the diameter change, height change, radian change, and turn-to-turn distance change of the winding are all not greater than the respective change thresholds, the transformer internal safety value is obtained based on the diameter and height of the winding; if the transformer internal safety value is less than the safety value threshold, then the transformer has an abnormality inside.

[0012] A transformer internal abnormality detection device comprises a processor and a memory, wherein the processor implements the above-mentioned transformer internal abnormality detection method when executing a computer program stored in the memory.

[0013] A computer-readable storage medium is used to store a computer program, wherein the computer program implements the above-mentioned transformer internal abnormality detection method when executed by a processor.

[0014] The beneficial effects of the present invention are: The present invention provides a method for detecting anomalies inside a transformer. First, a plurality of ultrasonic sensors are arranged around the transformer to obtain ultrasonic data in real time, and then filtering and alignment processing are performed to improve the quality of ultrasonic data and improve the calculation efficiency, so as to obtain a plurality of ultrasonic alignment data of the transformer. Then, the plurality of ultrasonic alignment data of the transformer are respectively converted into three-dimensional point cloud data and then aligned and spliced ​​to obtain a three-dimensional point cloud model of the transformer that can intuitively reflect the internal spatial structure characteristics of the transformer. Finally, based on the three-dimensional point cloud model of the transformer, parameters reflecting the shape characteristics (diameter, height, radian) and position characteristics (turn-to-turn distance) of the winding inside the transformer are accurately and quickly obtained, and after comparing with the corresponding parameters of the three-dimensional point cloud model of the standard transformer, the deformation degree of each winding can be quickly evaluated. When the deformation degree of the winding is not large, the cumulative stress conditions of all windings inside the transformer are calculated based on the diameter and radian of the winding, and the internal safety of the transformer is comprehensively analyzed to improve the accuracy and stability of the detection of abnormalities inside the transformer. DETAILED DESCRIPTION

[0015] This embodiment provides a method for detecting an abnormality inside a transformer, including the following operations: S1. Arrange a number of ultrasonic sensors around the transformer to obtain ultrasonic data in real time to obtain a number of transformer ultrasonic data; perform alignment processing on the several transformer ultrasonic data after filtering processing respectively to obtain a number of transformer ultrasonic alignment data; S2. Converting several transformer ultrasonic alignment data into point cloud data respectively to obtain several transformer point cloud data; performing registration and splicing processing on several transformer point cloud data to obtain a three-dimensional point cloud model of the transformer; S3. From the three-dimensional point cloud model of the transformer, obtain the diameter, height, radian and turn-to-turn distance of the winding, calculate the differences with the standard diameter, standard height, standard radian and standard turn-to-turn distance respectively, and obtain the diameter change, height change, radian change and turn-to-turn distance change of the winding; if the diameter change, height change, radian change, or turn-to-turn distance change of the winding are respectively greater than their respective change thresholds, there is an abnormality inside the transformer; if the diameter change, height change, radian change, and turn-to-turn distance change of the winding are all not greater than their respective change thresholds, obtain the internal safety value of the transformer based on the diameter and height of the winding; if the internal safety value of the transformer is less than the safety value threshold, there is an abnormality inside the transformer.

[0016] S1. Arrange a number of ultrasonic sensors around the transformer to acquire ultrasonic data in real time, and obtain a number of transformer ultrasonic data; after filtering the several transformer ultrasonic data, perform alignment processing to obtain a number of transformer ultrasonic alignment data.

[0017] By arranging several ultrasonic sensors around the transformer, the ultrasonic data is acquired in real time and then filtered and aligned to improve the quality of ultrasonic data and the calculation efficiency, and obtain several transformer ultrasonic alignment data.

[0018] First, several ultrasonic sensors with strong penetration ability and not easily affected by the external environment are arranged around the transformer to ensure that they can fully cover the key parts inside the transformer, such as windings and other areas. At the same time, ultrasonic data of the internal structure of the transformer is obtained in real time, and several transformer ultrasonic data are obtained for real-time monitoring of the transformer.

[0019] Then, after filtering processing (preferably wavelet filtering processing) of several transformer ultrasonic data respectively, the noise in the non-stationary signal can be effectively processed while the detailed information of the ultrasonic data can be well preserved, thereby obtaining several transformer ultrasonic filtering data.

[0020] Finally, the plurality of transformer ultrasonic filter data are aligned to obtain a plurality of transformer ultrasonic alignment data. The alignment operation can be achieved by performing spatial transformation on each transformer ultrasonic filter data and converting each transformer ultrasonic filter data from its own local coordinate system to a unified global coordinate system.

[0021] S2. Convert several transformer ultrasonic alignment data into point cloud data respectively to obtain several transformer point cloud data; perform registration and splicing processing on several transformer point cloud data to obtain a three-dimensional point cloud model of the transformer.

[0022] The ultrasonic alignment data of several transformers are converted into three-dimensional point cloud data and then aligned and spliced ​​to obtain a three-dimensional point cloud model of the transformer that can intuitively reflect the internal spatial structure characteristics of the transformer, which is conducive to the subsequent rapid and accurate detection of internal abnormalities of the transformer.

[0023] Firstly, several transformer ultrasonic alignment data are converted into point cloud data respectively to obtain several transformer point cloud data.

[0024] Taking the current transformer ultrasonic alignment data as an example, the operations of converting the current transformer ultrasonic alignment data into point cloud data are as follows: based on the position of each point in the current transformer ultrasonic alignment data, the coordinates of each point in the current transformer ultrasonic alignment data are obtained to obtain the local coordinates of each point; based on the distance between the ultrasonic sensor and the center of the transformer corresponding to the current transformer ultrasonic alignment data, the local coordinates of each point are converted to a coordinate system with the center of the transformer as the origin to obtain the global coordinates of each point, thereby forming the current transformer point cloud data; the current transformer point cloud data is stored as a three-dimensional point cloud file format (for example, saved in PLY, XYZ, etc. format).

[0025] The above operations are performed on the remaining transformer ultrasonic alignment data to obtain the remaining transformer point cloud data, which can intuitively reflect the internal three-dimensional structure of the transformer and is conducive to improving the accuracy of subsequent detection.

[0026] In addition, in order to improve the computational efficiency, several transformer point cloud data are downsampled to obtain several transformer point cloud downsampled data for subsequent registration operations. The downsampling operation is as follows: the current transformer point cloud data is divided into several point cloud units, the centroid point cloud in each point cloud unit is retained, and the current transformer point cloud downsampled data is obtained. The centroid point cloud is the average value of all point clouds in the corresponding point cloud unit.

[0027] Next, a plurality of transformer point cloud data or a plurality of transformer point cloud down-sampled data are registered to obtain a plurality of transformer point cloud registration data.

[0028] Taking the current transformer point cloud data (or the current transformer point cloud downsampled data) and the next set of transformer point cloud data (or the next set of transformer point cloud downsampled data) as an example, the registration processing operation is as follows: The positions of the two ultrasonic sensors corresponding to the current transformer point cloud data and the next set of transformer point cloud data are adjacent.

[0029] Step 1: In the next set of transformer point cloud data (or the next set of transformer point cloud downsampled data), obtain the corresponding point cloud of the feature point cloud of the current transformer point cloud data (or the current transformer point cloud downsampled data) to obtain several feature point cloud pairs.

[0030] Among them, the method for obtaining the characteristic point cloud of the current transformer point cloud data is specifically as follows: obtaining the neighborhood covariance matrix of each point cloud in the current transformer point cloud data (based on the coordinate difference between each point cloud and the point cloud within the corresponding neighborhood range); taking the corresponding point cloud whose eigenvalue ratio of the neighborhood covariance matrix is ​​greater than the eigenvalue ratio threshold as the characteristic point cloud.

[0031] In addition, in a feature point cloud pair, the cosine similarity between the feature point cloud in the next set of transformer point cloud data and the feature point cloud in the current transformer point cloud data is the maximum value of the cosine similarities of all point clouds in the next set of transformer point cloud data and the feature point cloud of the current transformer point cloud data.

[0032] Step 2. Based on several pairs of feature point clouds, construct several linear equations to form a linear equation group; the linear equation group is processed by the least squares method to obtain a rotation matrix and a translation vector; based on the rotation matrix and the translation vector, perform spatial transformation on the next set of transformer point cloud data (or the next set of transformer point cloud downsampled data) (which can be achieved by multiplying the next set of transformer point cloud data with the transformation matrix obtained based on the rotation matrix and the translation vector) to obtain the next set of transformer point cloud initial registration data.

[0033] Step 3: To further improve the registration accuracy, the next set of transformer point cloud registration data and the current transformer point cloud data (or the current transformer point cloud downsampled data) are processed by the iterative closest point (ICP) method to obtain the current transformer point cloud registration data and the next set of transformer point cloud registration data.

[0034] All transformer point cloud data (or all transformer point cloud downsampled data) execute the operations of step 1, step 2 and step 3 to obtain a number of transformer point cloud registration data.

[0035] Finally, the overlapping areas between two adjacent transformer point cloud registration data in several transformer point cloud registration data are fused by weighted average, and the point clouds in the non-overlapping areas are merged to achieve point cloud splicing, so as to obtain a three-dimensional point cloud model of the transformer that can intuitively reflect the internal spatial structure characteristics of the transformer.

[0036] S3. Obtain the diameter, height, radian and turn-to-turn distance of the winding from the three-dimensional point cloud model of the transformer, calculate the differences with the standard diameter, standard height, standard radian and standard turn-to-turn distance respectively, and obtain the diameter change, height change, radian change and turn-to-turn distance change of the winding; if the diameter change, height change, radian change, or turn-to-turn distance change of the winding are respectively greater than their respective change thresholds, there is an abnormality inside the transformer; if the diameter change, height change, radian change, and turn-to-turn distance change of the winding are all not greater than their respective change thresholds, obtain the internal safety value of the transformer based on the diameter and radian of the winding; if the internal safety value of the transformer is less than the safety value threshold, there is an abnormality inside the transformer.

[0037] Based on the three-dimensional point cloud model of the transformer, the parameters reflecting the shape characteristics (diameter, height, curvature) and position characteristics (turn-to-turn distance) of the transformer's internal windings can be accurately and quickly obtained. After comparing with the corresponding parameters of the standard transformer three-dimensional point cloud model, the deformation degree of each winding can be quickly evaluated. When the deformation degree of the winding is not large, the cumulative stress conditions of all windings inside the transformer will be calculated based on the diameter and curvature of the winding, and the internal safety of the transformer will be comprehensively analyzed to improve the accuracy of internal abnormality detection of the transformer.

[0038] First, the diameter, height, arc and turn-to-turn distance of the winding are obtained from the 3D point cloud model of the transformer.

[0039] Then, the diameter, height, radian and turn-to-turn distance of the winding in the transformer 3D point cloud model are calculated, and the differences with the standard diameter, standard height, standard radian and standard turn-to-turn distance in the standard transformer 3D point cloud model are obtained to obtain the diameter change, height change, radian change and turn-to-turn distance change of the winding in the transformer 3D point cloud model, which are used to analyze the changes of the winding in the horizontal, vertical and bilateral directions.

[0040] The winding diameter is the cross-sectional diameter of the winding area of ​​interest (preferably the middle area of ​​the winding) in the transformer 3D point cloud model. The winding height is the vertical height of the winding in the transformer 3D point cloud model. The winding curvature can be obtained by curve analysis of the vertical curve of the winding in the transformer 3D point cloud model. The winding turn distance is the minimum distance between adjacent windings in the transformer 3D point cloud model.

[0041] Finally, based on the diameter change, height change, arc change, turn-to-turn distance change and change amount of the winding, it is determined whether the winding deformation will cause internal abnormalities in the transformer.

[0042] If the change in the diameter, height, arc, or turn-to-turn distance of the winding is greater than the respective change thresholds, it proves that the winding is severely deformed and there is an abnormality inside the transformer.

[0043] If the change in the diameter, height, curvature, and turn-to-turn distance of the winding are not greater than their respective change thresholds, for comprehensive analysis, the winding stress condition is evaluated based on the winding diameter and curvature to obtain the internal safety value of the transformer.

[0044] The internal safety value of the transformer is obtained by the following calculation formula: , , , is the internal safety value of the transformer, is the coefficient, For the i The maximum allowable stress of a winding, For the i The mechanical stress of a winding, For the i The electromagnetic stress of the winding, is the total number of windings, For the i The elastic modulus of the winding, For the i The curvature of the winding (based on the height and arc), For the i The diameter of the winding, For the i The arc of the winding, For the i The current density of the winding, For the i The magnetic field strength of the winding.

[0045] If the internal safety value of the transformer is less than the safety value threshold, it proves that the greater the stress on the internal winding of the transformer, the easier it is to deform, and it is easy for danger to occur inside the transformer, which means that there is an abnormality inside the transformer.

[0046] This embodiment also provides a transformer internal abnormality detection system, which is used to implement the above transformer internal abnormality detection method, including: The transformer ultrasonic alignment data generation module is used to arrange a number of ultrasonic sensors around the transformer to obtain ultrasonic data in real time and obtain a number of transformer ultrasonic data; the several transformer ultrasonic data are respectively filtered and aligned to obtain a number of transformer ultrasonic alignment data; The transformer 3D point cloud model generation module is used to convert several transformer ultrasonic alignment data into point cloud data respectively to obtain several transformer point cloud data; perform registration and splicing processing on several transformer point cloud data to obtain the transformer 3D point cloud model; The transformer has an abnormality judgment module inside, which is used to obtain the diameter, height, radian and turn-to-turn distance of the winding from the three-dimensional point cloud model of the transformer, and calculate the differences with the standard diameter, standard height, standard radian and standard turn-to-turn distance respectively, to obtain the diameter change, height change, radian change and turn-to-turn distance change of the winding; if the diameter change, height change, radian change, or turn-to-turn distance change of the winding are respectively greater than the respective change thresholds, then the transformer has an abnormality inside; if the diameter change, height change, radian change, and turn-to-turn distance change of the winding are all not greater than the respective change thresholds, the transformer internal safety value is obtained based on the diameter and height of the winding; if the transformer internal safety value is less than the safety value threshold, then the transformer has an abnormality inside.

[0047] This embodiment further provides a transformer internal abnormality detection device, including a processor and a memory, wherein the processor implements the above-mentioned transformer internal abnormality detection method when executing a computer program stored in the memory.

[0048] This embodiment further provides a computer-readable storage medium for storing a computer program, wherein the computer program implements the above-mentioned transformer internal abnormality detection method when executed by a processor.

[0049] The present embodiment provides a method for detecting anomalies inside a transformer. First, a plurality of ultrasonic sensors are arranged around the transformer to obtain ultrasonic data in real time, and then filtering and alignment processing is performed to improve the quality of ultrasonic data and improve the calculation efficiency, so as to obtain a plurality of transformer ultrasonic alignment data. Then, the plurality of transformer ultrasonic alignment data are respectively converted into three-dimensional point cloud data and then registered and spliced ​​to obtain a three-dimensional point cloud model of the transformer that can intuitively reflect the internal spatial structure characteristics of the transformer. Finally, based on the three-dimensional point cloud model of the transformer, parameters reflecting the shape characteristics (diameter, height, curvature) and position characteristics (turn-to-turn distance) of the internal winding of the transformer are accurately and quickly obtained. After comparing with the corresponding parameters of the three-dimensional point cloud model of the standard transformer, the deformation degree of each winding can be quickly evaluated. When the deformation degree of the winding is not large, the cumulative stress of all windings inside the transformer is calculated based on the diameter and curvature of the winding, and the internal safety of the transformer is comprehensively analyzed to improve the accuracy and stability of the detection of abnormalities inside the transformer.

Claims

1. A method for detecting internal abnormality of a transformer, characterized in that: The following operations are included: S1. Arrange a number of ultrasonic sensors around the transformer to obtain ultrasonic data in real time to obtain a number of transformer ultrasonic data; perform alignment processing on the several transformer ultrasonic data after filtering processing respectively to obtain a number of transformer ultrasonic alignment data; S2. Converting several transformer ultrasonic alignment data into point cloud data respectively to obtain several transformer point cloud data; performing registration and splicing processing on several transformer point cloud data to obtain a three-dimensional point cloud model of the transformer; S3. Obtain the diameter, height, radian and turn-to-turn distance of the winding from the three-dimensional point cloud model of the transformer, calculate the difference with the standard diameter, standard height, standard radian and standard turn-to-turn distance respectively, and obtain the change in diameter, height, radian and turn-to-turn distance of the winding; If the change in the diameter, height, arc, or turn-to-turn distance of the winding is greater than the respective change thresholds, there is an abnormality inside the transformer; If the change in the diameter, height, curvature, and turn-to-turn distance of the winding are not greater than their respective change thresholds, the internal safety value of the transformer is obtained based on the diameter and height of the winding; if the internal safety value of the transformer is less than the safety value threshold, there is an abnormality inside the transformer.

2. The transformer internal abnormality detection method according to claim 1, characterized in that: In S2, the operation of converting the current transformer ultrasonic alignment data into point cloud data is specifically as follows: Based on the position of each point in the current transformer ultrasonic alignment data, the coordinates of each point in the current transformer ultrasonic alignment data are obtained to obtain the local coordinates of each point; based on the distance between the ultrasonic sensor and the center of the transformer corresponding to the current transformer ultrasonic alignment data, the local coordinates of each point are transformed into a coordinate system with the transformer center as the origin to obtain the global coordinates of each point, forming the current transformer point cloud data.

3. The transformer internal abnormality detection method according to claim 1, characterized in that: In S2, the operation of registering the current transformer point cloud data with the next set of transformer point cloud data is specifically as follows: In the next set of transformer point cloud data, the corresponding point cloud of the feature point cloud of the current transformer point cloud data is obtained to obtain a number of feature point cloud pairs; Based on several pairs of feature point clouds, several linear equations are constructed to form a linear equation group; the linear equation group is processed by the least square method to obtain a rotation matrix and a translation vector; based on the rotation matrix and the translation vector, the next set of transformer point cloud data is spatially transformed to obtain the next set of transformer point cloud initial registration data; The next set of transformer point cloud registration data and the current transformer point cloud data are processed by the nearest point iteration method to obtain the current transformer point cloud registration data and the next set of transformer point cloud registration data.

4. The transformer internal abnormality detection method according to claim 3, characterized in that: The method for obtaining the characteristic point cloud of the current transformer point cloud data is specifically as follows: obtaining the neighborhood covariance matrix of each point cloud in the current transformer point cloud data; and taking the corresponding point cloud whose eigenvalue ratio of the neighborhood covariance matrix is ​​greater than the eigenvalue ratio threshold as the characteristic point cloud.

5. The transformer internal abnormality detection method according to claim 1, characterized in that: In S3, the internal safety value of the transformer is obtained by the following calculation formula: ， ， ， is the internal safety value of the transformer, is the coefficient, For the i The maximum allowable stress of a winding, For the i Mechanical stress of a winding, For the i The electromagnetic stress of the winding, is the total number of windings, For the i The elastic modulus of the winding, For the i The curvature of the winding, For the i The diameter of the winding, For the i The arc of the winding, For the i The current density of the winding, For the i The magnetic field strength of the winding.

6. The transformer internal abnormality detection method according to claim 1, characterized in that: In S1, the alignment process can be implemented by performing spatial transformation on each transformer ultrasonic filter data, and converting each transformer ultrasonic filter data from its own local coordinate system to a global coordinate system.

7. The transformer internal abnormality detection method according to claim 1, characterized in that: The step S2 also includes downsampling the plurality of transformer point cloud data to obtain a plurality of transformer point cloud downsampled data for performing subsequent registration operations; The downsampling processing operation is specifically as follows: dividing the current transformer point cloud data into a number of point cloud units, retaining the centroid point cloud in each point cloud unit, and obtaining the current transformer point cloud downsampling data; The centroid point cloud is the average value of all point clouds in the corresponding point cloud unit.

8. A transformer internal abnormality detection system, used to implement the transformer internal abnormality detection method according to claim 1, characterized in that: include: The transformer ultrasonic alignment data generation module is used to arrange a number of ultrasonic sensors around the transformer to obtain ultrasonic data in real time and obtain a number of transformer ultrasonic data; the several transformer ultrasonic data are respectively filtered and aligned to obtain a number of transformer ultrasonic alignment data; The transformer 3D point cloud model generation module is used to convert several transformer ultrasonic alignment data into point cloud data respectively to obtain several transformer point cloud data; perform registration and splicing processing on several transformer point cloud data to obtain the transformer 3D point cloud model; The transformer has an abnormality judgment module inside, which is used to obtain the diameter, height, radian and turn-to-turn distance of the winding from the three-dimensional point cloud model of the transformer, and calculate the differences with the standard diameter, standard height, standard radian and standard turn-to-turn distance respectively, to obtain the diameter change, height change, radian change and turn-to-turn distance change of the winding; if the diameter change, height change, radian change, or turn-to-turn distance change of the winding are respectively greater than the respective change thresholds, then the transformer has an abnormality inside; if the diameter change, height change, radian change, and turn-to-turn distance change of the winding are all not greater than the respective change thresholds, the transformer internal safety value is obtained based on the diameter and height of the winding; if the transformer internal safety value is less than the safety value threshold, then the transformer has an abnormality inside.

9. A transformer internal abnormality detection device, characterized in that: The method comprises a processor and a memory, wherein the processor implements the transformer internal abnormality detection method according to any one of claims 1 to 7 when executing the computer program stored in the memory.

10. A computer-readable storage medium, characterized in that: Used to store a computer program, wherein when the computer program is executed by a processor, the transformer internal abnormality detection method according to any one of claims 1 to 7 is implemented.

Images