CN-121978137-A - GIS equipment X-ray detection method for realizing automatic space pose compensation

CN121978137ACN 121978137 ACN121978137 ACN 121978137ACN-121978137-A

Abstract

The invention provides a GIS equipment X-ray detection method for realizing automatic space pose compensation, which belongs to the technical field of GIS equipment detection, acquires GIS equipment point cloud data through multi-view scanning of a laser radar, eliminates noise by adopting self-adaptive multi-scale statistical filtering, realizes non-rigid point cloud registration by utilizing a space registration optimization model combining a depth expansion network and sparse coding, corrects the position of an internal conductor based on an electromagnetic field inverse problem inversion algorithm, automatically adjusts the space position and the ray angle of an X-ray machine through a pose compensation adjustment function according to pose deviation vectors, and solves the technical problem that the detection pose cannot be automatically compensated due to equipment deformation and positioning errors during GIS equipment X-ray detection.

Inventors

NIU BO
MA FEIYUE
Dai Longcheng
YU JIAYING
WANG BO
NI HUI
ZHANG CE
QIAN YONG
XU YUHUA

Assignees

国网宁夏电力有限公司电力科学研究院
国网宁夏电力有限公司

Dates

Publication Date: 20260505
Application Date: 20260311

Claims (10)

1. A GIS equipment X-ray detection method capable of achieving space pose automatic compensation is characterized by comprising the steps of adopting a laser radar to conduct multi-view scanning on GIS equipment to obtain an initial point cloud data set, recording space pose information of the laser radar through an RTK differential positioning system, conducting self-adaptive multi-scale statistical filtering processing on the initial point cloud data set to eliminate noise points generated by electromagnetic interference and metal reflection, obtaining a preprocessed point cloud data set, inputting the preprocessed point cloud data set and the space pose information into a space registration optimization model, outputting a complete GIS equipment digital model containing conductor contact space coordinates through the space registration optimization model, calculating material thickness distribution on an X-ray penetrating path based on the complete GIS equipment digital model, determining an initial exposure parameter set according to the material thickness distribution through an exposure parameter prediction function, acquiring probability distribution estimated values of an internal conductor position according to the external electromagnetic field distribution data through an electromagnetic field inverse problem inversion algorithm, correcting conductor contact space coordinates in the complete GIS equipment digital model according to the probability distribution estimated values, obtaining corrected contact coordinates, automatically calculating position deviation vectors according to the corrected contact and an X-ray machine current position pose, and automatically adjusting position vectors according to the position deviation vectors of the corrected contact and the X-ray machine, and automatically adjusting the position vectors according to the position vectors of the corrected position vectors.
2. The method of claim 1, wherein the RTK differential positioning system provides three-dimensional coordinates of centimeter-level accuracy for a mobile station mounted on a lidar and X-ray machine by receiving differential signals of a reference station and a satellite.
3. The method according to claim 2, wherein the adaptive multi-scale statistical filtering process specifically calculates a local density value and a curvature change value of each point in the preprocessed point cloud data set, divides the point cloud data into a high-density area and a low-density area according to the local density value, performs statistical analysis on the high-density area by using a small-scale neighborhood, and performs statistical analysis on the low-density area by using a large-scale neighborhood.
4. A method according to claim 3, wherein the adaptive multi-scale statistical filtering process further comprises calculating an average distance and a standard deviation from each point to its neighborhood point, and determining a noise point when the average distance of a certain point exceeds a dynamic threshold, wherein the dynamic threshold is adjusted according to a curvature change value of an area where the point is located.
5. The method of claim 4, wherein the spatial registration optimization model comprises an input layer, a first feature extraction module, a sparse coding layer, a depth expansion network layer, a second feature extraction module, a Riemann manifold registration module, a local affine transformation estimation module, a global optimization layer, and an output layer.
6. The method of claim 5, wherein the depth expansion network layer expands an iterative process of a fast iterative shrink threshold algorithm FISTA into 8 learnable network layers, each layer containing a soft threshold operator and a gradient descent update module.
7. The method of claim 6, wherein the Riemann manifold registration module embeds high-level semantic information into the Riemann manifold space and calculates the point-to-similarity matrix using geodesic distances calculated using Dijkstra's shortest path algorithm.
8. The method of claim 7, wherein the local affine transformation estimation module constrains neighboring region transformation consistency by graph laplace regularization, outputs transformation parameters for each point cloud mass, and the global optimization layer iteratively solves for an optimal transformation field on the lie manifold using a Log-Euclidean framework.
9. The method of claim 8, wherein the training of the spatial registration optimization model is performed end-to-end using an Adam optimizer, and the loss function is a weighted sum of registration error loss and sparse regularization loss, and the registration error loss uses a mean square error of the predicted coordinates and the tag data.
10. The method of claim 9, wherein the calculating of the exposure parameter prediction function includes extracting thickness values and material density values of all material layers on the X-ray penetration path, querying corresponding linear attenuation coefficients according to material types, calculating a total equivalent thickness of the penetration path, and querying an initial tube voltage value and an initial exposure time value according to the total equivalent thickness.

Description

GIS equipment X-ray detection method for realizing automatic space pose compensation Technical Field The invention belongs to the technical field of GIS equipment detection, and particularly relates to a GIS equipment X-ray detection method for realizing automatic space pose compensation. Background The GIS equipment X-ray detection technology is used for detecting the contact state and the defect of the internal conductor contact of the gas insulated switchgear, the traditional method relies on manual measurement and experience judgment to determine the detection position of an X-ray machine, the size of the equipment shell is measured through a laser range finder, the position of the internal contact is calculated according to a design drawing, and the angle of the X-ray machine is manually adjusted to scan. In the prior art, because the GIS equipment is subjected to non-rigid deformation under the influence of thermal stress and mechanical load in the long-term operation process, the deviation exists between a design drawing and an actual geometric state, and the laser range finder can only acquire the discrete point coordinates of the surface of the shell, a complete three-dimensional digital model cannot be constructed, and the calculation error of the position of an internal contact is larger. The traditional method needs to repeatedly adjust the position of the X-ray machine to perform heuristic scanning, judges whether the ray angle is aligned with the contact or not according to experience of a detector, and has low detection efficiency and is easy to miss key parts. That is, the technical problem that the detection pose cannot be automatically compensated due to the deformation and the positioning error of the GIS equipment during the X-ray detection in the prior art exists. Disclosure of Invention In view of the above, the invention provides a GIS equipment X-ray detection method for realizing automatic compensation of space pose, which can solve the technical problem that the detected pose cannot be automatically compensated due to equipment deformation and positioning errors in the prior art of GIS equipment X-ray detection. The invention provides a GIS equipment X-ray detection method for realizing automatic space pose compensation, which comprises the following steps of adopting a laser radar to carry out multi-view scanning on GIS equipment to obtain an initial point cloud data set, recording space pose information of the laser radar through an RTK differential positioning system, carrying out self-adaptive multi-scale statistical filtering processing on the initial point cloud data set to eliminate noise points generated by electromagnetic interference and metal reflection, obtaining a preprocessed point cloud data set, inputting the preprocessed point cloud data set and the space pose information into a space registration optimization model, outputting a complete GIS equipment digital model containing conductor contact space coordinates by the space registration optimization model, calculating material thickness distribution on an X-ray penetrating path based on the complete GIS equipment digital model, determining an initial exposure parameter set according to the material thickness distribution through an exposure parameter prediction function, acquiring probability distribution estimation values of internal conductor positions according to the external electromagnetic field distribution data through an electromagnetic field inverse problem inversion algorithm, correcting conductor contact space coordinates in the complete GIS equipment digital model according to the probability distribution estimation values, obtaining corrected contact coordinates, carrying out automatic position compensation according to corrected contact coordinates and an X-ray current pose vector, and carrying out automatic position compensation according to an initial pose compensation position and an X-ray imaging position vector, and carrying out automatic position compensation according to an initial pose compensation function. The RTK differential positioning system provides three-dimensional coordinates with centimeter-level precision for a mobile station installed on a laser radar and an X-ray machine by receiving differential signals of a reference station and a satellite. The self-adaptive multi-scale statistical filtering processing is specifically to calculate a local density value and a curvature change value of each point in the preprocessed point cloud data set, divide the point cloud data into a high-density area and a low-density area according to the local density value, perform statistical analysis on the high-density area by adopting a small-scale neighborhood, and perform statistical analysis on the low-density area by adopting a large-scale neighborhood. The adaptive multi-scale statistical filtering processing further comprises calculating the average distance and standard deviation from each