CN-121636632-B - Automatic acquisition and space-time correlation method for inspection data under Beidou space-time reference

CN121636632BCN 121636632 BCN121636632 BCN 121636632BCN-121636632-B

Abstract

The invention relates to the technical field of Beidou inspection data processing, and discloses an automatic acquisition and space-time correlation method for inspection data under a Beidou space-time reference. The method comprises the steps of acquiring historical working condition data of the inspection equipment based on nanosecond time synchronization and centimeter space positioning information of a Beidou system, constructing a space-time dynamic prediction model, acquiring measured values of equipment sensors, a moving track point set and a time stamp sequence in real time, outputting theoretical inspection values through the model, carrying out space-time multidimensional comparison on the theoretical values and the actual values, analyzing time accumulation errors, space deviation and data flow form consistency, generating difference characteristic tensors, inputting the tensors into a geocoding topological reasoning network, combining topographic map node attributes with equipment distribution, simulating abnormal diffusion, generating an abnormal probability cloud map and delineating an abnormal range, enabling enhanced acquisition on a high probability abnormal area according to the dynamic configuration inspection parameters of the cloud map, and executing data reliability inspection on peripheral equipment.

Inventors

WANG HAILI
HU XIAOMING
DUAN QIBIN
LI TONG
WANG ZHONGHONG
WU HUA
LIU JUNCHENG

Assignees

北京安心易维科技有限公司
山东科翼数字科技有限公司

Dates

Publication Date: 20260512
Application Date: 20251201

Claims (9)

1. The method for automatically acquiring and correlating the patrol data under the Beidou space-time reference is characterized by comprising the following steps of: Acquiring historical working condition data of the inspection equipment based on nanosecond time synchronization and centimeter space positioning information provided by a Beidou satellite navigation system, and constructing a space-time dynamic prediction model; Acquiring a sensor measured value, a moving track point set and a time stamp sequence of the inspection equipment in real time, and outputting a theoretical inspection value through the space-time dynamic prediction model; Carrying out space-time multidimensional comparison analysis on the theoretical inspection value and the actual inspection value, wherein analysis contents comprise accumulated errors on a time axis, deviation degree on space coordinates and form consistency on a data stream, and generating a difference characteristic tensor of each inspection device; inputting the difference characteristic tensor to a topological reasoning network of the geocode, simulating an abnormal data diffusion process by combining the node attribute of the topographic map and the equipment distribution relation, generating an abnormal probability cloud map, and defining an abnormal inspection range; Dynamically configuring inspection parameters according to the anomaly probability cloud image, wherein the inspection parameters comprise starting an enhanced acquisition mode for a high-probability anomaly region, and executing data credibility inspection for peripheral equipment; The construction of the space-time dynamic prediction model comprises the steps of carrying out multi-resolution analysis on historical working condition data, adopting ensemble empirical mode decomposition to extract the energy proportion of long-term trend and short-term fluctuation components of sensor measurement values, establishing the mapping relation between a moving track point set and a patrol road section through track segment clustering, and using a dynamic time warping algorithm to calibrate the time stamp sequence modes under different weather conditions; The construction of the space-time dynamic prediction model comprises the steps of inputting preprocessed historical working condition data into an integrated learning frame, wherein the integrated learning frame comprises a gating circulation unit prediction component based on an equipment degradation model and is used for generating a reference inspection predicted value, integrating a spatial network of a geographic attention mechanism, correcting prediction deviation caused by positioning errors, dynamically adjusting a prediction coefficient according to a moving track point set acquired in real time based on a waveform adapter, and synchronously acquiring signal-to-noise ratio and receiving intensity indexes of sensor measured values, longitude and latitude precision and time synchronization residual errors of moving tracks, clock drift and transmission delay parameters of a time stamp sequence through a Beidou time service acquisition module; the calculation of the theoretical inspection value comprises the step of inputting real-time acquisition data into the space-time dynamic prediction model to obtain the theoretical inspection value.
2. The method for automatically acquiring and correlating the patrol data under the Beidou space-time reference according to claim 1 is characterized in that the calculation of the theoretical patrol value comprises the steps of executing self-adaptive smoothing processing based on meteorological conditions, eliminating observation noise caused by wind speed sunlight, fusing correlation attributes of sensor measurement values, a moving track point set and a time stamp sequence through a space-time encoder, outputting the theoretical patrol value containing a normal fluctuation range, and updating the theoretical patrol value along with the health state of equipment in a self-adaptive manner.
3. The method for automatically acquiring and correlating the patrol data under the Beidou space-time reference according to claim 1, wherein the space-time multidimensional comparison analysis specifically comprises: Calculating time accumulated errors, namely comparing a theoretical inspection value with an actual inspection value in a rolling window mode in a set time period, aligning asynchronously sampled data streams by adopting a dynamic time warping algorithm, and calculating accumulated error quantity in each period to form a time error vector; space deviation degree detection, namely performing space transformation on a moving track point set of a theoretical value and an actual value, calculating Euclidean distance ratio of coordinate points, extracting deviation metrics of all position points, and constructing a space deviation vector; The data flow shape consistency assessment comprises the steps of calculating the time delay difference of data peak points based on the shape context algorithm matching theoretical value and the profile distribution of the actual data flow, quantifying the jessen divergence of the flow interval distribution, and generating a shape consistency vector; and generating a difference characteristic tensor, namely performing tensor superposition on the time error vector, the space deviation vector and the form consistency vector, eliminating scale difference through characteristic weight normalization processing, and outputting the difference characteristic tensor.
4. The method for automatically acquiring and correlating the patrol data under the Beidou space-time reference according to claim 3 is characterized in that the detection of the spatial deviation degree specifically comprises the steps of extracting the corresponding moving track coordinates in a theoretical patrol value and an actual patrol value in a space comparison stage, carrying out multi-scale grid division on each group of track points by using a density clustering algorithm, extracting the position density characteristics in a preset key geographic area, selecting a movement range covering a common patrol area in the key geographic area, carrying out quantitative evaluation on the distribution intensity of the theoretical data and the actual data in the key geographic area after the extraction is completed, extracting the deviation indexes of each path segment based on the relative difference degree of the theoretical patrol value and the actual patrol value, collecting the deviation results of all the segments, and constructing a spatial deviation vector.
5. The method for automatically acquiring and correlating the patrol data under the Beidou space-time reference according to claim 3, wherein in the evaluation of the data stream form consistency, a contour matching algorithm adopts a shape context distance algorithm to perform shape matching on extreme point sets in two data streams, and a delay difference is identified.
6. The method for automatically acquiring and correlating the patrol data under the Beidou space-time reference according to claim 1, wherein the step of correlating and locating the abnormal area specifically comprises the following steps: Geographic topology modeling, namely constructing a map grid connection topological graph according to the position information of the routing inspection equipment, marking path length parameters among grids, superposing propagation loss conditions caused by topography fluctuation in the topological graph, and generating a topology model comprising a path matrix and a grid reachability matrix; Performing abnormal diffusion deduction based on a graph convolution network, wherein the calculation of the abnormal diffusion deduction comprises calculating an attenuation coefficient of abnormal data according to a path length parameter, capturing a cross-grid abnormal association mode through a self-attention mechanism, and simulating a migration path of the abnormal data in the topology network by adopting a random walk method; Counting the occurrence frequency of abnormal data of each path in simulation diffusion, calculating the retention probability value of the abnormal data by combining the path length parameters, generating an abnormal probability cloud chart covering the whole graph, and labeling suspicious path groups with the probability value exceeding a set threshold value; and defining a physical range, namely executing density cluster analysis on the abnormal probability cloud image, identifying an abnormal probability dense region, and defining an abnormal inspection physical boundary according to the connection relation between the position of the inspection equipment and the geographic topology.
7. The method for automatically acquiring and associating the inspection data under the Beidou space-time reference according to claim 6 is characterized in that the step of associating and positioning the abnormal area further comprises outputting suspicious equipment identifiers and abnormal diffusion main paths, wherein the suspicious equipment identifiers are based on map grids connected by suspicious path groups, the map grids are associated with real inspection equipment to form a suspicious equipment identifier list to indicate possible abnormal data sources or infected equipment, the abnormal diffusion main paths record grid paths and sequences of the grid paths which are passed by each migration in the abnormal migration process of random walk simulation, the occurrence times of each path in all simulation paths are counted, a path sequence with the largest occurrence times is selected to serve as an abnormal diffusion main path, and the outputted abnormal diffusion main path sequence is an ordered grid list to reflect the main propagation route of abnormal data in a geographic space.
8. The method for automatically acquiring and associating the patrol data under the Beidou space-time reference according to claim 1 is characterized in that the step of dynamically configuring the patrol parameters specifically comprises the steps of sending an acquisition strategy updating instruction to an acquisition terminal to which a certain area belongs when an abnormal probability value of the area exceeds a set threshold value in an abnormal probability cloud picture, executing the step of increasing the data sampling rate to an initial multiplying power, simultaneously starting a track monitoring high-frequency mode, capturing a coordinate jump event of a moving track, installing an abnormal detector at a terminal side, recording data abnormal fragments, and executing the step of checking data credibility, namely applying multi-type data excitation to adjacent grid equipment with the most intense abnormal probability change in the abnormal probability cloud picture.
9. The method for automatically acquiring and correlating the patrol data under the Beidou space-time reference of claim 8, wherein the multi-type data excitation comprises: Injecting a sweep test waveform through a programmable signal generator; the method comprises the steps of calculating a theoretical response curve based on a topological model and a path matrix, acquiring and recording all grid data after test waveform injection by actual measurement response to obtain an actual measurement response curve, calculating the difference between the theoretical response curve and the actual measurement response curve by abnormal deviation, comparing the difference between the actual measurement response curve and the theoretical response curve, calculating the abnormal grid deviation rate, and marking as data abnormal association equipment when the abnormal grid deviation rate exceeds a limit value.

Description

Automatic acquisition and space-time correlation method for inspection data under Beidou space-time reference Technical Field The invention relates to the technical field of Beidou inspection data processing, in particular to an automatic acquisition and space-time correlation method for inspection data under a Beidou space-time reference. Background In the fields of operation and maintenance of industrial equipment, monitoring of infrastructure and the like, inspection is an important means for guaranteeing stable operation of equipment and finding potential faults in time. Along with the improvement of equipment complexity and the expansion of monitoring range, traditional inspection methods gradually expose a plurality of limitations, and the requirements of high-precision and high-efficiency inspection are difficult to meet. Traditional inspection data acquisition relies on manual recording or single positioning technology, time synchronization accuracy is usually stopped at millisecond level or even second level, space positioning error is also more than meter level, and accurate binding of data and space-time information cannot be achieved. The lack of precision causes that the data collected by different inspection equipment and different time lacks a unified space-time reference, the relativity among the data is weak, the complete equipment running state image is difficult to form, the problems of data dislocation, logic contradiction and the like are easy to occur in the subsequent analysis, and the accurate judgment of the equipment state is influenced. Traditional inspection is mainly carried out by post analysis of real-time data, depth utilization of historical working condition data is lacked, and an effective dynamic prediction mechanism is not constructed. In actual operation, the abnormality is often judged only when the actual inspection value exceeds a preset threshold value, the slight change of the running trend of the equipment cannot be perceived in advance, obvious hysteresis exists in abnormal discovery, the optimal opportunity for early fault intervention can be missed, and the equipment damage and the operation and maintenance cost are increased. In the data comparison and analysis link, the traditional method focuses on numerical value differences in a single dimension, and ignores cumulative effects of errors on a time axis, position deviation on a space coordinate and form consistency of a data stream. For example, only the sensor value at a certain moment is compared, the accumulated deviation of the value and the historical synchronous data is not considered, whether the movement track of the equipment deviates from a preset path or not is not analyzed, whether the fluctuation rule of the data flow accords with the normal operation characteristic or not is not checked, the difference identification is on one side, potential abnormal signals associated with multiple dimensions of the equipment are difficult to capture, and missing detection or false detection easily occurs. Traditional abnormal range defining and inspection parameter configuration lacks dynamic adjustment capability. During anomaly detection, the range is directly defined based on the anomaly data of a single device, the distribution relation of the node attribute of the non-combined topographic map and the peripheral device simulates an anomaly diffusion process, the deviation between the defined anomaly range and the actual diffusion condition is larger, and the follow-up inspection pertinence is insufficient. Meanwhile, the inspection parameters are fixedly arranged, no matter the abnormal probability of the area is high or low, the same acquisition frequency and inspection standard are adopted, so that insufficient data acquisition of the abnormal area with high probability is caused, resources in the low probability area are wasted, special inspection is not carried out on the data reliability of peripheral equipment in the abnormal area, and the reliability of the inspection result is further influenced. Disclosure of Invention The invention aims to provide an automatic acquisition and space-time correlation method for inspection data under the Beidou space-time reference so as to solve the problems in the background technology. In order to achieve the above purpose, the invention provides a method for automatically acquiring and correlating the patrol data under the Beidou space-time reference, which comprises the following steps: Acquiring historical working condition data of the inspection equipment based on nanosecond time synchronization and centimeter space positioning information provided by a Beidou satellite navigation system, and constructing a space-time dynamic prediction model; Acquiring a sensor measured value, a moving track point set and a time stamp sequence of the inspection equipment in real time, and outputting a theoretical inspection value through the space-time dynamic prediction m