Topological consistency risk analysis and monitoring early warning method for tower inclination state

Abstract

The invention relates to the technical field of data processing, in particular to a topological consistency risk analysis and monitoring early warning method for a tower inclination state, and the method carries out the time alignment and quality control of multi-modal data, such as an inclination angle, a wind speed and direction, temperature and humidity, positioning and an equipment state. Constructing a time-varying adjacency matrix based on line topology, geographical adjacency and wind field elements, and normalizing the time-varying adjacency matrix to form a graph model; the inclination angle or the trend of the inclination angle serves as a graph signal, space consistency measurement is calculated, and linkage with sliding window trend detection of a single rod and EWMA and ADWIN drift detection is carried out; identifying segment-level or region aggregation anomalies by synthesizing image smoothing energy, adjacent rod difference and spatial self-correlation; and outputting early warning grades under wind direction, wind speed, duration time window and data quality gating, and adaptively adjusting a threshold value, a time window and a model weight during drifting. The method can reduce false alarm and missing alarm, improves early warning accuracy and interpretability, and is suitable for centralization or edge cloud collaborative deployment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a topological consistency risk analysis and monitoring and early warning method for tower tilt state. BACKGROUND

[0002] The existing tower tilt monitoring mostly uses single rod body inclination sensor measurement or image point cloud recognition to monitor the tilt trend, and some systems integrate environmental factors such as wind speed, temperature, humidity, etc. to give early warning. However, these systems generally lack group trend comparison based on the topological relationship between rods or within the line feeder; they often rely on static threshold or only single rod event triggering, without considering the consistency of adjacent rods, and there are few multi-condition linkage judgments such as trend prediction + wind direction, wind speed, duration window, etc. based on uploaded multi-modal data. Therefore, there are more false positives or false negatives, and the stability and accuracy need to be improved. SUMMARY

[0003] In order to overcome the deficiencies in the prior art, a topological consistency risk analysis and monitoring and early warning method for tower tilt state is proposed.

[0004] A topological consistency risk analysis and monitoring and early warning method for tower tilt state is applied to a system composed of end-side monitoring devices and a background server, characterized by the following steps: S1, acquire multi-modal data including inclination or acceleration, wind speed and direction, temperature and humidity, positioning and device state, and perform unified time alignment and quality control on the data; S2, based on the line topological relationship, geographical proximity and wind field elements, construct a time-varying adjacency matrix and normalize it to obtain a graph model for spatial consistency analysis; S3, consider the inclination or trend of each tower as a graph signal, calculate the spatial consistency measure, and link it with the single rod time sequence detection and drift detection results; S4, output graded warning according to multi-condition strategy, and adaptively update the threshold, time window and model weight when drift is detected.

[0005] Further, in order to better implement the present application, in S2, the time-varying adjacency matrix is obtained by weighted summation of topological adjacency weight, geographical distance kernel weight and wind field coupling weight with non-negative coefficients, wherein the wind field coupling weight is determined according to the directional cosine of wind direction and the direction of rod connection line and the weight function which increases monotonically with wind speed.

[0006] Further, in order to better implement the present application, in S2, the adjacency matrix is symmetrically normalized or standardized to obtain a normalized adjacency or graph Laplacian; in S3, the graph smoothing energy and adjacent rod difference are used as the index threshold of spatial consistency measure.

[0007] Further, in order to better realize the present application, the S3 includes weighted regression of the inclination sequence in a sliding window to estimate the short-term trend and the change rate, and a dynamic threshold and a minimum duration window are adopted to determine the single-pole abnormal candidate, which is adaptive to the wind speed stage, temperature and humidity, and time period.

[0008] Further, in order to better realize the present application, the S3 adopts the exponential weighted moving average and the adaptive sliding window detection, when the mean deviation or distribution drift is detected, the adaptive update of the threshold, the time window and the adjacent weight parameter in the S4 is triggered and the version is archived.

[0009] Further, in order to better realize the present application, the S4 includes calculating the spatial autocorrelation statistic to identify the clustering anomaly, and combining the threshold results of the graph smoothing energy and the adjacent pole difference for segment-level or regional-level alarm promotion.

[0010] Further, in order to better realize the present application, the S4 adopts the wind direction and wind speed gating, the data quality gating and the duration window gating in the hierarchical early warning, and outputs according to the level rules of attention, early warning and alarm.

[0011] A tower tilt monitoring and early warning system for realizing the topological consistency risk analysis and monitoring and early warning method of the tower tilt state as described above, comprising an end-side monitoring device and a background server, characterized in that: The end-side monitoring device is used for collecting multi-modal data and reporting, and the background server is configured with a time alignment and quality control module, a space-time topological modeling module, a single-pole time sequence and drift detection module, a spatial consistency analysis module and a pre-warning decision module to execute the above steps.

[0012] A computer readable storage medium having a computer program stored thereon, the program realizing the topological consistency risk analysis and monitoring and early warning method of the tower tilt state as described above when executed by a processor.

[0013] The present application has the following beneficial effects: The line topology, geographical proximity and wind field coupling are included in the time-varying graph model, the segment-level regional clustering anomaly is identified through the graph smoothing energy, the adjacent pole difference and the spatial autocorrelation criterion, and the single-pole isolated false alarm is significantly reduced. The short-term trend captures the suddenness, the EWMA control chart and the ADWIN adaptive sliding window identify the slow drift, and once the drift is established, the threshold, the time window and the weight are automatically updated, and the long-term stability and the robustness are improved. The wind direction and wind speed gating, the duration window and the data quality gating are introduced, the single-pole and group criteria are combined to output the early warning, and good interpretability and traceability are achieved. The data alignment, the graph modeling, the detection and the decision are modularized, the centralized or edge-cloud collaborative deployment is supported, the parameters have default suggestions and can be updated online. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 The overall flowchart of the present application is shown. DETAILED DESCRIPTION

[0015] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings can be arranged and designed in various different configurations.

[0016] Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed application, but only represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0017] Figure 1 For a specific embodiment of the present application, the embodiment is a topological consistency risk analysis and monitoring and early warning method for tower inclination state, and the specific technical route is as follows: 1. Spatio-temporal topology modeling and multi-modal data alignment The background system first receives the inclination angle or acceleration measurement value, inclination angle trend, change rate, wind speed and direction, temperature and humidity, battery state, solar input condition, and device identification (pole ID) and its geographic coordinates (longitude and latitude + elevation, height) and installation direction uploaded from the device end. The backend performs timestamp verification and standardization processing on these data, resamples them into a unified time window (such as every 1 minute or 5 minutes), and performs interpolation or rejection processing on missing data and abnormal jump values, generating a multi-modal feature vector of each pole at any time .

[0018] Then, the background system constructs a graph structure using the position of each kind and the line connection relationship: the node represents the pole body, and the edge includes the line (topology) adjacency and the geographical adjacency (threshold or kernel function weighting according to the distance between poles).

[0019] Among them, the purpose of constructing the adjacency matrix A is to quantitatively reflect the spatial, topological factors such as the position between the tower, the line topology and the wind field, so as to calculate the spatial consistency and abnormal deviation in the graph signal processing stage. The adjacency matrix A is a symmetric matrix, and for any two poles , the element (not diagonal) is calculated by a combination of multiple sub-weights: Among them, For topological adjacency weight, set 1 if bar i and bar j are in the same line / feeder and are directly adjacent, otherwise set 0; If the geographical distance is less than a certain maximum distance threshold , otherwise set 0; Where is the current wind direction, is the azimuth angle of the line connecting the two bars, is the wind speed, is a weighting function or coefficient that increases with wind speed; the coefficient is a non-negative weighting coefficient used to adjust the influence of different types of adjacent sub-weights.

[0020] In order to ensure the numerical stability of the graph structure, A is often symmetrically standardized or normalized, for example, defined by the degree matrix D Then the normalized adjacency matrix can be calculated: Then form the graph Laplacian matrix: As a mathematical expression of spatial consistency / smoothing prior.

[0021] For the inclination or its trend, it can be regarded as the graph signal: Where is the inclination value or trend of bar i at time t.

[0022] At this point, the system has time-aligned multi-modal features as input, and with the help of spatial and topological graph construction adjacency relationship and smoothing measure, it can provide basic input and prior for subsequent trend prediction, anomaly detection and group consistency analysis.

[0023] 2. Single bar time series detection After completing the spatio-temporal topology modeling and multi-modal alignment, the background system will dynamically monitor the trend and abnormal behavior of each bar body, and detect long-term drift and mutation to timely alarm.

[0024] For short-term trends, the background first estimates the short-term trend of each bar i based on its inclination or order trend sequence Using the sliding window method, such as linear regression or least squares straight line fitting in the last time points: Then calculate the inclination rate (rate): If the two values exceed a static / dynamic threshold (determined by pre-set or adaptive according to wind speed level, ambient temperature, etc.), and the exceeding lasts longer than a pre-set time window , it will be marked as "abnormal candidate".

[0025] Meanwhile, to detect long-term drift and concept drift, the background will apply drift detection algorithms such as ADaptive WINdowing (ADWIN) and EWMA control chart. Among them, EWMA is used to monitor the mean deviation of the inclination sequence or trend sequence: let Combined with the mean and standard deviation of the historical static stable segment, if then trigger drift detection (slow change or offset).

[0026] ADWIN maintains a dynamic length window , which detects the distribution drift when it detects that the old sub-window and the new sub-window have significant differences in the average value. That is, find is divided into two parts and , so that: where is the threshold related to the confidence parameter . Such detection can automatically adapt to different drift rates and data fluctuation amplitudes.

[0027] Short-term trend and rate determination is responsible for quickly capturing mutations (such as sudden inclination jumps / structure damage sudden events), and drift detection is responsible for capturing slow-changing trends (such as foundation settlement / material fatigue / environmental temperature long-term effects). When the drift detection algorithm finds that the distribution changes or the trend deviates, the system automatically triggers threshold update or sliding window adjustment.

[0028] 3. Multi-rod group detection The background algorithm is based on node-level trend and drift detection, and also analyzes the behavior of rod groups in space / topology structure to determine whether a whole line or multiple adjacent rods tend to be abnormal at the same time, thereby improving the alarm level and reducing single rod false positives.

[0029] The background system will regard all the tower inclination values or trends at each moment as a graph signal , and the adjacency matrix A and the graph Laplacian (or normalized version) are used for spatial consistency measurement. Calculate the graph smoothing energy: If exceeds a certain threshold , it indicates that the overall deviation between some nodes and their neighbors is large, and there may be a group anomaly. At the same time, the difference index of each rod and its neighboring rods is calculated: If the difference index of multiple adjacent rods is greater than a certain threshold (for example, the average difference of adjacent rods exceeds the average value of the static stable section + several standard deviation multiples), it is determined to be a section-level anomaly.

[0030] The background can also use spatial autocorrelation statistical indicators such as Moranisl or local LISA (Local Indicators of Spatial Autocorrelation) to determine the spatial distribution characteristics of the anomaly cluster. For example, the calculation formula of Moranisl is: Wherein, When I is significantly higher than zero (according to statistical significance test), it indicates that the inclination trend is spatially clustered (multiple adjacent rods are simultaneously high or low), which is a signal of group anomaly.

[0031] Finally, it should be pointed out that the above embodiments are only used to illustrate the technical solutions of the present application and not to limit it. Other modifications or equivalent replacements to the technical solutions of the present application made by those skilled in the art should be covered within the scope of the claims of the present application as long as they do not deviate from the spirit and scope of the technical solutions of the present application.

Claims

1. A method for topology consistency risk analysis and monitoring early warning of tower tilting state, applied to a system consisting of end-side monitoring devices and a back-end server, characterized in that, Includes the following steps: S1 acquires multimodal data including tilt angle or acceleration, wind speed and direction, temperature and humidity, positioning and equipment status, and performs unified time alignment and quality control on the data. S2, based on the line topology, geographical proximity and wind field elements, a time-varying adjacency matrix is ​​constructed and normalized to obtain a graph model for spatial consistency analysis; S3 treats the tilt angle or trend of each tower as a graphical signal, calculates the spatial consistency metric, and links it with the results of single-tower timing detection and drift detection; S4 outputs hierarchical warnings based on a multi-condition strategy, and adaptively updates the threshold, time window, and model weights when drift is detected.

2. The method for topological consistency risk analysis and monitoring and early warning of tower tilt state according to claim 1, characterized in that: In S2, the time-varying adjacency matrix is ​​obtained by weighting the topological adjacency weight, the geographical distance kernel weight, and the wind field coupling weight with non-negative coefficients. The wind field coupling weight is determined based on the direction cosine of the wind direction and the orientation of the line connecting the poles, as well as a weight function that increases monotonically with wind speed.

3. The method for topological consistency risk analysis and monitoring and early warning of tower tilt state according to claim 1, characterized in that: In S2, the adjacency matrix is ​​symmetrically normalized or standardized to obtain normalized adjacency or graph Laplace. In S3, the graph smoothing energy and the difference between adjacent rods are used as the threshold index for measuring spatial consistency.

4. The method for topological consistency risk analysis and monitoring and early warning of tower tilt state according to claim 1, characterized in that: The S3 includes performing weighted regression on the tilt angle sequence within a sliding window to estimate short-term trends and rates of change, and using a dynamic threshold and minimum duration window that are adaptive to wind speed, temperature, humidity and time period to determine single-pole anomaly candidates.

5. The method for topological consistency risk analysis and monitoring and early warning of tower tilt state according to claim 1, characterized in that: In S3, exponentially weighted moving average and adaptive sliding window detection are used. When mean deviation or distribution drift is detected, the threshold, time window and adjacent weight parameters in S4 are adaptively updated and versioned.

6. The method for topological consistency risk analysis and monitoring and early warning of tower tilt state according to claim 1, characterized in that: S4 includes calculating spatial autocorrelation statistics to identify clustered anomalies, and combining them with the threshold results of graph smoothing energy and differences between adjacent bars for segment-level or region-level alarm enhancement.

7. The method for topological consistency risk analysis and monitoring and early warning of tower tilt state according to claim 1, characterized in that: The S4-based tiered early warning system employs wind direction and speed gating, data quality gating, and duration window gating, and outputs data according to the rules for attention, early warning, and alarm levels.

8. A tower tilt monitoring and early warning system for implementing the topology consistency risk analysis and monitoring and early warning method for the tower tilt state according to any one of claims 1-7, comprising an end-side monitoring device and a back-end server, characterized in that: The end-side monitoring device is used to collect and report multimodal data. The back-end server is configured with a time alignment and quality control module, a spatiotemporal topology modeling module, a single-pole timing and drift detection module, a spatial consistency analysis module, and an early warning decision module to execute the steps described in claims 1-7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the method for topological consistency risk analysis and monitoring and early warning of the tower tilt state as described in any one of claims 1-7.

Images