CN-121999305-A - Abnormal region detection method based on point location multi-phase atlas and storage medium

CN121999305ACN 121999305 ACN121999305 ACN 121999305ACN-121999305-A

Abstract

The application discloses an abnormal region detection method based on a point location multi-phase atlas, which comprises the steps of collecting multi-phase core data according to detection requirements, generating the point location multi-phase atlas through data registration, establishing a normal fluctuation model based on the point location multi-phase atlas, marking suspected abnormal points, calculating the point location abnormality degree through an abnormality clustering algorithm, identifying abnormal region boundaries, initially identifying an abnormal region range, constructing an energy function, combining the abnormal point location cluster and the abnormal region boundaries, constructing a diffusion region model to predict the diffusion trend of the abnormal region, determining whether the point location in the region accords with abnormal characteristics again according to the region boundaries and the diffusion trend, confirming a final abnormal region, integrating detection results, and generating an abnormal region report containing abnormal information, a visual chart and recommended measures. The method aims at solving the problems that the traditional method is difficult to accurately capture abnormal dynamic evolution, the adaptability of fusion characteristics is poor and the diffusion trend prediction is inaccurate.

Inventors

XUE HAI
YIN ZE
LIU ZIQUAN
LU YONGLING
WANG ZHEN
HU CHENGBO
HU YANJIE
LIU ZHENGYU
PAN JIANYA
YU CONGCONG

Assignees

国网江苏省电力有限公司电力科学研究院
国网江苏省电力有限公司
江苏省电力试验研究院有限公司

Dates

Publication Date: 20260508
Application Date: 20260409

Claims (8)

1. An abnormal region detection method based on a point location multi-phase atlas is characterized by comprising the following steps: acquiring multi-time-phase core data according to detection requirements, and generating a point-location multi-time-phase atlas through data registration; Establishing a normal fluctuation model based on the point location multi-phase atlas, and marking suspected abnormal point locations; calculating the point position abnormality degree through an abnormality clustering algorithm according to the suspected abnormal point positions, and identifying an abnormal point position cluster; constructing an energy function according to the abnormal point clusters, identifying the boundary of the abnormal region, and primarily identifying the range of the abnormal region; Combining the abnormal point clusters with the abnormal region boundary to construct a diffusion region model to predict the diffusion trend of the abnormal region; according to the regional boundary and the diffusion trend, determining whether the point positions in the region meet the abnormal characteristics again, and confirming the final abnormal region; And integrating the detection results to generate an abnormal region report containing abnormal information, a visual chart and suggested measures.
2. The abnormal region detection method based on the point-location multi-phase atlas according to claim 1, wherein the multi-phase core data is collected according to detection requirements, and the point-location multi-phase atlas is generated through data registration, and the abnormal region detection method comprises the following steps: acquiring multi-time-phase core data according to detection requirements, and executing space-time alignment and standardized preprocessing facing detection tasks; Based on the preprocessed multi-temporal core data, a point location multi-temporal image set is generated through fusion of feature point matching and a space-time registration algorithm.
3. The abnormal region detection method based on the point-location multi-phase atlas according to claim 1, wherein the normal fluctuation model is built based on the point-location multi-phase atlas, and the marking of the suspected abnormal point comprises: analyzing the multi-time phase diagram set sequence of each point location, and constructing a normal fluctuation model reflecting the normal change rule and fluctuation range of the point location; And comparing the latest time phase data of each point position with a dynamic reference model of the point position, and automatically identifying and marking the data points deviating from the normal fluctuation range as suspected abnormal points.
4. The abnormal region detection method based on the point location multi-phase atlas according to claim 1, wherein calculating the point location abnormality degree according to the suspected abnormal point location through an abnormality clustering algorithm, and identifying the abnormal point location cluster comprises: According to the feature data of the suspected abnormal points, the suspected abnormal points are initially grouped by adopting an abnormal clustering algorithm, and the local abnormal degree value of each point in the group to which the point belongs is calculated; And optimizing and screening out a cluster with high significance according to the local abnormality degree value of each point, and identifying the cluster as an abnormal point cluster with clear space aggregation characteristics.
5. The abnormal region detection method based on the point-location multi-phase atlas according to claim 1, wherein constructing an energy function according to an abnormal point-location cluster, identifying an abnormal region boundary, and primarily identifying an abnormal region range, comprises: Merging the statistical features and the geospatial attributes of the abnormal point clusters to construct an energy function for boundary identification; and solving the optimal segmentation boundary by minimizing an energy function, and preliminarily determining the accurate range of the abnormal region.
6. The abnormal region detection method based on the point multi-phase atlas according to claim 1, wherein the construction of the diffusion region model for predicting the diffusion trend of the abnormal region by combining the abnormal point clusters and the abnormal region boundary comprises the following steps: Analyzing the abnormal intensity gradient and external environment factors in the abnormal region, and determining diffusion driving factors and parameters; And constructing a diffusion model based on space-time geographic weighted regression, and simulating the diffusion trend and direction of the abnormal region under the action of the driving factor.
7. The abnormal region detection method based on the point-location multi-phase atlas according to claim 1, wherein determining whether the point location in the region again meets the abnormal feature according to the region boundary and the diffusion trend, and confirming the final abnormal region, comprises: feeding back the diffusion trend to an abnormality judging module, and carrying out enhanced rechecking on the point positions on the potential diffusion path; and (5) integrating the initial result, the boundary and the rechecking information, and optimizing and confirming the range and the confidence level of the final abnormal region.
8. A computer readable storage medium comprising one or more program instructions for execution by a processor of a method of anomaly region detection based on a point-wise multi-phase atlas as claimed in any one of claims 1-7.

Description

Abnormal region detection method based on point location multi-phase atlas and storage medium Technical Field The invention relates to the technical field of region detection, in particular to an abnormal region detection method based on a point location multi-phase atlas and a storage medium. Background In the fields of ecological environment monitoring, urban safety precaution and the like, abnormal region detection is a key link for guaranteeing system stability and safety. The traditional method relies on single-phase data or artificial experience to judge, and has obvious limitations. The single-phase data is difficult to capture the dynamic evolution process of the abnormality, gradual change type abnormality is easy to miss, and the manual judgment is greatly influenced by subjective factors, and has low efficiency and insufficient accuracy. With the development of technologies such as remote sensing and the internet of things, point-location multi-phase atlas is largely generated, and a data basis is provided for abnormal dynamic detection, but how to efficiently use the multi-phase atlas, accurately extract abnormal characteristics and detect abnormal areas becomes a technical problem to be solved. The method aims at solving the problems that the traditional method is difficult to accurately capture abnormal dynamic evolution, the adaptability of fusion characteristics is poor and the diffusion trend prediction is inaccurate. Disclosure of Invention The invention provides an abnormal region detection method based on a point location multi-phase atlas, which comprises the following steps: acquiring multi-time-phase core data according to detection requirements, and generating a point-location multi-time-phase atlas through data registration; Establishing a normal fluctuation model based on the point location multi-phase atlas, and marking suspected abnormal point locations; calculating the point position abnormality degree through an abnormality clustering algorithm according to the suspected abnormal point positions, and identifying an abnormal point position cluster; constructing an energy function according to the abnormal point clusters, identifying the boundary of the abnormal region, and primarily identifying the range of the abnormal region; Combining the abnormal point clusters with the abnormal region boundary to construct a diffusion region model to predict the diffusion trend of the abnormal region; according to the regional boundary and the diffusion trend, determining whether the point positions in the region meet the abnormal characteristics again, and confirming the final abnormal region; And integrating the detection results to generate an abnormal region report containing abnormal information, a visual chart and suggested measures. The abnormal region detection method based on the point location multi-phase atlas, as described above, wherein multi-phase core data is collected according to detection requirements, and the point location multi-phase atlas is generated through data registration, comprising: acquiring multi-time-phase core data according to detection requirements, and executing space-time alignment and standardized preprocessing facing detection tasks; Based on the preprocessed multi-temporal core data, a point location multi-temporal image set is generated through fusion of feature point matching and a space-time registration algorithm. The abnormal region detection method based on the point-location multi-phase atlas, as described above, wherein the normal fluctuation model is built based on the point-location multi-phase atlas, and suspected abnormal points are marked, comprising: analyzing the multi-time phase diagram set sequence of each point location, and constructing a normal fluctuation model reflecting the normal change rule and fluctuation range of the point location; And comparing the latest time phase data of each point position with a dynamic reference model of the point position, and automatically identifying and marking the data points deviating from the normal fluctuation range as suspected abnormal points. The abnormal region detection method based on the point location multi-phase atlas, as described above, wherein the point location abnormality degree is calculated according to the suspected abnormal point location through an abnormality clustering algorithm, and an abnormal point location cluster is identified, comprising: According to the feature data of the suspected abnormal points, the suspected abnormal points are initially grouped by adopting an abnormal clustering algorithm, and the local abnormal degree value of each point in the group to which the point belongs is calculated; And optimizing and screening out a cluster with high significance according to the local abnormality degree value of each point, and identifying the cluster as an abnormal point cluster with clear space aggregation characteristics. The abnormal region detection method b