CN-114328075-B - Intelligent power distribution room sensor multidimensional data fusion abnormal event detection method, system and computer readable storage medium

CN114328075BCN 114328075 BCN114328075 BCN 114328075BCN-114328075-B

Abstract

The invention discloses a detection method, a detection system and a computer readable storage medium for multi-dimensional data fusion abnormal events of an intelligent power distribution room sensor, wherein the method comprises the steps of S1, acquiring normal multi-dimensional sensing data of the power distribution room, establishing a normal sample database, S2, preprocessing multi-dimensional data in the normal sample database, establishing a multi-dimensional data correlation characteristic model, S3, preprocessing real-time operation data of the power distribution room, inputting the preprocessed multi-dimensional data correlation characteristic model, analyzing the real-time operation data, judging whether the real-time operation data is abnormal, and if the data is abnormal, turning to step S4, if the data is normal, ending the current detection flow, and detecting the data at the next moment, S4, comparing the abnormal data with a preset abnormal event library, if the abnormal data belongs to an abnormal event, outputting the abnormal event type, and if the abnormal data does not belong to the abnormal event type, updating the model. The invention fully utilizes the relevance of the multidimensional data and improves the utilization rate of the data and the detection efficiency of the abnormal event.

Inventors

ZHANG MIN
FANG JIAN
WANG YONG
HAO FANGZHOU
YANG FAN
HE JIAXING
LIN XIANG
YIN KUANG

Assignees

广东电网有限责任公司广州供电局
广东电网有限责任公司广州供电局

Dates

Publication Date: 20260421
Application Date: 20210909
Priority Date: 20210909

Claims (7)

1. The method for detecting the abnormal event of the multidimensional data fusion of the sensor of the intelligent power distribution room is characterized by comprising the following steps: s1, acquiring normal multidimensional sensing data of a power distribution room, and establishing a normal sample database; s2, preprocessing multidimensional data in the normal sample database, and establishing a multidimensional data correlation characteristic model; S3, preprocessing real-time operation data of the power distribution room, inputting the preprocessed real-time operation data into the multidimensional data correlation feature model, analyzing the real-time operation data, judging whether the real-time operation data is abnormal, if so, turning to the step S4, and if so, ending the current detection flow and detecting the data at the next moment; s4, comparing and judging the abnormal data with a preset abnormal event library, outputting the judged abnormal event type if the abnormal data belongs to the abnormal event, and updating the multidimensional data correlation characteristic model if the abnormal data does not belong to the abnormal event; The construction process of the multidimensional data correlation feature model in the step S2 is as follows: Defining fixed time window lengths L by the multidimensional data, and calculating pearson correlation coefficients among multidimensional variables in each time window length L; establishing a correlation coefficient matrix by using the calculated correlation coefficient, and determining each correlation coefficient threshold value in normal data; establishing a multidimensional data correlation characteristic model according to a correlation coefficient matrix and the correlation coefficient threshold; The analyzing the real-time operation data in step S3, and judging whether the real-time operation data is abnormal specifically includes: sequentially calculating the correlation coefficient of the multidimensional data in the window by adopting a sliding time window method for the real-time operation data; if the correlation coefficient between the multidimensional data is larger than the correlation coefficient threshold value in the model, the correlation coefficient is an abnormal data point; and if the correlation coefficient among the multidimensional data is smaller than or equal to the correlation coefficient threshold value in the model, judging the next time window.
2. The method for detecting the multidimensional data fusion abnormal event of the sensor of the intelligent power distribution room according to claim 1, wherein the preset abnormal event library in the step S4 is established according to expert experience, and the abnormal event library comprises event types including abnormality of detected equipment and abnormality of detected equipment.
3. The method for detecting the multidimensional data fusion abnormal event of the intelligent power distribution room sensor according to claim 1 is characterized in that step S4 further comprises the steps of manually judging whether abnormal data points are abnormal data points or not if abnormal data do not exist in a corresponding abnormal event library, defining the abnormal data points as new abnormal events to be stored in the abnormal event library if the abnormal data points are abnormal data points, and updating a multidimensional data correlation characteristic model if the abnormal data points are not abnormal data points.
4. The method for detecting abnormal event fusion of multidimensional data of intelligent power distribution room sensor according to claim 3, wherein if abnormal data is not abnormal data point in manual judgment, the specific process of updating the multidimensional data correlation characteristic model comprises the steps of adding a time sequence window of abnormal data judgment error into a monitoring normal database and updating a normal data sample database; Recalculating a correlation coefficient among the multidimensional samples; the correlation coefficient threshold is redetermined.
5. The system for detecting the multi-dimensional data fusion abnormal event of the intelligent power distribution room sensor is characterized by comprising a data acquisition module, a data processing module, a modeling module, an online research and judgment module, an abnormal event library and an evaluation module, wherein the data acquisition module is used for acquiring multi-dimensional sensing data of the power distribution room, the data processing module is used for processing the acquired multi-dimensional sensing data of the power distribution room, the modeling module is used for realizing establishment and update of a multi-dimensional data correlation characteristic model, the online research and judgment module is used for detecting abnormal data points and judging abnormal event types online, and the abnormal event library comprises abnormal events added during initial construction and abnormal event model update, and the evaluation module is used for evaluating accuracy of a multi-dimensional data correlation characteristic model and abnormal event model judgment accuracy.
6. The system for detecting abnormal event fusion of multidimensional data of intelligent power distribution room sensor according to claim 5, wherein the data processing module processes data in the manners of classification, cleaning, conversion, time sequence alignment, standardization and integration.
7. A computer readable storage medium, wherein the computer readable storage medium includes a program for detecting an abnormal event of multi-dimensional data fusion of an intelligent power distribution room sensor, and when the program for detecting the abnormal event of multi-dimensional data fusion of the intelligent power distribution room sensor is executed by a processor, the steps of the method for detecting the abnormal event of multi-dimensional data fusion of the intelligent power distribution room sensor are implemented.

Description

Intelligent power distribution room sensor multidimensional data fusion abnormal event detection method, system and computer readable storage medium Technical Field The invention relates to the technical field of intelligent detection of power distribution rooms, in particular to a method and a system for detecting multidimensional data fusion abnormal events of sensors of intelligent power distribution rooms and a computer readable storage medium. Background The power distribution room is an indispensable important part in the power grid system, and meanwhile, the power distribution room is widely distributed and large in quantity in the power grid, so that the management difficulty of the power distribution room is increased. The manual inspection is long in time consumption, low in efficiency and large in workload. Therefore, the intelligent online monitoring system and the inspection robot have been more developed and applied in the power distribution room. However, the data collected and obtained by the online monitoring system are rarely and effectively utilized, a large amount of multidimensional monitoring data only form data which are not related to each other, or only simply set normal upper and lower limits on the measured data, so that the related information among multidimensional data of a plurality of sensors is rarely and effectively utilized, and abnormal phenomena in a power distribution room are difficult to discover in time. In the prior art, the Chinese invention patent with publication number CN110690763A discloses an intelligent monitoring device of an electric power system and a monitoring method thereof in 2021, 1 and 14 days, wherein the intelligent monitoring device comprises a signal acquisition front end, a signal processing system, a central main control system, a data storage unit, a network unit and a power management unit, wherein the signal acquisition front end comprises a split type local discharge sensor, an environment humidity sensor and an environment temperature sensor. The scheme realizes the acquisition of various sensor data, but the acquired data are not correlated or fused, and the multidimensional data cannot be effectively utilized. Disclosure of Invention The invention provides a method, a system and a computer readable storage medium for detecting abnormal event fusion of multidimensional data of an intelligent power distribution room sensor, which are used for overcoming the defects that the existing power distribution room detection does not effectively utilize multidimensional monitoring data, does not realize association of multidimensional data, and has low data utilization rate and detection efficiency. The primary purpose of the invention is to solve the technical problems, and the technical scheme of the invention is as follows: The invention provides a method for detecting abnormal event of sensor multidimensional data fusion of an intelligent power distribution room, which comprises the following steps: s1, acquiring normal multidimensional sensing data of a power distribution room, and establishing a normal sample database; s2, preprocessing multidimensional data in the normal sample database, and establishing a multidimensional data correlation characteristic model; S3, preprocessing real-time operation data of the power distribution room, inputting the preprocessed real-time operation data into the multidimensional data correlation feature model, analyzing the real-time operation data, judging whether the real-time operation data is abnormal, if so, turning to the step S4, and if so, ending the current detection flow and detecting the data at the next moment; And S4, comparing and judging the abnormal data with a preset abnormal event library, outputting the judged abnormal event type if the abnormal data belongs to the abnormal event, and updating the multidimensional data correlation characteristic model if the abnormal data does not belong to the abnormal event. Further, the construction process of the multidimensional data correlation feature model in step S2 is as follows: defining fixed time window lengths L by the multidimensional data, and calculating correlation coefficients among multidimensional variables in each time window length L; establishing a correlation coefficient matrix by using the calculated correlation coefficient, and determining each correlation coefficient threshold value in normal data; And establishing a multidimensional data correlation characteristic model according to the correlation coefficient matrix and the correlation coefficient threshold value. Further, the correlation coefficient is pearson correlation coefficient. Further, in step S3, the real-time operation data is analyzed, and whether the real-time operation data is abnormal or not is determined specifically as follows: sequentially calculating the correlation coefficient of the multidimensional data in the window by adopting a sliding time window metho