CN-121388952-B - Intelligent transformer monitoring and fault diagnosis system and method based on multi-source data fusion

CN121388952BCN 121388952 BCN121388952 BCN 121388952BCN-121388952-B

Abstract

The invention is suitable for the technical field of power equipment monitoring, and provides a transformer intelligent monitoring and fault diagnosis system and method based on multi-source data fusion, wherein a difference mode reference library is generated according to seasons and load factor dimensions by integrating historical working condition data; the method comprises the steps of collecting real-time data, comparing the real-time data with reference data, outputting abnormal information, constructing a correlation map based on historical interaction data, calling the same-dimension monitoring data of adjacent equipment, comparing the correlation data with the reference data, and generating a fault report or an optimized reference library. According to the invention, through multidimensional reference setting, associated equipment collaborative analysis and a dynamic optimization mechanism, the accuracy and adaptability of fault diagnosis are improved, and the intelligent monitoring and diagnosis of the transformer are realized.

Inventors

QIAO XIAOQIANG
DING SHUHAI
WANG FEI
LIN ZHI

Assignees

国能四川天明发电有限公司
南京电研电力自动化股份有限公司

Dates

Publication Date: 20260512
Application Date: 20251224

Claims (5)

1. Transformer intelligent monitoring and fault diagnosis system based on multisource data fusion, which is characterized by comprising: the difference mode reference library generation module is used for integrating the historical working condition data of the transformer, dividing data subsets according to the seasonal dimension and the load factor dimension, extracting the normal fluctuation range of the monitoring parameters under each dimension combination and generating a difference mode reference library; the anomaly detection module is used for collecting real-time monitoring data of the target transformer, converting the real-time monitoring data into real-time data vectors, matching the reference data corresponding to the season dimension and the load factor dimension in the difference mode reference library, calculating the deviation degree of the real-time data vectors and the reference data, and outputting the anomaly equipment identification and the anomaly characteristic information when the deviation degree exceeds a set deviation threshold value; The association map construction module is used for constructing a transformer substation equipment association map based on transformer historical operation interaction data, positioning adjacent equipment sets by taking a transformer corresponding to the abnormal equipment identifier as a starting node, and calling monitoring data of the adjacent equipment sets in the same seasonal dimension and load factor dimension as abnormal characteristic information to be used as association monitoring data; The fault positioning and optimizing module is used for converting the association monitoring data into association data vectors, comparing the deviation states of the association data vectors and the reference data of corresponding dimensions in the difference mode reference library, generating a fault positioning report if association abnormality occurs in the adjacent equipment set, and starting the dimension weight optimization of the difference mode reference library if the association abnormality occurs in the adjacent equipment set; The generating the difference mode reference library specifically comprises the following steps: preprocessing integrated transformer historical working condition data, wherein the transformer historical working condition data comprises historical fault data and historical monitoring parameters; dividing a seasonal dimension and a load factor dimension respectively aiming at the historical monitoring parameters, and generating a seasonal dimension data subset and a load factor dimension data subset by cross combination; the monitoring parameters in the seasonal dimension data subset and the load factor dimension data subset are calculated to form a normal fluctuation interval, reverse verification is conducted by combining historical fault data, a reference fluctuation threshold under an extreme working condition is automatically corrected, and a difference mode reference library with dimension labels is generated; The construction of the transformer substation equipment association map specifically comprises the following steps: Aiming at the initial abnormality based on the historical working condition data of the transformer, extracting interaction data among transformers, and constructing a transformer substation equipment association map taking the transformers as nodes and the association strength as side weights; Taking a node corresponding to the abnormal equipment identifier as a center, screening a node with the top 5 side weight ranking in a transformer substation equipment association map as an adjacent transformer set, and marking the association type of each transformer; According to the time stamp in the abnormal characteristic information, identifying the seasonal dimension and the load factor dimension when the transformer is abnormal, wherein the initial abnormality is judged to occur, automatically calling the historical monitoring parameters and the real-time monitoring parameters of the adjacent transformer sets under the same dimension, and combining the historical monitoring parameters and the real-time monitoring parameters into associated monitoring data; The correlation strength is calculated by the following formula: ; Wherein, the Representation transformer And transformer The association strength between the two, namely the weight of the edge in the association graph, Is a transformer And transformer If the bus is directly shared, 1 is adopted, if the bus is indirectly connected, 0.5 is adopted, if the bus is not electrically connected, 0 is adopted, Is a transformer And transformer The physical distance between the two plates is set to be equal, For the frequency of co-occurrence of historical faults, i.e. transformers in the past when faults occur And transformer The ratio of the number of simultaneous anomalies to the total number of faults, Is a weight coefficient; The association map is expressed as: ; Is that Is a matrix of the (c) in the matrix, For the total number of transformers in the substation, Is the first in the matrix Line 1 Column elements, representing transformers And transformer The strength of the association between them; the comparing the deviation state of the associated data vector and the reference data of the corresponding dimension in the difference mode reference library specifically comprises the following steps: converting the associated monitoring data into associated data vectors, and calculating the deviation degree of each adjacent transformer associated data vector and corresponding dimension reference data; If more than 30% of transformers in the adjacent transformer sets all recognize that the deviation degree exceeds the abnormality judgment threshold, the adjacent transformers are defined as associated abnormalities, the fault conduction paths of the transformers judged to be initially abnormal are judged by combining the edge weight of the associated graphs, and a fault positioning report comprising a fault source and an influence range is automatically generated; If the adjacent transformers do not recognize the association abnormality, defining a judging result of the transformer judged to be the initial abnormality as misjudgment, analyzing the contribution degree of the seasonal dimension and the load factor dimension to the misjudgment, recognizing the key influence dimension, proportionally increasing the weight value of the key influence according to the contribution degree, and retraining the reference fluctuation threshold value of the seasonal dimension data subset and the load factor dimension data subset in the difference mode reference library.
2. The system of claim 1, wherein the method for implementing the intelligent transformer monitoring and fault diagnosis system based on multi-source data fusion comprises: integrating historical working condition data of the transformer, dividing data subsets according to seasonal dimensions and load factor dimensions, extracting normal fluctuation ranges of monitoring parameters under each dimension combination, and generating a difference mode reference library; Collecting real-time monitoring data of a target transformer, converting the real-time monitoring data into real-time data vectors, matching reference data corresponding to season dimensions and load factor dimensions in the difference mode reference library, calculating the deviation degree of the real-time data vectors and the reference data, and outputting abnormal equipment identification and abnormal characteristic information when the deviation degree exceeds a set deviation threshold; constructing a transformer substation equipment association map based on transformer historical operation interaction data, positioning adjacent equipment sets by taking a transformer corresponding to the abnormal equipment identifier as a starting node, and calling monitoring data of the adjacent equipment sets in the same season dimension and load rate dimension as abnormal characteristic information to be used as association monitoring data; And converting the association monitoring data into association data vectors, comparing the deviation states of the association data vectors and the reference data of corresponding dimensions in the difference mode reference library, generating a fault positioning report if association abnormality occurs in the adjacent equipment set, and otherwise, starting the dimension weight optimization of the difference mode reference library.
3. The system according to claim 1, wherein the outputting of the abnormal device identification and the abnormal feature information when the deviation exceeds a set deviation threshold value specifically includes: The real-time monitoring parameters acquired in real time are aggregated into time sequence data according to preset time intervals, and the time sequence data are converted into real-time data vectors consistent with the format of a differential mode reference library; Extracting a season label and a load rate label of a real-time data vector, matching reference data of corresponding dimensions in a difference mode reference library, and calculating a deviation degree; An abnormality determination threshold is set based on the history fault data, the calculated degree of deviation is compared with the abnormality determination threshold, if the degree of deviation exceeds the abnormality determination threshold, the initial abnormality is determined, and abnormality characteristic information including a transformer ID, an abnormality monitoring parameter, and an occurrence time is generated and outputted.
4. A system according to claim 3, wherein the matching difference pattern reference library matches reference data of corresponding dimensions and calculates a degree of deviation, in particular: ; Wherein, the Representing the degree of deviation of the real-time data vector from the reference data, Refer to the first of the real-time data vectors of the target transformer The normalized value of the individual monitored parameter(s), Is under the corresponding season dimension and load factor dimension in the difference mode reference library The reference average value of each of the monitored parameters, To monitor the number of parameters.
5. The system according to claim 1, wherein the analyzing the contribution degree of the seasonal dimension and the load factor dimension to the erroneous determination is specifically: ; ; Wherein, the The contribution degree of the seasonal dimension to the erroneous judgment is calculated, The contribution degree of the load factor dimension to the erroneous judgment is calculated, , In the erroneous judgment, the deviation degree of the target transformer in the current season dimension, The average deviation of the historical normal state in the seasonal dimension, Is the deviation degree of the target transformer in the dimension of the current load rate in the erroneous judgment, As the average deviation of the historical normal state in the load factor dimension, Represent the first Degree of deviation of target transformer in each dimension, when In the time-course of which the first and second contact surfaces, I.e. degree of deviation in seasonal dimension, when In the time-course of which the first and second contact surfaces, I.e., the degree of deviation in the load factor dimension, Represent the first Average degree of deviation of historical normal state in each dimension when In the time-course of which the first and second contact surfaces, I.e. degree of deviation in seasonal dimension, when In the time-course of which the first and second contact surfaces, ; To participate in the total number of dimensions for contribution analysis, Namely a seasonal dimension and a load factor dimension.

Description

Intelligent transformer monitoring and fault diagnosis system and method based on multi-source data fusion Technical Field The invention belongs to the technical field of power equipment monitoring, and particularly relates to an intelligent transformer monitoring and fault diagnosis system and method based on multi-source data fusion. Background The transformer is used as core equipment of the power system, and the running state of the transformer is directly related to the safety and stability of the power grid. With the development of smart grid technology, transformer monitoring and fault diagnosis have evolved from traditional manual inspection, single parameter monitoring to intelligent, multi-source data driving. Currently, the multi-source data fusion technology is widely applied to equipment state evaluation, and by integrating historical operation data, real-time monitoring parameters, environment data and the like, timeliness and accuracy of fault diagnosis are improved, so that the multi-source data fusion technology becomes an important development trend of intelligent management of power equipment. Aiming at the transformer monitoring technology, the prior art has the following problems that the dimension division of a reference library is single, the cross influence of seasons and load rates is not fully considered, so that the definition of normal parameter ranges under different working conditions is fuzzy, misjudgment is easy to occur, the equipment association analysis only depends on a simple connection relation, the association strength is not quantized, the fault conduction path is difficult to accurately position, the reference library lacks a dynamic optimization mechanism, the dimension weight cannot be adjusted according to misjudgment cases in actual operation, and the adaptability is insufficient, so that the intelligent diagnosis requirement under complex working conditions is difficult to meet. Disclosure of Invention The invention aims to provide a transformer intelligent monitoring and fault diagnosis system and method based on multi-source data fusion, and aims to solve the technical problems in the prior art determined in the background art. The invention is realized in such a way that the intelligent monitoring and fault diagnosis system of the transformer based on multi-source data fusion comprises: the difference mode reference library generation module is used for integrating the historical working condition data of the transformer, dividing data subsets according to the seasonal dimension and the load factor dimension, extracting the normal fluctuation range of the monitoring parameters under each dimension combination and generating a difference mode reference library; the anomaly detection module is used for collecting real-time monitoring data of the target transformer, converting the real-time monitoring data into real-time data vectors, matching the reference data corresponding to the season dimension and the load factor dimension in the difference mode reference library, calculating the deviation degree of the real-time data vectors and the reference data, and outputting the anomaly equipment identification and the anomaly characteristic information when the deviation degree exceeds a set deviation threshold value; The association map construction module is used for constructing a transformer substation equipment association map based on transformer historical operation interaction data, positioning adjacent equipment sets by taking a transformer corresponding to the abnormal equipment identifier as a starting node, and calling monitoring data of the adjacent equipment sets in the same seasonal dimension and load factor dimension as abnormal characteristic information to be used as association monitoring data; The fault positioning and optimizing module is used for converting the association monitoring data into association data vectors, comparing the deviation states of the association data vectors and the reference data of corresponding dimensions in the difference mode reference library, generating a fault positioning report if association abnormality occurs in the adjacent equipment set, and starting the dimension weight optimization of the difference mode reference library if the association abnormality occurs. Another object of the present invention is to provide a transformer intelligent monitoring and fault diagnosis method based on multi-source data fusion, the method comprising: integrating historical working condition data of the transformer, dividing data subsets according to seasonal dimensions and load factor dimensions, extracting normal fluctuation ranges of monitoring parameters under each dimension combination, and generating a difference mode reference library; Collecting real-time monitoring data of a target transformer, converting the real-time monitoring data into real-time data vectors, matching reference data corresponding to season dimens