CN-122020367-A - Multi-mode edge fusion power fault diagnosis method and system

CN122020367ACN 122020367 ACN122020367 ACN 122020367ACN-122020367-A

Abstract

The invention discloses a power failure diagnosis method and system with multi-mode edge fusion, and relates to the technical field of power failure diagnosis. The method comprises the steps of establishing a power equipment topology in a preset area, wherein the power equipment topology comprises a plurality of power equipment, respectively collecting electrical quantity parameters, non-electrical quantity parameters and peripheral equipment response parameters of each power equipment, calculating multi-type fault probabilities of each power equipment, generating a plurality of fault probability distributions according to the multi-type fault probabilities of each power equipment, and determining target fault equipment and target fault types based on distribution characteristics of the plurality of fault probability distributions. The invention effectively improves the accuracy and the anti-interference capability of the power fault diagnosis.

Inventors

MA CHUN
XU SEN
FENG CHUNXIA
YANG LIJIN
GUO WEILIANG
LI MENGZHE
YI XIAOTING
DONG ZHIYONG
WANG CHENYU
CHAI YUNFENG
CHENG YITAO
KANG WEI
WANG KAI
LI DING
GUO YU
MA YONGZHAN
DUAN YUFENG
WEI XUEFENG
ZHANG TIAN
MA SHAOKUN
WEN YUE

Assignees

国网山西省电力有限公司晋城供电分公司

Dates

Publication Date: 20260512
Application Date: 20260115

Claims (10)

1. The power failure diagnosis method for multi-mode edge fusion is characterized by comprising the following steps of: Establishing a power equipment topology in a preset area, wherein the power equipment topology comprises a plurality of power equipment; respectively acquiring an electrical quantity parameter, a non-electrical quantity parameter and a peripheral equipment response parameter of each electrical equipment, and calculating the multi-type fault probability of each electrical equipment; and generating a plurality of fault probability distributions according to the multi-type fault probability of each power equipment, and determining target fault equipment and target fault types based on the distribution characteristics of the plurality of fault probability distributions.
2. The method of claim 1, wherein separately collecting electrical quantity parameters, non-electrical quantity parameters, and peripheral response parameters for each of the electrical devices, calculating a multi-type fault probability for each of the electrical devices, comprising: Determining a first power device from the plurality of power devices, extracting a first electrical quantity parameter, a first non-electrical quantity parameter, and a first peripheral response parameter of the first power device; Invoking a fault probability assessment engine bound to the first electrical device, the fault probability assessment engine comprising a first evaluator, a second evaluator, and a third evaluator, each evaluator comprising an evaluation unit of a plurality of fault types; processing the first electric quantity parameter, the first non-electric quantity parameter and the first peripheral equipment response parameter based on the first evaluator, the second evaluator and the third evaluator respectively to obtain multi-type fault probability of the first electric equipment; And acquiring the multi-type fault probability of the rest power equipment according to the mode of acquiring the multi-type fault probability of the first power equipment, so as to acquire the multi-type fault probability of each power equipment.
3. The method of claim 2, wherein processing the first electrical quantity parameter, the first non-electrical quantity parameter, and the first peripheral response parameter based on the first, second, and third estimators, respectively, results in a multi-type fault probability for the first electrical device, comprising: According to the first electrical quantity parameter, respectively acquiring a first fault probability of each fault type through each evaluation unit of the first evaluator; According to the first non-electrical quantity parameter, respectively acquiring a second fault probability of each fault type through each evaluation unit of the second evaluator; According to the response parameters of the first peripheral equipment, respectively acquiring third fault probabilities of all fault types through all evaluation units of the third evaluator; Based on the first fault probability and the third fault probability of each fault type, obtaining a first weight and a second weight of each fault type; And weighting the first fault probability and the second fault probability of each fault type through the first weight and the second weight of each fault type to obtain multi-type fault probability of the first power equipment.
4. The method of claim 2, wherein the constructing step of the failure probability assessment engine bound to the first power device comprises: acquiring a device history monitoring record of the first power device, wherein the device history monitoring record comprises a plurality of groups of history monitoring data, and each group of history monitoring data comprises a history electric quantity parameter, a history non-electric quantity parameter, a history peripheral device response parameter and a corresponding fault identification parameter; based on multiple groups of historical monitoring data, a sample electric quantity parameter set, a sample non-electric quantity parameter set and a sample peripheral equipment response parameter set are constructed; Constructing a plurality of sample fault annotation sets according to the fault identification parameters and combining the plurality of fault types, wherein the plurality of fault types are in one-to-one correspondence with the plurality of sample fault annotation sets; Constructing a first evaluator based on the sample electrical quantity parameter set and the plurality of sample fault annotation sets; constructing a second evaluator based on the sample non-electrical quantity parameter set and the plurality of sample fault annotation sets; constructing a third evaluator based on the sample peripheral response parameter set and the plurality of sample fault annotation sets; and integrating the first evaluator, the second evaluator and the third evaluator to obtain a fault probability evaluation engine of the first power equipment.
5. The method of claim 4, wherein constructing a first evaluator based on the sample electrical quantity parameter set and the plurality of sample fault annotation sets comprises: determining a first fault type from the plurality of fault types, and determining a corresponding first sample fault annotation set from a plurality of sample fault annotation sets; Constructing a plurality of first evaluation subunit architectures; respectively training the first evaluation subunit architectures to be converged based on the sample electric quantity parameter set and the first sample fault labeling set to obtain a plurality of first evaluation subunits; integrating the plurality of first evaluation subunits to obtain a first evaluation unit; training the evaluation units of the rest fault types according to a mode of acquiring a first evaluation unit of a first fault type to obtain a plurality of evaluation units; and integrating the plurality of evaluation units to obtain the first evaluator.
6. The method of claim 5, wherein obtaining, by each evaluation unit of the first evaluator, a first failure probability for each failure type, respectively, based on the first electrical quantity parameter, comprises: inputting the first electrical quantity parameter into the first evaluator; The first evaluator distributes the first electric quantity parameters to a plurality of evaluation units in the interior at the same time, and each evaluation unit processes the first electric quantity parameters according to the plurality of evaluation subunits in the interior to obtain a plurality of evaluation result sets; and acquiring a first fault probability of each fault type according to the plurality of evaluation result sets.
7. A method according to claim 3, wherein deriving the first and second weights for each fault type based on the first and third fault probabilities for each fault type comprises: Determining a first fault type from a plurality of fault types; Calculating the matching degree between the first fault probability and the third fault probability corresponding to the first fault type; Taking the matching degree as a first weight corresponding to a first fault type, and determining a second weight corresponding to the first fault type based on the first weight; And respectively acquiring the first weight and the second weight corresponding to the rest fault types according to the mode of acquiring the first weight and the second weight of the first fault type.
8. The method of claim 1, wherein generating a plurality of fault probability distributions from the plurality of types of fault probabilities for each of the electrical devices and determining a target fault device and a target fault type based on distribution characteristics of the plurality of fault probability distributions comprises: Generating a plurality of fault probability distributions according to the multi-type fault probability of each power equipment and a plurality of fault types, wherein each fault probability distribution corresponds to one fault type; And inputting the multiple fault probability distributions into a pre-trained fault positioning engine, and outputting target fault equipment and target fault types.
9. The method of claim 8, wherein the training mode of the fault localization engine comprises: Acquiring a topology history monitoring record of the power equipment topology, and constructing a sample probability distribution set according to the topology history monitoring record, wherein the sample probability distribution set comprises a plurality of groups of sample probability distribution, and each group of sample probability distribution comprises a plurality of sample fault probability distribution; Performing fault labeling on each group of sample probability distributions based on the topology history monitoring records to obtain a sample fault labeling set; And training and generating the fault positioning engine by taking the sample probability distribution set as an input characteristic and the sample fault labeling set as a supervision label.
10. A multi-modal edge-fused power failure diagnosis system for implementing the multi-modal edge-fused power failure diagnosis method of any one of claims 1-9, the system comprising: the equipment topology construction module is used for establishing an electric equipment topology in a preset area, wherein the electric equipment topology comprises a plurality of electric equipment; the fault probability calculation module is used for respectively acquiring the electrical quantity parameter, the non-electrical quantity parameter and the peripheral equipment response parameter of each electrical equipment and calculating the multi-type fault probability of each electrical equipment; The fault positioning module is used for generating a plurality of fault probability distributions according to the multi-type fault probability of each power device and determining a target fault device and a target fault type based on the distribution characteristics of the plurality of fault probability distributions.

Description

Multi-mode edge fusion power fault diagnosis method and system Technical Field The invention relates to the technical field of power fault diagnosis, in particular to a multi-mode edge fusion power fault diagnosis method and system. Background As the power system develops to a large capacity and high intelligent direction, the topology structure of the power grid becomes more and more complex, and the safe and stable operation of the power equipment becomes a core link for guaranteeing the energy supply. The existing power fault diagnosis technology is based on single electrical quantity parameters such as voltage, current, power factor and the like, and fault detection is realized through threshold judgment or simple rule matching. However, in the traditional power fault diagnosis method, the dimension of the diagnosis data is single, the interference of complex scenes such as electromagnetic interference, equipment aging and the like is difficult to deal with only depending on the electric quantity parameters, the problems of false detection and missing detection are easy to occur, and the characteristic differences of different fault types cannot be adapted by adopting fixed weights, so that the accuracy of the diagnosis result is limited, and the fault equipment and the specific fault types are difficult to accurately identify. Disclosure of Invention The invention provides a multi-mode edge fusion power failure diagnosis method and system, and aims to solve the technical problems of insufficient accuracy and interference resistance of power failure diagnosis in the prior art. In view of the above problems, the present invention provides a method and a system for diagnosing power failure by multi-modal edge fusion. In a first aspect, the present invention provides a method for diagnosing a power failure by multi-modal edge fusion, including: Establishing a power equipment topology in a preset area, wherein the power equipment topology comprises a plurality of power equipment; respectively acquiring an electrical quantity parameter, a non-electrical quantity parameter and a peripheral equipment response parameter of each electrical equipment, and calculating the multi-type fault probability of each electrical equipment; and generating a plurality of fault probability distributions according to the multi-type fault probability of each power equipment, and determining target fault equipment and target fault types based on the distribution characteristics of the plurality of fault probability distributions. In a second aspect, the present invention provides a power failure diagnosis system for multi-modal edge fusion, comprising: the equipment topology construction module is used for establishing an electric equipment topology in a preset area, wherein the electric equipment topology comprises a plurality of electric equipment; the fault probability calculation module is used for respectively acquiring the electrical quantity parameter, the non-electrical quantity parameter and the peripheral equipment response parameter of each electrical equipment and calculating the multi-type fault probability of each electrical equipment; The fault positioning module is used for generating a plurality of fault probability distributions according to the multi-type fault probability of each power device and determining a target fault device and a target fault type based on the distribution characteristics of the plurality of fault probability distributions. One or more technical schemes provided by the invention have at least the following technical effects or advantages: The invention provides a multi-mode edge fusion power fault diagnosis method and a system, which construct a fault probability assessment engine of a multi-level integrated learning architecture by constructing power equipment topology, fusion electric quantity, non-electric quantity and peripheral equipment response three-dimensional data, innovatively adopt a matching degree dynamic weight adjustment mechanism, and combine a pre-training fault positioning engine to mine topology fault propagation rules, so that full-flow intellectualization of fault diagnosis is realized. The method has the advantages that the limitation of traditional single parameter diagnosis is broken through, a solid data base is laid for fault assessment through multi-mode data cross verification, a dynamic weight adjustment mechanism intelligently distributes weights according to the matching degree of electric quantity and peripheral equipment response, the anti-interference capability and accuracy of a diagnosis result are effectively improved, a positioning engine based on a topology fault propagation rule solves the problem of fuzzy positioning in the traditional method, target fault equipment and types can be accurately identified, and a multi-level integrated learning design can adapt to complex and variable operation conditions. Finally, a complete solution integrating f