CN-121979988-A - Intelligent question-answering and business guiding system oriented to multi-mode interaction in electric power field

CN121979988ACN 121979988 ACN121979988 ACN 121979988ACN-121979988-A

Abstract

The invention discloses an intelligent question-answering and service guiding system facing multi-mode interaction in the electric power field, which relates to the technical field of intelligent operation and maintenance of electric power and aims to solve the technical problems that the oil sludge content, the oil temperature and the oil level of an existing mineral-insulated oil transformer lack of collaborative correlation diagnosis, and the multi-mode interaction and the operation and maintenance guiding are disjointed, so that faults caused by the oil sludge are difficult to accurately identify and efficiently treat. The invention has the advantages of realizing fault multidimensional accurate diagnosis, multimode smooth interaction and customized operation and maintenance guidance, and improving the operation and maintenance accuracy and efficiency of the transformer.

Inventors

HUANG JINGJING
ZHANG TAO
XU WEI
GUO WEIWEI
LI YUNPENG
AN JING
HUANG XIAOHUA
MA SHAOLONG
GU XIAOYE
LI SHUXIA
WANG LELE
NA YU

Assignees

国网宁夏电力有限公司营销服务中心(国网宁夏电力有限公司计量中心)

Dates

Publication Date: 20260505
Application Date: 20260123

Claims (10)

1. The intelligent question-answering and service guiding system oriented to the multi-mode interaction in the electric power field is characterized by comprising a multi-mode data acquisition module, a data preprocessing module, a coupling diagnosis module, an intelligent guiding output module and a service linkage module; The multi-mode data acquisition module is used for acquiring multi-source data of the oil immersed transformer, wherein the multi-source data comprise infrared thermal imaging data, oil level gauge image data, oil liquid sampling image data and voice questioning data; The data preprocessing module is used for carrying out adaptation processing on the acquired multi-source data to obtain standardized data; The coupling diagnosis module calls a three-dimensional coupling model adapting to the mineral insulating oil, carries out association analysis on the standardized data and generates a fault diagnosis result; the intelligent guidance output module outputs multidimensional collaborative guidance information based on a fault diagnosis result; The service linkage module is linked with the transformer operation and maintenance ledger system to realize the closed loop connection of the diagnosis result and the operation and maintenance service.
2. The intelligent question-answering and business guiding system for multi-modal interaction in the electric power domain according to claim 1, wherein the multi-modal data acquisition module comprises an infrared acquisition unit, an image acquisition unit and a voice acquisition unit; the infrared acquisition unit is used for acquiring infrared thermal imaging data of the transformer oil tank and the winding area so as to acquire oil temperature information; the image acquisition unit comprises an oil level gauge image acquisition subunit and an oil liquid sampling image acquisition subunit, and is respectively used for acquiring an oil level gauge real-time image and an oil liquid sampled high-definition image so as to acquire oil level information and oil sludge content information; The voice acquisition unit is used for acquiring voice questioning data of operation and maintenance personnel, and the voice questioning data comprises oil sludge fault consultation and operation and maintenance operation questioning contents.
3. The intelligent question-answering and business guiding system for multi-modal interaction in the electric power domain according to claim 1, wherein the data preprocessing module comprises an image preprocessing sub-module and a voice preprocessing sub-module; the image preprocessing submodule performs directional processing on the image data of the oil level gauge, and the processing process comprises gray enhancement, oil sludge sedimentation region segmentation and oil level datum line calibration; the image preprocessing sub-module is used for preprocessing oil sampling image data to extract oil sludge distribution and content characteristic information; in order to realize the division of the oil sludge sedimentation area, the following algorithm formula is adopted to calculate the division threshold value of the oil sludge sedimentation area: ; Wherein, the A threshold is divided for the sludge settling zone, As the weight coefficient of the light-emitting diode, Is the gray average value of the background area of the oil level gauge image, And the gray average value of the suspected sludge area in the oil level gauge image.
4. The intelligent question-answering and service guiding system for multi-modal interaction in the electric power domain according to claim 3, wherein the voice preprocessing sub-module is used for performing adaptation processing on voice question data, and the processing process comprises power frequency noise cancellation and voice intention recognition; the power frequency noise cancellation is used for eliminating interference of a power frequency environment of the transformer substation on a voice signal; the voice intention recognition is realized based on a mineral insulating oil operation and maintenance intention base, and the intention base comprises intention types of oil sludge blocking, oil liquid filtering, oil change period adjustment and oil sludge cleaning operation; In order to realize power frequency noise cancellation, the following algorithm formula is adopted: ; ; ; Wherein, the Is the first The adaptive noise cancellation coefficient of the moment in time, Is the first The adaptive noise cancellation coefficient of the moment in time, In order for the convergence factor to be a factor, Is the first The noise at the moment cancels the post-error signal, In the case of a noisy speech signal, Is the first The estimated noise signal at the moment in time, Is the first The reference noise signal at the moment in time, Is the first The transposition of the time-of-day noise cancellation coefficients, To represent reference noise signals Is a binary norm of (c).
5. The intelligent question-answering and service guiding system for multi-modal interaction in the electric power field according to claim 1, wherein a three-dimensional coupling model in the coupling diagnosis module is a model adapting to mineral insulating oil, and the associated dimensions of the three-dimensional coupling model comprise an oil sludge content dimension, an oil temperature dimension and an oil level dimension; inputting standardized oil sludge content data, oil temperature data and oil level data into a three-dimensional coupling model, analyzing the association relation of the three through the model, judging whether oil level display deviation and heat dissipation efficiency reduction faults caused by oil sludge exist or not, and determining fault grades; In order to realize three-dimensional association analysis, the coupling association degree of the oil sludge content, the oil temperature and the oil level is calculated by adopting the following algorithm formula: ; ; ; ; Wherein, the For the degree of association of the three-dimensional coupling, 、、 Respectively the weight coefficients of the oil sludge content, the oil temperature and the oil level, Is the correlation coefficient of the content of the oil sludge, Is the oil temperature correlation coefficient, As the oil level correlation coefficient, Is the standard value of the normal oil sludge content of the mineral insulating oil, Is the actual measurement value of the content of the oil sludge, Is the standard value of the normal operation oil temperature of the transformer, As the measured value of the oil temperature, Is a normal oil level standard value of the transformer, As the actual measured value of the oil level, As a function of the maximum value.
6. The intelligent question-answering and business guiding system for multi-modal interaction in the electric power domain according to claim 5, wherein the coupling diagnosis module further comprises a question-answering matching sub-module, the question-answering matching sub-module is used for matching the preprocessed voice question data with a fault diagnosis result to generate matched question-answering response contents; The question-answer response content comprises fault cause explanation, fault influence explanation and preliminary treatment direction guide; To realize the matching of the voice questioning data and the fault diagnosis result, the coupling relevance is based And calculating the matching similarity of the questions and the answers, and adopting the following algorithm formula: ; Wherein, the For the matching of the questions and the answers to the similarity, Feature vectors for speech questioning data And fault diagnosis result feature vector Is arranged at the lower end of the cylinder, In order to penalize the coefficients, For the feature vector of the speech quiz data, For the feature vector of the fault diagnosis result, Is a feature vector And (3) with Is used for the two norms of (2), As a function of the maximum value.
7. The intelligent question-answering and business guiding system for multi-modal interaction in the electric power domain according to claim 1, wherein the intelligent guiding output module comprises an AR guiding unit, a voice output unit and a text output unit; the AR guiding unit is used for superposing marking information on AR equipment worn by operation and maintenance personnel based on a fault diagnosis result, wherein the marking information comprises an oil level gauge sludge attachment position, an oil tank bottom sludge deposition area and a fault core area; the voice output unit is used for outputting the question-answer response content and the fault level information in a voice broadcasting mode; the text output unit is used for outputting the customized operation and maintenance step text.
8. The intelligent question-answering and service guiding system for multi-mode interaction in the electric power field according to claim 7, wherein the customized operation and maintenance step text output by the text output unit is the operation and maintenance content of the adaptive mineral insulating oil, and comprises an oil filtering operation flow, an oil sludge cleaning operation specification, an oil change period adjustment suggestion and a corresponding tool list; The text output unit is also used for outputting a fault diagnosis report, and the fault diagnosis report comprises oil sludge content analysis, fault association factors and treatment priority information.
9. The intelligent question-answering and service guiding system for multi-modal interaction in the electric power domain according to claim 1, wherein the service linkage module comprises a ledger association subunit and a case retrieval subunit; The standing book association subunit is used for linking the operation and maintenance standing book system of the mineral oil transformer and synchronizing fault diagnosis results, question and answer records and operation and maintenance guide contents to the operation and maintenance standing book of the corresponding transformer; the case calling sub-unit is used for calling historical cleaning cases and oil detection reports of similar oil sludge faults from the operation and maintenance ledger system based on the current fault type and transformer parameters.
10. The intelligent question-answering and service guiding system for multi-mode interaction in the electric power field according to claim 1, further comprising a model adapting module, wherein the model adapting module is used for dynamically adapting the association logic of the three-dimensional coupling model according to the type and the operation age of mineral insulating oil used by the transformer; the model adaptation module is used for carrying out iterative optimization on the three-dimensional coupling model according to the historical operation and maintenance data and the fault treatment effect; To realize iterative optimization of the three-dimensional coupling model, similarity is matched based on questions and answers And (3) adjusting model parameters, and adopting the following algorithm formula: ; ; Wherein, the To the optimized first Instead of the parameters of the three-dimensional coupling model, To be the first before optimization Instead of the parameters of the three-dimensional coupling model, In order for the rate of learning to be high, Is the first Model loss function Is used for the gradient of (a), Is the first The loss function of the generation model, Is the first The questions and answers corresponding to the generation model are matched with the similarity.

Description

Intelligent question-answering and business guiding system oriented to multi-mode interaction in electric power field Technical Field The invention relates to the technical field of intelligent operation and maintenance of electric power, in particular to an intelligent question-answering and service guiding system oriented to multi-mode interaction in the electric power field. Background The oil immersed transformer is used as core equipment of a power transmission and distribution link of a power system, and the safe and stable operation of the oil immersed transformer is directly related to the reliability of power supply of a power grid. Mineral insulating oil has good insulating property and heat dissipation property, and is widely applied to oil immersed transformers, but is influenced by factors such as temperature change, oxidation reaction and the like in the long-term operation process, so that the mineral insulating oil is easy to generate sludge precipitation. The oil sludge not only can be attached to the surface of the oil level gauge to cause oil level observation deviation, but also can block a heat dissipation channel to cause abnormal rise of oil temperature, further cause cascading failures such as oil level display deviation, heat dissipation efficiency reduction and the like, and cause the shutdown of a transformer when serious, thereby bringing great economic loss to an electric power system. In the current power field, aiming at the operation and maintenance process of the transformer, the traditional monitoring mode mostly adopts single-dimensional data acquisition and diagnosis, such as oil temperature monitoring through infrared equipment, oil level manual visual inspection, oil level sampling detection, oil sludge content detection and the like, and the oil sludge content is not subjected to collaborative correlation analysis with key operation parameters such as oil temperature, oil level and the like, so that faults caused by the oil sludge are difficult to accurately identify in an early stage. Meanwhile, the existing multi-mode interaction technology is lack of pertinence in the transformer operation and maintenance scene, voice interaction is easily interfered by power frequency noise of a transformer substation, image processing is not suitable for a visual interference scene caused by oil sludge adhesion, multi-mode data processing and fault diagnosis and operation and maintenance guiding links are mutually disjointed, and full-link coordination of data acquisition-diagnosis analysis-interaction question-answer-operation and maintenance treatment cannot be realized. In addition, fault guidance of the existing system is mostly generalized prompt, a customized operation and maintenance scheme is not provided by combining the characteristics of mineral insulating oil, and the operation and maintenance ledger system cannot be linked to realize data closed loop update and historical case multiplexing, so that operation and maintenance personnel operate fussy, the fault handling efficiency is low, and the requirements of the power field on the accuracy and the high efficiency of the operation and the maintenance of the transformer are difficult to meet. Disclosure of Invention The invention aims to provide an intelligent question-answering and service guiding system facing multi-mode interaction in the electric power field, which aims to solve the technical problems that the oil sludge content, the oil temperature and the oil level of the existing mineral insulating oil transformer lack of collaborative correlation diagnosis, and the multi-mode interaction and operation and maintenance guiding are disjoint, so that the faults caused by the oil sludge are difficult to accurately identify and efficiently treat. In order to solve the technical problems, the invention provides the following technical scheme that the intelligent question-answering and service guiding system facing the multi-mode interaction in the electric power field comprises a multi-mode data acquisition module, a data preprocessing module, a coupling diagnosis module, an intelligent guiding output module and a service linkage module; The multi-mode data acquisition module is used for acquiring multi-source data of the oil immersed transformer, wherein the multi-source data comprise infrared thermal imaging data, oil level gauge image data, oil liquid sampling image data and voice questioning data; The data preprocessing module is used for carrying out adaptation processing on the acquired multi-source data to obtain standardized data; The coupling diagnosis module calls a three-dimensional coupling model adapting to the mineral insulating oil, carries out association analysis on the standardized data and generates a fault diagnosis result; the intelligent guidance output module outputs multidimensional collaborative guidance information based on a fault diagnosis result; The service linkage module is linked with t