CN-122020532-A - Language and image information fused power exchange station equipment fault recognition system

CN122020532ACN 122020532 ACN122020532 ACN 122020532ACN-122020532-A

Abstract

The invention discloses a power exchange station equipment fault recognition system integrating language and image information, which belongs to the technical field of power equipment fault recognition and comprises a data acquisition module, an image analysis module, a language text analysis module and an early warning module, wherein the data acquisition module is used for acquiring image data and language text data of power exchange station equipment, the image analysis module is used for carrying out image analysis on the image data by utilizing a preset image recognition algorithm to obtain an image analysis result, the language text analysis module is used for analyzing the language text data by utilizing a power exchange station equipment fault tree to obtain a text analysis result, and the early warning module is used for triggering a grading treatment strategy according to the image analysis result and the text analysis result, so that a more accurate power exchange station equipment fault recognition result can be obtained by combining the image analysis and the language text analysis.

Inventors

SHI YANHUI
HUANG HUA
Peng teng
LIU YUCHAO
ZHANG ZHEN
WANG NING
ZHANG BO
HU MENGZHU
SONG YU
ZHENG XING
KONG WEIQI
XU CHENG
YU JUNSONG
ZHANG WEN
Ruan Yanjun
LIAO YI
WANG HAO
ZHAO HANGHANG
GU ZHIPENG
MAO XIONG

Assignees

中国南方电网有限责任公司超高压输电公司广州局

Dates

Publication Date: 20260512
Application Date: 20260126

Claims (8)

1. A battery exchange station equipment fault recognition system integrating language and image information, comprising: the data acquisition module is used for acquiring image data and language text data of the power exchange station equipment; the image analysis module is used for carrying out image analysis on the image data by utilizing a preset image recognition algorithm to obtain an image analysis result; The language text analysis module is used for analyzing the language text data by using a power exchange station equipment fault tree to obtain a text analysis result; and the early warning module carries out grading early warning according to the image analysis result and the text analysis result.
2. The system for recognizing the fault of the power exchange station equipment by fusing language and image information according to claim 1, wherein the image analysis of the image data by using a preset image recognition algorithm to obtain an image analysis result comprises: carrying out appearance analysis on the image data of the power exchange station equipment to obtain an appearance defect analysis result of the power exchange station equipment; analyzing the image data of the indicator lights of the power exchange station equipment to obtain a status result of the indicator lights; and performing thermal imaging analysis on the image data of the power exchange station equipment to obtain a temperature analysis result of the power exchange station equipment.
3. The system for recognizing fault of power exchange station equipment based on fusion of language and image information according to claim 2, wherein after analyzing the image data of the indicator lamp of the power exchange station equipment to obtain the status result of the indicator lamp, the system further comprises adjusting the status result of the indicator lamp to obtain the status result of the indicator lamp after adjustment, specifically comprising: acquiring a voltage value and a current value of the power exchange station equipment in each phase in a certain time period, and according to the voltage value and the current value, corresponding current period data of the power exchange station equipment; Constructing a periodic characteristic vector library, and matching the current periodic data with the periodic data in the periodic characteristic vector library to obtain a matching result; And correcting the state result of the indicator lamp by using the matching result, and obtaining an adjusted state result of the indicator lamp according to the correction result.
4. A power plant equipment fault identification system incorporating language and image information according to claim 3, wherein said constructing a periodic feature vector library comprises: classifying the power exchange station equipment according to the function type of the power exchange station equipment; Collecting multi-mode data from the classified power exchange station equipment, and carrying out time domain and frequency domain analysis on the multi-mode data to obtain time domain characteristics and frequency domain characteristics; and generating a periodic feature vector library for the time domain features and the frequency domain features based on a time sequence mode of normal operation of LSTM network learning power exchange station equipment.
5. The system for identifying a fault in a power exchange station device for fusing language and image information as set forth in claim 1, wherein said analyzing said language text data using a fault tree of the power exchange station device to obtain a text analysis result comprises: Removing stop words from the language text, and performing word segmentation processing to obtain a plurality of target words; And constructing a fault tree of the power exchange station equipment, and obtaining a text analysis result according to the target word and the fault tree of the power exchange station equipment.
6. The system for identifying a fault in a power plant that fuses language and image information as recited in claim 5, wherein said building a fault tree for a power plant comprises: constructing a substation sub-equipment fault tree according to the function type of the substation equipment, wherein each substation sub-equipment fault tree corresponds to one piece of sub-equipment; acquiring historical fault information, carrying out semantic recognition according to the historical fault information, extracting a plurality of keywords, and mapping the keywords to fault tree nodes; And determining a fault path by utilizing causal analysis according to the mapped fault nodes.
7. The system for identifying a fault in a power plant that fuses language and image information as defined in claim 6, wherein the causal analysis comprises: inquiring the root node upwards from the fault node, and identifying a fault propagation path; Inquiring the leaf nodes downwards from the fault node, and positioning the fault source; And verifying the rationality of the fault path through the association relation between the events at the same level.
8. The system for identifying a fault in a power exchange station device, which fuses language and image information according to claim 1, wherein said step of performing hierarchical early warning according to said image analysis result and said text analysis result comprises: dynamic weight distribution is carried out according to the image analysis result and the text analysis result, and according to the image analysis result, the text analysis result and the corresponding weight; Presetting a fault criterion library, and determining confidence scores corresponding to the image analysis result and the text analysis result according to the image analysis result, the text analysis result and the fault criterion library; And calculating to obtain comprehensive confidence scores according to the confidence scores corresponding to the image analysis results and the text analysis results and weights corresponding to the image analysis results and the text analysis results, and carrying out hierarchical early warning according to the comprehensive confidence scores.

Description

Language and image information fused power exchange station equipment fault recognition system Technical Field The invention relates to the technical field of power equipment fault identification, in particular to a power exchange station equipment fault identification system integrating language and image information. Background The power exchange station is an energy infrastructure for providing rapid battery replacement service for the electric automobile through centralized storage, charging and distribution of power batteries. The core mode is to separate the vehicle from the battery, the user can complete the energy supplementing by directly replacing the full-charge battery without waiting for charging, and the process is usually completed within 3-5 minutes. The failure of the power exchange station equipment can cause power exchange failure, a user needs to wait for maintenance or transfer to other stations to directly influence the charging efficiency, potential failures can be found in advance by the failure identification of the power exchange station equipment, the unplanned downtime is reduced, the existing failure identification of the power exchange station equipment only depends on image identification, and the power exchange station equipment has certain limitation and low accuracy. Disclosure of Invention Aiming at the prior art, the invention aims to provide a power exchange station equipment fault identification system integrating language and image information, which mainly solves the technical problems in the background art. In order to achieve the above object, the technical solution of the embodiment of the present invention is as follows: a battery exchange station equipment fault identification system integrating language and image information, comprising: the data acquisition module is used for acquiring image data and language text data of the power exchange station equipment; the image analysis module is used for carrying out image analysis on the image data by utilizing a preset image recognition algorithm to obtain an image analysis result; The language text analysis module is used for analyzing the language text data by using a power exchange station equipment fault tree to obtain a text analysis result; and the early warning module carries out grading early warning according to the image analysis result and the text analysis result. Optionally, the image analysis is performed on the image data by using a preset image recognition algorithm to obtain an image analysis result, including: carrying out appearance analysis on the image data of the power exchange station equipment to obtain an appearance defect analysis result of the power exchange station equipment; analyzing the image data of the indicator lights of the power exchange station equipment to obtain a status result of the indicator lights; and performing thermal imaging analysis on the image data of the power exchange station equipment to obtain a temperature analysis result of the power exchange station equipment. Optionally, after the analyzing the indicator light image data of the power exchange station device to obtain an indicator light state result, the analyzing further includes adjusting the indicator light state result to obtain an adjusted indicator light state result, specifically includes: acquiring a voltage value and a current value of the power exchange station equipment in each phase in a certain time period, and according to the voltage value and the current value, corresponding current period data of the power exchange station equipment; Constructing a periodic characteristic vector library, and matching the current periodic data with the periodic data in the periodic characteristic vector library to obtain a matching result; And correcting the state result of the indicator lamp by using the matching result, and obtaining an adjusted state result of the indicator lamp according to the correction result. Optionally, the constructing the periodic feature vector library includes: classifying the power exchange station equipment according to the function type of the power exchange station equipment; Collecting multi-mode data from the classified power exchange station equipment, and carrying out time domain and frequency domain analysis on the multi-mode data to obtain time domain characteristics and frequency domain characteristics; and generating a periodic feature vector library for the time domain features and the frequency domain features based on a time sequence mode of normal operation of LSTM network learning power exchange station equipment. Optionally, the analyzing the language text data by using a fault tree of the power exchange station equipment to obtain a text analysis result includes: Removing stop words from the language text, and performing word segmentation processing to obtain a plurality of target words; And constructing a fault tree of the power exchange station equipment, and ob