CN-121981854-A - Intelligent inspection system and method for power grid equipment

CN121981854ACN 121981854 ACN121981854 ACN 121981854ACN-121981854-A

Abstract

The invention discloses an intelligent inspection system and method for power grid equipment, the system comprises a data acquisition module, a defect intelligent processing module, an environment adaptation adjusting module, a dynamic path planning module, an augmented reality interaction module and a data collaborative management module. The method comprises the steps that multi-mode data and space position data of power grid equipment are obtained through a data acquisition module, an environment adaptation adjustment module dynamically optimizes acquisition parameters, a defect intelligent processing module fuses an algorithm to identify defects and judge hazard grades, a dynamic path planning module combines defect priorities and real-time barriers to generate an optimized path, an augmented reality interaction module realizes visual presentation and interaction, and a data collaborative management module guarantees multi-terminal data synchronization and storage. The invention solves the problems of low traditional inspection efficiency, inaccurate defect identification, path disjointing, data lag and the like, realizes accurate defect identification, path dynamic optimization and convenient and fast operation coordination, and promotes the intelligent, efficient and standardized development of power grid equipment inspection.

Inventors

CHENG JIAN
CHEN CHUNLONG
CHEN TENGBIAO
QIAN CEN
Lai Jieheng
HUANG ZONGZE
Man Chenxi
HAN LEI
MENG XIN
ZHANG HANPU

Assignees

深圳供电局有限公司

Dates

Publication Date: 20260505
Application Date: 20260115

Claims (10)

1. An intelligent patrol system for power grid equipment, comprising: The data acquisition module is used for acquiring visible light images, infrared thermal images and spatial position data of the power grid equipment; The defect intelligent processing module is connected with the data acquisition module and is used for identifying equipment defects based on the visible light images and the infrared thermal images, outputting defect confidence degrees and judging the hazard level of the defects based on defect information and equipment operation data; The environment adaptation adjusting module is used for acquiring the patrol environment parameters and the real-time obstacle information of the patrol environment in real time, and automatically adjusting the working parameters of the data acquisition module or extracting the effective area of the equipment according to the illumination intensity, the air humidity, the visibility and the equipment shielding condition; The dynamic path planning module is respectively connected with the data acquisition module, the defect intelligent processing module and the environment adaptation adjusting module and is used for generating and adjusting and optimizing a patrol path in real time based on the distribution information of the power grid equipment, the hazard level of the defect, the spatial position data and the real-time obstacle information; The augmented reality interaction module is connected with the dynamic path planning module and the defect intelligent processing module and is used for visually presenting the optimized tour path, the equipment state and the defect information through augmented reality equipment and receiving interaction instructions of tour personnel; And the data collaborative management module is respectively connected with the data acquisition module, the defect intelligent processing module, the environment adaptation adjusting module, the dynamic path planning module and the augmented reality interaction module, and is used for storing data generated by each module and realizing bidirectional data synchronization and remote instruction interaction between the augmented reality equipment and the background monitoring terminal.
2. The system of claim 1, wherein the defect intelligent processing module is internally provided with an improved convolutional neural network model, device characteristics are extracted through a convolutional layer and a pooling layer, the device characteristics are compared with a standard device characteristic library by combining a SIFT characteristic matching algorithm, the defect confidence is greater than or equal to 0.85, the defect hazard level is calculated by the formula H=ω 1 T+ω 2 S+ω 3 V, wherein H is defect hazard level, ω 1 is defect type weight, T is defect influence duration, ω 2 is influence range weight, S is defect influence device number, ω 3 is development speed weight, and V is defect development speed.
3. The system according to claim 1, wherein the environment adaptation adjustment module presets an environment parameter threshold, when the illumination intensity is lower than a preset value, the data acquisition module is controlled to turn on a light supplementing lamp and optimize an image white balance parameter, when the air humidity or the visibility exceeds a preset range, a defogging algorithm and a lens anti-fog control are started, and when equipment shielding exists, an effective area of the equipment is extracted through an image segmentation algorithm.
4. The system of claim 1, wherein the dynamic path planning module employs a modified Dijkstra algorithm to generate an initial path with a shortest total tour duration and a priority coverage of a high priority device as an objective function, calculates an alternative path synthesis cost and selects a path with the smallest synthesis cost as an optimized tour path when a sudden obstacle or a newly added emergency defect occurs, and the calculation formula of the alternative path synthesis cost is as follows: Wherein, the For the integrated cost of the path, The start time is adjusted for the path and, For the path adjustment completion time to be sufficient, As the distance weight is used for the distance, As a result of the deviation of the total distance of the path, As the weight of the time in question, For the duration of the tour delay, In order for the obstacle avoidance weights to be appropriate, Number of evasion for obstacle.
5. The system of claim 1, wherein the hardware carrier of the augmented reality interaction module is AR glasses, a voice interaction assembly is configured, defect marks, path guidance and equipment parameters are superimposed in the AR field of view, defect confirmation and path adjustment instructions of the patrolling personnel are received through voice instructions or touch operation, and voice prompts are sent when the patrolling personnel approach the target equipment.
6. The system of claim 1, wherein the data collaborative management module adopts a distributed database and a fragment storage strategy, bidirectional data synchronization is realized based on a 5g+ edge computing architecture, data transmission adopts an AES-256 encryption algorithm, and the synchronous data comprises an image video, a defect recognition result, a patrol path execution progress and a patrol personnel real-time position.
7. The system of claim 1, further comprising an energy consumption optimization module for dynamically adjusting the operating parameters of the data acquisition module and the augmented reality interaction module, wherein the data acquisition module is used for reducing the frame rate of the camera and turning off the infrared thermal imaging assembly when no power grid equipment is detected, restoring the normal frame rate of the camera and turning on the infrared thermal imaging assembly when equipment is detected, restoring the normal frame rate of the camera after a short time of raising the frame rate of the camera when a defect is identified, and reducing the screen brightness and enabling the voice interaction assembly to enter a sleep mode when the augmented reality interaction module is not operated.
8. The system of claim 1, further comprising a historical data comparison module for retrieving the historical data stored by the data collaborative management module, performing comparison analysis with the current defect identification result and the path planning result, wherein the defect comparison adopts a cosine similarity algorithm, outputting a historical processing scheme, a processing period and a processing effect of the similar defects, the path comparison calculates a time length difference value between the current path and a historical optimal path, and a high-priority equipment coverage efficiency difference value, and outputs a path optimization improvement suggestion, and the comparison result is presented in a graph form through the augmented reality interaction module.
9. The system of claim 1, further comprising an equipment state early warning module, wherein the equipment state early warning module is used for monitoring the defect confidence coefficient, defect hazard coefficient and equipment operation parameters output by the defect intelligent processing module in real time, triggering three-level early warning when the defect confidence coefficient is more than or equal to 0.9, the defect hazard coefficient H is more than or equal to 8 or the equipment parameters exceed a preset threshold value, and warning information is used for reminding patrolling personnel through the augmented reality interaction module and is synchronously pushed to a background monitoring terminal.
10. An intelligent power grid equipment inspection method is characterized by comprising the following steps: step S1, visible light images, infrared thermal images and space position data of power grid equipment are obtained, and inspection environment parameters and real-time obstacle information are synchronously acquired; step S2, adjusting data acquisition parameters or extracting an effective area of the equipment according to illumination intensity, air humidity, visibility and equipment shielding conditions; Step S3, identifying equipment defects based on the visible light images and the infrared thermal images, outputting defect confidence, and judging the hazard level of the defects by combining defect information and equipment operation data; Step S4, generating an optimized patrol path based on the distribution information of the power grid equipment, the hazard level of the defects, the spatial position data and the real-time obstacle information; Step S5, visually presenting the optimized patrol path, the equipment state and the defect information through the augmented reality equipment, and receiving and responding to an interaction instruction of patrol personnel; and S6, storing the data generated in each step, and realizing bidirectional data synchronization and remote instruction interaction between the augmented reality equipment and the background monitoring terminal.

Description

Intelligent inspection system and method for power grid equipment Technical Field The invention relates to the technical field of power system equipment monitoring, in particular to an intelligent power grid equipment inspection system and method. Background The power grid equipment is a core carrier for the stable operation of the power system, and daily inspection and defect identification of the power grid equipment are important for guaranteeing the power supply reliability. At present, most power enterprises still mainly adopt a traditional manual inspection mode. In the mode, the patrol personnel need to carry handheld infrared thermometer, camera and other equipment to check the equipment such as transformer, circuit breaker, insulator one by one, and the whole process is highly dependent on personal experience-for example, whether the insulator is damaged or not is judged through naked eye observation, whether oil leakage of the sealing surface of the transformer oil tank is checked in a short distance, and the contact temperature of the circuit breaker is measured point by using the handheld infrared meter. The mode has lower efficiency, and usually, a single person can only complete inspection of 10 to 15 pieces of equipment, and is easily interfered by environmental factors, namely, fine defects of the equipment are difficult to distinguish under strong light conditions, effective distance of infrared temperature measurement can be shortened due to rain and fog, round trip time of personnel can be increased due to outdoor complex topography, inspection period of partial remote equipment is prolonged, and potential defects are difficult to discover in time. In addition, the manually recorded data needs to be arranged and recorded in a system after the fact, so that the problem of data lag exists, and if personnel misjudgment or neglect is caused, the decision quality of subsequent defect processing is affected. With the application of the intelligent technology in the electric power field, part of enterprises begin to introduce vision recognition and path planning technology to assist in inspection, but the existing scheme still has obvious limitations. In the aspect of defect identification, most schemes only depend on a single visible light image and do not combine infrared thermal imaging data, so that internal defects (such as local overheating) of equipment are difficult to accurately detect, and meanwhile, the anti-interference capability of an identification model on a complex background is weak and the misjudgment rate is high. In the aspect of path planning, a traditional shortest path algorithm is generally adopted, only the shortest distance is used as a target, the defect priority of equipment, real-time environmental obstacle and actual walking speed difference of patrol personnel are not comprehensively considered, so that the planned path and the site execution are disjointed, part of high-priority equipment is checked after the path bypasses, and the patrol personnel can deviate from a preset path due to the site obstacle avoidance. In addition, the data storage and the calling of the existing system adopt a centralized architecture, and when the number of the patrol area equipment is large, the data transmission and the query response are slow, so that the requirements of real-time defect identification and path dynamic adjustment are difficult to support. In recent years, augmented Reality (AR) technology is gradually applied to power inspection, but most of the existing AR auxiliary schemes still stay on the information superposition display level, and cannot realize the whole-flow integration from data acquisition, defect identification, path planning to early warning feedback. For example, part of AR equipment can only superimpose equipment model and parameters in the field of view, infrared thermal imaging data and accurate position information cannot be acquired automatically, part of the scheme has a basic defect marking function, but defect grade evaluation still needs manual intervention, inspection sequence cannot be adjusted dynamically according to hazard degree, the scheme lacks multi-terminal coordination capability, on-site inspection data needs to be transmitted to the background in an off-line mode (such as USB flash disk copy), so that a monitoring center cannot master inspection progress in real time, and emergency processing instructions are difficult to issue in time when equipment is in sudden defect. The problems cause the prior AR scheme to not fully exert technical advantages, a plurality of links still depend on manual intervention, and the intelligent and efficient promotion of the inspection and defect identification of the power grid equipment cannot be fundamentally realized. Disclosure of Invention The technical problem to be solved by the embodiment of the invention is to provide an intelligent power grid equipment inspection system and method for r