Substation robot inspection and maintenance method and system based on augmented reality

Abstract

This invention discloses a substation robot inspection and maintenance method and system based on augmented reality, relating to the field of power equipment inspection technology. The augmented reality-based substation robot inspection and maintenance system includes: an inspection fault identification module, an inspection fault maintenance module, and an inspection plan update module. This invention enables real-time monitoring of substation equipment status and timely identification and response to potential faults through frequent and regular inspections according to a preset cycle. The automated inspection process reduces manual intervention, thereby improving inspection efficiency. Simultaneously, the use of an augmented reality inspection model provides clear visual guidance for maintenance personnel, effectively improving the efficiency and quality of maintenance work. Furthermore, the adoption of a swarm optimization algorithm to adjust the inspection plan based on the actual operating conditions of the equipment enables more effective coverage of all equipment, and by increasing the number of inspections of faulty substation equipment, it prevents missed faults.

Description

Technical Field

[0001] The present invention relates to the technical field of power equipment inspection, and in particular to a substation robot inspection and maintenance method and system based on augmented reality. Background Art

[0002] As an important part of the power system, substation plays a key role in the transmission and distribution of electric energy. Its operation directly affects the stability of the power system and the reliability of power supply. Therefore, regular and irregular inspections of substations are one of the necessary measures to ensure their efficient and stable operation.

[0003] In the past, inspections of substations were mainly carried out manually. Workers needed to enter the substation to observe and judge the appearance, sound, temperature, etc. of the equipment. Due to the existence of dangerous factors such as high voltage and high current in the substation, manual inspections faced great safety risks. When performing live operations, if the operation is not careful, electric shock accidents may occur, threatening the lives of personnel. Manual inspections rely on personal experience and subjective judgment, which may vary from person to person, which to a certain extent affects the efficiency and accuracy of the inspections.

[0004] Therefore, there is an urgent need to provide a substation robot inspection and maintenance method and system based on augmented reality that improves inspection efficiency and ensures safety. Summary of the invention

[0005] The present invention aims to provide a substation robot inspection and maintenance method and system based on augmented reality, which improves the inspection efficiency and accuracy and significantly enhances the safety of on-site operations.

[0006] Substation robot inspection and maintenance method based on augmented reality, including:

[0007] S1. Every preset period, the robot performs inspection according to the real-time inspection plan to obtain a real-time inspection image. The real-time inspection plan includes the inspection area, the inspection equipment sequence and the inspection path; based on the real-time inspection image and the fault recognition model, a fault signal is obtained; if the fault signal is no fault, the robot continues to inspect according to the real-time inspection plan; otherwise, the real-time inspection image is used as the fault inspection image; in the current cycle, the corresponding substation equipment in the fault inspection image is marked as a faulty substation equipment, and the faulty substation equipment number is recorded;

[0008] S2. Input the fault inspection image into the inspection augmented reality model for processing to obtain an augmented reality image of the faulty substation equipment; provide maintenance guidance for the faulty substation equipment based on the augmented reality image of the faulty substation equipment and the fault signal to obtain a fault maintenance instruction; based on the fault maintenance instruction, the robot performs maintenance on the faulty substation equipment, and after the maintenance is completed, the robot continues to perform inspections according to the real-time inspection plan; after the robot inspection is completed in the current cycle, the real-time inspection plan is updated using the swarm optimization algorithm to obtain a new real-time inspection plan for the next cycle, and the substation equipment marked as faulty substation equipment is unmarked.

[0009] As a preferred technical solution of the present invention, the fault recognition model in step S1 includes an image preprocessing layer, a feature recognition layer and a result output layer;

[0010] The image preprocessing layer is used to preprocess the real-time inspection image to obtain a preprocessed real-time inspection image;

[0011] The feature recognition layer is used to perform image fault feature recognition on the pre-processed real-time inspection image to obtain the fault signal;

[0012] The resulting output layer is used to output fault signals.

[0013] As a preferred technical solution of the present invention, the specific steps of training the fault identification model include:

[0014] Collect several real-time inspection images with target value annotations; the target value is the fault signal corresponding to the real-time inspection image; combine several real-time inspection images with target value annotations to obtain a fault recognition training set;

[0015] The CNN model is used as the basic model of the fault recognition model to obtain the initial fault recognition model; the fault recognition training set is used to train the initial fault recognition model with the target value as the target to obtain the model evaluation result; if the model evaluation result is passed, the trained initial fault recognition model is used as the fault recognition model; otherwise, the fault recognition training set is used to continue training the initial fault recognition model.

[0016] As a preferred technical solution of the present invention, the inspection augmented reality model in step S2 includes an image processing layer, an image update layer and an image output layer;

[0017] The image processing layer is used to extract the feature points of the fault inspection image according to the SURF algorithm to obtain the feature points of the fault inspection image; extract the augmented reality image corresponding to the fault substation equipment number in the substation equipment augmented reality image library to obtain the augmented reality image of the substation equipment to be updated; extract the feature points of the augmented reality image of the substation equipment to be updated to obtain the feature points of the augmented reality image of the substation equipment to be updated; the substation equipment augmented reality image library contains all substation equipment numbers and corresponding augmented reality images;

[0018] The image update layer is used to perform image fusion based on the feature points of the fault inspection image and the feature points of the augmented reality image of the substation equipment to be updated, so as to obtain the augmented reality image of the faulty substation equipment;

[0019] The image output layer is used to output augmented reality images of faulty substation equipment.

[0020] As a preferred technical solution of the present invention, the specific steps of performing image fusion in the image update layer include:

[0021] Q1. Compare the fault inspection image and the augmented reality image of the substation equipment to be updated based on the feature points of the fault inspection image and the feature points of the augmented reality image of the substation equipment to be updated to obtain the significant change feature points;

[0022] Q2. Update the augmented reality image of the substation equipment to be updated according to the significant change feature points, merge the fault inspection image with the augmented reality image of the substation equipment to be updated, and obtain the augmented reality image of the fault substation equipment to be processed;

[0023] Q3. Perform image adjustment on the augmented reality image of the faulty substation equipment to be processed to obtain an augmented reality image of the faulty substation equipment.

[0024] As a preferred technical solution of the present invention, the specific steps of updating the real-time inspection plan using the swarm optimization algorithm in step S2 include:

[0025] In the current cycle, the substation equipment marked as faulty substation equipment is recorded as important substation equipment;

[0026] Construct J inspection plan individuals G j , j=1,2,…,J; each inspection plan individual G j There is a set of real-time inspection plans, each of which has different inspection device sequences and inspection paths; in the individual inspection plan G j The inspection frequency of important substation equipment is P j times, inspection equipment sequence interval Q j ; J inspection plan individuals G jCombination, get the inspection plan iteration population; set the maximum number of iterations, P j and Q j The values ​​of are all within the set threshold range;

[0027] According to the inspection plan iteration population, the inspection plan individual G j Perform path simulation to obtain the simulated inspection time T j , the simulated inspection time T j The reciprocal of is taken as the inspection plan individual G j The fitness S j ;

[0028] Based on all inspection schemes individual G j The fitness S j Update the iterative population of all inspection schemes;

[0029] When the maximum number of iterations is reached, the inspection plan individual G corresponding to the maximum fitness is output j , the inspection plan individual G corresponding to the maximum fitness j As the optimal inspection plan individual, the real-time inspection plan in the optimal inspection plan individual is used as a new real-time inspection plan, and the real-time inspection plan is updated.

[0030] As a preferred technical solution of the present invention, the swarm optimization algorithm in the real-time inspection scheme updated by the swarm optimization algorithm in step S2 is a genetic algorithm.

[0031] Substation robot inspection and maintenance system based on augmented reality, including:

[0032] The inspection fault identification module is used to deploy the fault identification model. Every preset period, the robot performs inspection according to the real-time inspection plan to obtain a real-time inspection image. The real-time inspection plan includes the inspection area, the inspection equipment sequence and the inspection path. Based on the real-time inspection image and the fault identification model, a fault signal is obtained. If the fault signal is no fault, the robot continues to inspect according to the real-time inspection plan. Otherwise, the real-time inspection image is used as the fault inspection image. In the current cycle, the corresponding substation equipment in the fault inspection image is marked as a fault substation equipment, and the fault substation equipment number is recorded.

[0033] The inspection fault maintenance module inputs the fault inspection image into the inspection augmented reality model for processing, and obtains the augmented reality image of the faulty substation equipment; provides maintenance guidance for the faulty substation equipment based on the augmented reality image of the faulty substation equipment and the fault signal, and obtains the fault maintenance instruction; based on the fault maintenance instruction, the robot maintains the faulty substation equipment, and after the maintenance is completed, the robot continues to inspect according to the real-time inspection plan;

[0034] The inspection plan update module is used to update the real-time inspection plan by using the swarm optimization algorithm after the robot inspection is completed in the current cycle, obtain a new real-time inspection plan for the next cycle, and unmark the substation equipment marked as faulty substation equipment.

[0035] The present invention has the following advantages:

[0036] 1. The present invention can monitor the status of substation equipment in real time and identify and respond to potential faults in a timely manner by performing frequent and regular inspections according to preset cycles; through the automated inspection process, manual intervention is reduced, thereby improving the efficiency of inspections; at the same time, the inspection augmented reality model is used to provide maintenance personnel with clear visual guidance, effectively improving the efficiency and quality of maintenance work; in addition, the use of a swarm optimization algorithm to adjust the inspection plan according to the actual operating conditions of the equipment can more effectively cover all equipment, and by increasing the number of inspections of faulty substation equipment, the missed detection of faults is prevented; robots perform most of the high-risk operations, reducing the risk of power workers being exposed to dangerous environments and ensuring the personal safety of workers.

[0037] 2. The present invention uses the SURF algorithm to extract feature points from fault inspection images, so as to quickly and accurately locate the fault point, which is crucial for quickly responding to and handling faults. By fusing the fault feature points with the images of the corresponding equipment in the augmented reality image library, a detailed fault augmented reality image can be generated, which provides intuitive visual assistance for maintenance personnel, helping them to more easily understand the problem and maintenance steps. Rapid and accurate fault location and efficient maintenance guidance can significantly reduce equipment downtime and maintenance costs, and improve the overall operating efficiency of the substation. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a schematic diagram of the structure of the substation robot inspection and maintenance system based on augmented reality adopted in an embodiment of the present invention. DETAILED DESCRIPTION

[0039] In order to enable persons skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0040] In embodiment 1, there are multiple robots in a substation, and each robot performs inspection according to the inspection area, inspection equipment sequence and inspection path according to the corresponding real-time inspection plan. In the embodiment of the present invention, the inspection and maintenance method of one of the robots in the substation is described. The substation robot inspection and maintenance method based on augmented reality includes:

[0041] S1. Every preset period, the robot performs inspection according to the real-time inspection plan to obtain a real-time inspection image. The real-time inspection plan includes the inspection area, the inspection equipment sequence and the inspection path; based on the real-time inspection image and the fault recognition model, a fault signal is obtained; if the fault signal is no fault, the robot continues to inspect according to the real-time inspection plan; otherwise, the real-time inspection image is used as the fault inspection image; in the current cycle, the corresponding substation equipment in the fault inspection image is marked as a faulty substation equipment, and the faulty substation equipment number is recorded;

[0042] The preset cycle f is set by professional technicians based on actual conditions. Every time a real-time inspection image is obtained, the real-time inspection image is monitored for faults until the robot completes the inspection according to the real-time inspection plan within the current cycle; the real-time inspection image is obtained by the robot photographing the substation equipment; the shooting angle and direction are set by professional technicians based on actual model needs;

[0043] The robot conducts inspections according to preset cycles to ensure that substation equipment is continuously monitored, reducing faults that may be missed due to long monitoring intervals; by combining real-time inspection images with fault recognition models, it can detect and identify equipment faults in real time, which helps to take measures in advance to prevent the faults from causing greater impact; after identifying the fault, the faulty substation equipment number is recorded, which is conducive to the tracking management and historical record analysis of the faulty equipment; the automated inspection process greatly improves the inspection efficiency, is not affected by human factors, and ensures the continuity and stability of the inspection;

[0044] All substation equipment is assigned a unique number, and each number corresponds to a specific device, which facilitates tracking and recording of the equipment's historical maintenance, inspection and operation records;

[0045] The fault recognition model in step S1 includes an image preprocessing layer, a feature recognition layer and a result output layer;

[0046] The image preprocessing layer is used to preprocess the real-time inspection image to obtain a preprocessed real-time inspection image;

[0047] The feature recognition layer is used to perform image fault feature recognition on the pre-processed real-time inspection image to obtain the fault signal;

[0048] The result output layer is used to output the fault signal;

[0049] The specific steps for training the fault recognition model include:

[0050] Collect several real-time inspection images with target value annotations; the target value is the fault signal corresponding to the real-time inspection image; combine several real-time inspection images with target value annotations to obtain a fault recognition training set;

[0051] The CNN model is used as the basic model of the fault recognition model to obtain the initial fault recognition model; the fault recognition training set is used to train the initial fault recognition model with the target value as the target to obtain the model evaluation result; if the model evaluation result is passed, the trained initial fault recognition model is used as the fault recognition model; otherwise, the fault recognition training set is used to continue training the initial fault recognition model;

[0052] The steps for obtaining the model evaluation result are as follows: obtaining the fault identification value according to the fault identification training set, comparing the fault identification value with the target value, and obtaining the identification accuracy. If the identification accuracy is within the error standard range, the model evaluation result is passed; otherwise, the model evaluation result is failed.

[0053] By applying the fault identification process based on the CNN model, the accuracy of fault detection is significantly improved, the probability of false alarms and missed alarms is reduced, and small problems are prevented from evolving into major faults in a timely manner. At the same time, maintenance costs are saved, equipment downtime is reduced, maintenance processes are optimized, and work efficiency is improved;

[0054] S2. Input the fault inspection image into the inspection augmented reality model for processing to obtain an augmented reality image of the faulty substation equipment; provide maintenance guidance for the faulty substation equipment based on the augmented reality image of the faulty substation equipment and the fault signal to obtain a fault maintenance instruction; based on the fault maintenance instruction, the robot maintains the faulty substation equipment, and after the maintenance is completed, the robot continues to inspect according to the real-time inspection plan; after the robot inspection is completed in the current cycle, the real-time inspection plan is updated using the swarm optimization algorithm to obtain a new real-time inspection plan for the next cycle, and the substation equipment marked as a faulty substation equipment is unmarked;

[0055] Professional technicians receive augmented reality images of faulty substation equipment at the monitoring terminal, provide maintenance guidance based on the information and fault signals in the augmented reality images of the faulty substation equipment, and output fault maintenance instructions to the robot; the robot receives the fault maintenance instructions and maintains the faulty substation equipment according to the fault maintenance instructions;

[0056] The inspection augmented reality model in step S2 includes an image processing layer, an image update layer and an image output layer;

[0057] The image processing layer is used to extract the feature points of the fault inspection image according to the SURF algorithm to obtain the feature points of the fault inspection image; extract the augmented reality image corresponding to the fault substation equipment number in the substation equipment augmented reality image library to obtain the augmented reality image of the substation equipment to be updated; extract the feature points of the augmented reality image of the substation equipment to be updated to obtain the feature points of the augmented reality image of the substation equipment to be updated; the substation equipment augmented reality image library contains all substation equipment numbers and corresponding augmented reality images;

[0058] The accurate extraction of SURF feature points is helpful for the subsequent image fusion operation, and can provide technical support for subsequent professional technicians to provide maintenance guidance for faulty substation equipment based on the augmented reality images and fault signals of the faulty substation equipment;

[0059] The steps of constructing the augmented reality image library of substation equipment are to obtain the numbers of all substation equipment in the substation, and perform three-dimensional modeling on all substation equipment in turn. After the three-dimensional modeling is completed, the three-dimensional model needs to be converted into an augmented reality image to obtain the numbers of all substation equipment and the corresponding augmented reality images;

[0060] The image update layer is used to perform image fusion based on the feature points of the fault inspection image and the feature points of the augmented reality image of the substation equipment to be updated, so as to obtain the augmented reality image of the faulty substation equipment;

[0061] The image output layer is used to output augmented reality images of faulty substation equipment;

[0062] The specific steps of image fusion in the image update layer include:

[0063] Q1. Compare the fault inspection image and the augmented reality image of the substation equipment to be updated based on the feature points of the fault inspection image and the feature points of the augmented reality image of the substation equipment to be updated to obtain the significant change feature points;

[0064] Q2. Update the augmented reality image of the substation equipment to be updated according to the significant change feature points, merge the fault inspection image with the augmented reality image of the substation equipment to be updated, and obtain the augmented reality image of the fault substation equipment to be processed;

[0065] Q3. Adjust the augmented reality image of the faulty substation equipment to be processed to obtain an augmented reality image of the faulty substation equipment;

[0066] The augmented reality images of faulty substations provide clear visual guidance for on-site personnel or automated robots, helping to quickly identify and locate faults. The intuitive prompts in the augmented reality images ensure the accuracy of maintenance work while avoiding misoperations that may pose safety risks to workers. The automation of image fusion ensures that changes are quickly reflected in the augmented reality guidance, making maintenance strategies more adaptive and dynamic.

[0067] The specific steps of updating the real-time inspection plan using the swarm optimization algorithm in step S2 include:

[0068] In the current cycle, the substation equipment marked as faulty substation equipment is recorded as important substation equipment;

[0069] Construct J inspection plan individuals Gj , j=1,2,…,J; each inspection plan individual G j There is a set of real-time inspection plans, each of which has different inspection device sequences and inspection paths; in the individual inspection plan G j The inspection frequency of important substation equipment is P j times, inspection equipment sequence interval Q j ; J inspection plan individuals G j Combination, get the inspection plan iteration population; set the maximum number of iterations, P j and Q j The values ​​of are all within the set threshold range;

[0070] Inspection plan individual G j A set of real-time inspection plans is included, which includes inspection area, inspection equipment sequence and inspection path. For the same robot, the inspection area remains unchanged, and only the inspection equipment sequence and inspection path are changed. For example, in the inspection plan individual G a In the inspection, the inspection equipment sequence and inspection path are expressed as A1-A3-A5-A1-A4-A6-A1-A2-A7-A8-A9, where A1 represents important substation equipment, P j The value of Q is 3. j The value is 2; the maximum number of iterations and the threshold range are set by professional technicians based on actual conditions; b In the j The value of Q is 2, change j If the value of is 4, the inspection equipment sequence and inspection path are expressed as A1-A3-A5-A4-A6-A1-A7-A2-A8-A9;

[0071] According to the inspection plan iteration population, the inspection plan individual G j Perform path simulation to obtain the simulated inspection time T j , the simulated inspection time T j The reciprocal of is taken as the inspection plan individual G j The fitness S j ;

[0072] In the simulation software, the individual G of the inspection plan is j The real-time inspection scheme in the simulation is used to perform a simulated inspection. During the simulation, it is stipulated that all substation equipment is fault-free and the robot operates normally, and the simulated inspection time T is obtained. j ;

[0073] By simulating in an idealized environment, a standard performance benchmark can be set for the inspection scheme. In this case, any increase in inspection time can be directly attributed to the design of the inspection scheme, rather than external uncontrollable factors such as random failures of equipment or abnormal operation of robots. Simulating under fault-free conditions can accurately measure the time required to complete the inspection task, which is a key indicator for evaluating the efficiency of the inspection scheme. By comparing the individual G of different inspection schemes, the performance of the inspection scheme can be improved. j Required simulation inspection time T j , which can intuitively reflect the pros and cons of the inspection route design;

[0074] Based on all inspection schemes individual G j The fitness S j Update the iterative population of all inspection schemes;

[0075] In step S2, the population optimization algorithm used to update the real-time inspection plan is a genetic algorithm. In the genetic algorithm, the methods for updating the iterative population of the inspection plan include selection and recombination;

[0076] Select the operation as individual G in the inspection plan j The fitness S j Based on this, the inspection plan individual G in the inspection plan iteration population is selected by the roulette wheel selection algorithm. j Make a selection and choose a better inspection plan individual G j Form a new inspection plan iterative population;

[0077] The reorganization operation is to randomly select two inspection scheme individuals G in the inspection scheme iteration population j Perform multi-point crossover changes, that is, randomly select different parameters, such as different inspection equipment orders; randomly select two inspection plan individuals G j Interchange elements in the selected inspection equipment sequence, such as exchanging the sequence of any several substation equipment;

[0078] When the maximum number of iterations is reached, the inspection plan individual G corresponding to the maximum fitness is output j , the inspection plan individual G corresponding to the maximum fitness j As the optimal inspection plan individual, the real-time inspection plan in the optimal inspection plan individual is used as a new real-time inspection plan to update the real-time inspection plan;

[0079] By giving priority to and increasing the number of inspections of important substation equipment in the inspection plan, the operating reliability of key equipment is ensured. At the same time, by optimizing the inspection equipment sequence and path, the total inspection time is reduced, thereby improving the inspection efficiency; using group optimization algorithms, such as genetic algorithms, to periodically update the inspection plan, adjust the inspection strategy in real time to adapt to changing operation and maintenance needs and environmental changes, and ensure the real-time and flexibility of the inspection plan; individualizing the inspection plan and testing the implementation effect of each plan in simulation, with the plan with the greatest adaptability as the optimal choice, which helps to fine-tune the maintenance of substation equipment, especially the management of important substation equipment; by optimizing the inspection path and frequency, redundant inspection operations are reduced, reducing the waste of manpower and material resources, which in the long run helps to significantly reduce the operation and maintenance costs; regular and systematic optimization of the inspection plan ensures the stable operation of substation equipment, reduces the risk of power supply interruption due to equipment failure, and enhances the overall stability and reliability of the substation.

[0080] Example 2, substation robot inspection and maintenance system based on augmented reality, see Figure 1 As shown, including:

[0081] The inspection fault identification module is used to deploy the fault identification model. Every preset period, the robot performs inspection according to the real-time inspection plan to obtain a real-time inspection image. The real-time inspection plan includes the inspection area, the inspection equipment sequence and the inspection path. Based on the real-time inspection image and the fault identification model, a fault signal is obtained. If the fault signal is no fault, the robot continues to inspect according to the real-time inspection plan. Otherwise, the real-time inspection image is used as the fault inspection image. In the current cycle, the corresponding substation equipment in the fault inspection image is marked as a fault substation equipment, and the fault substation equipment number is recorded.

[0082] The inspection fault maintenance module inputs the fault inspection image into the inspection augmented reality model for processing, and obtains the augmented reality image of the faulty substation equipment; provides maintenance guidance for the faulty substation equipment based on the augmented reality image of the faulty substation equipment and the fault signal, and obtains the fault maintenance instruction; based on the fault maintenance instruction, the robot maintains the faulty substation equipment, and after the maintenance is completed, the robot continues to inspect according to the real-time inspection plan;

[0083] The inspection plan update module is used to update the real-time inspection plan by using the swarm optimization algorithm after the robot inspection is completed in the current cycle, obtain a new real-time inspection plan for the next cycle, and unmark the substation equipment marked as faulty substation equipment.

[0084] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all such improvements and changes should fall within the scope of protection of the appended claims of the present invention. Parts not described in detail in this specification belong to the prior art known to those skilled in the art.

Claims

1. A substation robot inspection and maintenance method based on augmented reality, characterized in that: include: S1. Every preset period f robot performs inspection according to the real-time inspection program to obtain a real-time inspection image. The real-time inspection program includes the inspection area, inspection equipment sequence and inspection path; Based on the real-time inspection image and fault recognition model, the fault signal is obtained; If the fault signal indicates no fault, the robot continues to inspect according to the real-time inspection plan; Otherwise, the real-time inspection image is used as the fault inspection image; In the current cycle, the corresponding substation equipment in the fault inspection image is marked as a faulty substation equipment, and the faulty substation equipment number is recorded; S2. Input the fault inspection image into the inspection augmented reality model for processing to obtain an augmented reality image of the faulty substation equipment; According to the augmented reality image and fault signal of the faulty substation equipment, maintenance guidance is provided for the faulty substation equipment to obtain fault maintenance instructions; based on the fault maintenance instructions, the robot performs maintenance on the faulty substation equipment, and after the maintenance is completed, the robot continues to perform inspections according to the real-time inspection plan; after the robot inspection is completed in the current cycle, the real-time inspection plan is updated using the swarm optimization algorithm to obtain a new real-time inspection plan for the next cycle, and the substation equipment marked as faulty substation equipment is unmarked; The inspection augmented reality model in step S2 includes an image processing layer, an image update layer and an image output layer; The image processing layer is used to extract the feature points of the fault inspection image according to the SURF algorithm to obtain the feature points of the fault inspection image; extract the augmented reality image corresponding to the fault substation equipment number in the substation equipment augmented reality image library to obtain the augmented reality image of the substation equipment to be updated; Extracting feature points of the augmented reality image of the substation equipment to be updated, and obtaining feature points of the augmented reality image of the substation equipment to be updated; The substation equipment augmented reality image library contains all substation equipment numbers and corresponding augmented reality images; The image update layer is used to perform image fusion based on the feature points of the fault inspection image and the feature points of the augmented reality image of the substation equipment to be updated, so as to obtain the augmented reality image of the faulty substation equipment; The image output layer is used to output augmented reality images of faulty substation equipment; The specific steps of image fusion in the image update layer include: Q1. Compare the fault inspection image and the augmented reality image of the substation equipment to be updated based on the feature points of the fault inspection image and the feature points of the augmented reality image of the substation equipment to be updated to obtain the significant change feature points; Q2. Update the augmented reality image of the substation equipment to be updated according to the significant change feature points, merge the fault inspection image with the augmented reality image of the substation equipment to be updated, and obtain the augmented reality image of the fault substation equipment to be processed; Q3. Adjust the augmented reality image of the faulty substation equipment to be processed to obtain an augmented reality image of the faulty substation equipment; The specific steps of updating the real-time inspection plan using the swarm optimization algorithm in step S2 include: In the current cycle, the substation equipment marked as faulty substation equipment is recorded as important substation equipment; Construct J inspection plan individuals G j , j = 1, 2, ..., J; each inspection plan individual G j There is a set of real-time inspection plans, each of which has different inspection device sequences and inspection paths; in the individual inspection plan G j The inspection frequency of important substation equipment is P j times, inspection equipment sequence interval Q j ; J inspection plan individuals G j Combination, get the inspection plan iteration population; set the maximum number of iterations, P j and Q j The values ​​of are all within the set threshold range; According to the inspection plan iteration population, the inspection plan individual G j Perform path simulation to obtain the simulated inspection time T j , the simulated inspection time T j The reciprocal of is taken as the inspection plan individual G j The fitness S j ; Based on all inspection schemes individual G j The fitness S j Update the iterative population of all inspection plans; When the maximum number of iterations is reached, the inspection plan individual G corresponding to the maximum fitness is output j , the inspection plan individual G corresponding to the maximum fitness j As the optimal inspection plan individual, the real-time inspection plan in the optimal inspection plan individual is used as a new real-time inspection plan to update the real-time inspection plan; In step S2, the swarm optimization algorithm used to update the real-time inspection plan is a genetic algorithm.

2. The substation robot inspection and maintenance method based on augmented reality according to claim 1 is characterized in that: The fault recognition model in step S1 includes an image preprocessing layer, a feature recognition layer and a result output layer; The image preprocessing layer is used to preprocess the real-time inspection image to obtain a preprocessed real-time inspection image; The feature recognition layer is used to perform image fault feature recognition on the pre-processed real-time inspection image to obtain the fault signal; The resulting output layer is used to output fault signals.

3. The substation robot inspection and maintenance method based on augmented reality according to claim 2 is characterized in that: The specific steps for training the fault recognition model include: Collect several real-time inspection images with target value annotations; the target value is the fault signal corresponding to the real-time inspection image; Combine several real-time inspection images with target value annotations to obtain a fault recognition training set; The CNN model is used as the basic model of the fault recognition model to obtain the initial fault recognition model; the fault recognition training set is used to train the initial fault recognition model with the target value as the target to obtain the model evaluation result; if the model evaluation result is passed, the trained initial fault recognition model is used as the fault recognition model; otherwise, the fault recognition training set is used to continue training the initial fault recognition model.

4. Substation robot inspection and maintenance system based on augmented reality, characterized in that: The system applies the substation robot inspection and maintenance method based on augmented reality as described in any one of claims 1 to 3 above, including: The inspection fault identification module is used to deploy the fault identification model. Every preset period, the robot performs inspection according to the real-time inspection plan to obtain a real-time inspection image. The real-time inspection plan includes the inspection area, the inspection equipment sequence and the inspection path. Based on the real-time inspection image and the fault identification model, a fault signal is obtained. If the fault signal is no fault, the robot continues to inspect according to the real-time inspection plan. Otherwise, the real-time inspection image is used as the fault inspection image. In the current cycle, the corresponding substation equipment in the fault inspection image is marked as a fault substation equipment, and the fault substation equipment number is recorded. The inspection fault maintenance module inputs the fault inspection image into the inspection augmented reality model for processing, and obtains the augmented reality image of the faulty substation equipment; provides maintenance guidance for the faulty substation equipment based on the augmented reality image of the faulty substation equipment and the fault signal, and obtains the fault maintenance instruction; based on the fault maintenance instruction, the robot maintains the faulty substation equipment, and after the maintenance is completed, the robot continues to inspect according to the real-time inspection plan; The inspection plan update module is used to update the real-time inspection plan by using the swarm optimization algorithm after the robot inspection is completed in the current cycle, obtain a new real-time inspection plan for the next cycle, and unmark the substation equipment marked as faulty substation equipment.

Images