Unmanned aerial vehicle airport-based power grid construction supervision method and system, and medium

Abstract

The application provides a power grid construction supervision method and system based on a UAV airport, and a medium, and relates to the technical field of operation supervision. The method comprises the following steps: obtaining supervision target information corresponding to a power grid construction area, and planning a target supervision path of a UAV corresponding to each airport according to the supervision target information, wherein the target supervision path is a path of the UAV from the airport to the power grid construction area; obtaining target image data generated by the UAV during execution of the supervision path; inputting the target image data into a pre-trained anomaly detection model to obtain an anomaly detection result; and supervising the power grid construction according to the anomaly detection result. The method of the application improves the efficiency and accuracy of power grid construction supervision, thereby improving the safety of on-site operation.

Description

Technical Field

[0001] This application relates to the field of operation monitoring technology, and in particular to a method, system and medium for monitoring power grid construction based on unmanned aerial vehicle (UAV) airports. Background Technology

[0002] With the gradual rollout of power grid construction across entire lines and areas, the numerous and widespread work sites have become increasingly prominent, leading to a severe shortage of supervisory personnel. Therefore, effectively supervising power grid construction by covering all work sites has become crucial.

[0003] In related technologies, remote video monitoring equipment such as surveillance cameras and enforcement recorders can be installed in power grid construction areas. Supervisory personnel can then manually identify any abnormal behaviors or conditions during power grid construction based on the video images collected by these devices. However, the placement of surveillance cameras and enforcement recorders is determined by humans, which can easily lead to blind spots in supervision. Furthermore, manual identification of anomalies is not only inefficient but also prone to errors, resulting in poor supervision and affecting the safety of on-site operations.

[0004] Therefore, there is a need for a power grid construction supervision scheme that can improve the efficiency and accuracy of power grid construction supervision, thereby enhancing the safety of on-site operations. Summary of the Invention

[0005] This application provides a method, system, and medium for supervising power grid construction based on unmanned aerial vehicle (UAV) airports, which can improve the efficiency and accuracy of power grid construction supervision, thereby enhancing the safety of on-site operations.

[0006] In a first aspect, embodiments of this application provide a method for supervising power grid construction based on unmanned aerial vehicle (UAV) airports, including:

[0007] Obtain the monitoring target information corresponding to the power grid construction area, and plan the target monitoring path for each airport's drones based on the monitoring target information. The target monitoring path is the path taken by the drone from its airport to the power grid construction area.

[0008] Acquire target image data generated by the drone during execution according to the monitoring path;

[0009] The target image data is input into a pre-trained anomaly detection model to obtain anomaly detection results;

[0010] The construction of the power grid is supervised based on the anomaly detection results.

[0011] Secondly, embodiments of this application provide a job monitoring system, including:

[0012] The path planning module is used to obtain the monitoring target information corresponding to the power grid construction area, and plan the target monitoring path for each airport's drones based on the monitoring target information. The target monitoring path is the path taken by the drone from its airport to the power grid construction area.

[0013] An anomaly detection module is used to acquire target image data generated by the UAV during execution according to the monitoring path; input the target image data into a pre-trained anomaly detection model to obtain anomaly detection results; and monitor power grid construction based on the anomaly detection results.

[0014] Thirdly, embodiments of this application provide another job monitoring system, including:

[0015] The processor, and the memory that is communicatively connected to the processor;

[0016] Memory is used to store instructions that the computer executes;

[0017] The processor is configured to execute computer execution instructions stored in memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0018] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect described above.

[0019] Fifthly, embodiments of this application provide a computer program product, including a computer program, which, when executed by a processor, is used to implement the first aspect and / or various possible implementations of the first aspect as described above.

[0020] This application provides a method, system, and medium for supervising power grid construction based on unmanned aerial vehicle (UAV) airports. It can plan target supervision paths for UAVs at each airport based on the supervision target information corresponding to the power grid construction area. Using these target supervision paths, a full-view supervision of the power grid construction area can be achieved, minimizing blind spots. When UAVs at each airport execute according to the supervision path, they can acquire target image data. This target image data is input into a pre-trained anomaly detection model for anomaly detection and localization. Based on the anomaly detection results, power grid construction can be supervised quickly and accurately, greatly improving the efficiency and accuracy of power grid construction supervision, avoiding errors caused by manual identification, and thus improving the safety of on-site operations. Attached Figure Description

[0021] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0022] Figure 1 This is a system architecture diagram of an embodiment of this application;

[0023] Figure 2 A flowchart illustrating a power grid construction monitoring method based on an unmanned aerial vehicle (UAV) airport according to an embodiment of this application;

[0024] Figure 3 This is a schematic diagram of the structure of a neural network model according to an embodiment of this application;

[0025] Figure 4 This is a schematic diagram of the structure of a work monitoring system according to an embodiment of this application;

[0026] Figure 5 This is a schematic diagram of the structure of a job monitoring system according to another embodiment of this application.

[0027] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0028] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application.

[0029] The terms “first,” “second,” “third,” “fourth,” etc., as used herein (if applicable), are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0030] The collection, storage, use, processing, transmission, provision, and disclosure of financial data or user data involved in the technical solution of this application all comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0031] It should be noted that in the embodiments of this application, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of this application. However, it does not mean that the applicant has used or necessarily used the solution.

[0032] It should also be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with relevant laws, regulations and standards, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0033] The power grid construction supervision method, system, and medium based on unmanned aerial vehicle (UAV) airports of this application can be used in the field of operation supervision, or in any field other than operation supervision, such as automation. The application fields of the power grid construction supervision method, system, and medium based on UAV airports of this application are not limited.

[0034] The power grid construction supervision method, system, and medium based on unmanned aerial vehicle (UAV) airports of this application can be applied to scenarios that utilize UAV airports for supervision, such as scenarios that utilize UAV airports to supervise power grid construction areas. As long as supervision is involved, the power grid construction supervision method, system, and medium based on UAV airports of this application can be applied.

[0035] With the gradual rollout of power grid construction across entire lines and areas, the numerous and widespread work sites have become increasingly prominent, leading to a severe shortage of supervisory personnel. Therefore, effectively supervising power grid construction by covering all work sites has become crucial.

[0036] Among related technologies, remote video supervision equipment such as surveillance cameras and law enforcement recorders can be set up in the power grid construction area. Supervisory personnel can manually identify whether there are any abnormal behaviors or conditions that violate regulations during the power grid construction process based on the video images collected by these devices.

[0037] However, the placement of devices such as surveillance cameras and law enforcement recorders is determined by humans, which can easily lead to blind spots in supervision. Furthermore, manual identification of anomalies is not only inefficient but also prone to errors, resulting in poor supervision and affecting the safety of on-site operations.

[0038] Based on the above-mentioned technical problems, the inventive concept of this application is to provide a power grid construction supervision scheme that can improve the efficiency and accuracy of power grid construction supervision, thereby improving the safety of on-site operations.

[0039] This application provides a method, system, and medium for supervising power grid construction based on unmanned aerial vehicle (UAV) airports. It can plan target supervision paths for UAVs at each airport based on the supervision target information corresponding to the power grid construction area. Using these target supervision paths, a full-view supervision of the power grid construction area can be achieved, minimizing blind spots. When UAVs at each airport execute according to the supervision path, they can acquire target image data. This target image data is input into a pre-trained anomaly detection model for anomaly detection and localization. Based on the anomaly detection results, power grid construction can be supervised quickly and accurately, greatly improving the efficiency and accuracy of power grid construction supervision, avoiding errors caused by manual identification, and thus improving the safety of on-site operations.

[0040] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0041] Figure 1 This is a system architecture diagram of one embodiment of this application, such as... Figure 1 As shown, the operation supervision system can communicate with the task scheduling systems of various drone airports. The operation supervision system can obtain the supervision target information corresponding to the power grid construction area, plan the target supervision path for each drone at each airport based on the supervision target information, and send the target supervision path to the task scheduling system. The task scheduling system can then schedule each drone in the airport according to the target supervision path to execute flight missions. The drones are equipped with cameras that can communicate with the operation supervision system, sending the target image data collected by the drones during flight according to the supervision path to the operation supervision system in real time. The operation supervision system can also input the target image data into a pre-trained anomaly detection model to obtain anomaly detection results; and supervise the power grid construction based on the anomaly detection results.

[0042] Figure 2 This is a flowchart illustrating a method for supervising power grid construction based on unmanned aerial vehicle (UAV) airports, according to an embodiment of this application. This embodiment describes the method using an operational monitoring system as the implementing entity. Figure 2 As shown, the method for supervising power grid construction based on unmanned aerial vehicle (UAV) airports may include the following steps:

[0043] S201: Obtain the regulatory target information corresponding to the power grid construction area, and plan the target regulatory path for each airport's drones based on the regulatory target information.

[0044] In this embodiment, the target monitoring path can be the path taken by the drone from its airport to the power grid construction area.

[0045] It should be noted that the airports in the embodiments of this application are all unmanned aerial vehicle (UAV) airports (UAV nests), which are base facilities that provide one-stop services for UAVs such as take-off and landing, parking, charging / battery swapping, data exchange, scheduling management and maintenance.

[0046] In this embodiment, the operation monitoring system can be a terminal device such as a computer or tablet, such as an edge device for power grid construction, or a server-side device such as a physical server or cloud server, or an integrated terminal-server device. There are no restrictions here, as long as the operation monitoring system can be loaded with an anomaly detection model and can communicate with the task scheduling system of each airport.

[0047] In this embodiment, the number and location of the drone airports can be flexibly set by those skilled in the art according to actual needs, as long as each drone airport (and the drones within it) can achieve full-view monitoring of the power grid construction area, and no restrictions are imposed here.

[0048] In this embodiment, the monitoring target information may be the latitude and longitude information of the power grid construction area, or other information that can characterize the specific location of the power grid construction area.

[0049] In this embodiment, the target monitoring path for each airport's drones can be the mileage, time, and flight sequence for each airport's drones to reach the power grid construction area for monitoring. It is necessary to ensure that when each airport's drones execute the target monitoring path, they can essentially cover the entire field of view of the power grid construction area, i.e., minimizing monitoring blind spots. For example, when drone A's battery level drops below 20%, drone B automatically takes over the monitoring task. Alternatively, if the coverage area is exceeded, other drones can monitor the power grid construction operation.

[0050] S202: Acquire target image data generated by the drone during its execution according to the regulatory path.

[0051] In this embodiment, the task scheduling system of the UAV airport can communicate with the operation supervision system. After the operation supervision system plans the target supervision path, it can send the target supervision path to the task scheduling system. The task scheduling system can schedule each UAV in the airport according to the target supervision path to execute flight missions.

[0052] In this embodiment, the drone may be equipped with a camera, which can communicate with the operation monitoring system to send the target image data collected during the flight to the operation monitoring system in real time.

[0053] S203: Input the target image data into the pre-trained anomaly detection model to obtain the anomaly detection result.

[0054] In this embodiment, those skilled in the art can pre-train an anomaly detection model based on sample data from the power grid construction process, enabling the model to identify anomalies such as unsafe human behavior and unsafe conditions of objects. After acquiring the target image data collected in real time, the target image data is input into the anomaly detection model to obtain the anomaly detection result.

[0055] In this embodiment, the anomaly detection result can be whether an anomaly exists, and the location of the anomaly if it exists.

[0056] S204: Supervise power grid construction based on anomaly detection results.

[0057] In this embodiment, if no abnormality is found, the operation monitoring system can continue to perform abnormality detection based on the target image data acquired in real time; if an abnormality is found, the operation monitoring system can capture an image based on the location of the abnormality to accurately locate the abnormal area, and generate an abnormality prompt in the form of an image visualization and send it to the monitoring terminal to prompt the supervisory personnel to take corresponding measures in a timely manner to avoid the occurrence of safety accidents.

[0058] In this embodiment, based on the regulatory target information corresponding to the power grid construction area, target monitoring paths for drones at each airport can be planned. These paths enable comprehensive monitoring of the power grid construction area, minimizing blind spots. When drones at each airport execute the monitoring paths, they acquire target image data. This data is then input into a pre-trained anomaly detection model for anomaly detection and localization. Based on the anomaly detection results, power grid construction can be monitored quickly and accurately, significantly improving the efficiency and accuracy of power grid construction monitoring, avoiding errors caused by manual identification, and thus enhancing the safety of on-site operations.

[0059] In one possible implementation, the monitoring target information may include the latitude and longitude information of the power grid construction area, and the step S201 above, which involves planning the target monitoring path for each airport's UAV based on the monitoring target information, may include:

[0060] S11: Input the latitude and longitude information of the power grid construction area and the UAV airport information into the path planning system for initialization to obtain the initial monitoring path for each UAV at each airport. The initial monitoring path includes the flight time and mileage of the UAV to the power grid construction area.

[0061] S12: Optimize the initial monitoring path using a genetic algorithm to obtain the target monitoring path with the minimum loss when drones from each airport travel to the power grid construction area.

[0062] The information on drone airports includes the number of drone airports, their locations and identifiers, the number of drones in each drone airport, and the locations and identifiers of the drones in each drone airport.

[0063] In this embodiment, the initial monitoring path for each airport's drone needs to achieve full-view monitoring of the power grid construction area, minimizing blind spots in monitoring. There can be one or more initial monitoring paths.

[0064] In this embodiment, the steps of initializing using a path planning system to obtain the initial monitoring paths for drones at each airport can be referred to in relevant existing technologies and will not be elaborated here.

[0065] In this embodiment, losses may include flight time and flight mileage, and the losses of the monitoring path can be expressed using the following formula (1):

[0066] Sum(k)={T(k)×Wt+S(k)×Ws+ΔT} (1)

[0067] Where Sum(k) represents the loss of the k-th path, T(k) represents the total monitoring time for drones from each airport to reach the power grid construction area in the k-th path, S(k) represents the total monitoring distance for drones from each airport to reach the power grid construction area in the k-th path, and k represents the number of paths; Wt and Ws are weighting coefficients, 0≤Wt≤1, 0≤Ws≤1, and Wt+Ws=1; ΔT is the charging time of drones at each airport.

[0068] In this embodiment, after obtaining the initial regulatory path, the genetic algorithm can be used to optimize the initial regulatory path to obtain multiple regulatory paths. The loss of each regulatory path can be calculated using the above formula (1), and the path with the smallest loss is the target regulatory path.

[0069] In this embodiment, after obtaining the initial monitoring path that can realize the monitoring of the entire power grid construction area from all angles, it is also necessary to optimize the initial monitoring path with the goal of minimizing losses using a genetic algorithm, and select the target monitoring path with the minimum flight loss, effectively avoiding problems such as redundant routes and detours, and minimizing the power consumption and mechanical losses of the UAV.

[0070] In one possible implementation, the above-described anomaly detection model can be trained in the following manner:

[0071] S21: Construct a database related to power grid construction and preprocess the data in the database to obtain a sample dataset.

[0072] S22: Based on the sample dataset and the truncated L1 loss function, iteratively train the neural network model to obtain a pre-trained anomaly detection model.

[0073] In this embodiment, the database may include various image data related to power grid construction, and the sample dataset may include positive samples and negative samples. Positive samples may be image data of normal power grid construction, and negative samples may be data of abnormal power grid construction.

[0074] In this embodiment, the neural network model can be a deep learning model or other classification models, and no restrictions are imposed here.

[0075] In this embodiment, by constructing and preprocessing a dedicated database for power grid construction, a high-quality, highly consistent sample dataset is formed, significantly improving the availability and reliability of the data and providing a precise data foundation for subsequent model training. The use of a truncated L1 loss function for iterative training of the neural network model effectively suppresses the interference of extreme outliers in the power grid construction data on model training. This results in a more robust and generalizable anomaly detection model, capable of more accurately capturing latent anomaly features during power grid construction, and improving the model's detection accuracy and stability in complex construction scenarios.

[0076] In one possible implementation, step S21 above involves constructing a database related to power grid construction and preprocessing the data in the database to obtain a sample dataset, which may include:

[0077] S31: Construct a database related to power grid construction. The database includes sample image data of drones monitoring power grid construction. The sample image data includes positive and negative samples. The negative samples include unsafe human behavior and unsafe conditions of objects.

[0078] S32: Preprocess the sample image data to obtain preprocessed sample data. Preprocessing includes one or more of the following: damaged image deletion, image correction, image scaling, random cropping, data augmentation, and standardization.

[0079] S33: Divide the preprocessed sample data into a training set, a validation set, and a test set. The training set includes only positive samples, the validation set includes only negative samples, and the test set includes both positive and negative samples.

[0080] In this embodiment, negative samples may include image data of unsafe acts of people and unsafe conditions of objects that violate the power grid's "On-site Operation Violation Deduction Clauses" and "Safety Regulations," etc.; positive samples may include image data of safe acts of people and safe conditions of objects that comply with various clauses and regulations of power grid construction.

[0081] In this embodiment, preprocessing may include, but is not limited to, damaged image deletion, image correction, image scaling, random cropping, data augmentation, and standardization. Any processing that can improve the quality and quantity of sample data can be used as preprocessing.

[0082] In this embodiment, the training set can be used to generate normal features during model training, the validation set can be used to generate abnormal features during model training, and the test set can be used for performance testing after model training.

[0083] In this embodiment, deleting damaged images may involve checking whether sample images are damaged or incomplete, and deleting images that are unusable.

[0084] In this embodiment, image correction can be used to correct image distortion caused by drone movement or camera angle, including lens distortion correction, perspective correction, and color correction.

[0085] In this implementation, image scaling, random cropping, data augmentation, and standardization can effectively increase the amount of data samples, helping the model to better generalize to new data.

[0086] In this embodiment, the sample image data can include both positive and negative samples, ensuring comprehensive coverage and clear categories. Preprocessing the sample image data significantly improves its usability, consistency, and richness. A hierarchical partitioning strategy—with the training set containing only positive samples, the validation set containing only negative samples, and the test set containing both positive and negative samples—achieves precise and targeted optimization of the model training process. The pure positive samples in the training set allow the model to fully learn the image features of normal power grid construction scenarios, quickly building a benchmark feature library. The pure negative samples in the validation set accurately verify the model's ability to identify unsafe behaviors / states during iterative training, guiding the model to focus on capturing abnormal features and avoiding feature learning confusion caused by mixed positive and negative sample training. The mixed samples in the test set comprehensively verify the model's overall recognition performance in real-world scenarios, significantly improving training efficiency, shortening the model convergence cycle, and reducing false positive and false negative rates.

[0087] In one possible implementation, step S22 above, which iteratively trains the neural network model based on the sample dataset and the truncated L1 loss function to obtain a pre-trained anomaly detection model, may include:

[0088] S41: Input the training set into the large-scale visual base model of the neural network model for feature extraction, and input the extracted features into the adapter of the neural network model to obtain the normal adaptation features.

[0089] S42: Input the validation set into the large-scale visual base model of the neural network model for feature extraction, and input the extracted features into the adapter of the neural network model to obtain the abnormal fitting features.

[0090] S43: Input the abnormal fitting features into the generator of the neural network model to obtain the abnormal fitting features after noise perturbation.

[0091] S44: Input the normal adaptation features and the abnormal adaptation features after noise perturbation into the discriminator of the neural network model, and iteratively train the neural network model according to the output of the discriminator and the truncated L1 loss function to obtain the pre-trained anomaly detection model.

[0092] In this embodiment, Figure 3 This is a schematic diagram of the structure of a neural network model according to an embodiment of this application, as shown below. Figure 3 As shown, a neural network model can include a large-scale vision base model, an adapter, a generator, and a discriminator. The large-scale vision base model can be used to extract features at different depths of an image. The adapter can be used to transfer the extracted features to UAV airport data. The generator can be used to generate anomalous features. The discriminator can be used to determine whether the features are normal or anomalous.

[0093] In this implementation, the large-scale visual foundation model can extract features at different depths from an image, with the outputs of different layers representing features at different levels of abstraction. Shallower layers typically capture lower-level local features, while deeper layers capture more complex global features. Specifically,

[0094] ① The input image can be segmented into fixed-size blocks, each block is flattened into a one-dimensional vector, and mapped to a high-dimensional space through a linear transformation. These vectors are then augmented with location information (location embedding) to preserve spatial information.

[0095] ② Each input vector computes its relationship with all other vectors through a multi-head self-attention mechanism. This mechanism captures global information, helping the model understand the relationships between different parts of the image. Multi-head attention allows the model to learn different feature representations in different subspaces. After self-attention, each vector undergoes a non-linear transformation through a feedforward neural network. This process is somewhat similar to the activation layers in a convolutional network.

[0096] In this implementation, after the large-scale visual base model provides features at different scales to the adapter, the feature pyramid network first samples the deepest high-level features layer by layer upwards, and then fuses them with the corresponding low-level features through lateral connections. This not only preserves high-level semantic information, but also combines low-level spatial details, thereby generating a pyramid hierarchical structure with rich feature representations, and thus transferring the extracted features to UAV airport data.

[0097] In this embodiment, the generator can add noise perturbation to the abnormal adaptation features output by the adapter to generate a greater number of abnormal features.

[0098] In this embodiment, the discriminator acts as a normality scorer, directly estimating whether each location in the image features is normal. During training, the discriminator receives normal features and abnormal features perturbed by Gaussian noise as input, aiming to output positive values ​​for normal features and negative values ​​for abnormal features. The discriminator uses a two-layer multilayer perceptron structure, and the output normality score represents the degree of normality of the features, used to distinguish between normal and abnormal situations.

[0099] In this implementation, a large-scale visual foundation model is used to extract features from power grid construction images. Combined with the scene adaptation capability of the adapter, general features are accurately converted into power grid construction-specific features, significantly improving the representation and recognition capabilities of the features. Noise perturbations are added to the negative sample adaptation features by the generator to simulate abnormal feature patterns under complex interference factors in power grid construction scenarios, expanding the diversity and coverage of abnormal samples. This allows the model to learn more comprehensive abnormal patterns, significantly improving its ability to identify latent anomalies under complex working conditions and reducing the risk of missed detections due to the scarcity of abnormal samples. An iterative training method using a generative adversarial architecture combined with a truncated L1 loss function is employed. The discriminator accurately distinguishes between normal and abnormal features by comparing normal adaptation features with noise-perturbed abnormal adaptation features. The truncated L1 loss function effectively suppresses the interference of extreme noise perturbation samples on model training, avoiding discrimination bias caused by overfitting to extreme abnormal data, significantly improving the model's robustness and generalization ability, shortening the model convergence cycle, and reducing the false detection rate and overfitting risk.

[0100] In one possible implementation, step S203 above, which inputs the target image data into a pre-trained anomaly detection model to obtain anomaly detection results, may include:

[0101] S51: Input the target image data into the large-scale visual base model of the anomaly detection model for feature extraction to obtain features at different scales.

[0102] S52: Input features of different scales into the adapter of the anomaly detection model for upsampling and lateral connection fusion to obtain adapted features.

[0103] S53: Input the adaptation features into the discriminator of the anomaly detection model to obtain the anomaly detection score corresponding to the target image data.

[0104] In this embodiment, the anomaly detection model removes the generator from the training process, including the large-scale visual base model, adapter, and discriminator.

[0105] In this embodiment, after providing the adaptation features to the discriminator, an anomaly score map can be obtained. Then, interpolation processing is performed to adjust the size of the anomaly score map relative to the input. Figure 1 The score at each position represents the anomaly of the corresponding pixel in the input image. Gaussian filtering (σ=4) can be used to smooth the boundaries.

[0106] In this embodiment, since there is a most sensitive point for any size abnormal region, the highest score of the abnormal mapping can be used as the abnormal detection score corresponding to the target image data.

[0107] In this embodiment, after the target image data is input into the anomaly detection model, steps such as image feature extraction, feature scene transfer and adaptation, anomaly recognition and detection are performed in sequence. The anomaly detection score corresponding to the target image data can then be output. This anomaly detection score can be used to determine whether there is an anomaly in the target image data, and to locate the anomaly when it exists, thereby improving the accuracy of power grid construction supervision.

[0108] In one possible implementation, step S204, which involves monitoring power grid construction based on anomaly detection results, may include:

[0109] S61: Determine whether the anomaly detection score is less than the preset score threshold.

[0110] S62: If the score is less than the threshold, then the target image data is determined to be free of job anomalies.

[0111] S63: If the score is greater than or equal to the threshold, it is determined that there is an operational anomaly in the target image data, and the target image data and operational anomaly prompt information are output to the monitoring terminal.

[0112] In this embodiment, the fractional threshold can be flexibly set by those skilled in the art according to actual conditions, and no restrictions are imposed here.

[0113] In this embodiment, the abnormal operation prompt information may include one or more of the following: sound information, image information, text information, and light information.

[0114] In this embodiment, abnormal regions (identification boxes, etc., automatically generated by the abnormal detection model) corresponding to the abnormal detection scores can be identified in the target image data. The visualization of the image allows regulatory personnel to quickly locate abnormal locations.

[0115] In this embodiment, if the anomaly detection score is less than the score threshold, it indicates that there is no operational anomaly in the target image data, and the step of obtaining the target image data generated by the UAV during its execution according to the monitoring path can continue.

[0116] In this embodiment, after obtaining the anomaly detection score, the score is compared with a preset score threshold to quickly and accurately determine whether there is an operational anomaly in the target image data. If an operational anomaly is found, the target image data and operational anomaly alert information can be output to the monitoring terminal to prompt supervisory personnel to take timely measures based on the visualized anomaly location, thus preventing safety accidents from occurring.

[0117] The following is a specific embodiment illustrating the power grid construction supervision method based on unmanned aerial vehicle (UAV) airports of this application.

[0118] In one specific embodiment, power grid construction is being carried out in a certain area, and a drone airport is set up around the power grid construction area. The monitoring process for the power grid construction area is as follows:

[0119] The first step is to input the latitude and longitude information of the power grid construction area and the information of the UAV airport into the path planning system for initialization, so as to obtain the initial monitoring path for each UAV at each airport.

[0120] The second step involves the operation monitoring system using a genetic algorithm to optimize the initial monitoring path, obtaining the target monitoring path for each airport's drones to reach the power grid construction area with the least loss.

[0121] The third step involves supervisors collecting sample image data from drones used for monitoring power grid construction and transmitting the sample image data to the operation monitoring system to build a database related to power grid construction.

[0122] The fourth step involves the operation monitoring system preprocessing the sample image data to obtain preprocessed sample data; the preprocessed sample data is then divided into training set, validation set, and test set.

[0123] Fifth, the operation monitoring system inputs the training set into the large-scale visual base model of the neural network model for feature extraction, and inputs the extracted features into the adapter of the neural network model to obtain normal adaptation features; inputs the validation set into the large-scale visual base model of the neural network model for feature extraction, and inputs the extracted features into the adapter of the neural network model to obtain abnormal adaptation features.

[0124] The sixth step involves the operation monitoring system inputting the negative sample adaptation features into the generator of the neural network model to obtain the abnormal adaptation features after noise perturbation; inputting the normal adaptation features and the abnormal adaptation features after noise perturbation into the discriminator of the neural network model, and iteratively training the neural network model based on the output of the discriminator and the truncated L1 loss function to obtain the pre-trained anomaly detection model.

[0125] The seventh step is for the operation monitoring system to send the target monitoring path to the task scheduling system. The task scheduling system then schedules each UAV in the airport according to the target monitoring path to execute flight missions.

[0126] The eighth step involves the operation monitoring system acquiring target image data generated by the drone during its execution according to the monitoring path; inputting the target image data into a pre-trained anomaly detection model to obtain the corresponding anomaly detection score.

[0127] In the ninth step, if the operation monitoring system detects that the anomaly detection score corresponding to the target image data A is greater than the preset score threshold, it determines that there is an operation anomaly in the target image data A and outputs the target image data A and the operation anomaly prompt information to the monitoring terminal.

[0128] Figure 4 This is a schematic diagram of the structure of a work monitoring system according to an embodiment of this application, as shown below. Figure 4 As shown, the operation supervision system includes: a path planning module 41, used to acquire the supervision target information corresponding to the power grid construction area, and plan the target supervision path corresponding to the UAVs at each airport according to the supervision target information, wherein the target supervision path is the path of the UAV from its airport to the power grid construction area; an anomaly detection module 42, used to acquire the target image data generated by the UAV during the execution of the supervision path; input the target image data into a pre-trained anomaly detection model to obtain the anomaly detection result; and supervise the power grid construction according to the anomaly detection result.

[0129] The operation monitoring system provided in this application embodiment can execute the technical solutions shown in the above method embodiments. Its implementation principle and beneficial effects are similar, and will not be described again here.

[0130] Figure 5 This is a schematic diagram of the structure of a work monitoring system according to another embodiment of this application, as shown below. Figure 5 As shown, the operation monitoring system includes: a processor 501 and a memory 502 communicatively connected to the processor 501; the memory 502 stores computer execution instructions; the processor 501 executes the computer execution instructions stored in the memory 502 to implement the steps of the power grid construction monitoring method based on the UAV airport in the above method embodiments.

[0131] In the aforementioned job monitoring system, the memory 502 and the processor 501 are electrically connected directly or indirectly to enable data transmission or interaction. For example, these components can be electrically connected to each other via one or more communication buses or signal lines, such as a bus connection. The memory 502 stores computer execution instructions that implement data access control methods, including at least one software function module that can be stored in the memory 502 in the form of software or firmware. The processor 501 executes various functional applications and data processing by running the software programs and modules stored in the memory 502.

[0132] The memory 502 may be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), etc. The memory 502 stores programs, which are then executed by the processor 501 upon receiving execution instructions. Furthermore, the software programs and modules within the memory 502 may include an operating system, which may include various software components and / or drivers for managing system tasks (e.g., memory management, storage device control, power management, etc.) and can communicate with various hardware or software components to provide an operating environment for other software components.

[0133] Processor 501 can be an integrated circuit chip with signal processing capabilities. The aforementioned processor 501 can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor.

[0134] An embodiment of this application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the steps of the various method embodiments of this application.

[0135] An embodiment of this application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the various method embodiments of this application.

[0136] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to this application.

[0137] It should be further noted that although the steps in the flowchart are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowchart may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0138] It should be understood that the above-described device embodiments are merely illustrative, and the device of this application can also be implemented in other ways. For example, the division of units / modules in the above embodiments is only a logical functional division, and there may be other division methods in actual implementation. For example, multiple units, modules, or components may be combined, or integrated into another system, or some features may be ignored or not executed.

[0139] Furthermore, unless otherwise specified, the functional units / modules in the various embodiments of this application can be integrated into one unit / module, or each unit / module can exist physically separately, or two or more units / modules can be integrated together. The integrated units / modules described above can be implemented in hardware or as software program modules.

[0140] In the above embodiments, the descriptions of each embodiment have their own emphasis. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments. The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0141] Finally, it should be noted that other embodiments of this application will readily conceive of by those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes may be made without departing from its scope.

Claims

1. A method for supervising power grid construction based on unmanned aerial vehicle (UAV) airports, characterized in that, include: Obtain the monitoring target information corresponding to the power grid construction area, and plan the target monitoring path for each airport's drones based on the monitoring target information. The target monitoring path is the path taken by the drone from its airport to the power grid construction area. Acquire target image data generated by the drone during execution according to the monitoring path; The target image data is input into a pre-trained anomaly detection model to obtain anomaly detection results; The construction of the power grid is supervised based on the anomaly detection results.

2. The method for supervising power grid construction based on unmanned aerial vehicle (UAV) airports according to claim 1, characterized in that, The monitoring target information includes the latitude and longitude information of the power grid construction area, and the step of planning the target monitoring path for each airport's drones based on the monitoring target information includes: The latitude and longitude information of the power grid construction area and the UAV airport information are input into the path planning system for initialization to obtain the initial monitoring path corresponding to the UAVs at each airport. The initial monitoring path includes the flight time and mileage of the UAVs to the power grid construction area. The initial monitoring path is optimized using a genetic algorithm to obtain the target monitoring path with the minimum loss when the drones from each airport travel to the power grid construction area; The drone airport information includes the number of drone airports, the location and identification of the drone airports, the number of drones in each drone airport, and the location and identification of the drones in each drone airport.

3. The method for supervising power grid construction based on unmanned aerial vehicle (UAV) airports according to claim 1 or 2, characterized in that, The step of inputting the target image data into a pre-trained anomaly detection model to obtain anomaly detection results includes: The target image data is input into the large-scale visual base model of the anomaly detection model for feature extraction to obtain features at different scales; The features at different scales are input into the adapter of the anomaly detection model for upsampling and lateral connection fusion to obtain adaptive features; The adaptation features are input into the discriminator of the anomaly detection model to obtain the anomaly detection score corresponding to the target image data.

4. The method for supervising power grid construction based on unmanned aerial vehicle (UAV) airports according to claim 3, characterized in that, The supervision of power grid construction based on the anomaly detection results includes: Determine whether the anomaly detection score is less than a preset score threshold; If the score is less than the threshold, it is determined that the target image data does not have any operational anomalies. If the score is greater than or equal to the threshold, it is determined that there is an operational anomaly in the target image data, and the target image data and operational anomaly prompt information are output to the monitoring terminal.

5. The method for supervising power grid construction based on unmanned aerial vehicle (UAV) airports according to claim 1 or 2, characterized in that, The anomaly detection model was trained in the following manner: A database related to power grid construction was constructed, and the data in the database was preprocessed to obtain a sample dataset; Based on the sample dataset and the truncated L1 loss function, the neural network model is iteratively trained to obtain a pre-trained anomaly detection model.

6. The method for supervising power grid construction based on unmanned aerial vehicle (UAV) airports according to claim 5, characterized in that, The process involves constructing a database related to power grid construction and preprocessing the data in the database to obtain a sample dataset, including: A database related to power grid construction is constructed. The database includes sample image data of drones monitoring power grid construction. The sample image data includes positive samples and negative samples. The negative samples include unsafe human behaviors and unsafe conditions of objects. The sample image data is preprocessed to obtain preprocessed sample data. The preprocessing includes one or more of the following: damaged image deletion, image correction, image scaling, random cropping, data augmentation, and normalization. The preprocessed sample data is divided into a training set, a validation set, and a test set. The training set includes only positive samples, the validation set includes only negative samples, and the test set includes both positive and negative samples.

7. The method for supervising power grid construction based on unmanned aerial vehicle (UAV) airports according to claim 6, characterized in that, The step of iteratively training the neural network model based on the sample dataset and the truncated L1 loss function to obtain a pre-trained anomaly detection model includes: The training set is input into the large-scale visual base model of the neural network model for feature extraction, and the extracted features are input into the adapter of the neural network model to obtain normal adaptation features. The validation set is input into the large-scale visual base model of the neural network model for feature extraction, and the extracted features are input into the adapter of the neural network model to obtain abnormal adaptation features; The abnormal fitting features are input into the generator of the neural network model to obtain the abnormal fitting features after noise perturbation. The normal adaptation features and the abnormal adaptation features after noise perturbation are input into the discriminator of the neural network model, and the neural network model is iteratively trained according to the output of the discriminator and the truncated L1 loss function to obtain a pre-trained anomaly detection model.

8. A work monitoring system, characterized in that, include: The path planning module is used to obtain the monitoring target information corresponding to the power grid construction area, and plan the target monitoring path for each airport's drones based on the monitoring target information. The target monitoring path is the path taken by the drone from its airport to the power grid construction area. An anomaly detection module is used to acquire target image data generated by the UAV during execution according to the monitoring path; input the target image data into a pre-trained anomaly detection model to obtain anomaly detection results; and monitor power grid construction based on the anomaly detection results.

9. A work monitoring system, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory is used to store computer-executed instructions; The processor is used to execute computer execution instructions stored in the memory, causing the processor to execute the power grid construction supervision method based on unmanned aerial vehicle airport as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the power grid construction supervision method based on any one of claims 1-7.

Images