Method and apparatus for anomaly detection and analysis of power transmission line, electronic device and medium

Abstract

The present application provides a method and apparatus for anomaly detection and analysis of a power transmission line, an electronic device, and a medium. The present application relates to the technical field of electric power. An acquired power transmission line image of a surrounding environment of a power transmission line and prompt information are inputted into an anomaly detection and analysis model to obtain an anomaly analysis result of the power transmission line. The anomaly detection and analysis model is pre-stored, and the anomaly detection and analysis model is obtained by performing parameter adjustment on an initial anomaly detection and analysis model by means of power transmission line sample images, sample prompt information, and sample response information. Each power transmission line sample image contains a labeled anomaly feature and / or confusion feature, the sample prompt information is used for indicating a response type that needs to be obtained, and the sample response information refers to a response corresponding to the prompt information. Thus, the obtained anomaly detection and analysis model can accurately identify an anomaly in the power transmission line, thereby reducing the misjudgment rate of the anomaly, prompting the potential risk of the power transmission line in a timely manner, and improving the safety of the power transmission line.

Description

Methods, devices, electronic equipment and media for detecting and analyzing anomalies in power transmission lines

[0001] This application claims priority to Chinese Patent Application No. 2025102993074, filed on March 13, 2025, entitled “Method, Apparatus, Electronic Device and Medium for Detection and Analysis of Transmission Line Anomalies”, the entire contents of which are incorporated herein by reference. Technical Field

[0002] This application relates to the field of power technology, and in particular to a method, apparatus, electronic device and medium for detecting and analyzing anomalies in power transmission lines. Background Technology

[0003] Overhead transmission lines are power transmission lines that use insulators to fix the transmission conductors to towers that stand upright on the ground in order to transmit electrical energy.

[0004] The external environment of overhead transmission line corridors is complex, making them susceptible to weather and environmental influences, which can lead to faults. It is necessary to identify anomalies in overhead transmission lines to detect risks promptly and facilitate timely risk management.

[0005] However, the current method for identifying anomalies in overhead transmission lines has a high false alarm rate. Summary of the Invention

[0006] This application provides a method, apparatus, electronic device, and medium for detecting and analyzing anomalies in transmission lines, in order to solve the problem of high false alarm rate in anomaly identification of overhead transmission lines.

[0007] In a first aspect, this application provides a method for detecting and analyzing anomalies in transmission lines, the method comprising:

[0008] Acquire images of power transmission lines and provide prompts; the images of power transmission lines are images of the surrounding environment of the power transmission lines; the prompts are used to indicate the type of response required based on the images of power transmission lines.

[0009] The transmission line image and the prompt information are input into the anomaly detection and analysis model to obtain the anomaly analysis result corresponding to the prompt information; the anomaly analysis result is used to respond to the prompt information.

[0010] The anomaly detection and analysis model is obtained by adjusting the parameters of an initial anomaly detection and analysis model using transmission line sample images, sample prompt information, and sample response information. The initial anomaly detection and analysis model is based on a multimodal large model. Each transmission line sample image contains anomaly features and / or confusing features. The anomaly features refer to features that pose a risk to the transmission line. The confusing features are features whose similarity to the anomaly features is greater than a preset threshold. The sample prompt information is used to indicate the type of response required, and the sample response information refers to the response corresponding to the prompt information.

[0011] In some embodiments, the abnormal features include at least one of the following: foreign object features on the transmission line, fire features, smoke features, first crane features, and hazardous material features; the confusing features include at least one of the following: structural features of the transmission line, light features, cloud features, fog features, and second crane features; the sample prompt information may include at least one of the following: foreign object detection prompt on the transmission line; light, cloud, and fire detection prompt; fog and smoke detection prompt; crane scene detection and analysis prompt; and hazardous material prompt.

[0012] In some embodiments, inputting the transmission line image and the prompt information into an anomaly detection and analysis model to obtain the anomaly analysis result corresponding to the prompt information includes:

[0013] The prompt information is broken down into multiple sub-prompt information stages;

[0014] According to the sequential order of the sub-prompt information of the multiple stages, the transmission line image and the sub-prompt information of the multiple stages are input into the anomaly detection and analysis model in stages to obtain the anomaly analysis results corresponding to the prompt information.

[0015] In some embodiments, the initial anomaly detection and analysis model is based on knowledge distillation, using the first anomaly detection and analysis model as the teacher model to obtain the student model through distillation.

[0016] In some embodiments, the method further includes:

[0017] If the anomaly analysis results of the transmission line indicate that there is an anomaly in the transmission line, an anomaly alarm message is output, which includes information about the transmission line and anomaly type information.

[0018] In some embodiments, the method further includes:

[0019] Acquire sample images, sample prompts, and sample responses for power transmission lines;

[0020] The parameters of the initial anomaly detection and analysis model are adjusted using sample images of transmission lines, sample prompt information, and sample response information to obtain the anomaly detection and analysis model.

[0021] In some embodiments, the method further includes:

[0022] Based on knowledge distillation, the first anomaly detection and analysis model is used as the teacher model, and the student model obtained by distillation is used as the initial anomaly detection and analysis model.

[0023] Secondly, this application provides a method for adjusting a transmission line anomaly detection and analysis model, the method comprising:

[0024] Acquire sample images of transmission lines, sample prompt information, and sample response information; each sample image of a transmission line contains abnormal features and / or confusing features; the abnormal features refer to features that pose a risk to the transmission line; the confusing features are features whose similarity to the abnormal features is greater than a preset threshold; the sample prompt information is used to indicate the type of response required, and the sample response information refers to the response corresponding to the prompt information;

[0025] The parameters of the initial anomaly detection analysis model are adjusted using the transmission line sample images, sample prompt information, and sample response information to obtain the anomaly detection analysis model; the initial anomaly detection analysis model is established based on a multimodal large model.

[0026] In some embodiments, the method further includes:

[0027] Based on knowledge distillation, the first anomaly detection and analysis model is used as the teacher model, and the student model obtained by distillation is used as the initial anomaly detection and analysis model.

[0028] Thirdly, this application provides a transmission line anomaly detection and analysis device, comprising:

[0029] An acquisition module is used to acquire images of transmission lines and prompt information; the images of transmission lines are images of the surrounding environment of the transmission lines; the prompt information is used to indicate the type of response required by viewing the images of transmission lines.

[0030] The processing module is used to input the transmission line image and the prompt information into an anomaly detection and analysis model to obtain the anomaly analysis result corresponding to the prompt information; the anomaly analysis result is used to respond to the prompt information; wherein, the anomaly detection and analysis model is obtained by adjusting the parameters of an initial anomaly detection and analysis model through transmission line sample images, sample prompt information, and sample response information; the initial anomaly detection and analysis model is established based on a multimodal large model; each transmission line sample image contains anomaly features and / or confusing features; the anomaly features refer to features that pose a risk to the transmission line; the confusing features are features whose similarity to the anomaly features is greater than a preset threshold; the sample prompt information is used to indicate the type of response required, and the sample response information refers to the response corresponding to the prompt information.

[0031] Fourthly, this application provides an electronic device including a memory and a processor, wherein the memory stores a computer program executable on the processor, and the processor executes the program to implement the method described in the first aspect above.

[0032] Fifthly, this application provides an electronic device including a memory and a processor, wherein the memory stores a computer program executable on the processor, and the processor executes the program to implement the method described in the second aspect above.

[0033] In a sixth aspect, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in the first aspect above.

[0034] In a seventh aspect, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in the second aspect above.

[0035] The transmission line anomaly detection and analysis method, apparatus, electronic device, and medium provided in this application embodiment involve a server inputting acquired transmission line images and prompt information into an anomaly detection and analysis model to obtain transmission line anomaly analysis results. The server 102 pre-stores the anomaly detection and analysis model, which is obtained by adjusting the parameters of a pre-trained initial anomaly detection and analysis model using transmission line sample images, sample prompt information, and sample response information. The transmission line sample images contain anomalous features and / or confusing features. Anomalous features are those that pose a risk to the transmission line, while confusing features are those with a high degree of similarity to anomalous features. The sample prompt information indicates the type of response required, and the sample response information refers to the response corresponding to the prompt information. By inputting sample images and sample prompts from transmission lines into an initial anomaly detection and analysis model built on a multimodal large model, the initial anomaly detection and analysis model can quickly learn the differences between abnormal features and similar confusing features in the transmission line. It can also learn rapidly through sample prompts, better stimulating the model's own potential. The resulting anomaly detection and analysis model can accurately identify abnormal conditions in the transmission line, reduce the false alarm rate, ensure that risks to the transmission line can be handled in a timely manner, and improve the safety of the transmission line. Attached Figure Description

[0036] Figure 1 is a schematic diagram of the structure of a power transmission line anomaly detection and analysis system provided in an embodiment of this application;

[0037] Figure 2 is a flowchart illustrating a method for detecting and analyzing anomalies in transmission lines provided in an embodiment of this application;

[0038] Figure 3 is a flowchart illustrating a method for adjusting a transmission line anomaly detection and analysis model according to an embodiment of this application;

[0039] Figure 4 is a schematic diagram of the structure of a power transmission line anomaly detection and analysis device provided in an embodiment of this application. Detailed Implementation

[0040] In this application, "at least one" means one or more, and "more than one" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c alone can mean: a alone, b alone, c alone, a combination of a and b, a combination of a and c, a combination of b and c, or a, b, and c, where a, b, and c can be single or multiple. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0041] The terms “center,” “longitudinal,” “lateral,” “up,” “down,” “left,” “right,” “front,” and “rear,” etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0042] The terms "connected" and "connected" should be interpreted broadly. For example, in circuit structures, "connected" or "connected" can refer not only to physical connections but also to electrical or signal connections. This could be a direct connection (physical connection) or an indirect connection via at least one intermediate component, as long as the circuit is connected. It could also refer to the internal connection between two components. Similarly, a signal connection can refer to a connection via a circuit or a medium, such as radio waves. Those skilled in the art will understand the specific meaning of these terms in this application based on the specific circumstances.

[0043] The external environment of overhead transmission line corridors is complex, with risks such as fire, external damage, and hanging objects, which may cause faults in overhead transmission lines. In this application, overhead transmission lines are referred to simply as transmission lines.

[0044] The transmission line anomaly detection and analysis method provided in this application can be applied to transmission line anomaly detection and analysis systems. A transmission line anomaly detection and analysis system is described below.

[0045] Please refer to Figure 1, which is a schematic diagram of the structure of a power transmission line anomaly detection and analysis system provided in this embodiment of the application. The power transmission line anomaly detection and analysis system provided in this embodiment includes: a camera device 101 and a server 102. The camera device is installed around the power transmission line to capture images of the power transmission line and its surrounding environment. There can be one or more camera devices 101. Figure 1 exemplarily shows one camera device; it is understood that this is only an example and does not constitute a limitation of this application. The camera device 101 can be a webcam, which can be a webcam fixedly installed on the ground around the power transmission line or a webcam mounted on a drone. The server 102 is communicatively connected to each camera device 101. It is understood that the communication connection between the server 102 and each camera device 101 can be a wired connection or a wireless connection. The server 102 can be a single server or a server cluster consisting of two or more servers.

[0046] The camera device 101 can capture images of the transmission line in real time or periodically and send these images to the server 102. The server 102 inputs the acquired transmission line images and prompt information into the anomaly detection and analysis model to obtain the anomaly analysis results. The server 102 pre-stores the anomaly detection and analysis model, which is obtained by adjusting the parameters of a pre-trained initial anomaly detection and analysis model using transmission line sample images, sample prompt information, and sample response information. The transmission line sample images contain anomalous features and / or confusing features. Anomalous features are those that pose a risk to the transmission line, while confusing features are those with a high degree of similarity to anomalous features. The sample prompt information indicates the type of response required, and the sample response information is the corresponding reply to the prompt information. By inputting sample images and sample prompts from transmission lines into an initial anomaly detection and analysis model built on a multimodal large model, the initial anomaly detection and analysis model can quickly learn the differences between abnormal features and similar confusing features in the transmission line. It can also learn rapidly through sample prompts, better stimulating the model's own potential. The resulting anomaly detection and analysis model can accurately identify abnormal conditions in the transmission line, reduce the false alarm rate, ensure that risks to the transmission line can be handled in a timely manner, and improve the safety of the transmission line.

[0047] The technical solutions provided in this application will be described in detail below with reference to specific embodiments.

[0048] Please refer to Figure 2, which is a flowchart illustrating a method for detecting and analyzing anomalies in transmission lines according to an embodiment of this application. As shown in Figure 2, the method provided in this embodiment is executed by a server, which can be server 102 in the embodiment shown in Figure 1 above. The method provided in this embodiment may include steps 201 and 202.

[0049] Step 201: Obtain images and prompts of the power transmission line.

[0050] Among them, the transmission line image is an image of the surrounding environment of the transmission line. The transmission line image can be taken by a camera device, for example, the camera device 101 shown in Figure 1 above.

[0051] The prompt information indicates the type of response required when viewing the transmission line image. The prompt information is equivalent to posing a question to the model and waiting for a corresponding response. The response type can be yes / no. For example, the prompt information could be: Is there a fire in the current transmission line image?

[0052] Step 202: Input the transmission line image and prompt information into the anomaly detection and analysis model to obtain the anomaly analysis results of the transmission line corresponding to the prompt information.

[0053] Among them, the anomaly analysis results of transmission lines are used to indicate whether there are any anomalies in the transmission lines.

[0054] Furthermore, when the anomaly analysis results of the transmission line indicate that there is an anomaly in the transmission line, the anomaly analysis results of the transmission line can also include information such as the anomaly type of the transmission line.

[0055] The anomaly detection and analysis model is pre-stored on the server. It can be trained by the server itself or trained by other electronic devices and then sent to the server. The methods for obtaining the anomaly detection and analysis model are described below.

[0056] An initial anomaly detection and analysis model is established based on a multimodal large model. Here, "large model" refers to a "large parameter" model trained using massive amounts of data and powerful computing capabilities. This initial anomaly detection and analysis model can be a pre-trained model, typically trained using large-scale unlabeled datasets through unsupervised learning. This initial anomaly detection and analysis model is a base model.

[0057] By using sample images of transmission lines and sample prompts, the parameters of the initial anomaly detection and analysis model are adjusted to obtain the anomaly detection and analysis model.

[0058] Each of the aforementioned transmission line sample images contains anomalous features and / or confusing features. Anomalous features are those that pose a risk to the transmission line, while confusing features are those whose similarity to anomalous features exceeds a preset threshold. The preset threshold is a pre-defined value. Confusing features are typically those that are very similar to a particular anomalous feature, and the model may easily misidentify them as anomalous features when identifying them. For example, lights and fires are similar in images. Lights do not affect transmission lines, while fires may pose a safety risk. Therefore, during anomaly detection, fires in the transmission line sample images are defined as anomalous features, and lights are defined as confusing features. For example, anomalous and confusing features can be manually labeled on the transmission line sample images.

[0059] Furthermore, the abnormal characteristics may include, but are not limited to, at least one of the following: foreign object characteristics on the transmission line, fire / smoke characteristics, smoke characteristics, and characteristics of the first crane. Specifically, foreign object characteristics on the transmission line refer to the presence of foreign objects on or around the transmission line. Fire / smoke characteristics refer to the presence of fire / smoke on or around the transmission line. Smoke characteristics refer to the characteristics of smoke produced when a fire breaks out on or around the transmission line. Characteristics of the first crane refer to the characteristics of a crane that may come into contact with the transmission line; if the crane comes into contact with the transmission line, it may cause damage to the transmission line.

[0060] Accordingly, the confusing features include at least one of the following: transmission line structural features, lighting features, cloud features, fog features, and a second crane feature. Among these, transmission line structural features refer to the characteristics of each structure of the transmission line; lighting features refer to the presence of lights on and around the transmission line; cloud features refer to the presence of clouds above the transmission line; fog features refer to the presence of fog around the transmission line; and the second crane feature refers to the characteristics of a crane that will not come into contact with the transmission line.

[0061] It can be understood that the confusing features corresponding to the foreign object characteristics of the transmission line are the structural characteristics of the transmission line; the confusing features corresponding to the fire feature are the light features or cloud features, etc.; the confusing features corresponding to the smoke feature are the fog features, etc.; and the confusing features corresponding to the first crane feature are the second crane feature.

[0062] The sample prompts mentioned above indicate the type of response required, and the sample response refers to the response corresponding to the prompt. Sample prompts can also be used to distinguish between anomalous features and confusing features, and / or the risks posed by anomalous features to transmission lines. Sample prompts can be in text or voice format. It should be noted that anomalous features may be of multiple types; correspondingly, the sample prompts refer to the distinguishing information between any given anomalous feature and its corresponding confusing features.

[0063] Furthermore, the sample prompt information may include at least one of the following: foreign object detection prompts for power transmission lines; light, cloud, fog and fire detection prompts; crane scene detection and analysis prompts; and hazard warnings.

[0064] For example, fog sometimes exists in the environment. Fog refers to tiny water droplets formed by the condensation of water vapor in the air upon cooling, floating near the ground. Fog is a natural phenomenon and usually does not pose a risk to power transmission lines. However, in the event of a fire, smoke features may be present on images of power transmission lines. In this application, smoke can also be referred to as haze, specifically the smoke produced during a fire. Smoke and fog are often easily confused. Considering that fog usually floats in the air while smoke usually rises from the ground, sample prompts can indicate the difference between smoke and fog, including the location of the feature, thereby determining whether it is fog or smoke.

[0065] For example, firelight and smoke often coexist. Considering that clouds or lights typically do not carry smoke, sample cue information could indicate the difference between firelight and clouds / lights, including whether smoke is present around the feature. If smoke is detected around a light feature, then the currently detected light feature can be determined to be a firelight feature.

[0066] For example, if a crane is present in the image of a power transmission line, the sample prompt information can also be information indicating the terrain features of the reference power transmission line and the location of the crane, so that the initial anomaly detection analysis model can learn how to distinguish whether a crane is the first crane feature or the second crane feature if a crane is present in the image.

[0067] Furthermore, there are several ways to adjust the initial anomaly detection and analysis model. One approach is to fine-tune the model parameters, such as using a large model fine-tuning method (Low-Rank Adaptation, or LoRa for short).

[0068] Since the transmission line sample images contain abnormal features that may cause transmission line anomalies, confusing features similar to abnormal features, and sample prompt information indicating the difference between abnormal features and confusing features, these can be input into the initial anomaly detection analysis model for fine-tuning to obtain the anomaly detection analysis model.

[0069] In this embodiment, the server inputs the acquired transmission line images and prompt information into the anomaly detection and analysis model to obtain the transmission line anomaly analysis results. The server 102 pre-stores the anomaly detection and analysis model, which is obtained by adjusting the parameters of a pre-trained initial anomaly detection and analysis model using transmission line sample images, sample prompt information, and sample response information. The transmission line sample images contain anomalous features and / or confusing features. Anomalous features are those that pose a risk to the transmission line, while confusing features are those with high similarity to anomalous features. Sample prompt information indicates the type of response required, and sample response information refers to the response corresponding to the prompt information. By inputting the transmission line sample images and sample prompt information into the initial anomaly detection and analysis model based on a multimodal large model, the initial anomaly detection and analysis model can quickly learn the differences between anomalous features and confusing features similar to anomalous features in the transmission line. Through rapid learning using sample prompt information, the model's potential is better stimulated. The resulting anomaly detection and analysis model can accurately identify transmission line anomalies, reducing the false positive rate, ensuring timely handling of transmission line risks, and improving the safety of the transmission line.

[0070] In some embodiments, step 202 can be implemented in the following manner:

[0071] Step 2021: Decompose the prompt information to obtain multi-stage sub-prompt information.

[0072] Furthermore, the prompts can be broken down according to their type. The prompt type is determined by categorizing the prompts based on their complexity.

[0073] Step 2022: According to the order of the sub-prompt information of multiple stages, input the transmission line image and the sub-prompt information of multiple stages into the anomaly detection and analysis model in stages to obtain the anomaly analysis results corresponding to the prompt information.

[0074] In this context, "multi-stage" refers to processing information step by step according to the type of prompt. Each stage processes one type of prompt.

[0075] In this embodiment, the anomaly detection and analysis model is typically a multimodal model with billions of steps. Since the complexity of the prompts it can handle is limited, the original prompts can be broken down and modified into combinations of prompts for the multimodal model with billions of steps. Prompts are then provided according to these combinations, and reasoning is performed on the multimodal model with billions of steps. That is, the prompts are categorized according to their complexity based on their type, and the anomaly detection and analysis model is prompted step-by-step to obtain the output results at each stage.

[0076] For example, the first step provides a preliminary identification prompt, while subsequent steps provide further complex identification and analysis based on the results of the previous step. For instance, the first step might prompt the identification of whether there is light of a special color, while the second step might prompt the identification and analysis of whether it is light / smoke; or the first step might prompt the identification of whether there is a crane, while the second step might prompt the analysis of the crane's distance.

[0077] In this embodiment, due to the complexity of transmission line scenarios, the prompt information can also be more complex than in other fields. Therefore, in order to obtain better output results through the anomaly detection and analysis model, the prompt information can be decomposed into multi-stage sub-prompt information. According to the order of the multi-stage sub-prompt information, the transmission line image and the multi-stage sub-prompt information are input into the anomaly detection and analysis model in stages to obtain the anomaly analysis results corresponding to the prompt information. This allows the anomaly detection and analysis model to fully understand the prompt information, thereby accurately identifying abnormal conditions of the transmission line, reducing the false alarm rate, predicting potential risks of the transmission line in a timely manner, ensuring that risks of the transmission line can be handled in a timely manner, and improving the safety of the transmission line.

[0078] In some embodiments, step 202 may be followed by steps 203 and 204.

[0079] Step 203: Determine whether the anomaly analysis results of the first transmission line indicate that there is an anomaly in the first transmission line.

[0080] If yes, proceed to step 204; otherwise, return to step 201.

[0081] Step 204: Output abnormal alarm information.

[0082] The abnormal alarm information includes information about the power transmission line and the type of abnormality.

[0083] The server may need to handle anomalies in multiple power transmission lines simultaneously. Therefore, the anomaly alarm information needs to include information about the power transmission line currently experiencing an anomaly. This information can be the transmission line's identification information or the identification information of the camera device. Different power transmission lines typically have different names or numbers, and the location of the anomaly can be determined using this identification information. Camera devices are usually fixedly installed around the power transmission lines. The server can identify the sender of the image of the anomaly, i.e., the identification information of the camera device that sent the image. This identification information can also be used to determine the power transmission line currently experiencing an anomaly.

[0084] The abnormal alarm information may also include abnormality type information, which refers to the type of abnormality currently detected. For example, the current abnormality type information may be the presence of fire.

[0085] In this embodiment, when the server determines that the abnormality analysis results of the transmission line indicate that there is an anomaly in the transmission line, it outputs an anomaly alarm message containing information about the transmission line and the type of anomaly. This promptly alerts the user to the risks present in the current transmission line and accurately locates the location and type of the risk. As a result, the user can quickly locate the transmission line with the risk and conduct risk investigation, ensuring that the risks of the transmission line can be dealt with in a timely manner, thus improving the safety of the transmission line.

[0086] In some embodiments, the initial anomaly detection analysis model is based on knowledge distillation, using the first anomaly detection analysis model as the teacher model to obtain the student model through distillation.

[0087] Knowledge distillation is a method of model compression. It typically involves building a lightweight, smaller model and training it using supervision information from a larger, more powerful model, aiming for better performance and accuracy. The larger model can be called the teacher model, and the smaller model the student model. The supervision information from the teacher model's output is called knowledge, and the process by which the student model learns and transfers this supervision information is called distillation.

[0088] In this embodiment, the first anomaly detection and analysis model is typically a large multimodal model with tens of billions, hundreds of billions, or even trillions of data points. In practical applications, this can lead to significant resource consumption. Therefore, based on knowledge distillation, the first anomaly detection and analysis model can be used as the teacher model, and the resulting billion-level student model can be used as the initial anomaly detection and analysis model. This makes the initial anomaly detection and analysis model more lightweight, thus making the final anomaly detection and analysis model more lightweight as well. This improves the processing efficiency of the anomaly detection and analysis model and also increases the efficiency of using the anomaly detection and analysis model for transmission line anomaly detection and analysis.

[0089] In some scenarios, anomaly detection and analysis models are obtained through model training. The following describes a method for adjusting a transmission line anomaly detection and analysis model.

[0090] Please refer to Figure 3, which is a flowchart illustrating a method for adjusting a transmission line anomaly detection and analysis model according to an embodiment of this application. As shown in Figure 3, the method provided in this embodiment is executed by an electronic device, which can be the server shown in the embodiment of Figure 2 above, or other electronic devices different from those shown in Figure 2. The method provided in this embodiment may include steps 301 and 302. If the embodiment shown in Figure 3 is combined with the embodiment shown in Figure 2, steps 301 and 302 are executed before step 202.

[0091] Step 301: Obtain the transmission line sample image, sample prompt information, and sample response information.

[0092] The transmission line sample images contain labels for anomalous features and labels for confusing features; confusing features are features similar to anomalous features.

[0093] Among them, the sample hint information is used to indicate the distinction between abnormal features and confusing features.

[0094] Step 302: Adjust the parameters of the initial anomaly detection analysis model using the transmission line sample images, sample prompt information, and sample response information to obtain the anomaly detection analysis model.

[0095] The initial anomaly detection and analysis model is built upon a large multimodal model. This initial anomaly detection and analysis model can be a pre-trained model, typically trained using a large-scale unlabeled dataset through unsupervised learning. This initial anomaly detection and analysis model is a base model.

[0096] Each of the aforementioned transmission line sample images contains anomalous features and / or confusing features. Anomalous features are those that pose a risk to the transmission line, while confusing features are those whose similarity to anomalous features exceeds a preset threshold. The preset threshold is a pre-defined value. Confusing features are typically those that are very similar to a particular anomalous feature, and the model may easily misidentify them as anomalous features when identifying them. For example, lights and fires are similar in images. Lights do not affect transmission lines, while fires may pose a safety risk. Therefore, during anomaly detection, fires in the transmission line sample images are defined as anomalous features, and lights are defined as confusing features. For example, anomalous and confusing features can be manually labeled on the transmission line sample images.

[0097] Furthermore, the abnormal characteristics may include, but are not limited to, at least one of the following: foreign object characteristics on the transmission line, fire / smoke characteristics, smoke characteristics, and characteristics of the first crane. Specifically, foreign object characteristics on the transmission line refer to the presence of foreign objects on or around the transmission line. Fire / smoke characteristics refer to the presence of fire / smoke on or around the transmission line. Smoke characteristics refer to the characteristics of smoke produced when a fire breaks out on or around the transmission line. Characteristics of the first crane refer to the characteristics of a crane that may come into contact with the transmission line; if the crane comes into contact with the transmission line, it may cause damage to the transmission line.

[0098] Accordingly, the confusing features include at least one of the following: transmission line structural features, lighting features, cloud features, fog features, and a second crane feature. Among these, transmission line structural features refer to the characteristics of each structure of the transmission line; lighting features refer to the presence of lights on and around the transmission line; cloud features refer to the presence of clouds above the transmission line; fog features refer to the presence of fog around the transmission line; and the second crane feature refers to the characteristics of a crane that will not come into contact with the transmission line.

[0099] It can be understood that the confusing features corresponding to the foreign object characteristics of the transmission line are the structural characteristics of the transmission line; the confusing features corresponding to the fire feature are the light features or cloud features, etc.; the confusing features corresponding to the smoke feature are the fog features, etc.; and the confusing features corresponding to the first crane feature are the second crane feature.

[0100] The sample prompts mentioned above indicate the type of response required, and the sample response refers to the response corresponding to the prompt. Sample prompts can also be used to distinguish between anomalous features and confusing features, and / or the risks posed by anomalous features to transmission lines. Sample prompts can be in text or voice format. It should be noted that anomalous features may be of multiple types; correspondingly, the sample prompts refer to the distinguishing information between any given anomalous feature and its corresponding confusing features.

[0101] Furthermore, the sample prompt information may include at least one of the following: foreign object detection prompts for power transmission lines; light, cloud, fog and fire detection prompts; crane scene detection and analysis prompts; and hazard warnings.

[0102] For example, fog sometimes exists in the environment. Fog refers to tiny water droplets formed by the condensation of water vapor in the air upon cooling, floating near the ground. Fog is a natural phenomenon and usually does not pose a risk to power transmission lines. However, in the event of a fire, smoke features may be present on images of power transmission lines. In this application, smoke can also be referred to as haze, specifically the smoke produced during a fire. Smoke and fog are often easily confused. Considering that fog usually floats in the air while smoke usually rises from the ground, sample prompts can indicate the difference between smoke and fog, including the location of the feature, thereby determining whether it is fog or smoke.

[0103] For example, firelight and smoke often coexist. Considering that clouds or lights typically do not carry smoke, sample cue information could indicate the difference between firelight and clouds / lights, including whether smoke is present around the feature. If smoke is detected around a light feature, then the currently detected light feature can be determined to be a firelight feature.

[0104] For example, if a crane is present in the image of a power transmission line, the sample prompt information can also be information indicating the terrain features of the reference power transmission line and the location of the crane, so that the initial anomaly detection analysis model can learn how to distinguish whether a crane is the first crane feature or the second crane feature if a crane is present in the image.

[0105] Furthermore, there are several ways to adjust the initial anomaly detection and analysis model, such as by fine-tuning the model parameters, for example, using LoRa.

[0106] In this embodiment, the transmission line sample image contains anomalous features and / or confusing features. Anomalous features refer to features that pose a risk to the transmission line, while confusing features are features with a high degree of similarity to anomalous features. Sample prompt information is used to indicate the type of response required, and sample response information refers to the response corresponding to the prompt information. The transmission line sample image and sample prompt information are jointly input into an initial anomaly detection and analysis model built based on a multimodal large model. This allows the initial anomaly detection and analysis model to quickly learn the differences between anomalous features and confusing features similar to anomalous features in the transmission line, and to learn rapidly through sample prompt information, better stimulating the model's own potential. The resulting anomaly detection and analysis model can accurately identify abnormal situations in the transmission line, reducing the false positive rate, ensuring timely handling of risks to the transmission line, and improving the safety of the transmission line.

[0107] In some embodiments, the method provided in this embodiment may further include step 300 before step 302.

[0108] Step 300: Based on knowledge distillation, the student model obtained by distillation, using the first anomaly detection and analysis model as the teacher model, is used as the initial anomaly detection and analysis model.

[0109] In this embodiment, the first anomaly detection and analysis model is typically a large multimodal model with tens of billions, hundreds of billions, or even trillions of data points. In practical applications, this can lead to significant resource consumption. Therefore, based on knowledge distillation, the first anomaly detection and analysis model can be used as the teacher model, and the resulting billion-level student model can be used as the initial anomaly detection and analysis model. This makes the initial anomaly detection and analysis model more lightweight, thus making the final anomaly detection and analysis model more lightweight as well. This improves the processing efficiency of the anomaly detection and analysis model and also increases the efficiency of using the anomaly detection and analysis model for transmission line anomaly detection and analysis.

[0110] Please refer to Figure 4, which is a schematic diagram of the structure of a transmission line anomaly detection and analysis device provided in an embodiment of this application. The device provided in this embodiment includes:

[0111] The acquisition module 401 is used to acquire images of transmission lines; the images of transmission lines are images of the surrounding environment of the transmission lines.

[0112] The processing module 402 is used to input the transmission line image into the anomaly detection and analysis model to obtain the anomaly analysis results of the transmission line.

[0113] Among them, the anomaly analysis results of transmission lines are used to indicate whether there are any anomalies in the transmission lines; the anomaly detection analysis model is obtained by adjusting the parameters of the initial anomaly detection analysis model through transmission line sample images and sample prompt information; the initial anomaly detection analysis model is established based on a multimodal large model; the transmission line sample images contain labels for anomaly features and labels for confusing features; confusing features are features similar to anomaly features; the sample prompt information is used to indicate the difference between anomaly features and confusing features, and / or the risks that anomaly features bring to the transmission lines.

[0114] In some embodiments, the anomalous features include at least one of the following: foreign object features on the transmission line, fire features, features of the first crane, and features of hazardous materials; the confusing features include at least one of the following: structural features of the transmission line, light features, cloud features, fog features, and features of the second crane.

[0115] In some embodiments, the anomaly detection analysis model is obtained in the following way: by adjusting the parameters of the initial anomaly detection analysis model using transmission line sample images and sample prompt information, a fine-tuned initial anomaly detection analysis model is obtained, and the fine-tuned initial anomaly detection analysis model is further adjusted in multiple stages based on the sample prompt information type.

[0116] In some embodiments, the initial anomaly detection analysis model is based on knowledge distillation, using the first anomaly detection analysis model as the teacher model to obtain the student model through distillation.

[0117] In some embodiments, the device further includes:

[0118] The anomaly alarm module is used to output anomaly alarm information when the anomaly analysis results of the transmission line indicate that there is an anomaly in the transmission line. The anomaly alarm information includes information about the transmission line and the anomaly type.

[0119] In some embodiments, the device further includes:

[0120] The training module is used to acquire sample images and sample prompts of transmission lines; the parameters of the initial anomaly detection analysis model are adjusted using the sample images and sample prompts to obtain the anomaly detection analysis model.

[0121] The device in this embodiment is similar in principle and effect to the above embodiments, and will not be described again here.

[0122] This application provides an electronic device including a memory and a processor. The memory stores a computer program that can run on the processor, and the processor executes the program to implement the steps of the method as described in any of the above embodiments.

[0123] Based on the transmission line anomaly detection and analysis method described in any of the above embodiments, this application also provides a computer-readable storage medium. For example, a non-transitory computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a CD-ROM, magnetic tape, a floppy disk, or an optical data storage device, etc. This storage medium stores computer instructions for executing the transmission line anomaly detection and analysis method described in any of the above embodiments, which will not be elaborated further here.

[0124] Based on the adjustment method of the transmission line anomaly detection and analysis model described in any of the above embodiments, this application also provides a computer-readable storage medium. For example, a non-transitory computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a CD-ROM, magnetic tape, a floppy disk, or an optical data storage device, etc. This storage medium stores computer instructions for executing the adjustment method of the transmission line anomaly detection and analysis model described in any of the above embodiments, which will not be elaborated further here.

[0125] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware, or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0126] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the application 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. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.

Claims

1. A power line anomaly detection and analysis method, characterized by, The method includes: Acquire images of power transmission lines and provide prompts; the images of power transmission lines are images of the surrounding environment of the power transmission lines; the prompts are used to indicate the type of response required based on the images of power transmission lines. The transmission line image and the prompt information are input into the anomaly detection and analysis model to obtain the anomaly analysis result corresponding to the prompt information; the anomaly analysis result is used to respond to the prompt information. The anomaly detection and analysis model is obtained by adjusting the parameters of an initial anomaly detection and analysis model using transmission line sample images, sample prompt information, and sample response information. The initial anomaly detection and analysis model is based on a multimodal large model. Each transmission line sample image contains anomaly features and / or confusing features. The anomaly features refer to features that pose a risk to the transmission line. The confusing features are features whose similarity to the anomaly features is greater than a preset threshold. The sample prompt information is used to indicate the type of response required, and the sample response information refers to the response corresponding to the prompt information.

2. The method of claim 1, wherein, The abnormal features include at least one of the following: foreign object features on the transmission line, fire features, smoke features, features of the first crane, and hazardous material features; the confusing features include at least one of the following: structural features of the transmission line, light features, cloud features, fog features, and features of the second crane; the sample prompt information may include at least one of the following: foreign object detection prompts on the transmission line; light, cloud, and fire detection prompts; fog and smoke detection prompts; crane scene detection and analysis prompts; and hazardous material prompts.

3. The method of claim 1, wherein, The step of inputting the transmission line image and the prompt information into the anomaly detection and analysis model to obtain the anomaly analysis result corresponding to the prompt information includes: The prompt information is broken down into multiple sub-prompt information stages; According to the sequential order of the sub-prompt information of the multiple stages, the transmission line image and the sub-prompt information of the multiple stages are input into the anomaly detection and analysis model in stages to obtain the anomaly analysis results corresponding to the prompt information.

4. The method of claim 1, wherein, The initial anomaly detection and analysis model is based on knowledge distillation, using the first anomaly detection and analysis model as the teacher model to obtain the student model through distillation.

5. The method of claim 1, wherein, The method further includes: If the anomaly analysis results of the transmission line indicate that there is an anomaly in the transmission line, an anomaly alarm message is output, which includes information about the transmission line and anomaly type information.

6. The method according to any one of claims 1 to 5, characterized in that, The method further includes: Acquire sample images, sample prompts, and sample responses for power transmission lines; The parameters of the initial anomaly detection and analysis model are adjusted using sample images of transmission lines, sample prompt information, and sample response information to obtain the anomaly detection and analysis model.

7. A method of adjusting a power transmission line anomaly detection analysis model, characterized by, The method includes: Acquire sample images of transmission lines, sample prompt information, and sample response information; each sample image of a transmission line contains abnormal features and / or confusing features; the abnormal features refer to features that pose a risk to the transmission line; the confusing features are features whose similarity to the abnormal features is greater than a preset threshold; the sample prompt information is used to indicate the type of response required, and the sample response information refers to the response corresponding to the prompt information; The parameters of the initial anomaly detection analysis model are adjusted using the transmission line sample images, sample prompt information, and sample response information to obtain the anomaly detection analysis model; the initial anomaly detection analysis model is established based on a multimodal large model.

8. The method of claim 7, wherein, The method further includes: Based on knowledge distillation, the first anomaly detection and analysis model is used as the teacher model, and the student model obtained by distillation is used as the initial anomaly detection and analysis model.

9. A power transmission line anomaly detection and analysis apparatus characterized by comprising: include: The acquisition module is used to acquire images and prompts of power transmission lines; The transmission line image is a photograph of the surrounding environment of the transmission line; The prompt information is used to indicate the type of response required when viewing the transmission line image; The processing module is used to input the transmission line image and the prompt information into the anomaly detection and analysis model to obtain the anomaly analysis result corresponding to the prompt information; The anomaly analysis results are used to respond to the prompt information; wherein, the anomaly detection analysis model is obtained by adjusting the parameters of an initial anomaly detection analysis model through transmission line sample images, sample prompt information, and sample response information; the initial anomaly detection analysis model is established based on a multimodal large model; each transmission line sample image contains anomaly features and / or confusing features; the anomaly features refer to features that pose a risk to the transmission line; the confusing features are features whose similarity to the anomaly features is greater than a preset threshold; the sample prompt information is used to indicate the type of response required, and the sample response information refers to the response corresponding to the prompt information.

10. An electronic device comprising a memory and a processor, said memory storing a computer program operable on said processor, characterized in that, When the processor executes the program, it implements the steps of the method according to any one of claims 1 to 8.

11. A computer readable storage medium having stored thereon a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 8.

Images