Electric power potential safety hazard behavior identification and distance measurement method, system and device and medium

Abstract

The invention relates to the technical field of electric power potential safety hazard behavior recognition, and discloses an electric power potential safety hazard behavior recognition and distance measurement method, system and device and a medium, and the method comprises the steps: obtaining first historical parameter data related to a to-be-detected electric power facility, and carrying out the preprocessing operation, thereby guaranteeing the quality and accuracy of input data; and a reliable basis is provided for subsequent behavior identification and judgment. According to the method, the behavior recognition and judgment model is established, and the preprocessed data is used for training, so that unsafe behaviors can be automatically recognized and judged, and the real-time performance and accuracy of monitoring are improved. If unsafe behaviors are recognized, safe distance measurement and judgment are carried out immediately, and the severity degree and the emergency degree of hidden dangers can be further confirmed through accurate measurement and judgment. The electric power potential safety hazard behavior is determined according to the safety distance measurement and judgment result, a risk level judgment mechanism is established, corresponding processing measures can be taken for risks of different levels, and safety accidents are effectively prevented.

Description

Technical Field

[0001] This invention relates to the field of power safety hazard behavior identification technology, and in particular to a power safety hazard behavior identification and ranging method, system, equipment and medium. Background Technology

[0002] With the continuous expansion and modernization of power infrastructure networks, safety hazards in power facilities are receiving increasing attention. To ensure the stable operation of power facilities and prevent accidents caused by human activity, traditional power facility safety monitoring relies on fixed-location sensors and manual inspections. These sensors typically include devices for monitoring current, voltage, temperature, and humidity, aiming to acquire real-time information on the operating status and environmental conditions of the power facilities. In these traditional technologies, monitoring data is mostly centrally analyzed through a central processing system to provide decision support.

[0003] Besides the application of sensors, video surveillance is also an important means of monitoring the safety of power facilities. Especially when dynamic monitoring of the environment surrounding power facilities is required, cameras and image processing technology are widely used. Video surveillance can not only observe the operational status of power facilities in real time, but also capture unsafe behaviors in the surrounding environment, such as unauthorized entry into power facility protection zones, unauthorized climbing of power towers, and illegal construction. In recent years, with the rapid development of deep learning technology, deep learning-based video analysis technology has been gradually applied to the automatic detection of power safety hazards. These technologies, by training deep neural networks (DNNs) or convolutional neural networks (CNNs), can automatically extract features from video images and identify potential unsafe behaviors.

[0004] Meanwhile, with the advancement of big data and cloud computing technologies, power facility monitoring systems are gradually evolving towards distributed networking. Existing monitoring systems can not only collect and analyze large amounts of data from sensors and video surveillance equipment in real time, but also store and process data through cloud platforms, thereby improving the speed and accuracy of data processing.

[0005] Traditional safety monitoring methods typically rely on manual inspections to identify potential safety hazards. While manual inspections can uncover some problems, they are inefficient and have limited coverage. Especially in remote areas or at night, manual inspections struggle to provide comprehensive coverage, potentially overlooking some hazards. Furthermore, manual inspections are limited by the experience and judgment of the inspectors, often introducing subjectivity and error, making it difficult to guarantee that all potential hazards are detected in a timely manner. Therefore, when dealing with large-scale and complex power facilities, manual inspections are insufficient to meet the real-time and comprehensive requirements of modern safety management. Summary of the Invention

[0006] In view of the aforementioned existing problems, the present invention is proposed.

[0007] Therefore, this invention provides a method, system, device, and medium for identifying and measuring potential power safety hazards, which can solve the problems of real-time performance and comprehensiveness in traditional power facility safety monitoring methods and improve the safety management level of power facilities.

[0008] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0009] In a first aspect, the present invention provides a method for identifying and measuring potential electrical safety hazards, comprising:

[0010] Acquire the first historical parameter data related to the power facility under test, and perform preprocessing operations on the first historical parameter data;

[0011] Establish a behavior recognition and judgment model, and train the behavior recognition and judgment model based on the first historical parameter data after the preprocessing operation;

[0012] The behavior recognition and judgment model is used to determine whether unsafe behavior has been identified based on the input.

[0013] If unsafe behavior is detected, a safe distance will be measured and assessed.

[0014] The potential electrical safety hazards are determined based on the results of safety distance measurements and assessments.

[0015] As a preferred embodiment of the power safety hazard behavior identification and ranging method described in this invention, the establishment of the behavior identification and judgment model includes:

[0016] Establish a set of unsafe behaviors of the power facility under test and a set of features for the set of unsafe behaviors;

[0017] Perform feature extraction operations on the first historical parameter data regarding the feature set;

[0018] The data after the feature extraction operation is used as the input to the model, and the set of unsafe behaviors is used as the output of the model.

[0019] As a preferred embodiment of the power safety hazard behavior identification and distance measurement method of the present invention, the step of measuring and judging the safe distance if an unsafe behavior is identified includes:

[0020] The reference distance of the object in the unsafe act is obtained based on the disparity map; the parameter distance is the distance of the object in the unsafe act from a certain preset fixed position;

[0021] Set a set of judgment thresholds based on the preset fixed position;

[0022] The safe distance is measured and judged based on the set of judgment thresholds for the output of the behavior recognition and judgment model.

[0023] This preferred solution improves the accuracy and efficiency of safe distance judgment. By obtaining the reference distance of an object through a disparity map, the distance between the object and the preset fixed position during unsafe behavior can be determined more accurately, avoiding errors caused by manual measurement or estimation in traditional methods. Simultaneously, setting a set of judgment thresholds based on the preset fixed position allows for flexible adjustment according to different safety requirements, improving the applicability of safe distance judgment. Furthermore, this preferred solution simplifies the safe distance measurement process, reduces manpower and time costs, and improves overall work efficiency.

[0024] As a preferred embodiment of the power safety hazard behavior identification and ranging method described in this invention, the step of determining the power safety hazard behavior based on the safety distance measurement and judgment results includes:

[0025] Establish a risk level assessment mechanism;

[0026] The risk level is determined based on the results of the safety distance measurement and assessment.

[0027] The risk level assessment mechanism includes quantifying and scoring the results of safety distance measurement and assessment, and classifying different risk levels based on the scoring results.

[0028] Each risk level corresponds to different types of safety hazards and corresponding handling measures.

[0029] As a preferred embodiment of the power safety hazard behavior identification and distance measurement method described in this invention, the quantitative scoring of the safety distance measurement and judgment results includes:

[0030] Establish scoring criteria, which are comprehensively formulated based on the specific nature of the unsafe behavior, the degree of deviation between the object and the preset fixed position, and the magnitude of the potential hazard;

[0031] The results of the safety distance measurement and judgment are compared with the preset scoring criteria to obtain a quantitative score value;

[0032] The higher the quantitative score, the higher the risk level of the potential power safety hazard.

[0033] As a preferred embodiment of the power safety hazard behavior identification and ranging method described in this invention, the preprocessing operation of the first historical parameter data includes:

[0034] The first historical parameter data includes video image data and depth perception data;

[0035] The preprocessing operations include frame-by-frame denoising and brightness adjustment of the video image data, and the creation of a disparity map from the depth-sensing data.

[0036] As a preferred embodiment of the power safety hazard behavior identification and ranging method of the present invention, the method further includes updating the behavior identification and judgment model by retraining the behavior identification and judgment model based on the latest collected second historical parameter data; the second historical parameter data is parameter data newly generated during the operation of power facilities, including new video image data and depth perception data.

[0037] Secondly, the present invention provides a power safety hazard behavior identification and ranging system, comprising:

[0038] The data acquisition and processing module is used to acquire first historical parameter data related to the power facility under test, and to perform preprocessing operations on the first historical parameter data.

[0039] The model building module is used to build a behavior recognition and judgment model, and to train the behavior recognition and judgment model based on the first historical parameter data after the preprocessing operation.

[0040] The behavior recognition and judgment model is used to determine whether unsafe behavior has been identified based on the input.

[0041] The first judgment module is used to measure and judge the safe distance if unsafe behavior is detected.

[0042] The second judgment module is used to determine potential electrical safety hazards based on the results of safety distance measurement and judgment.

[0043] Thirdly, the present invention provides an electronic device including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method described above.

[0044] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0045] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention proposes a method for identifying and measuring the behavior of potential power safety hazards. First, by acquiring and preprocessing the first historical parameter data related to the power facility under test, the quality and accuracy of the input data can be ensured, providing a reliable foundation for subsequent behavior identification and judgment. Second, by establishing a behavior identification and judgment model and training it using the preprocessed data, unsafe behaviors can be automatically identified and judged, greatly improving the real-time performance and accuracy of monitoring. Third, if an unsafe behavior is identified, a safe distance is immediately measured and judged. Through precise measurement and judgment, the severity and urgency of the hazard can be further confirmed. Finally, based on the results of the safe distance measurement and judgment, the power safety hazard behavior is determined, and a risk level judgment mechanism is established. This allows for corresponding measures to be taken for different levels of risk, effectively preventing safety accidents. Attached Figure Description

[0046] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 The present invention provides a flowchart of a method for identifying and measuring electrical safety hazards, which is an embodiment of the present invention. Detailed Implementation

[0048] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0049] Example 1, referring to Figure 1 This is the first embodiment of the present invention, which provides a method for identifying and measuring electrical safety hazard behaviors, including:

[0050] Existing technologies have several limitations. For instance, the safety monitoring of power facilities often relies on traditional sensors and manual inspections, which are insufficient in terms of real-time performance and comprehensiveness. Especially in remote areas or at night, manual inspections are inefficient and struggle to provide complete coverage, potentially leading to overlooking safety hazards. Furthermore, traditional methods are limited by the experience and judgment of inspection personnel, introducing subjectivity and error, making it difficult to guarantee that all potential hazards will be detected in a timely manner.

[0051] This invention provides a method that can effectively solve the problems mentioned above. The following will describe in detail how to implement the power safety hazard behavior identification and ranging method with reference to several embodiments.

[0052] Figure 1 A flowchart of a method for identifying and measuring potential electrical safety hazards is shown, including:

[0053] S101, acquire the first historical parameter data related to the power facility under test, and perform preprocessing operations on the first historical parameter data;

[0054] It should be noted that safety hazards in power facilities typically stem from various violations, such as fishing in ponds, flying kites, and unauthorized climbing of power towers. These activities can damage power facilities or cause safety accidents. However, traditional security monitoring systems often fail to efficiently identify these potential threats and, in some cases, rely on manual inspections, resulting in blind spots and response delays. To overcome these challenges, this invention acquires and analyzes first-historical parameter data related to the power facility under test, ultimately deriving a method for identifying and measuring power safety hazard behaviors.

[0055] In some specific implementations, video image data and depth perception data related to the power facility under test are collected. This data may come from devices such as surveillance cameras and LiDAR installed around the power facility. The data should cover as many scenes and time periods as possible so that the model can learn normal and unsafe behaviors in various environments.

[0056] In this embodiment of the invention, the preprocessing operation for the first historical parameter data includes:

[0057] The first historical parameter data includes video image data and depth perception data;

[0058] Preprocessing operations include frame-by-frame denoising and brightness adjustment of video image data, and the creation of a disparity map from depth-sensing data.

[0059] For example, a continuous video stream is split into individual frames. This step is typically achieved using video processing software or libraries such as OpenCV. Noise is then identified in each frame. Noise can be randomly occurring pixels, snow effects, or other interfering factors.

[0060] Use filters (such as Gaussian filtering and median filtering) to smooth the image and remove isolated noise. For moving scenes, a frame-based difference approach can be used to compare changes between adjacent frames, retaining only the actual moving parts while suppressing background noise.

[0061] Analyze the brightness level of each frame to determine if adjustments are needed. This can be done by calculating the average brightness value or through histogram analysis.

[0062] It directly changes the brightness value of the entire image, suitable for situations where the overall image is too bright or too dark.

[0063] Brightness compensation is applied to specific areas (such as shadowy or bright areas) to improve visibility and contrast in these areas.

[0064] Histogram equalization is used to improve the overall contrast of an image, making details clearer.

[0065] Specifically, the detailed steps for creating a disparity map from depth-sensing data are as follows:

[0066] 1. If a binocular camera system is used, stereo calibration is first required for the images captured by the left and right cameras to ensure that the scan lines of the same object are aligned in the two images.

[0067] 2. Calculate the matching cost for each pixel between the left and right images. Commonly used matching methods include SAD (sum of absolute differences) and SSD (sum of squared differences).

[0068] 3. Based on the matching cost, find the best matching point for each pixel and calculate the disparity (i.e., pixel displacement) between it and the reference point.

[0069] 4. Advanced algorithms such as dynamic programming, graph cut algorithm, or semi-global matching (SGM) can be used to optimize the quality of disparity maps.

[0070] 5. Filter and smooth the generated initial disparity map to eliminate mismatches and discontinuities, thereby improving the accuracy and visual effect of the disparity map.

[0071] In some specific implementations, a multi-sensor system consisting of high-definition cameras and depth-sensing devices can be used to monitor the surrounding environment of power facilities in real time. Cameras capture image data, while depth-sensing devices provide three-dimensional spatial information, measuring the distance between objects and power facilities. This data is transmitted in real time to a central processing unit via a data transmission system for data fusion and processing. In image processing, traditional image enhancement methods such as noise reduction and brightness adjustment are used to ensure that the acquired images are sharp enough for subsequent deep learning processing. The depth-sensing technology used in the system can be described by the following formula:

[0072]

[0073] Where D is the actual distance from the object to the device, f is the focal length of the camera, B is the baseline distance between the two cameras, and d is the parallax. This formula can be used to obtain the distance information between the object and the device.

[0074] It should be noted that acquiring and preprocessing the initial historical parameter data of the power facility under test can improve the accuracy and efficiency of subsequent processing. Preprocessing removes noise and outliers, ensuring data reliability and consistency. By preprocessing historical parameter data, we can better understand and analyze the operating status of the power facility, providing more accurate foundational data for subsequent behavior recognition and ranging. This not only helps improve the accuracy of identifying potential power safety hazards but also optimizes the performance of ranging algorithms, ensuring the stable operation of the entire system in complex and ever-changing power environments.

[0075] S102, Establish a behavior recognition and judgment model, and train the behavior recognition and judgment model based on the first historical parameter data after preprocessing.

[0076] In this embodiment of the invention, the behavior recognition and judgment model is used to determine whether an unsafe behavior is recognized based on the input.

[0077] In some specific implementations, behavior recognition and judgment models can use deep learning algorithms, such as convolutional neural networks (CNNs) or recurrent neural networks (RNNs), which are capable of extracting features from complex data and automatically learning behavior patterns.

[0078] In some specific implementations, the behavior recognition and judgment model can also use transfer learning techniques, using models that have been trained in other fields as pre-trained models, and then fine-tuning them for behaviors that pose potential power safety hazards, thereby accelerating the training process and improving the model's generalization ability.

[0079] In some specific implementations, when using transfer learning techniques, the following steps can be followed:

[0080] Step 1: Select a deep learning model that has already been trained on a large-scale dataset as the base model. Commonly used pre-trained models include convolutional neural networks (CNNs) trained on the ImageNet dataset, such as VGG, ResNet, or Inception.

[0081] Step 2: Collect video image data and depth perception data related to the power facility under test.

[0082] The collected data is preprocessed, including frame-by-frame noise reduction, brightness adjustment, and disparity mapping, to ensure data quality.

[0083] Step 3: Use the selected pre-trained model to extract features from the pre-processed image data. Typically, the last one or more classification layers of the pre-trained model are removed, and only the preceding convolutional layers are retained for feature extraction.

[0084] Each frame of image is input into a pre-trained model to obtain high-level feature representations.

[0085] Step 4: Add new fully connected layers or other types of layers to the pre-trained model to adapt to the new task of identifying power safety hazard behaviors.

[0086] Based on the set of unsafe behaviors of the power facility under test and the set of features for the set of unsafe behaviors, an output layer structure is designed so that the model can output the types of unsafe behaviors identified.

[0087] Step 5: Fine-tune the entire model (including the pre-trained part and the newly added part) using the pre-processed first historical parameter data. This step requires adjusting hyperparameters such as the learning rate according to the actual situation to avoid overfitting.

[0088] One approach is to freeze some layers, that is, only update the weights of the newly added layers while keeping the pre-trained model unchanged, or gradually unfreeze more layers for more refined adjustments.

[0089] In this embodiment of the invention, establishing a behavior recognition and judgment model includes:

[0090] Establish a set of unsafe behaviors of the power facility under test and a set of features for the set of unsafe behaviors;

[0091] Perform feature extraction on the first historical parameter data regarding the feature set;

[0092] The data after feature extraction is used as the input to the model, and the set of unsafe behaviors is used as the output of the model.

[0093] Specifically, defining the set of unsafe behaviors first requires clarifying which behaviors are considered unsafe in electrical facilities. This may include, but is not limited to, climbing power towers, unauthorized entry into protected areas, and improper operation.

[0094] Furthermore, a feature set is defined for each unsafe behavior. These features can be visual (such as the location, size, and color of an object) or based on depth-sensing data (such as distance to power facilities). For example, the behavioral characteristics of climbing power towers might include a person's location, posture, and movement trajectory. The behavioral characteristics of illegally entering a protected area might involve the approach path and speed of people or vehicles.

[0095] Furthermore, feature extraction operations are performed on the first historical parameter data regarding the feature set, and the collected historical video image data is processed to extract relevant features. This typically involves using computer vision techniques, such as object detection, tracking, and pose estimation, to analyze each frame of the image and extract the features defined above. Using data acquired from depth cameras or other depth-sensing devices, distance information of objects relative to power facilities is obtained by calculating disparity maps, etc. The raw features extracted from the video images and depth-sensing data are further processed, such as through normalization and feature selection, so that these features can be used more effectively during subsequent model training.

[0096] Furthermore, the data after feature extraction is used as the model's input, and the set of unsafe behaviors is used as the model's output. An appropriate model architecture is chosen based on the problem's complexity; for example, Convolutional Neural Networks (CNNs) are used for image feature processing, while Recurrent Neural Networks (RNNs) or Long Short-Term Memory (LSTM) networks are used for time series data processing. Using the feature-extracted data as input and the set of unsafe behaviors as labels (i.e., the model's target output), the model is trained through supervised learning. During training, model parameters are adjusted to ensure the model can accurately predict the corresponding unsafe behaviors given the input. During training, the model's performance is periodically evaluated using a validation set, and hyperparameters, such as the learning rate and batch size, are adjusted accordingly to improve the model's generalization ability and accuracy.

[0097] In this embodiment of the invention, an adaptive deep learning network is selected for establishing the behavior recognition and judgment model. The adaptive deep learning network is one of the core technologies of this invention, and its application in identifying behaviors that pose safety hazards in the power industry solves the problem of insufficient accuracy in dynamic scenarios using traditional monitoring methods. Compared with traditional deep learning models, the adaptive deep learning network introduced in this invention can automatically adjust model parameters according to environmental changes and different data characteristics, thereby achieving higher recognition accuracy and stronger environmental adaptability.

[0098] It should be noted that in traditional deep learning systems, the network structure and parameters are usually fixed during training. While such fixed-structure deep learning models can be trained and recognized well in static environments, their recognition performance is often significantly affected by changes in the environment surrounding power facilities (such as weather changes, lighting changes, and crowd activities) in dynamic environments or complex scenarios. To address this challenge, this invention proposes an adaptive deep learning network model that can automatically adjust its structure, weights, and parameters according to changes in the environment and the characteristics of the input data, thereby improving the accuracy of identifying potential hazards around power facilities.

[0099] It's important to note that the adaptability of adaptive deep learning networks is reflected in their ability to dynamically update the network's training process based on real-time data, continuously optimizing their learning strategy. For example, in power facility monitoring, when the system detects changes in the surrounding environment (such as changes in lighting conditions or increased ambient noise), the network can adjust its training weights and network layers to ensure the model can cope with the newly emerging changing conditions. This dynamic learning and optimization process greatly enhances the model's flexibility and accuracy in practical applications.

[0100] It should be noted that in adaptive deep learning networks, convolutional neural networks (CNNs) are used for feature extraction from image data. CNNs, with their powerful local feature extraction capabilities, occupy an important position in image recognition.

[0101] In this invention, CNNs are used to extract feature information related to safety hazards from real-time video images around power facilities, such as the trajectory of people, abnormal object movement, and illegal activities. CNNs extract information from images progressively from low to high levels through a combination of multiple convolutional and pooling layers. The formula for the convolution operation is as follows:

[0102]

[0103] Where S(i,j) represents the result after the convolution operation, I(m,n) is the pixel value in the input image, and K(i,j) is the convolution kernel, representing the local features extracted by the filter as it slides across the image. Through convolution operations, the network can identify targets such as people, vehicles, and objects in the image and generate feature maps, providing a foundation for subsequent behavior recognition.

[0104] However, relying solely on CNNs for static feature extraction may lead to recognition errors in complex or dynamic environments. Therefore, this invention introduces an Adaptive Neural Network (ANN) to further enhance the system's adaptability to complex scenarios surrounding power facilities. By weighting the input feature data, the ANN can dynamically adjust its parameters according to different data types or environmental changes, optimizing the network's output. In this way, the system can adapt to different environmental changes and adjust its recognition strategy in a timely manner to ensure accurate identification of potentially unsafe behaviors.

[0105] It should be noted that the ANN, through its adaptive adjustment process, can avoid recognition errors caused by environmental changes (such as changes in lighting, shadows, cluttered backgrounds, etc.). When monitoring the area around power facilities, the environment may vary significantly due to seasonal changes, weather changes, or different times of day. Traditional deep learning models often fail to effectively identify these changes. However, the adaptive neural network of this invention can automatically adjust its network parameters based on new data and feedback when environmental conditions change, optimizing its understanding of the environment surrounding power facilities and thus improving the accuracy of safety hazard identification.

[0106] It should be noted that another advantage of adaptive deep learning networks lies in their ability to "memorize" environmental changes. Traditional deep learning models typically rely on fixed datasets during training and cannot dynamically learn new changes and environments during application. In contrast, the adaptive network used in this invention can gradually "memorize" environmental changes from real-time data and optimize its behavior recognition model through online learning and incremental training mechanisms.

[0107] In this embodiment of the invention, when new dynamic behaviors occur around power facilities, the system can automatically identify and learn how to respond to these new behavioral patterns. The model training process is performed using the following optimized loss function formula:

[0108]

[0109] Where L(θ) is the loss function, θ is the model parameter, and y i It is the actual label. λ is the model's predicted output, N is the number of training samples, λ is the regularization term, and M is the number of parameters in the model. By minimizing this loss function, the system can optimize its model parameters, thereby gradually improving its recognition ability and effectively reducing overfitting and false positives / false negatives.

[0110] Furthermore, the adaptive deep learning network of this invention not only excels in static image analysis but also processes temporal information in videos, thereby enabling real-time recognition of dynamic behaviors. For example, the system can extract information such as the movement trajectory of people and abnormal activities from video streams around power facilities by combining CNN and ANN, further enhancing the system's ability to recognize dynamic behaviors. In this way, the system can monitor the surrounding environment of power facilities in real time, identify potential safety hazards, and provide timely warnings.

[0111] S103, if unsafe behavior is detected, a safe distance is measured and judged;

[0112] It should be noted that safe distance measurement and judgment can be performed by calculating the actual distance between the object and the power facility and comparing it with a preset safe distance threshold. The key to this step is accurate measurement and rapid judgment to ensure timely action can be taken when unsafe behavior is detected, preventing potential accidents. To achieve this goal, various technologies can be employed, such as deep learning-based object detection and tracking algorithms and 3D reconstruction technology, to improve the accuracy and efficiency of the measurement. Simultaneously, by considering the specific conditions of the power facility and safety regulations, a reasonable safe distance threshold can be set to ensure the accuracy and reliability of the judgment results. Based on the safe distance measurement and judgment, the system can automatically generate early warning information and promptly notify relevant personnel for handling, thereby effectively reducing the risk of potential power safety hazards.

[0113] In this embodiment of the invention, if an unsafe behavior is detected, the measurement and judgment of the safe distance includes:

[0114] The reference distance of objects in unsafe acts is obtained based on the disparity map; the parameter distance is the distance of the object in the unsafe act from a certain preset fixed position;

[0115] Set a set of judgment thresholds based on preset fixed positions;

[0116] The safe distance is measured and judged based on the set of judgment thresholds for the output of the behavior recognition and judgment model.

[0117] Specifically, in the aforementioned steps, a disparity map has been created from the depth-sensing data. The disparity map reflects the distance information of each pixel in the scene relative to the camera.

[0118] Furthermore, after the behavior recognition and judgment model outputs the recognition results, the system identifies objects (such as people, vehicles, etc.) with unsafe behaviors in the current frame or video segment and locates their regions in the image (e.g., through object detection boxes or segmentation masks).

[0119] Furthermore, by utilizing the corresponding pixel values ​​of the object in the disparity map, combined with the camera's intrinsic and extrinsic parameters, the disparity values ​​are converted into physical distances in actual space. This ultimately yields the reference distance of the object from a predetermined fixed location (such as the center point of a power facility, the boundary of a warning line, etc.).

[0120] Furthermore, the preset fixed location can be a key part of the power facility (such as high-voltage equipment, tower base, substation entrance, etc.) or a designated safety boundary line.

[0121] Furthermore, set multi-level safety distance thresholds: Based on safety management needs, set multiple levels of judgment thresholds to form a judgment threshold set, for example:

[0122] Level 1 threshold (warning threshold): An alert is triggered when the distance is less than 5 meters;

[0123] Level 2 threshold (danger threshold): A high-risk alarm is triggered when the distance is less than 2 meters;

[0124] Level 3 threshold (emergency threshold): The emergency response mechanism is triggered when the distance is less than 0.5 meters.

[0125] These thresholds can be flexibly configured according to different types of power facilities and different application scenarios.

[0126] Furthermore, the safe distance measurement and judgment of the output of the behavior recognition and judgment model based on the judgment threshold set can be determined according to the following steps:

[0127] Step 1: Input information integration. The behavior type output by the behavior recognition and judgment model (such as "illegal approach" or "climbing"), the location information of the target object, and its corresponding reference distance are all used as input.

[0128] Step 2: Distance judgment logic. If the reference distance is greater than the first-level threshold, it is considered to be within a safe range and no alarm is needed.

[0129] If the reference distance is between the first and second levels → trigger a low-level warning (such as a voice prompt or monitoring pop-up);

[0130] If the reference distance is less than the second-level threshold, a higher-level alarm will be triggered (such as a linked alarm or notification to security personnel).

[0131] If the reference distance is less than the third-level threshold, activate the emergency plan (such as automatic recording, remote announcement, power outage protection, etc.).

[0132] Step 3: Dynamic update mechanism. For moving objects, the system can track their trajectory in real time and continuously update the reference distance to achieve dynamic distance judgment. This can be combined with time-based judgment (e.g., "continuously approaching a certain threshold for more than 5 seconds") to avoid false alarms.

[0133] It should be noted that if unsafe behavior is detected, measuring and judging the safe distance can provide real-time response to potential safety risks, improving the efficiency and accuracy of safety management. By accurately measuring and judging the distance between unsafe behavior and power facilities, the system can quickly identify whether there are potential safety hazards, thereby taking timely early warning or alarm measures. This not only effectively prevents accidents from occurring but also minimizes safety hazards caused by false alarms or missed alarms, providing strong protection for the safe operation of power facilities.

[0134] S104. Determine potential electrical safety hazards based on the results of safety distance measurement and judgment.

[0135] In this embodiment of the invention, determining potential electrical safety hazards based on safety distance measurement and judgment results includes:

[0136] Establish a risk level assessment mechanism;

[0137] Risk level is determined based on the results of safe distance measurement and assessment;

[0138] The risk level assessment mechanism includes quantifying and scoring the results of safety distance measurement and assessment, and classifying different risk levels based on the scoring results;

[0139] Each risk level corresponds to different types of safety hazards and corresponding handling measures.

[0140] In this embodiment of the invention, the quantitative scoring of the safety distance measurement and judgment results includes:

[0141] Establish scoring criteria, which are comprehensively formulated based on the specific nature of the unsafe behavior, the degree of deviation between the object and the preset fixed position, and the magnitude of the potential hazard;

[0142] The results of the safety distance measurement and judgment are compared with the preset scoring criteria to obtain a quantitative score value;

[0143] The higher the quantitative score, the higher the risk level of the electrical safety hazard.

[0144] Specifically, establish a risk level assessment mechanism and set scoring standards, specifically:

[0145] The specific nature of unsafe acts: Different types of unsafe acts have different levels of potential threat. For example, climbing a high-voltage tower is more dangerous than accidentally entering a protected area.

[0146] The degree of distance deviation between an object and a preset fixed position: the closer to critical facilities (such as high-voltage equipment), the higher the risk factor. The distance can be divided into multiple intervals, each corresponding to a different score.

[0147] Potential hazard level: Assess the extent of damage this action may cause to power facilities. For example, damaging cables is more dangerous than simply entering a restricted area.

[0148] Furthermore, the results of safety distance measurement and judgment are quantitatively scored, and specific scoring rules are set, including:

[0149] Behavioral characteristics can be scored as follows:

[0150] Climbing a high-voltage tower: 50 points; illegally entering a protected area: 30 points; operating at close range but without direct contact: 20 points; other minor violations: 10 points;

[0151] Distance deviation scoring can be done as follows:

[0152] Distance less than 1 meter: 50 points; distance 1 to 5 meters: 40 points; distance 5 to 10 meters: 30 points; distance 10 to 20 meters: 20 points; distance more than 20 meters: 10 points;

[0153] Potential hazard rating can be as follows:

[0154] Potential for widespread power outages or severe equipment damage: 50 points; Potential for localized malfunctions: 30 points; Minor impact, may cause temporary shutdown: 20 points; Almost no impact: 10 points;

[0155] For example, suppose there is a specific case where someone illegally enters the protected area 5 meters away from the high-voltage tower and attempts to climb it.

[0156] According to the above scoring rules, the score for the nature of the behavior is: climbing a high-voltage tower (50 points) + illegally entering a protected area (30 points) = 80 points;

[0157] Distance deviation is scored as 5 to 10 meters (30 points);

[0158] The potential hazard score is 50 points, which could lead to widespread power outages or serious equipment damage.

[0159] Overall score = 80 + 30 + 50 = 160 points;

[0160] Furthermore, different risk levels are determined based on the scoring results.

[0161] A total score between 0 and 50 indicates a low-risk situation, suggesting a minor potential hazard that does not require immediate attention.

[0162] For medium-risk cases, the total score is between 51 and 100 points, and it is recommended to strengthen monitoring and take preventive measures.

[0163] When the total score is between 101 and 150, indicating a high risk level, immediate action is required to prevent an accident from occurring.

[0164] When the total score exceeds 150 points in cases of extremely high risk, an emergency response mechanism is triggered, such as automatic alarm or power cut-off.

[0165] It should be noted that in the above case, the overall score was 160 points, which is considered an extremely high risk level.

[0166] Furthermore, each risk level corresponds to different types of safety hazard behaviors and corresponding handling measures, including:

[0167] When the risk is low, the safety hazard behavior type is a minor violation, such as approaching but not entering the danger zone. The remedial action is to send a warning notice to the relevant management personnel and increase the frequency of monitoring the area.

[0168] When the risk level is moderate, the type of security hazard involves more serious violations, such as approaching critical equipment but not yet posing a direct threat. The handling measures include activating the early warning system, notifying security personnel to inspect the site, and recording the incident for subsequent analysis.

[0169] When the risk is high, the safety hazard type refers to actions involving extremely dangerous behaviors, such as directly climbing high-voltage towers or operating within restricted areas. The response measure is to immediately trigger the emergency response mechanism, such as automatic alarms and dispatching an emergency response team to the scene.

[0170] When the risk is extremely high, the safety hazard behavior type is considered an extremely dangerous act, such as performing an operation that could cause a major accident. The response measure is to immediately trigger the highest level of emergency response mechanism, such as automatic alarm, power cut-off, notification of all relevant personnel and departments, and dispatch of an emergency response team to the scene for emergency handling.

[0171] It should be noted that this detailed scoring and grading mechanism allows for the precise assessment of the risk level of each unsafe behavior, enabling the implementation of corresponding measures to prevent and address potential safety hazards. This approach not only improves the efficiency and accuracy of power facility safety management but also effectively reduces the occurrence of safety accidents.

[0172] In this embodiment of the invention, the behavior recognition and judgment model can also be updated by retraining the behavior recognition and judgment model based on the latest collected second historical parameter data; the second historical parameter data is parameter data newly generated during the operation of power facilities, including new video image data and depth perception data.

[0173] In summary, this invention proposes a method for identifying and measuring the behavior of potential power safety hazards. First, by acquiring and preprocessing historical parameter data related to the power facility under test, the quality and accuracy of the input data are ensured, providing a reliable foundation for subsequent behavior identification and judgment. Second, a behavior identification and judgment model is established and trained using the preprocessed data, enabling automatic identification and judgment of unsafe behaviors, significantly improving the real-time performance and accuracy of monitoring. Third, if an unsafe behavior is identified, a safe distance is immediately measured and judged. Precise measurement and judgment further confirm the severity and urgency of the hazard. Finally, based on the safe distance measurement and judgment results, the power safety hazard behavior is determined, and a risk level judgment mechanism is established. This allows for corresponding measures to be taken for different risk levels, effectively preventing safety accidents.

[0174] In a preferred embodiment, the real-time response and automatic optimization mechanism of this invention is one of its key innovations. Through adaptive optimization technology, the system can dynamically adjust itself based on changes in the surrounding environment of power facilities and the types of unsafe behaviors, thereby improving the efficiency and accuracy of the monitoring system. Different environmental conditions, climate changes, and potential safety hazards pose different challenges to the safety of power facilities. To address these challenges, this invention does not rely solely on fixed preset strategies but optimizes its operation through deep learning and real-time data updates, ensuring that the system can maintain efficient monitoring and timely response under any circumstances.

[0175] Whenever the system identifies an unsafe behavior through its adaptive deep learning network, it can quickly assess the risk level of that behavior and make a precise judgment based on the behavior type, the specific location of the occurrence, and real-time data (such as the safe distance between the target and power facilities, and the type of action). For example, when the system detects that people are flying kites within the protection zone of power facilities, it can not only identify the dangerous distance between the kite and the power facilities but also dynamically adjust its response strategy according to changes in the on-site environment. The system acquires precise three-dimensional spatial data through depth sensing devices (such as lidar and infrared sensing devices) to determine whether a safety threat exists. If a kite is detected approaching or entering a dangerous area, the system will immediately trigger an alarm system, send a warning signal to relevant management personnel, and automatically activate the emergency response mechanism.

[0176] This real-time response mechanism relies on the adaptive deep learning model employed in this invention. This model continuously learns and optimizes, enabling not only real-time behavior recognition around power facilities but also automatic optimization of model parameters during the recognition process. For example, when the environment around power facilities changes (such as weather changes, differences in lighting, increased human activity, etc.), the system can automatically adjust its learning strategy, updating model weights and parameters to ensure efficient operation under new environmental conditions. Over time, the system accumulates a large amount of monitoring data, continuously optimizing the accuracy and precision of its recognition. For instance, as more monitoring data is fed back to the system, the model can better adapt to various climatic conditions, geographical environments, and different types of unsafe behaviors, thereby continuously improving its response capabilities to different types of hazards.

[0177] This automatic optimization and adaptive adjustment mechanism significantly reduces the need for manual intervention and enhances the intelligence level of power facility safety monitoring. Traditional power facility monitoring systems typically rely on manual judgment and intervention, but in the face of complex and dynamically changing environments, manual intervention is often inefficient and difficult to respond in real time. This invention, through a highly automated intelligent decision-making system, can react rapidly without human intervention, greatly improving monitoring efficiency and safety response speed. For example, in emergency situations, the system can automatically adjust its response strategy based on the type of hazard and its risk priority, quickly determining the optimal countermeasures. This mechanism not only improves response speed but also ensures efficient allocation of resources and manpower, enabling faster and more accurate emergency response to safety incidents.

[0178] Furthermore, the system can continuously optimize its alarm strategies and emergency plans by learning from and accumulating monitoring data. Each alarm event becomes an opportunity for the system to further learn and optimize, constantly providing feedback based on new scenarios to improve its processing procedures. For example, the system can optimize the sensitivity and warning time of the alarm system based on the type of hazard in different environments (such as potential damage to power facilities during severe weather or changes in personnel behavior), thereby avoiding missed alarms and reducing false alarms.

[0179] Example 3: This example also provides a power safety hazard behavior identification and ranging system, including:

[0180] The data acquisition and processing module is used to acquire the first historical parameter data related to the power facility under test, and to perform preprocessing operations on the first historical parameter data;

[0181] The model building module is used to build a behavior recognition and judgment model, and trains the behavior recognition and judgment model based on the first historical parameter data after preprocessing.

[0182] The behavior recognition and judgment model is used to determine whether unsafe behavior has been identified based on the input.

[0183] The first judgment module is used to measure and judge the safe distance if unsafe behavior is detected.

[0184] The second judgment module is used to determine potential electrical safety hazards based on the results of safety distance measurement and judgment.

[0185] The above-mentioned unit modules can be embedded in the processor of the electronic device in hardware form or independent of it, or they can be stored in the memory of the electronic device in software form, so that the processor can call and execute the corresponding operations of the above modules.

[0186] This embodiment also provides an electronic device, which can be a terminal. The electronic device includes a processor, memory, communication interface, display screen, and input device connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a method for identifying and measuring electrical safety hazards. The display screen can be a liquid crystal display (LCD) or an e-ink display. The input device can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the device's casing, or an external keyboard, touchpad, or mouse.

[0187] This embodiment also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, it performs the following steps:

[0188] Acquire the first historical parameter data related to the power facility under test, and perform preprocessing operations on the first historical parameter data;

[0189] Establish a behavior recognition and judgment model, and train the behavior recognition and judgment model based on the first historical parameter data after preprocessing.

[0190] The behavior recognition and judgment model is used to determine whether unsafe behavior has been identified based on the input.

[0191] If unsafe behavior is detected, a safe distance will be measured and assessed.

[0192] The potential electrical safety hazards are determined based on the results of safety distance measurements and assessments.

[0193] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0194] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0195] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for identifying and measuring behaviors that pose potential power safety hazards, characterized in that, include: Acquire the first historical parameter data related to the power facility under test, and perform preprocessing operations on the first historical parameter data; Establish a behavior recognition and judgment model, and train the behavior recognition and judgment model based on the first historical parameter data after the preprocessing operation; The behavior recognition and judgment model is used to determine whether unsafe behavior has been identified based on the input. If unsafe behavior is detected, a safe distance will be measured and assessed. The potential electrical safety hazards are determined based on the results of safety distance measurements and assessments.

2. The method for identifying and measuring electrical safety hazard behaviors as described in claim 1, characterized in that, The establishment of the behavior recognition and judgment model includes: Establish a set of unsafe behaviors of the power facility under test and a set of features for the set of unsafe behaviors; Perform feature extraction operations on the first historical parameter data regarding the feature set; The data after the feature extraction operation is used as the input to the model, and the set of unsafe behaviors is used as the output of the model.

3. The method for identifying and measuring electrical safety hazard behaviors as described in claim 2, characterized in that, The step of measuring and judging a safe distance if an unsafe behavior is detected includes: The reference distance of the object in the unsafe act is obtained based on the disparity map; the parameter distance is the distance of the object in the unsafe act from a certain preset fixed position; Set a set of judgment thresholds based on the preset fixed position; The safe distance is measured and judged based on the set of judgment thresholds for the output of the behavior recognition and judgment model.

4. The method for identifying and measuring electrical safety hazard behaviors as described in claim 3, characterized in that, The actions taken to determine potential power safety hazards based on the results of safety distance measurement and judgment include: Establish a risk level assessment mechanism; The risk level is determined based on the results of the safety distance measurement and assessment. The risk level assessment mechanism includes quantifying and scoring the results of safety distance measurement and assessment, and classifying different risk levels based on the scoring results. Each risk level corresponds to different types of safety hazards and corresponding handling measures.

5. The method for identifying and measuring electrical safety hazard behaviors as described in claim 4, characterized in that, The quantitative scoring of the safety distance measurement and judgment results includes: Establish scoring criteria, which are comprehensively formulated based on the specific nature of the unsafe behavior, the degree of deviation between the object and the preset fixed position, and the magnitude of the potential hazard; The results of the safety distance measurement and judgment are compared with the preset scoring criteria to obtain a quantitative score value; The higher the quantitative score, the higher the risk level of the potential power safety hazard.

6. The method for identifying and measuring electrical safety hazard behaviors as described in claim 5, characterized in that, The preprocessing operation on the first historical parameter data includes: The first historical parameter data includes video image data and depth perception data; The preprocessing operations include frame-by-frame denoising and brightness adjustment of the video image data, and the creation of a disparity map from the depth-sensing data.

7. The method for identifying and measuring electrical safety hazard behaviors as described in claim 6, characterized in that, It also includes updating the behavior recognition and judgment model by retraining the model based on the latest collected second historical parameter data; the second historical parameter data is parameter data newly generated during the operation of power facilities, including new video image data and depth perception data.

8. A power safety hazard behavior identification and ranging system, using the method described in any one of claims 1 to 7, characterized in that, include: The data acquisition and processing module is used to acquire first historical parameter data related to the power facility under test, and to perform preprocessing operations on the first historical parameter data. The model building module is used to build a behavior recognition and judgment model, and to train the behavior recognition and judgment model based on the first historical parameter data after the preprocessing operation. The behavior recognition and judgment model is used to determine whether unsafe behavior has been identified based on the input. The first judgment module is used to measure and judge the safe distance if unsafe behavior is detected. The second judgment module is used to determine potential electrical safety hazards based on the results of safety distance measurement and judgment.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the power safety hazard behavior identification and ranging method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the power safety hazard behavior identification and ranging method according to any one of claims 1 to 7.

Images