Electricity-related safety behavior hidden danger identification method, system, equipment and medium
Patent Information
- Application Number
- CN202510912269.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-03
- Publication Date
- 2025-11-21
Smart Images
Figure CN120996552A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of electrical safety behavior hidden danger identification, and particularly relates to an electrical safety behavior hidden danger identification method, system, device and medium. BACKGROUND
[0002] With the increasing complexity of power distribution networks and the popularity of electrical facilities, public safety hazards related to electrical facilities have become more serious. The power distribution network not only carries a large amount of power flow, but also involves many safety issues related to electricity, such as external interference and operational errors in the operation of electrical equipment, which can all become potential public safety hazards. Therefore, timely identification and elimination of safety hazards around electrical facilities is crucial to prevent accidents.
[0003] Currently, the public safety hazard monitoring technology of the power system mainly relies on sensor technology, video monitoring technology and intelligent analysis system. Traditional monitoring systems mostly use sensors (such as temperature and humidity, air pressure, load sensors, etc.) and monitoring cameras to collect data, combined with image analysis and pattern recognition technology to evaluate safety hazards. However, the existing technology has the following significant defects:
[0004] Traditional electrical facility safety hazard monitoring systems rely on a single data source, such as only video monitoring or sensor monitoring. Although these technologies have some effect in certain scenarios, single modal data often cannot fully reflect complex and changing safety hazards. In the environment of power distribution network, the safety hazards of electrical equipment and external environmental factors often interweave with each other, and a single data source cannot effectively identify all potential risks.
[0005] The response speed of existing monitoring systems is often limited by data processing capacity, resulting in the inability to respond quickly when safety hazards are discovered. Especially in the event of electrical equipment failure or sudden safety accidents, traditional systems cannot achieve timely warning, thereby increasing the probability of accidents.
[0006] Due to environmental noise and data interference, traditional safety hazard monitoring systems are prone to false positives or false negatives. When sensors are disturbed by environmental changes (such as temperature, humidity, light, etc.), the system often makes mistakes in judgment, fails to accurately identify hazards, or mistakenly determines normal conditions as safety hazards, resulting in unnecessary waste of resources. SUMMARY
[0007] In view of the above existing problems, the present application is proposed.
[0008] Therefore, the application provides an electrical safety behavior hidden danger identification method, system, device and medium, which can solve the problems of traditional monitoring systems in data sources, response speed and false positives and false negatives, and improve the safety hidden danger identification accuracy and response efficiency of power facilities.
[0009] To solve the above technical problems, the application provides the following technical solutions.
[0010] In a first aspect, the application provides an electrical safety behavior hidden danger identification method, comprising:
[0011] Obtaining first historical parameter data related to a target power facility, the first historical parameter data related to the target power facility including video monitoring data, environmental data and behavior data;
[0012] Performing a preprocessing operation on the first historical parameter data related to the target power facility;
[0013] Training a first model, the first model performing behavior action identification and safety distance judgment on input data;
[0014] The training data set of the first model is obtained through the first historical parameter data after the preprocessing operation;
[0015] According to the first model, combining the first real-time parameter data related to the target power facility to perform risk assessment;
[0016] If the risk assessment result is that there is a hidden danger, then performing a corresponding response operation according to the risk assessment result.
[0017] As a preferred scheme of the electrical safety behavior hidden danger identification method, the preprocessing operation includes:
[0018] The first historical parameter data related to the target power facility is divided into several modalities;
[0019] Different preprocessing operations are performed on different modal data;
[0020] A dynamic weight adjustment mechanism for different preprocessing operations for different modal data is established.
[0021] As a preferred scheme of the electrical safety behavior hidden danger identification method, the training of the first model includes:
[0022] The first historical parameter data after the preprocessing operation is divided into sets;
[0023] The first model includes a behavior identification module and a safety distance judgment module;
[0024] The first model comprises a risk assessment module, and the risk assessment module comprises a risk assessment mechanism based on behavior and safety distance judgment logic.
[0025] The preferred scheme can improve the accuracy and robustness of the model. By dividing the preprocessed data into sets, the data can be more effectively utilized, improving the training effect of the model. The behavior recognition module can accurately identify the behavior of personnel around the power facility, and the safety distance determination module can determine whether the safety distance between the personnel and the power facility meets the regulations. The risk assessment module based on the behavior and safety distance judgment logic can comprehensively assess the real-time parameter data and identify potential safety hazards in a timely manner, so as to take corresponding response operations to ensure the safe operation of the power facility.
[0026] As a preferred scheme of the power-related safety behavior hidden danger identification method, the first model further comprises:
[0027] The behavior recognition module is used to predict the behavior state of any object in the first real-time parameter data of the power grid to be tested.
[0028] The safety distance determination module is used to set a plurality of distance thresholds and determine the final safety distance based on the first real-time parameter data of the power grid to be tested.
[0029] As a preferred scheme of the power-related safety behavior hidden danger identification method, the corresponding response operation based on the risk assessment result comprises:
[0030] Risk assessment based on the behavior state of any object and the final safety distance determination result;
[0031] A set of preset response operations and a risk assessment range threshold for the set of response operations;
[0032] Determine the risk assessment range of the risk assessment result, and determine the corresponding response operation according to the risk assessment range.
[0033] As a preferred scheme of the power-related safety behavior hidden danger identification method, the dynamic weight adjustment mechanism for different preprocessing operations for different modal data comprises:
[0034] Establish a set of possible scenarios for different modal data and a weight distribution logic for a plurality of possible scenarios;
[0035] Establish a judgment standard for a single possible scenario based on the set of possible scenarios;
[0036] According to the first real-time parameter data related to the target power facility, and combining the judgment standard of the single possible scenario, the dynamic weight adjustment is performed.
[0037] As a preferred scheme of the power-related safety behavior hidden danger identification method, wherein: the different preprocessing operations for different modal data include:
[0038] The different modal data includes video monitoring modal data, environmental modal data and behavior modal data;
[0039] Feature extraction is performed on the video monitoring modal data, and personnel, vehicle or equipment state recognition is performed in combination with the feature extraction result;
[0040] Safety evaluation of power equipment is performed on the environmental modal data;
[0041] Primary safety distance judgment is performed on the behavior modal data.
[0042] In a second aspect, the present application provides a power-related safety behavior hidden danger identification system, comprising:
[0043] A data acquisition module is configured to acquire first historical parameter data related to a target power facility, wherein the first historical parameter data related to the target power facility includes video monitoring data, environmental data and behavior data;
[0044] A preprocessing module is configured to perform preprocessing operations on the first historical parameter data related to the target power facility;
[0045] A model training module is configured to train a first model, wherein the first model is configured to perform behavior action recognition and safety distance judgment on input data;
[0046] The training data set of the first model is obtained through the first historical parameter data after preprocessing operations;
[0047] A risk assessment module is configured to perform risk assessment in combination with first real-time parameter data related to the target power facility according to the first model;
[0048] If the risk assessment result indicates that there is a hidden danger, corresponding response operations are performed according to the risk assessment result.
[0049] In a third aspect, the present application provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.
[0050] In a fourth aspect, the present application provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the steps of the above method.
[0051] Compared with the prior art, the beneficial effects of the present application are: the present application proposes an electrical safety behavior hidden danger identification method, first, by obtaining comprehensive historical parameter data of the target power facility, including video monitoring data, environmental data and behavior data, providing rich and accurate information basis for model training and subsequent risk assessment. Secondly, the preprocessing operation can effectively clean and organize the data, improve the data quality, and lay a good foundation for model training. Thirdly, the training of the first model can realize the accurate behavior action recognition and safety distance judgment of the input data, which is the key step of identifying safety hazards. Then, combined with real-time parameter data for risk assessment, potential safety hazards can be found in time, improving the response speed and accuracy. Finally, according to the risk assessment result, the corresponding response operation can effectively prevent and control the occurrence of safety accidents, and ensure the safe operation of the power facility. In summary, the electrical safety behavior hidden danger identification method proposed by the present application not only improves the accuracy and efficiency of safety hazard identification, but also enhances the safety and reliability of the power facility. BRIEF DESCRIPTION OF DRAWINGS
[0052] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.
[0053] Figure 1 A method flow chart of an electrical safety behavior hidden danger identification method provided by an embodiment of the present application.
[0054] Figure 2 An internal structure diagram of an electronic device of an electrical safety behavior hidden danger identification method provided by an embodiment of the present application. DETAILED DESCRIPTION
[0055] In order to make the above-mentioned objects, features and advantages of the present application more apparent and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should be within the scope of protection of the present application.
[0056] Embodiment 1, refer to Figure 1 For the first embodiment of the present application, the embodiment provides an electrical safety behavior hidden danger identification method, comprising:
[0057] In the prior art, there are some problems, for example, the traditional monitoring system usually only relies on video monitoring or sensor data, and this way often cannot comprehensively capture and analyze the security risks around the power facilities. Especially in the complex and changeable power distribution network environment, security risks may be caused by the combined action of many factors such as personnel behavior, environmental factors and power equipment state.
[0058] The present application provides a method that can effectively solve the above-mentioned problems, and the following will be combined with multiple embodiments to explain in detail how to realize the power-related security behavior risk identification method.
[0059] Figure 1 A method flowchart of a power-related security behavior risk identification method is shown, which includes:
[0060] S101, acquiring target power facility related first historical parameter data, the target power facility related first historical parameter data including video monitoring data, environmental data and behavior data;
[0061] It should be noted that in the prior art, most systems only rely on a single data source (such as video monitoring or sensors), which is difficult to fully reflect the complex and changeable security risks around the power facilities. The current security risk monitoring system mostly uses traditional data fusion methods, and lacks effective analysis of the relevance and importance between different data sources. The existing system generally has the problem of response lag, and cannot respond in time in the early stage of the risk. Due to environmental noise and data interference, the existing system often faces the problems of false alarm and missed alarm in complex environments, resulting in poor risk identification effect or resource waste. The power distribution network is often in a complex and changeable environment, and the existing technology lacks adaptability in different environments and cannot cope with various external disturbances.
[0062] Therefore, it is necessary to design a specific method that can process multi-modal data and identify power-related security behavior risks through multi-modal data.
[0063] In the embodiments of the present application, the multi-modal data considered by the present application includes target power facility related first historical parameter data, and the target power facility related first historical parameter data includes video monitoring data, environmental data and behavior data.
[0064] In some specific embodiments, video monitoring data, environmental data, and behavior data can be collected and processed in real time by sensors, cameras, and behavior recognition modules. Video monitoring data captures real-time images of target power facilities and their surrounding environment through high-definition cameras, environmental data collects key indicators such as temperature, humidity, and air pressure through environmental sensors, and behavior data is analyzed by behavior recognition modules to extract relevant information such as personnel activities and object movements. These data will be integrated into a multi-modal database to provide a comprehensive and accurate information base for subsequent safety behavior hazard identification.
[0065] It should be noted that obtaining the first historical parameter data related to the target power facility can provide a comprehensive, accurate and diverse source of information for subsequent safety behavior hazard identification. Video monitoring data can visually display the real-time status of power facilities and their surroundings, helping to identify abnormal behavior or potential hazards; environmental data can reflect the external conditions of facility operation, which is important for assessing facility status and predicting potential risks; behavior data reveals the relationship between personnel activities and power facility safety through detailed analysis, providing a basis for developing targeted safety management measures. Integrating these multi-modal data can build a more complete and detailed safety behavior hazard identification framework, thereby improving the accuracy and efficiency of identification.
[0066] S102, pre-processing operation is performed on the first historical parameter data related to the target power facility;
[0067] It should be noted that after obtaining the first historical parameter data, some redundant and useless data are contained in these data, therefore, the first historical parameter data needs to be pre-processed, and because the above-mentioned first historical parameter data obtained is multi-modal data, a unified pre-processing method cannot meet the use requirements, therefore, different pre-processing means are designed for different modal data.
[0068] In the embodiments of the present application, the pre-processing operation includes:
[0069] The first historical parameter data related to the target power facility is divided into several modalities;
[0070] Different pre-processing operations are performed on different modal data;
[0071] A dynamic weight adjustment mechanism for different pre-processing operations for different modal data is established.
[0072] It should be noted that the several modalities can include video monitoring modality data, feature extraction is performed on the video monitoring data, and personnel, vehicle or equipment state recognition is performed in combination with the feature extraction result; environmental modality data, power equipment safety evaluation is performed on the environmental data, considering factors such as temperature, humidity and other conditions that may affect the normal operation of power equipment; behavior modality data, primary safety distance judgment is performed, and behavior data is analyzed to preliminarily judge whether a safe distance is maintained between personnel and power facilities.
[0073] It should be noted that the dynamic weight adjustment mechanism is to better adapt to data changes in different scenarios and improve the flexibility and accuracy of preprocessing operations. This mechanism can dynamically adjust the weights of each modality data in the preprocessing process according to the importance, reliability and real-time performance of different modality data, to ensure that the preprocessing result can more truly reflect the safety situation of the target power facility. For example, in a harsh environment or frequent personnel activity scenario, the weights of video monitoring modality data and behavior modality data can be appropriately increased to more accurately identify potential safety hazards. By implementing this dynamic weight adjustment mechanism, the present application can further improve the accuracy and efficiency of the safety behavior hidden danger identification related to electricity.
[0074] In an embodiment of the present application, different preprocessing operations are performed on different modality data, including:
[0075] Different modality data includes video monitoring modality data, environmental modality data and behavior modality data;
[0076] Feature extraction is performed on the video monitoring modality data, and personnel, vehicle or equipment state recognition is performed in combination with the feature extraction result;
[0077] Safety evaluation of power equipment is performed on the environmental modality data;
[0078] Primary safety distance judgment is performed on the behavior modality data.
[0079] Specifically, the method of the present application is based on multi-modal data acquisition technology, which comprehensively and real-time perceives and monitors power facilities and their surrounding environment through video monitoring, environmental sensors and behavior recognition devices. This process is the basis for ensuring the safe operation of power equipment, and is also an important prerequisite for subsequent hidden danger identification and early warning. The multi-modal data acquisition of the method not only involves different types of data sources, but also focuses on the collaborative processing of data to ensure that the safety hidden danger related to electricity in complex environments can be accurately reflected.
[0080] To effectively extract key information from video data, the present invention uses advanced image processing techniques, particularly the Convolutional Neural Network (CNN) algorithm. CNN can extract deep features from images, including object shape, position, action, etc., helping the method to identify whether there is a potential electrical safety threat around the power equipment. The mathematical representation of the image recognition process is:
[0081] F = CNN(I)
[0082] Where I is the input image, and F is the feature vector extracted by the Convolutional Neural Network. CNN extracts features from images through multiple convolution operations, gradually extracting features from images, and finally uses these features to identify personnel, vehicles or equipment status in the image.
[0083] In some specific embodiments, in terms of behavior monitoring, not only can it be identified whether personnel are close to power equipment, but also whether they are performing illegal operations, such as climbing poles, touching power equipment, etc. Once these behaviors are identified, the system will immediately issue an alarm to alert personnel or relevant personnel of the danger.
[0084] In addition to video monitoring data, environmental data collection is also an important part of the method. By deploying environmental sensors (such as temperature and humidity sensors, wind speed sensors, etc.), the method can monitor the environmental conditions around the power facilities in real time. These environmental data can effectively reflect the potential impact of the surrounding environment on the safety of power facilities, especially in conditions of weather changes, climate abnormalities, etc. The operation safety of power facilities is often significantly affected by external environment.
[0085] In some specific embodiments, when the environmental temperature is too high or the wind speed is too large, the operation risk of power facilities will significantly increase, leading to equipment failure or causing accidents such as fire. Therefore, the system evaluates the safety of power equipment through various data obtained by environmental sensors. The environmental data collection process can be represented as:
[0086] S = {S1, S2,..., S n}
[0087] Where S i represents a specific data item obtained from each environmental sensor (such as temperature, humidity, wind speed, etc.), and S is the set of all collected environmental data. By analyzing these data, the system can determine whether there is a potential safety hazard of power facilities caused by external environmental factors (such as temperature anomalies, high wind speed). In this way, the system can warn of potential environmental risks in advance and take appropriate measures.
[0088] In embodiments of the present invention, behavior data collection is one of the important innovations, mainly through the combination of behavior recognition in video stream and environmental sensor data to determine whether personnel, vehicles, etc. are approaching power facilities and entering dangerous areas. The key technology of behavior data collection is to combine sensor and image recognition technology to obtain and process the behavior patterns of personnel or vehicles in real time, to determine whether there is an electric shock danger or other safety hazards. Safe distance d safe is the preset distance standard, representing that personnel or objects may pose a threat to the safety of power facilities when entering this range, especially near high-voltage power lines.
[0089] The specific calculation method of behavior recognition and safe distance determination is as follows:
[0090] d = Sensor(I)
[0091] Where d represents the distance between the personnel and the power facility measured by the sensor, and I is the real-time data provided by the sensor. The system compares the real-time measured distance with the preset safe distance. If d ≤ d safe , the system considers that there is a potential safety risk and immediately triggers an alarm. This process ensures timely response when personnel and vehicles approach power facilities, preventing electric shock or other safety accidents.
[0092] In some specific embodiments, the behavior recognition system can also combine deep learning technology to recognize and predict possible dangerous behaviors by learning various behavior patterns. For example, the system can not only detect whether personnel are approaching power equipment, but also identify their behavior types, such as climbing power poles, entering high-voltage power areas, etc., which provides strong technical support for power safety management.
[0093] In embodiments of the present invention, by combining infrared sensors, ultrasonic sensors, and video monitoring data, the system can comprehensively monitor and provide real-time warnings for the safety situation around power facilities under more extensive environmental conditions. This multi-dimensional data collection and processing mode greatly enhances the system's hazard identification ability and warning accuracy in complex environments.
[0094] In embodiments of the present invention, the dynamic weight adjustment mechanism for different preprocessing operations of different modal data includes:
[0095] Establishing a set of possible scenarios for different modal data and a weight distribution logic for several possible scenarios;
[0096] Establishing a judgment standard for a single possible scenario based on the set of possible scenarios;
[0097] According to the first real-time parameter data related to the target power facility, combining the judgment standard for a single possible scenario, dynamic weight adjustment is performed.
[0098] Specifically, the specific operation of the dynamic weight adjustment mechanism can be as follows:
[0099] In the field of electrical safety monitoring, traditional single data source monitoring methods often struggle to provide accurate and comprehensive hazard identification due to complex and changing environments and various potential safety hazards. The present application introduces cross-modal data fusion and optimization technology, comprehensively utilizes data from video monitoring, environmental sensors, behavior recognition, and other data sources, and dynamically adjusts the weight of each modality according to its characteristics and importance using a cross-modal attention mechanism, thereby improving the accuracy and efficiency of hazard identification.
[0100] After the method receives data from different modalities, the data is first weighted and fused. Each modality's data is assigned different weights according to the current environmental needs and task priorities, so that the most relevant and important information is given priority. This weighted fusion can effectively integrate information from different data sources, avoiding the shortcomings of single modalities that cannot fully identify hazards. This weighted fusion can be represented by the following formula:
[0101]
[0102] where D i is the data of the i-th modality, α i is the weight coefficient of the modality data, and D fused is the weighted and fused data. In this process, each modality's data is given different weights according to its contribution to the final identification result. For example, in some specific power facility environments, video data may be more important than temperature and humidity data, so its weight may be higher.
[0103] It should be noted that the advantage of weighted fusion is that it can effectively integrate information from different data sources, avoiding the shortcomings of single modalities that cannot fully identify hazards. By weighting each modality's data, the system can integrate information from multiple angles, resulting in more accurate and comprehensive hazard assessment. For example, a temperature sensor may detect signs of overheating of a device, but only video monitoring can capture the behavior of people around the device. Combining these two pieces of information, the system can draw more reliable conclusions for early warning.
[0104] Based on the above weighted fusion operation, dynamic weight adjustment is performed. As the environment changes, the method calculates the relevance of each modality's data and its importance in the current scene in real time, automatically adjusting the weights of each modality. This weight adjustment mechanism allows the method to dynamically optimize data fusion based on actual conditions, improving the accuracy of hazard identification and the ability to respond to emergencies. The process of dynamically adjusting weights can be achieved through a cross-modal attention mechanism, and the specific calculation method is as follows:
[0105]
[0106] Among them, score i It is the correlation score calculated for the i-th modality of data, α i This refers to the dynamic weight of the modality. The formula uses a normalized exponential function to calculate the importance of each modality data point and dynamically adjusts its weight by comparing the scores of different modalities. This mechanism ensures that the system can prioritize the most relevant modality data for hazard identification under different environments.
[0107] For example, in some situations, environmental sensors may provide critical information when they detect extreme weather conditions, while the weight of video surveillance data will automatically increase when abnormal behavior occurs near power equipment.
[0108] It should be noted that preprocessing the initial historical parameter data related to the target power facility can improve the accuracy and efficiency of subsequent hazard identification. Preprocessing removes noise and redundant information from the data, making it clearer and more accurate, thereby enhancing the system's identification capabilities. Simultaneously, preprocessing historical data provides the system with more comprehensive and in-depth information on the power facility's status, helping the system better understand the facility's operating modes and potential risks, thus enabling more accurate judgments in real-time hazard identification. This preprocessing operation is an indispensable part of the electrical safety behavior hazard identification system, providing strong support for the system's stable operation and efficient identification.
[0109] It's important to note that the system can automatically assess the contribution of each modality of information based on real-time acquired environmental changes, personnel behavior, or equipment status. For example, when the system detects personnel approaching power facilities, the weight of video surveillance data increases rapidly because video data is crucial for determining potential safety hazards. Environmental data (such as temperature and humidity) may no longer be the most important factor in this situation, and therefore its weight may decrease. This dynamic adjustment capability is particularly important when handling the safety monitoring of power facilities in complex environments. The environment surrounding power facilities is constantly changing; for example, differences in lighting between day and night, weather changes, and equipment usage status all affect the system's weighting of data modalities. Through this automatic adjustment mechanism, the system can effectively identify potential safety hazards at any time and in any environment.
[0110] Unlike traditional fixed weight methods, the dynamic weight adjustment of the present invention can adaptively optimize the identification process in complex environments. For example, under normal weather conditions, temperature sensors and video surveillance may work together, but in extreme weather (such as storm or high temperature weather), the system will automatically increase the weight of environmental sensors (such as wind speed, humidity, etc.) to better assess the impact of the environment on power facilities. In areas with high population density around the equipment, the weight of video data will automatically increase to accurately monitor whether the personnel are close to high-voltage equipment and identify potential electric shock risks.
[0111] This context-adaptive cross-modal data fusion approach not only improves the accuracy of hazard identification, but also enables the system to better cope with various challenges in complex and changing environments. The introduction of this technology greatly enhances the intelligent level of power facility safety management, enabling the system to respond quickly, judge accurately, and warn in time, providing solid technical support for the safe operation of power systems.
[0112] S103, training a first model, the first model performing behavior action recognition and safety distance judgment on input data;
[0113] In some specific embodiments, the first model can be constructed using deep learning frameworks such as TensorFlow or PyTorch. These deep learning frameworks provide rich neural network construction tools and training algorithms, enabling the first model to quickly and accurately learn to extract key features from input data and perform behavior action recognition and safety distance judgment.
[0114] In some specific embodiments, the first model can also use transfer learning and pre-trained models to further improve its performance. Transfer learning allows the first model to use weights that have been trained on similar tasks as initial weights, which can greatly reduce training time and improve the model's generalization ability. Pre-trained models are pre-trained on large datasets and then fine-tuned for specific tasks.
[0115] In the embodiment of the present invention, the training data set of the first model is obtained by pre-processing the first historical parameter data;
[0116] In the embodiment of the present invention, training the first model includes:
[0117] The pre-processed first historical parameter data is divided into sets;
[0118] The first model includes a behavior recognition module and a safety distance judgment module;
[0119] The first model comprises a risk assessment module, which comprises a risk assessment mechanism based on behavior and safe distance judgment logic.
[0120] In embodiments of the application, the first model further comprises:
[0121] The behavior recognition module is configured to predict the behavior state of any object in the first real-time parameter data of the power grid under test.
[0122] The safe distance determination module is configured to set a plurality of distance thresholds and make a final safe distance determination based on the first real-time parameter data of the power grid under test.
[0123] For example, the behavior recognition module is configured to recognize the behavior state of any object (such as a person or a vehicle) in the video monitoring data. The specific implementation steps of the behavior recognition model can be as follows:
[0124] Receive pre-processed video monitoring modal data.
[0125] Use a convolutional neural network (CNN) to extract image features from video frames. CNN can automatically learn hierarchical features in images, such as edges, textures, shapes, etc.
[0126] Perform optical flow analysis on consecutive frames to capture motion information.
[0127] Input the extracted features into a pre-trained deep learning model (such as LSTM, 3D-CNN or Transformer), which can predict the current behavior category (such as normal walking, climbing a pole, approaching power equipment, etc.) based on historical behavior patterns.
[0128] The model outputs a probability distribution for each behavior category, and selects the category with the highest probability as the final behavior prediction result.
[0129] For example, the safe distance determination module is configured to set a plurality of distance thresholds and determine whether the safe distance between the object and the power facility meets the regulations based on the real-time parameter data. The specific implementation steps can be as follows:
[0130] Receive pre-processed environmental modal data and behavior modal data, as well as the location coordinates of the target power facility.
[0131] Calculate the actual distance between the object and the power facility based on the object's location coordinates in the video and the known location coordinates of the power facility.
[0132] Camera calibration techniques may be required to convert two-dimensional image coordinates to three-dimensional space coordinates.
[0133] A plurality of safety distance thresholds are preset, which can be adjusted according to different scenarios and types of power facilities (for example, the safety distance of high-voltage power lines is usually larger than that of low-voltage power lines).
[0134] If the calculated actual distance is less than the set safety distance threshold, it is marked as "unsafe"; otherwise, it is marked as "safe".
[0135] If an unsafe situation is detected, the system will generate an alarm and record relevant information for subsequent analysis.
[0136] For example, the risk assessment module is used to comprehensively assess potential risks based on the results of the behavior recognition module and the safety distance determination module. The specific implementation steps can be as follows:
[0137] Obtain the behavior state prediction result from the behavior recognition module and the safety distance judgment result from the safety distance determination module.
[0138] Assign appropriate risk weights to different behavior states (for example, the risk weight of climbing a power pole is higher than that of normal walking).
[0139] Similarly, different distance intervals are also assigned appropriate risk weights (for example, the closer to the danger zone, the higher the risk weight).
[0140] According to the comprehensive risk score, compare it with the preset risk assessment range threshold to determine the corresponding response operation (such as issuing a warning, notifying maintenance personnel, starting an emergency plan, etc.).
[0141] As the real-time data changes, continuously update the risk assessment result and adjust the response strategy according to the latest situation.
[0142] Among them, the comprehensive risk score can be calculated by weighted average or other appropriate algorithms, combining the risk weight of the behavior state and the risk weight of the safety distance.
[0143] In the embodiments of the present application, through deep learning algorithms (such as long short-term memory network LSTM), the method can analyze the behavior of personnel obtained from the video and determine whether to touch or approach the power facility. The behavior state b output by the model can be calculated by the following formula:
[0144] b=LSTM(I t ,I t-1 ,...,I t-n )
[0145] Where I t represents the video image data at the current time, and b is the behavior state predicted by the model (for example, normal, close, contact, etc.).
[0146] Further, the distance d between the personnel and the power facility is measured using an infrared sensor or an ultrasonic sensor, and compared with a preset safe distance d safe If d≤d safe , it is considered as a potential danger, and the system will automatically alarm.
[0147] It should be noted that training the first model can improve the recognition accuracy and efficiency of the system. Through training on a large amount of historical video image data, the first model can learn various electrical safety behavior features, so as to quickly and accurately identify potential safety hazards in real-time monitoring. In addition, the trained model can also adapt to video images under different lighting, angles and occlusion conditions, enhancing the robustness and practicality of the system.
[0148] S104, according to the first model, combining the first real-time parameter data related to the target power facility to perform risk assessment;
[0149] If the risk assessment result is that there is a hidden danger, a corresponding response operation is performed according to the risk assessment result.
[0150] In the embodiments of the present application, the corresponding response operation according to the risk assessment result includes:
[0151] According to the behavior state of the arbitrary object and the final safe distance determination result, the risk assessment is performed;
[0152] A set of preset response operations and a risk assessment range threshold for the set of response operations are provided.
[0153] The risk assessment range of the risk assessment result is determined, and the corresponding response operation is determined according to the risk assessment range.
[0154] In some specific embodiments, the risk assessment according to the behavior state of the arbitrary object and the final safe distance determination result includes:
[0155] The behavior state of all detected objects in the current video frame is obtained from the behavior recognition module. For example, whether a person is climbing a pole, whether a vehicle is approaching a power facility, etc.
[0156] The actual distance between each object and the power facility is obtained from the safe distance determination module, and it is determined whether these distances meet the preset safe distance threshold.
[0157] The behavior state and the safe distance of each object are respectively given corresponding risk weights (such as a higher risk weight for climbing a pole, and a lower risk weight for normal walking) through comprehensive risk score calculation.
[0158] In some specific embodiments, the specific steps of the set of preset response operations and the risk assessment range threshold can be as follows:
[0159] The preset response operation set includes low risk, sending a regular reminder or recording a log. Medium risk, issuing a warning notification to relevant personnel, increasing monitoring frequency. High risk, immediately taking emergency measures such as cutting off power, notifying the maintenance team, starting the emergency plan, etc.
[0160] The risk assessment range threshold is designed as follows (the numerical value is calculated by weighted average):
[0161] Low risk range: 0 ≤ comprehensive risk score < 30;
[0162] Medium risk range: 30 ≤ comprehensive risk score < 70;
[0163] High risk range: 70 ≤ comprehensive risk score ≤ 100;
[0164] In some specific embodiments, the specific steps of judging the risk assessment result and determining the corresponding response operation can be as follows:
[0165] According to the comprehensive risk score calculated in the previous step, determine the risk range it belongs to.
[0166] Perform the corresponding response operation:
[0167] When low risk, send a regular reminder to relevant management personnel. Record the event in the system log for subsequent analysis.
[0168] When medium risk, send a warning notification to relevant management personnel, prompting them to pay attention to potential hazards. Increase the monitoring frequency of the area to ensure timely detection of any changes.
[0169] When high risk, immediately take emergency measures such as cutting off power to prevent accidents. Notify the maintenance team through SMS, phone, etc. and require them to arrive at the scene quickly to handle the problem. Start the emergency plan to ensure that all necessary resources and personnel are in place to quickly respond to emergencies.
[0170] For example, suppose the system detects that a person is approaching a high-voltage power device, the specific steps are as follows:
[0171] The behavior recognition module detects that the person is approaching the power facility.
[0172] The behavior risk weight is set to 50 (the risk weight of climbing a power pole is 80, and the risk weight of normal walking is 20).
[0173] The safe distance determination module calculates that the actual distance between the person and the power facility is 3 meters.
[0174] The distance risk weight is set to 60 (the risk weight is 60 when the safe distance threshold is less than 4 meters, and the risk weight is 10 when the safe distance threshold is greater than or equal to 4 meters).
[0175] The comprehensive risk score is calculated as follows:
[0176] Assuming that wbehavior=0.6 and wdistance=0.4, the comprehensive risk score (weighted average calculation) is 54.
[0177] The comprehensive risk score is 54, which belongs to the medium risk range (30≤comprehensive risk score<70).
[0178] Send a warning notice to the relevant management personnel to remind them to pay attention to the situation of the person approaching the power facility.
[0179] Increase the monitoring frequency of the area to ensure that any further changes can be discovered in a timely manner.
[0180] In summary, the present application proposes a method for identifying potential safety hazards related to electrical safety behavior. First, by obtaining comprehensive historical parameter data of the target power facility, including video monitoring data, environmental data and behavior data, it provides a rich and accurate information base for model training and subsequent risk assessment. Second, the preprocessing operation can effectively clean and organize the data, improve the data quality, and lay a good foundation for model training. Third, the training of the first model can achieve accurate behavior action recognition and safety distance judgment of the input data, which is a key step in identifying safety hazards. Then, combined with real-time parameter data for risk assessment, potential safety hazards can be discovered in a timely manner, improving response speed and accuracy. Finally, according to the risk assessment results, corresponding response operations can be carried out to effectively prevent and control the occurrence of safety accidents, ensuring the safe operation of power facilities. In summary, the method for identifying potential safety hazards related to electrical safety behavior proposed in the present application not only improves the accuracy and efficiency of safety hazard identification, but also enhances the safety and reliability of power facilities.
[0181] In one preferred embodiment, the system can improve the accuracy of identifying various safety hazards of power facilities through continuous optimization and learning. For example, if the system initially fails to identify a certain special behavior pattern of personnel, it will adjust itself according to feedback information and gradually update the identification algorithm to improve the accuracy of subsequent identification. This self-learning process not only improves the reliability of hazard identification, but also gradually optimizes according to different historical data, avoiding system failure when facing unknown or sudden dangerous behavior.
[0182] Due to the distribution of power facilities in different environments such as urban and rural areas and mountainous areas, the system has strong environmental adaptability. Through self-learning and deep learning algorithms, the system can learn and adapt to sensor data and behavior patterns in different environments to provide accurate hazard identification under various environmental conditions.
[0183] In urban environments, there are usually more people and vehicle activities around power facilities, which requires the system to have high sensitivity to personnel approaching power equipment. When identifying safety hazards of power facilities, the system will prioritize the analysis of personnel behavior, especially those approaching power facilities and potentially causing electric shock. For example, the system will monitor whether personnel climb power poles or illegally touch high-voltage lines through image recognition technology, and will immediately issue a warning signal once an anomaly is detected.
[0184] Compared with urban environments, power facilities in mountainous or remote areas face different challenges. In these environments, power facilities may be more affected by environmental factors such as temperature, humidity, wind speed, etc., and there are fewer people in these areas, so the system's hazard identification strategy needs to pay more attention to the operating status of the equipment. For example, power facilities may be overloaded or malfunction due to adverse weather, so the system will use temperature sensors, wind speed sensors, etc. to determine whether the equipment is operating normally, avoiding safety accidents caused by equipment failure.
[0185] In summary, the system can dynamically adjust the hazard identification strategy by learning and accumulating data from different regions. Specifically, the system will automatically adjust the sensitivity of the model according to the environmental conditions of the equipment to identify the more common hazard types in that environment. For example, in mountainous areas, environmental sensor data such as temperature and humidity may play a greater role in identifying potential risks of equipment overheating or external environmental changes, while in urban environments, the system will increase its sensitivity to personnel behavior such as illegal contact with power facilities. In addition, the system can automatically adjust its data collection frequency and processing accuracy according to environmental changes. In some extreme weather conditions such as heavy rain and snow, the system may automatically increase the monitoring frequency of power facility equipment to ensure timely detection of equipment failure or anomalies caused by weather changes. Such environmental adaptability enables the system to effectively address safety hazards of power facilities in different regions and under different environmental conditions, further enhancing the system's identification ability and response speed in complex and changing environments.
[0186] Embodiment 3, refer to Figure 2 The embodiment also provides a safety behavior hazard identification system related to electricity, which includes:
[0187] The data acquisition module is configured to acquire first historical parameter data related to the target power facility, the first historical parameter data related to the target power facility including video monitoring data, environmental data, and behavior data.
[0188] The preprocessing module is configured to perform a preprocessing operation on the first historical parameter data related to the target power facility.
[0189] The model training module is configured to train a first model, the first model being configured to perform behavior action recognition and safety distance judgment on input data.
[0190] The training data set of the first model is obtained through the first historical parameter data after the preprocessing operation.
[0191] The risk assessment module is configured to perform risk assessment according to the first model in combination with first real-time parameter data related to the target power facility.
[0192] If the risk assessment result is that there is a hidden danger, a corresponding response operation is performed according to the risk assessment result.
[0193] The above-mentioned various unit modules can be embedded in or independent of the processor in the electronic device in hardware form, or can be stored in the memory in the electronic device in software form, so as to call and execute the operations corresponding to the above-mentioned various modules by the processor.
[0194] The embodiment also provides an electronic device, which can be a terminal, and an internal structure diagram of the electronic device can be as shown in Figure 2 The electronic device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. The processor of the electronic device is configured to provide calculation and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The communication interface of the electronic device is configured to perform wired or wireless communication with an external terminal. The wireless communication can be achieved through WIFI, an operator network, NFC (near field communication), or other technologies. The computer program is executed by the processor to implement an electrical safety behavior hidden danger identification method. The display screen of the electronic device can be a liquid crystal display screen or an electronic ink display screen. The input device of the electronic device can be a touch layer overlaid on the display screen, or can be a key, trackball, or touchpad arranged on the shell of the electronic device. In addition, the input device can be an external keyboard, touchpad, or mouse, etc.
[0195] The embodiment also provides a computer readable storage medium having a computer program stored thereon, and the computer program is executed by the processor to implement the following steps:
[0196] Obtaining first historical parameter data related to the target power facility, the first historical parameter data related to the target power facility including video monitoring data, environmental data and behavior data;
[0197] Performing a preprocessing operation on the first historical parameter data related to the target power facility;
[0198] Training a first model, the first model performing behavior action recognition and safety distance judgment on input data;
[0199] The training data set of the first model is obtained through the first historical parameter data after the preprocessing operation;
[0200] Performing risk assessment according to the first model in combination with first real-time parameter data related to the target power facility;
[0201] If the risk assessment result is that there is a hidden danger, then performing a corresponding response operation according to the risk assessment result.
[0202] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and are not limiting. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the present application, and all should be covered in the scope of the claims of the present application.
[0203] Although the preferred embodiments of the present application have been described, those skilled in the art can make further changes and modifications to these embodiments once they know the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.
[0204] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application also intends to include these modifications and variations.
Claims
1. A method for identifying electrical safety behavior hazards, the method comprising: receiving a plurality of electrical safety behavior data; and identifying an electrical safety behavior hazard based on the plurality of electrical safety behavior data. The method comprises the following steps: acquiring first historical parameter data related to a target power facility, wherein the first historical parameter data related to the target power facility comprises video monitoring data, environmental data and behavior data; performing a preprocessing operation on the first historical parameter data related to the target power facility; training a first model, wherein the first model is used for behavior action recognition and safety distance judgment of input data; the training data set of the first model is obtained through the first historical parameter data after the preprocessing operation; performing risk assessment according to the first model and in combination with first real-time parameter data related to the target power facility; if the risk assessment result is that there is a hidden danger, performing a corresponding response operation according to the risk assessment result.
2. The method of claim 1, wherein the method further comprises: The preprocessing operation comprises the following steps: dividing the first historical parameter data related to the target power facility into several modalities; performing different preprocessing operations on different modal data respectively; establishing a dynamic weight adjustment mechanism for different preprocessing operations of different modal data.
3. The method of claim 2, wherein the method further comprises: The training of the first model comprises the following steps: performing set division on the first historical parameter data after the preprocessing operation; the first model comprises a behavior recognition module and a safety distance judgment module; the first model comprises a risk assessment module, wherein the risk assessment module comprises a risk assessment mechanism based on behavior and safety distance judgment logic.
4. The method of claim 3, wherein the method further comprises: The first model further comprises: the behavior recognition module is used for predicting the behavior state of any object in the first real-time parameter data of the power grid under test; the safety distance judgment module is used for setting several distance threshold values and performing final safety distance judgment according to the first real-time parameter data of the power grid under test.
5. The method of claim 4, wherein the method further comprises: The corresponding response operation according to the risk assessment result comprises the following steps: performing risk assessment according to the behavior state of any object and the final safety distance judgment result; presetting a response operation set and a risk assessment range threshold value for the response operation set; judging the risk assessment range of the risk assessment result, and determining the corresponding response operation according to the risk assessment range.
6. The method of claim 5, wherein the method further comprises: The establishment of the dynamic weight adjustment mechanism for different preprocessing operations of different modal data comprises the following steps: establishing a possible scene set for different modal data and a weight distribution logic for several possible scenes; establishing a judgment standard for a single possible scene based on the possible scene set; performing dynamic weight adjustment according to the first real-time parameter data related to the target power facility in combination with the judgment standard for the single possible scene.
7. The method of claim 6, wherein the method further comprises: The different preprocessing operations for different modal data comprise the following steps: the different modal data comprise video monitoring modal data, environmental modal data and behavior modal data; performing feature extraction on the video monitoring modal data, and performing personnel, vehicle or equipment state recognition in combination with the feature extraction result; performing power equipment safety assessment on the environmental modal data; performing primary safety distance judgment on the behavior modal data.
8. An electrical safety behavior hazard identification system applying the method according to any one of claims 1 to 7, characterized in that The method comprises the following steps: a data acquisition module is used for acquiring first historical parameter data related to a target power facility, wherein the first historical parameter data related to the target power facility comprises video monitoring data, environmental data and behavior data; A preprocessing module is configured to perform a preprocessing operation on first historical parameter data related to the target power facility; A model training module is configured to train a first model, which is configured to perform behavior action recognition and safety distance judgment on input data; The training data set of the first model is obtained from the first historical parameter data after the preprocessing operation; A risk assessment module is configured to perform risk assessment according to the first model in combination with first real-time parameter data related to the target power facility; If the risk assessment result indicates that there is a hidden danger, a corresponding response operation is performed according to the risk assessment result. 9.An electronic device comprising a memory and a processor, the memory storing a computer program, wherein, The processor executes the computer program to implement the steps of the method for identifying hidden dangers of electrical safety behavior according to any one of claims 1-7.
10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method for identifying hidden dangers of electrical safety behavior according to any one of claims 1-7.