Operation early warning method and device for electric power operation site, terminal equipment and storage medium

Abstract

The invention discloses an operation early warning method and device for an electric power operation site, terminal equipment and a storage medium, and belongs to the technical field of operation early warning, and the method comprises the steps: obtaining multi-dimensional data, such as environment data, site monitoring image data, electric power equipment state data and human body biological characteristic data of an operator; according to the method, the tool falling track and the movement track of the operator are obtained through the multi-dimensional data, quantitative calculation of the specific tool falling track and the movement track of the operator is achieved through collection of the multi-dimensional data, and the specific tool falling position and the movement area of the operator can be accurately obtained in advance; therefore, operation early warning is carried out according to the tool falling track and the personnel activity area, the accuracy of electric power operation site risk early warning is remarkably improved, and the problem that in the prior art, due to the fact that quantitative calculation cannot be carried out on the tool falling track and the personnel movement track in combination with multi-aspect data, risk early warning is not accurate enough can be solved.

Description

Technical Field

[0001] The present invention relates to the field of operation warning technology, and in particular to an operation warning method, device, terminal equipment and storage medium for an electric power operation site. Background Art

[0002] Monitoring and providing early warnings for tool drop risks and personnel safety are crucial at power operation sites. By monitoring and providing early warnings for tool drop trajectories and personnel movement trajectories at the operation site, potential hazards can be identified in advance, ensuring the safety of workers and the normal operation of power equipment.

[0003] However, traditional on-site warning technology for power operations has many shortcomings. When wind speeds exceed a preset wind speed threshold, it typically relies on a single dimension of data to determine risk. For example, it only monitors personnel positions through cameras, or uses simple sensor data to determine tool status for risk warning. This lacks accurate prediction of tool fall trajectories and the movement trajectories of operators. Because it relies on a single dimension of data, traditional methods are unable to quantitatively calculate the specific trajectory of tool falls and the movement trajectories of operators by combining multiple factors such as environmental data, tool location, power equipment status, and real-time biometrics of personnel. As a result, in actual operations, the specific location of tool falls and the activity area of ​​personnel cannot be accurately determined in advance, resulting in inaccurate risk warnings and the inability of operators to respond to sudden dangerous situations in a timely manner. Summary of the Invention

[0004] The embodiments of the present invention provide an operation warning method, device, terminal equipment and storage medium for an electric power operation site. By collecting multi-dimensional data and performing accurate trajectory prediction based on the multi-dimensional data, the quantitative calculation of the specific trajectory of the tool falling and the movement trajectory of the personnel is realized. The specific location of the tool falling and the activity area of ​​the personnel can be accurately obtained in advance, which significantly improves the reliability of the risk warning at the electric power operation site. It can effectively solve the problem in the existing technology that the risk warning is not accurate enough due to the inability to collect and combine various data and the inability to quantify the specific trajectory of the tool falling and the movement trajectory of the personnel.

[0005] An embodiment of the present invention provides an operation warning method for an electric power operation site, comprising:

[0006] Acquire environmental data and operational data corresponding to the power operation site; wherein the environmental data includes: wind speed greater than a preset wind speed threshold; the operational data includes: on-site monitoring image data, power equipment status data, and biometric data of operators; the on-site monitoring image data includes: tool position, operator position, and operator body movements;

[0007] Inputting the operation data into a preset operation warning model so that the operation warning model generates a tool fall trajectory based on the tool position, power equipment status data, and environmental data;

[0008] Generate a movement trajectory of the operator based on the operator's position, the operator's body movements, environmental data, and the human biometric data; wherein the operator's movement trajectory corresponds to an activity area of ​​the operator;

[0009] When it is determined that the falling end point corresponding to the falling trajectory of the tool coincides with the human activity area, generating a first operation warning information; wherein the first operation warning information is used to indicate that the operator is at risk of being hit by the tool;

[0010] When it is determined that the falling end point corresponding to the falling trajectory of the tool does not coincide with the personnel activity area, second operation warning information is generated to indicate that there is a falling risk of the tool.

[0011] Preferably, the power equipment status data includes: the vibration spectrum of each device in the power operation site; the environmental data also includes: air density and electric field strength for quantifying the field strength distribution of high-voltage equipment;

[0012] The operation warning model generates a tool fall trajectory based on the tool position, power equipment status data, and environmental data, including:

[0013] The operation warning model extracts resonance features for characterizing the resonance risk between the tool and the equipment based on the vibration spectrum of each equipment;

[0014] Extracting air density features, wind speed features, and electric field strength features based on the wind speed, air density, and electric field strength;

[0015] generating a direction correction factor for characterizing the falling direction of the correction tool according to the air density characteristics, wind speed characteristics, and electric field strength characteristics;

[0016] The tool falling trajectory is predicted according to the tool position, the resonance characteristics and the direction correction factor.

[0017] Preferably, the human biometric data includes: muscle fatigue and head posture; the operator's body movements correspond to an action type;

[0018] Generating the operator's motion trajectory according to the operator's position, the operator's body movements, environmental data, and the human biometric data includes:

[0019] generating a motion feature for correcting the movement speed according to the operator's body movements, motion types, and muscle fatigue;

[0020] generating a posture feature for correcting a movement direction according to the head posture and muscle fatigue;

[0021] Generating a trajectory deviation feature according to the electric field strength and the posture feature; wherein the trajectory deviation feature is used to measure the degree of influence of the electromagnetic force on the trajectory deviation;

[0022] The movement trajectory of the operator is predicted based on the operator's position, action characteristics, posture characteristics and trajectory deviation characteristics.

[0023] Preferably, the job warning model includes: a backbone network, a spatiotemporal alignment module, a weighted fusion module, a feature pyramid, and a fully connected layer;

[0024] Generating motion features for correcting movement speed based on the operator's limb movements, motion types, and muscle fatigue includes:

[0025] Extracting multi-scale spatial features corresponding to the limb movements through a backbone network, and generating a visual feature map corresponding to the limb movements based on the multi-scale spatial features;

[0026] The spatiotemporal alignment module is used to align the action types and muscle fatigue levels on a temporal axis. The weighted fusion module is then used to perform weighted fusion of the time-aligned action types and muscle fatigue levels to generate a non-visual feature vector.

[0027] Through the feature pyramid, the visual feature map and the non-visual feature vector are spliced ​​at the detection head stage to generate a spliced ​​feature map;

[0028] The action feature is generated according to the spliced ​​feature map through the fully connected layer.

[0029] Preferably, the on-site monitoring image data further includes: the spacing distance between the operator and the energized equipment;

[0030] The operation early warning method for the power operation site also includes:

[0031] Generating a field intensity gradient feature according to the electric field intensity and the separation distance; wherein the field intensity gradient feature is used to quantify the rate of change of the field intensity with the separation distance;

[0032] Generating a risk signature based on the head posture and field intensity gradient characteristics; wherein the risk signature is used to quantify a person's ability to perceive changes in field intensity;

[0033] Generate a safety distance violation probability based on the field intensity gradient characteristics, risk characteristics, wind speed characteristics, and the operator's movement trajectory; wherein the safety distance violation probability is used to represent the probability that the distance between the operator's movement trajectory and the live equipment is less than the safety distance;

[0034] When it is determined that the safety distance violation probability is greater than a preset violation probability threshold, generating third warning information for indicating that the operator is at risk of electric shock;

[0035] When it is determined that the safety distance violation probability is not greater than the preset violation probability threshold, first prompt information is generated to indicate that the operator needs to pay attention to the location of the energized equipment.

[0036] Preferably, after generating the first operation warning information, the method further includes:

[0037] Obtaining a preset bounce distance corresponding to the tool; wherein the preset bounce distance is used to represent the distance between the fall end point corresponding to the fall trajectory of the tool and the collision point when the tool first collides with the ground;

[0038] With the falling end point corresponding to the falling trajectory of the tool as the center and the preset bouncing distance as the radius, a warning area is generated to warn operators not to approach.

[0039] Preferably, after generating the third operation warning information, the method further includes:

[0040] sending a vibration instruction and a second prompt message to the tactile glove worn by the operator, so that the tactile glove vibrates and displays the second prompt message;

[0041] The second prompt message is: Please prohibit operation and evacuate immediately.

[0042] Based on the above method embodiments, the present invention provides corresponding device embodiments.

[0043] An embodiment of the present invention provides an operation warning device for an electric power operation site, comprising: a data acquisition module and an operation warning information generation module;

[0044] The data acquisition module is configured to acquire environmental data and operation data corresponding to the power operation site; wherein the environmental data includes: wind speed greater than a preset wind speed threshold; the operation data includes: on-site monitoring image data, power equipment status data, and operator biometric data; the on-site monitoring image data includes: tool position, operator position, and operator body movements;

[0045] The operation warning information generation module is used to input the operation data into a preset operation warning model, so that the operation warning model generates a tool fall trajectory based on the tool position, power equipment status data, and environmental data; generates an operator movement trajectory based on the operator position, operator body movements, environmental data, and human biometric data; generates first operation warning information when it is determined that the fall end point corresponding to the tool fall trajectory coincides with the operator activity area; and generates second operation warning information for indicating that the tool has a fall risk when it is determined that the fall end point corresponding to the tool fall trajectory does not coincide with the operator activity area.

[0046] The movement trajectory of the operator corresponds to a personnel activity area; the first operation warning information is used to indicate that the operator is at risk of being hit by a tool.

[0047] Based on the above method embodiments, the present invention provides corresponding terminal device embodiments.

[0048] Another embodiment of the present invention provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the operation warning method for an electric power operation site described in the above-mentioned embodiment of the invention.

[0049] Based on the above method embodiment, the present invention provides a corresponding storage medium embodiment.

[0050] Another embodiment of the present invention provides a storage medium, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute an operation warning method for an electric power operation site as described in the above-mentioned embodiment of the invention.

[0051] The following beneficial effects are achieved by implementing the present invention:

[0052] The embodiment of the present invention provides an operation warning method, device, terminal device and storage medium for an electric power operation site. When the wind speed is greater than a preset wind speed threshold and a risk warning is required, the present invention can obtain multi-dimensional data, such as environmental data, on-site monitoring image data, power equipment status data and human biometric data of the operator; in terms of trajectory prediction, the present invention can use the tool position, power equipment status data and environmental data, comprehensively consider the influence of wind speed, the tool's own position characteristics and equipment operating status factors, and accurately obtain the tool's falling trajectory and falling end point. At the same time, based on the operator's position, body movements, environmental data and human biometric data, the operator's position, movement, human biometrics and environmental interference can be fully considered to generate the operator's motion trajectory, and the personnel activity area can be delineated accordingly. The present invention can accurately generate the tool's falling trajectory, falling position and personnel activity area through multi-source data. In the risk warning link, the present invention can combine the predicted tool falling trajectory and personnel trajectory for early warning. When the falling end point corresponding to the tool falling trajectory coincides with the personnel activity area, a first operation warning message is generated, which can clearly inform the operator that there is a high risk of being hit by the tool, so as to remind the operator to take risk avoidance measures immediately; if the two do not coincide, a second operation warning message will also be generated, indicating that the tool is at risk of falling, prompting relevant personnel to take precautions in advance to avoid tool falling accidents. Compared with the existing technology, the present invention realizes the quantitative calculation of the specific trajectory of the tool falling and the movement trajectory of the personnel through multi-dimensional data collection and accurate trajectory prediction based on multi-dimensional data. The specific location of the tool falling and the personnel activity area can be accurately obtained in advance, significantly improving the reliability of risk warnings at the power operation site, so that the operator can respond to sudden dangerous situations in a timely manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 The present invention is a flowchart of an operation warning method for an electric power operation site provided by an embodiment of the present invention.

[0054] Figure 2 1 is a hierarchical diagram of a dangerous behavior analysis and monitoring system for power field operations provided by an embodiment of the present invention.

[0055] Figure 3 The figure is a schematic structural diagram of an operation warning device for an electric power operation site provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0056] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0057] like Figure 1 As shown, in order to solve the problem that the existing technology cannot combine multiple factors such as environmental data, tool location, power equipment status, and real-time biometric characteristics of personnel to quantitatively calculate the specific trajectory of tool falls and personnel movement trajectories, resulting in inaccurate risk warnings, an embodiment of the present invention provides an operation warning method for power operation sites, including:

[0058] Step S1: Acquire environmental data and operation data corresponding to the power operation site; wherein the environmental data includes: wind speed greater than a preset wind speed threshold; the operation data includes: on-site monitoring image data, power equipment status data, and operator's human biometric data; the on-site monitoring image data includes: tool position, operator position, and operator's body movements;

[0059] Illustratively, the present invention can acquire environmental data from power operation sites when providing early warnings for power operation sites, and can also perform risk assessments and early warnings when wind speeds exceed a preset threshold. Unlike conventional technologies, the present invention can also simultaneously collect visual data, environmental data, equipment status data, and human biometric data. Visual data refers to on-site monitoring image data from monitoring power operation sites.

[0060] Specifically, the on-site monitoring image data not only collects the personnel location information and personnel body movements monitored by the camera, but also collects the tool location information, which can accurately grasp the real-time status of the operating personnel and tools. The present invention also collects power equipment status data to determine the potential impact of equipment operation on tools and operations; by collecting real-time biometric data of personnel, the effect of personnel physical condition on the operation motion trajectory can be further evaluated. Therefore, through all-round data collection and multi-dimensional data collection, the present invention can provide a rich and comprehensive data foundation for subsequent trajectory generation.

[0061] Step S2: inputting the operation data into a preset operation warning model, so that the operation warning model generates a tool fall trajectory according to the tool position, power equipment status data and environmental data;

[0062] Generate a movement trajectory of the operator based on the operator's position, the operator's body movements, environmental data, and the human biometric data; wherein the operator's movement trajectory corresponds to an activity area of ​​the operator;

[0063] When it is determined that the falling end point corresponding to the falling trajectory of the tool coincides with the human activity area, generating a first operation warning information; wherein the first operation warning information is used to indicate that the operator is at risk of being hit by the tool;

[0064] When it is determined that the falling end point corresponding to the falling trajectory of the tool does not coincide with the personnel activity area, second operation warning information is generated to indicate that there is a falling risk of the tool.

[0065] Illustratively, the present invention utilizes a pre-defined operational warning model, combined with tool location, power equipment status data, and environmental data, to accurately calculate the tool's fall trajectory. For example, by factoring in the influence of wind speed on the tool's fall direction and speed, and the influence of power equipment status on the probability of a tool falling, the model can simulate the tool's parabolic fall trajectory in complex environments and accurately locate the fall endpoint.

[0066] At the same time, the operation warning model can also generate the operator's movement trajectory based on the operator's position, body movements, environmental data, and human biometric data, and delineate the corresponding personnel activity area. For example, the operator's biometric data affects the human body's behavioral characteristics, and environmental data affects the deviation of the operator's trajectory. Therefore, the present invention can accurately simulate the operator's future movement trajectory based on the operator's position, body movements, environmental data, and human biometric data, thereby obtaining the operator's activity area.

[0067] During the early warning generation phase, the present invention generates different types of early warning messages by determining whether the tool's fall endpoint coincides with the area where people are active. If so, a first early warning message is generated, clearly indicating the risk of the worker being struck by the tool and prompting the worker to take emergency measures. If not, a second early warning message is generated, indicating the risk of the tool falling, prompting the worker to reinforce the tool or adjust their work schedule in advance.

[0068] Therefore, the embodiments of the present invention break the limitations of the single data dimension of traditional technology, realize the precise quantitative calculation of tool falling trajectories and personnel movement trajectories, significantly improve the accuracy of risk warnings, and can effectively respond to sudden dangerous situations at power operation sites, providing strong protection for power operation safety.

[0069] In a preferred embodiment, Figure 2The hierarchical diagram of the dangerous behavior analysis and monitoring system for power field operations shown in the figure, the power field operation early warning method of the present invention can be applied to a dangerous behavior analysis and monitoring system, and the dangerous behavior analysis and monitoring system includes: a multi-source heterogeneous perception layer, an edge intelligent processing layer, a cloud intelligent analysis layer, and an augmented reality early warning layer connected in sequence from bottom to top;

[0070] The multi-source heterogeneous perception layer is used to collect multi-source physical signals at the work site. The multi-source heterogeneous perception layer includes a visual perception unit, an environmental perception unit, a wearable perception unit, and an equipment status perception unit.

[0071] Schematically, the visual perception unit is used to collect on-site monitoring image data, the environmental perception unit is used to collect environmental data, the wearable perception unit is used to collect human biometric data of operators, and the equipment status perception unit is used to collect power equipment status data.

[0072] Furthermore, the operation warning model of the present invention is set in the edge intelligent processing layer, the cloud intelligent analysis layer and the augmented reality warning layer, and then:

[0073] The edge intelligent processing layer is used to achieve spatiotemporal alignment and feature extraction of sensor data, and includes a multimodal data fusion engine and a lightweight AI inference module. Schematically, the edge intelligent processing layer can perform feature extraction on tool position, power equipment status data, and environmental data, and extract features on operator position, operator body movements, environmental data, and human biometric data, thereby obtaining features for generating tool fall trajectories and operator motion trajectories, respectively.

[0074] The cloud-based intelligent analysis layer is used to construct a risk evolution path, which includes a dangerous behavior knowledge graph and a spatiotemporal joint prediction model. Schematically, through the cloud-based intelligent analysis layer, the tool fall trajectory and the operator's movement trajectory can be predicted and inferred based on the extracted features and the preset dangerous behavior knowledge graph.

[0075] The augmented reality warning layer, used to implement multimodal risk warnings, includes a holographic projection warning module and a tactile feedback system. Illustratively, the augmented reality warning layer can compare the tool's fall trajectory with the personnel activity area to determine whether the two coincide. If so, a first operational warning message is generated; if not, a second operational warning message is generated.

[0076] Therefore, the present invention can achieve millimeter-level motion recognition of dangerous behaviors in power operation scenarios, 5-second risk prediction and holographic precise alarm, thereby improving the accuracy of accident identification and shortening the response delay. It also supports adaptive detection and causal chain blocking of illegal operations in complex environments, thereby reducing the operation accident rate.

[0077] For step S1, in a preferred embodiment, the power equipment status data includes: a vibration spectrum of each device in the power operation site; the environmental data also includes: air density and electric field strength for quantifying the field strength distribution of high-voltage equipment.

[0078] The human biometric data includes: muscle fatigue and head posture; the operator's body movements correspond to an action type.

[0079] The on-site monitoring image data also includes: the distance between the operator and the live equipment.

[0080] Schematically, on-site monitoring image data is collected through the visual perception unit, which includes an ultra-low illumination CMOS sensor that can capture the operator's body movements under no lighting conditions, and also includes a binocular depth camera that can capture the tool position, the operator's position, and the distance between the operator and the live equipment.

[0081] Environmental data is collected through environmental sensing, which includes distributed micro-weather stations and electric field strength sensors. The distributed micro-weather stations monitor sudden changes in wind speed and air humidity, while the electric field strength sensors collect electric field strength, which can quantify the field strength distribution around high-voltage equipment.

[0082] Human biometric data is collected through a wearable sensing unit, which includes a smart helmet and a bioelectric monitoring vest. The smart helmet is used to detect the posture of the head; the bioelectric monitoring vest is used to identify the muscle fatigue of the human body through electromyographic signals, and can predict the probability of operational errors.

[0083] Power equipment status data is collected through the device status sensing unit, which includes a vibration fingerprint analysis module and a voiceprint recognition microphone array. The vibration fingerprint analysis module extracts the vibration spectrum of equipment (such as a transformer) to identify loose mechanical faults, while the voiceprint recognition microphone array captures partial discharge ultrasonic signals to locate insulation defects.

[0084] In another preferred embodiment, the visual perception unit also includes a 360° panoramic infrared thermal imaging camera for real-time monitoring of the temperature distribution in the working area, identifying abnormal heat sources such as equipment overheating and arc discharge, and thus providing equipment abnormality warnings.

[0085] The environmental sensing unit also includes a gas composition analyzer, which is used to detect SF6 gas leaks and locate the source of the leak, thereby providing early warning of gas leakage risks. SF6 gas is the abbreviation of sulfur hexafluoride gas. It is a colorless, odorless, non-toxic and non-flammable inert gas with excellent insulation and arc extinguishing properties. It is widely used in power equipment such as SF6 circuit breakers and GIS (gas insulated switchgear). Its molecular structure is stable and its electronegativity is strong. It can quickly adsorb electrons to form low-mobility negative ions, effectively inhibiting the development of arcs and improving the insulation strength and breaking capacity of the equipment. At the power operation site, SF6 gas leakage may cause equipment failure or the risk of suffocation of personnel, so a gas composition analyzer is required for leak detection and positioning.

[0086] Schematically, in power field operations, data collected by multi-source heterogeneous perception layers is closely and multi-dimensionally correlated with risk warnings. Visual perception data provides a spatial and dynamic foundation for risk warnings. Real-time 3D coordinates and the distance between personnel and energized equipment captured by binocular depth cameras can be used to calculate personnel movement speed, direction, and safe distance, thereby predicting risks such as personnel approaching equipment or tools falling and covering personnel. Ultra-low-light CMOS sensors capture human body movements, identifying abnormal postures or illegal operations, and can be used to correct personnel motion trajectory predictions. Infrared thermal imaging captures equipment temperature distribution. Abnormal temperatures can trigger mechanical vibration or insulation aging, indirectly leading to loose tools and the fall of personnel or the emergency evacuation of personnel.

[0087] Furthermore, environmental perception data quantifies the dynamic impact of the external environment on risk. Wind speed data can increase the deviation of tool drop trajectories and affect personnel balance, thereby correcting risk levels and assisting in the prediction of tool drop and personnel movement trajectories. Electric field strength data can calculate the deflection effect of electromagnetic forces on personnel or tools to predict the risk of breaching safety distances or tools falling while energized. SF6 gas leakage data can be used for obstacle avoidance path planning and correcting movement trajectory predictions.

[0088] Wearable sensor data captures a person's physiological state and behavioral risks. Abnormal head posture can indicate distraction or imbalance, thereby revising the predicted direction of movement. Muscle fatigue data can indicate reduced movement speed and increased error probability, which can be used to adjust the speed parameters of the person's movement trajectory.

[0089] Equipment status sensing data also reveals the causal relationship between equipment failure and risk. Abnormal transformer vibration spectra can indicate mechanical looseness, leading to tool resonant drop or equipment failure, prompting emergency evacuation. Partial discharge soundprint data can reveal partial discharge caused by insulation defects, posing the risk of equipment failure and prompting personnel to adjust their work paths or evacuate.

[0090] Furthermore, multi-source data can also be coupled with risk associations. For example, the linkage of environmental, equipment, and personnel data can amplify risk probability. Once these data associations trigger an early warning, the causal chain can be broken through the augmented reality early warning layer. In summary, the embodiments of the present invention, through the integration of multi-source data collection, feature extraction, trajectory simulation modeling, and risk early warning logic, achieve accurate prediction and effective early warning of power operation risks, improving the accuracy of accident identification.

[0091] Regarding step S2, in a preferred embodiment, the tool falling trajectory is generated as follows:

[0092] The operation warning model extracts resonance features for characterizing the resonance risk between the tool and the equipment based on the vibration spectrum of each equipment;

[0093] Extracting air density features, wind speed features, and electric field strength features based on the wind speed, air density, and electric field strength;

[0094] generating a direction correction factor for characterizing the falling direction of the correction tool according to the air density characteristics, wind speed characteristics, and electric field strength characteristics;

[0095] The tool falling trajectory is predicted according to the tool position, the resonance characteristics and the direction correction factor.

[0096] Schematically, the resonance risk can be analyzed by comparing the matching degree of the equipment's vibration spectrum with the tool's natural frequency, thereby generating a resonance signature. If the difference between the equipment's vibration frequency and the tool's natural frequency is less than 5Hz and the spectrum offset is greater than 10%, the probability of the tool falling due to resonance loosening increases significantly (for example, when a transformer's main vibration frequency of 100Hz matches a wrench's natural frequency of 98Hz, resonance can cause bolts to loosen), thus affecting the tool's falling trajectory.

[0097] Schematically, resonance doesn't directly change the trajectory shape, but by increasing the weight of the fall probability, the trajectory prediction model prioritizes the fall scenario. For example, the probability of a tool falling is 40% without resonance, but increases to 70% with resonance. The trajectory prediction can be adjusted from the tool remaining on the platform to falling to the ground.

[0098] Schematically, the operation warning model includes a physical simulation module, which can be used to simulate the tool falling trajectory. The module is based on Newton's mechanics equations and combines the influence of environmental force fields and equipment vibration to achieve dynamic prediction of tool trajectory.

[0099] Furthermore, for the generation of the operator's motion trajectory, there are:

[0100] generating a motion feature for correcting the movement speed according to the operator's body movements, motion types, and muscle fatigue;

[0101] generating a posture feature for correcting a movement direction according to the head posture and muscle fatigue;

[0102] Generating a trajectory deviation feature according to the electric field strength and the posture feature; wherein the trajectory deviation feature is used to measure the degree of influence of the electromagnetic force on the trajectory deviation;

[0103] The movement trajectory of the operator is predicted based on the operator's position, action characteristics, posture characteristics and trajectory deviation characteristics.

[0104] Schematically, the operation warning model can determine the spatial reference and initial state of the trajectory prediction through the position of the operator, and can quantify the attenuation effect of the physiological state on the movement speed through the action characteristics; through the posture characteristics, the human body's attention and balance ability, and the interference on the movement direction, can be obtained; through the trajectory offset characteristics, the physical offset effect of the electromagnetic force field on the trajectory can be quantified. The present invention can obtain the final operator's movement trajectory through multi-feature fusion.

[0105] In a preferred embodiment, the operation warning model can be based on a spatiotemporal joint prediction model (Transformer-TCN+physical simulation), substitute the position coordinates, velocity attenuation value, direction deviation angle, and electromagnetic force offset into the kinematic equation, and iteratively calculate the motion trajectory for the next 5 seconds (such as the end point coordinates = initial position + velocity × time + electromagnetic force offset), thereby outputting the final motion trajectory of the operator.

[0106] In a preferred embodiment, the job warning model of the present invention is a target detection model corresponding to the improved YOLOv8-Tiny, where Tiny is the suffix in the YOLOv8-Tiny model. YOLOv8-Tiny represents a lightweight variant of the YOLOv8 (YouOnly Look Once v8) target detection framework, which mainly reduces the number of parameters and computational complexity through technologies such as channel pruning and model compression, and can adapt to low computing resource scenarios of edge devices.

[0107] The job warning model can process the raw data of the visual perception unit (RGB images, infrared thermal images, depth maps), and adopts a feature-level fusion strategy for other perception data:

[0108] For visual data (RGB-D images, infrared heat maps): directly input the model, and the backbone network extracts spatial features (such as helmet outline, tool position, body movements, etc.);

[0109] For non-visual data (distance, temperature, posture, etc.), it can be pre-processed by the multimodal data fusion engine to generate auxiliary feature vectors, which are then combined with the visual feature map in the detection head. The multimodal data fusion engine includes a spatiotemporal alignment module and a weighted fusion module.

[0110] Schematically, the spatiotemporal alignment module adopts a heterogeneous data synchronization mechanism based on the dynamic time warping (DTW) algorithm to make the time alignment error less than 10ms; the adaptive weighted fusion model can dynamically adjust the fusion weight according to the sensor confidence. The weight calculation formula is: in represents the measurement error variance of the i-th sensor, and n is the number of sensors.

[0111] Then, the job warning model includes: a backbone network, a spatiotemporal alignment module, a weighted fusion module, a feature pyramid and a fully connected layer;

[0112] Generating motion features for correcting movement speed based on the operator's limb movements, motion types, and muscle fatigue includes:

[0113] Extracting multi-scale spatial features corresponding to the limb movements through a backbone network, and generating a visual feature map corresponding to the limb movements based on the multi-scale spatial features;

[0114] The spatiotemporal alignment module is used to align the action types and muscle fatigue levels on a temporal axis. The weighted fusion module is then used to perform weighted fusion of the time-aligned action types and muscle fatigue levels to generate a non-visual feature vector.

[0115] Through the feature pyramid, the visual feature map and the non-visual feature vector are spliced ​​at the detection head stage to generate a spliced ​​feature map;

[0116] The action feature is generated according to the spliced ​​feature map through the fully connected layer.

[0117] Illustratively, the embodiment of the present invention can align and weightedly fuse the data of non-visual images through the spatiotemporal alignment module and the weighted fusion module, thereby generating non-visual feature vectors, so as to further realize the splicing and fusion of visual feature maps and non-visual feature vectors, solve the problem of temporal consistency of multi-source data, and improve the accuracy of feature fusion.

[0118] Furthermore, visual feature maps (such as limb movement contours) provide spatial position information, and non-visual feature vectors (such as fatigue level and field strength) supplement physiological state and environmental influences. The two are spliced ​​together to form multi-dimensional features that can form spatial, physiological, and environmental features, thereby improving the recognition accuracy of motion features.

[0119] It can be understood that when the present invention extracts the tool position, the resonance characteristics, the direction correction factor, the operator position, the action characteristics, the posture characteristics and the trajectory offset characteristics, if there are also steps of extracting and fusing non-visual features, the above-mentioned network hierarchical extraction principle is adopted to realize the splicing of the visual feature map and the non-visual feature vector at the detection head stage. By splicing multimodal features, the semantic information is enriched, thereby improving the trajectory prediction ability of the model.

[0120] The operation warning model of the present invention achieves accurate prediction and real-time blocking of power operation risks through multi-source data fusion, lightweight model design and edge-cloud collaborative architecture.

[0121] First, a multi-source heterogeneous perception layer is used to collect visual data (three-dimensional coordinates of binocular depth cameras, limb movements of CMOS sensors), environmental data (wind speed, electric field strength), wearable data (head posture, muscle fatigue) and equipment status data (transformer vibration spectrum, partial discharge signals), providing full-dimensional input for the operation warning model.

[0122] The non-visual data is processed by the spatiotemporal alignment module (DTW algorithm) and weighted fusion module of the multimodal data fusion engine to generate auxiliary non-visual feature vectors. The non-visual feature vectors are spliced ​​with the visual feature maps in the detection head stage to solve the problems of data timing inconsistency and noise interference.

[0123] In a preferred embodiment, the operation warning method for the power operation site further includes:

[0124] Generating a field intensity gradient feature according to the electric field intensity and the separation distance; wherein the field intensity gradient feature is used to quantify the rate of change of the field intensity with the separation distance;

[0125] Generating a risk signature based on the head posture and field intensity gradient characteristics; wherein the risk signature is used to quantify a person's ability to perceive changes in field intensity;

[0126] Generate a safety distance violation probability based on the field intensity gradient characteristics, risk characteristics, wind speed characteristics, and the operator's movement trajectory; wherein the safety distance violation probability is used to represent the probability that the distance between the operator's movement trajectory and the live equipment is less than the safety distance;

[0127] When it is determined that the safety distance violation probability is greater than a preset violation probability threshold, generating third warning information for indicating that the operator is at risk of electric shock;

[0128] When it is determined that the safety distance violation probability is not greater than the preset violation probability threshold, first prompt information is generated to indicate that the operator needs to pay attention to the location of the energized equipment.

[0129] In a preferred embodiment, after generating the third operation warning information, the method further includes:

[0130] sending a vibration instruction and a second prompt message to the tactile glove worn by the operator, so that the tactile glove vibrates and displays the second prompt message;

[0131] The second prompt message is: Please prohibit operation and evacuate immediately.

[0132] Illustratively, this embodiment of the present invention generates a field strength gradient feature based on electric field strength and separation distance, which can reflect the rate of change of field strength with distance in real time. For example, when a person approaches energized equipment, the field strength gradient increases, visually demonstrating a sharp increase in the risk of electric shock. This provides a more dynamic representation of risk changes than a single field strength value.

[0133] Combining head posture with field intensity gradients to generate risk signatures can quantify a person's ability to perceive field intensity changes. For example, when the head posture is abnormal (such as bowing the head for an extended period), the person's sensitivity to field intensity changes decreases, and the risk signature value increases. This allows for coupled risk assessment of physiological status and the electric field environment, avoiding misjudgments due to insufficient human perception.

[0134] By integrating field intensity gradients, risk characteristics, wind speed characteristics, and personnel movement trajectories to generate safe distance violation probabilities, physical field intensity data can be combined with personnel behavior and environmental factors to achieve quantitative risk prediction. For example, strong winds can affect a person's balance. Incorporating field intensity gradients can more accurately predict the probability of safe distance violation, providing a more scientific basis for early warning decisions.

[0135] When the probability of breaking the safety distance is greater than the threshold, the third warning information (risk of electric shock) is generated, triggering a high-priority warning; when the probability is not greater than the threshold, the first prompt information (pay attention to the location of the equipment) is generated, realizing risk-graded response, allowing operators to focus more accurately on high-risk scenarios and improve overall response efficiency.

[0136] Furthermore, after generating the third warning message, a vibration command and a second prompt message, "No operation, evacuate immediately," can be sent to the tactile glove. This dual stimulation of tactile feedback and visual prompts can enhance the worker's perception of high risks. For example, in a strong electric field environment, the vibration of the tactile glove can help the worker quickly realize the danger, compensating for the lack of visual or auditory warnings that may be interfered with by the environment.

[0137] In a preferred embodiment, after generating the first operation warning information, the method further includes:

[0138] Obtaining a preset bounce distance corresponding to the tool; wherein the preset bounce distance is used to represent the distance between the fall end point corresponding to the fall trajectory of the tool and the collision point when the tool first collides with the ground;

[0139] With the falling end point corresponding to the falling trajectory of the tool as the center and the preset bouncing distance as the radius, a warning area is generated to warn operators not to approach.

[0140] It's understandable that this embodiment of the present invention takes into account the preset bounce distance of a fallen tool, generating a warning zone centered around the end point of the fall and with the bounce distance as its radius, rather than focusing solely on the initial drop point. This is more consistent with the actual scenario of a tool falling. For example, a metal tool may bounce after falling to the ground, and the expansion of the warning zone can cover the potential range of secondary damage, improving the comprehensiveness of the warning.

[0141] By defining a warning area and explicitly prohibiting workers from approaching, secondary injuries caused by falling and bouncing tools can be effectively prevented. Compared to simply warning of fall risks, this invention can proactively block the spread of risks spatially, reducing the probability of accidents.

[0142] Therefore, the present invention can build a human-machine collaborative safety protection system through multimodal warning methods such as holographic projection and tactile feedback. It not only uses technical means to achieve accurate warning, but also ensures timely response of operators through human sensory stimulation, thereby improving the overall safety protection level.

[0143] Indicatively, the present invention can also improve the YOLOv8-Tiny job warning model, such as through channel pruning (such as compressing the number of C3 module channels from 256 to 128) and dynamic depth adjustment (skipping non-critical layers when resources are insufficient), compressing the parameter amount to less than 5M, adapting to the low computing power scenario of edge devices, and the inference delay ≤50ms.

[0144] The backbone network of the operation warning model (CSPDarknet-Tiny) can also extract multi-scale visual features (such as the outline of the helmet and the position of the tool). The feature pyramid (FPN) fuses the visual feature map with the non-visual feature vector to generate an enhanced feature vector containing spatial, physiological and environmental information.

[0145] In a preferred embodiment, the cloud-based intelligent analysis layer can construct risk evolution paths and implement trajectory prediction. It forms a complete risk prediction chain through data and knowledge graph fusion, spatiotemporal joint modeling, and dynamic risk blocking. The specific steps are as follows:

[0146] S100, deep integration of data and knowledge graphs:

[0147] S101. Define the knowledge graph structure. The power operation ontology library defines three entities: "personnel, equipment, and environment." These entities are associated with action nodes (e.g., "climbing a tower" and "operating an insulating rod") and risk rules (e.g., "not wearing a helmet + being less than 1 meter from live equipment → risk of electric shock"). For example, the personnel entity includes attributes such as "helmet status" and "distance from equipment," while the equipment entity includes attributes such as "vibration spectrum" and "insulation status."

[0148] S102. Establish data mapping rules: Visual data → bind to the location and action attributes of the "personnel" entity; Environmental data → bind to the wind speed and electric field strength attributes of the "environment" entity; Equipment data → bind to the vibration and insulation status attributes of the "equipment" entity;

[0149] In practical applications, it can be as follows: when the binocular camera detects that "person A is 2 meters away from the transformer" and the vibration sensor shows that "the transformer vibration spectrum is abnormal", the knowledge graph automatically triggers the causal chain of "mechanical failure → person approaching → fall risk";

[0150] S200, core processing flow of the spatiotemporal joint prediction model:

[0151] S201. Time series alignment: Integrate multi-source data (such as personnel coordinates, wind speed, and vibration spectrum) through a sliding window (5 seconds in length); perform feature normalization to normalize heterogeneous data such as electric field strength (kV / m) and wind speed (m / s) to the [0, 1] range.

[0152] S202. Spatiotemporal feature extraction: Capture long-term temporal dependencies (such as the continuous movement trend of personnel) through the Transformer encoder; extract local temporal features (such as the acceleration change of a falling tool) through the TCN (temporal convolutional network).

[0153] S203, physical simulation module intervention: The tool falling trajectory modeling is based on Newton's mechanics equations, and the falling path is calculated in combination with wind speed and air density (environmental data); the personnel motion trajectory correction introduces electric field gradient data to predict the displacement of personnel affected by electromagnetic force, such as when the field strength is 25kV / m, the displacement is 0.3125m within 5 seconds).

[0154] S300, Dynamic Generation and Blocking of Risk Paths:

[0155] S301, risk probability calculation: output the probability of each risk event in the next 5 seconds (e.g., "probability of tool falling: 72%", "probability of personnel electric shock: 65%");

[0156] S302, Causal chain blocking strategy: If it is predicted that "the tool falling trajectory covers the personnel activity area", the holographic projection warning box is immediately triggered; if it is detected that "personnel fatigue > threshold + close to live equipment", a strong vibration command is sent through the tactile gloves to forcibly interrupt the operation.

[0157] It is understandable that the entity attributes mapped by the knowledge graph (such as personnel location and equipment vibration) serve as inputs to the spatiotemporal joint prediction model, supporting the initial conditions for trajectory prediction (such as the coordinates of the personnel starting point and the risk of tool resonance caused by equipment vibration). The trajectory coordinates (such as the end point of the tool fall and the movement trajectory of the personnel) and risk probabilities output by the spatiotemporal joint prediction model serve as the direct basis for S300 to generate warning and blocking strategies (such as whether the end point coincides with the personnel activity area).

[0158] In a preferred embodiment, the present invention can also be applied to the following scenario: the power operation site is a high-voltage tower maintenance operation site, and person B is not wearing insulating gloves and is close to live equipment;

[0159] First, the binocular camera detects that the distance between person B and the device is 0.8m; the wearable device detects that person B's muscle fatigue is >70%;

[0160] Then, reasoning is performed based on the knowledge graph. For example, the risk chain of not wearing gloves + too close distance + fatigue → electric shock is triggered; through Transformer-TCN, it is predicted that person B's hand will touch a live part in 5 seconds, thereby generating a holographic red warning box covering the equipment area, and finally sending a "no operation" pulse vibration to the tactile glove.

[0161] In a preferred embodiment, when the target detection result of the operation warning model is to output a bounding box and category confidence (such as "not wearing a hard hat: 0.92" and "illegally placed tools: 0.85"), 512-dimensional semantic features (including multi-dimensional information such as space, posture, and environment) can be extracted from the backbone network terminal and uploaded to the cloud-based intelligent analysis layer to issue warnings for not wearing a hard hat and illegally placed tools;

[0162] The cloud can also perform associative reasoning based on the knowledge graph of dangerous behaviors. For example, if the feature vector contains "illegal tool placement + personnel distance from live equipment <1m", it triggers the "electric shock risk" causal chain, thereby generating an early warning of electric shock risk. Furthermore, the present invention can also combine illegal tool placement and environmental data (such as wind speed >13m / s) to predict the tool's falling trajectory.

[0163] like Figure 3 As shown, based on the above-mentioned various embodiments of the operation warning method for the power operation site, the present invention provides corresponding device embodiments;

[0164] An embodiment of the present invention provides an operation warning device for an electric power operation site, comprising: a data acquisition module and an operation warning information generation module;

[0165] The data acquisition module is configured to acquire environmental data and operation data corresponding to the power operation site; wherein the environmental data includes: wind speed greater than a preset wind speed threshold; the operation data includes: on-site monitoring image data, power equipment status data, and operator biometric data; the on-site monitoring image data includes: tool position, operator position, and operator body movements;

[0166] The operation warning information generation module is used to input the operation data into a preset operation warning model, so that the operation warning model generates a tool fall trajectory based on the tool position, power equipment status data, and environmental data; generates an operator movement trajectory based on the operator position, operator body movements, environmental data, and human biometric data; generates first operation warning information when it is determined that the fall end point corresponding to the tool fall trajectory coincides with the operator activity area; and generates second operation warning information for indicating that the tool has a fall risk when it is determined that the fall end point corresponding to the tool fall trajectory does not coincide with the operator activity area.

[0167] The movement trajectory of the operator corresponds to a personnel activity area; the first operation warning information is used to indicate that the operator is at risk of being hit by a tool.

[0168] It should be noted that the device embodiments described above are merely illustrative, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, and may be located in one place or distributed across multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. In addition, in the accompanying drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. Those of ordinary skill in the art can understand and implement the present invention without inventive effort.

[0169] Those skilled in the art can clearly understand that, for the sake of convenience and brevity, the specific working process of the device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0170] Based on the above-mentioned embodiments of the operation warning method for various power operation sites, the present invention provides corresponding embodiments of terminal equipment items.

[0171] An embodiment of the present invention provides a terminal device, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements an operation warning method for an electric power operation site as described in any method embodiment of the present invention.

[0172] The terminal device may be a computing terminal device such as a desktop computer, a notebook computer, a palmtop computer, a cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0173] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the terminal device, connecting various parts of the entire terminal device using various interfaces and lines.

[0174] The memory can be used to store the computer program, and the processor implements various functions of the terminal device by running or executing the computer program stored in the memory and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required for a function, etc.; the data storage area can store data created based on the use of the mobile phone, etc. In addition, the memory can include a high-speed random access memory and can also include a non-volatile memory, such as a hard disk, internal memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device or other volatile solid-state storage device.

[0175] Based on the above-mentioned various embodiments of the operation warning method for power operation sites, the present invention provides corresponding embodiments of storage media items.

[0176] An embodiment of the present invention provides a storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute an operation warning method for an electric power operation site as described in any method embodiment of the present invention.

[0177] The storage medium is a computer-readable storage medium, and the computer program is stored in the computer-readable storage medium. When the computer program is executed by the processor, it can implement the steps of the above-mentioned various method embodiments. The computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device that can carry the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0178] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for early warning of power operation sites, characterized in that: include: Acquire environmental data and operational data corresponding to the power operation site; wherein the environmental data includes: wind speed greater than a preset wind speed threshold; the operational data includes: on-site monitoring image data, power equipment status data, and biometric data of operators; the on-site monitoring image data includes: tool position, operator position, and operator body movements; Inputting the operation data into a preset operation warning model so that the operation warning model generates a tool fall trajectory based on the tool position, power equipment status data, and environmental data; Generate a movement trajectory of the operator based on the operator's position, the operator's body movements, environmental data, and the human biometric data; wherein the operator's movement trajectory corresponds to an activity area of ​​the operator; When it is determined that the falling end point corresponding to the falling trajectory of the tool coincides with the human activity area, generating a first operation warning information; wherein the first operation warning information is used to indicate that the operator is at risk of being hit by the tool; When it is determined that the falling end point corresponding to the falling trajectory of the tool does not coincide with the personnel activity area, second operation warning information is generated to indicate that there is a falling risk of the tool.

2. The method for early warning of an electric power operation site according to claim 1, characterized in that: The power equipment status data includes: the vibration spectrum of each device in the power operation site; the environmental data also includes: air density and electric field strength for quantifying the field strength distribution of high-voltage equipment; The operation warning model generates a tool fall trajectory based on the tool position, power equipment status data, and environmental data, including: The operation warning model extracts resonance features for characterizing the resonance risk between the tool and the equipment based on the vibration spectrum of each equipment; Extracting air density features, wind speed features, and electric field strength features based on the wind speed, air density, and electric field strength; generating a direction correction factor for characterizing the falling direction of the correction tool according to the air density characteristics, wind speed characteristics, and electric field strength characteristics; The tool falling trajectory is predicted according to the tool position, the resonance characteristics and the direction correction factor.

3. The method for early warning of an electric power operation site according to claim 2, characterized in that: The human biometric data includes: muscle fatigue and head posture; the operator's body movements correspond to an action type; Generating the operator's motion trajectory according to the operator's position, the operator's body movements, environmental data, and the human biometric data includes: generating a motion feature for correcting the movement speed according to the operator's body movements, motion types, and muscle fatigue; generating a posture feature for correcting a movement direction according to the head posture and muscle fatigue; Generating a trajectory deviation feature according to the electric field strength and the posture feature; wherein the trajectory deviation feature is used to measure the degree of influence of the electromagnetic force on the trajectory deviation; The movement trajectory of the operator is predicted based on the operator's position, action characteristics, posture characteristics and trajectory deviation characteristics.

4. The method for early warning of an electric power operation site according to claim 3, characterized in that: The job warning model includes: a backbone network, a spatiotemporal alignment module, a weighted fusion module, a feature pyramid, and a fully connected layer; Generating motion features for correcting movement speed based on the operator's limb movements, motion types, and muscle fatigue includes: Extracting multi-scale spatial features corresponding to the limb movements through a backbone network, and generating a visual feature map corresponding to the limb movements based on the multi-scale spatial features; The spatiotemporal alignment module is used to align the action types and muscle fatigue levels on a temporal axis. The weighted fusion module is then used to perform weighted fusion of the time-aligned action types and muscle fatigue levels to generate a non-visual feature vector. Through the feature pyramid, the visual feature map and the non-visual feature vector are spliced ​​at the detection head stage to generate a spliced ​​feature map; The action feature is generated according to the spliced ​​feature map through the fully connected layer.

5. The method for early warning of an electric power operation site according to claim 4, characterized in that: The on-site monitoring image data also includes: the distance between the operator and the live equipment; The operation early warning method for the power operation site also includes: Generating a field intensity gradient feature according to the electric field intensity and the separation distance; wherein the field intensity gradient feature is used to quantify the rate of change of the field intensity with the separation distance; Generating a risk signature based on the head posture and field intensity gradient characteristics; wherein the risk signature is used to quantify a person's ability to perceive changes in field intensity; Generate a safety distance violation probability based on the field intensity gradient characteristics, risk characteristics, wind speed characteristics, and the operator's movement trajectory; wherein the safety distance violation probability is used to represent the probability that the distance between the operator's movement trajectory and the live equipment is less than the safety distance; When it is determined that the safety distance violation probability is greater than a preset violation probability threshold, generating third warning information for indicating that the operator is at risk of electric shock; When it is determined that the safety distance violation probability is not greater than the preset violation probability threshold, first prompt information is generated to indicate that the operator needs to pay attention to the location of the energized equipment.

6. The method for early warning of an electric power operation site according to claim 5, characterized in that: After generating the first operation warning information, the method further includes: Obtaining a preset bounce distance corresponding to the tool; wherein the preset bounce distance is used to represent the distance between the fall end point corresponding to the fall trajectory of the tool and the collision point when the tool first collides with the ground; With the falling end point corresponding to the falling trajectory of the tool as the center and the preset bouncing distance as the radius, a warning area is generated to warn operators not to approach.

7. The method for early warning of an electric power operation site according to claim 6, characterized in that: After generating the third operation warning information, the following steps are also included: sending a vibration instruction and a second prompt message to the tactile glove worn by the operator, so that the tactile glove vibrates and displays the second prompt message; The second prompt message is: Please prohibit operation and evacuate immediately.

8. An operation warning device for an electric power operation site, characterized in that: include: Data acquisition module and operation warning information generation module; The data acquisition module is configured to acquire environmental data and operation data corresponding to the power operation site; wherein the environmental data includes: wind speed greater than a preset wind speed threshold; the operation data includes: on-site monitoring image data, power equipment status data, and operator biometric data; the on-site monitoring image data includes: tool position, operator position, and operator body movements; The operation warning information generation module is used to input the operation data into a preset operation warning model, so that the operation warning model generates a tool fall trajectory based on the tool position, power equipment status data, and environmental data; generates an operator movement trajectory based on the operator position, operator body movements, environmental data, and human biometric data; generates first operation warning information when it is determined that the fall end point corresponding to the tool fall trajectory coincides with the operator activity area; and generates second operation warning information for indicating that the tool has a fall risk when it is determined that the fall end point corresponding to the tool fall trajectory does not coincide with the operator activity area. The movement trajectory of the operator corresponds to a personnel activity area; the first operation warning information is used to indicate that the operator is at risk of being hit by a tool.

9. A terminal device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the method implements an operation warning method for an electric power operation site as described in any one of claims 1 to 7.

10. A storage medium, characterized in that: The storage medium includes a stored computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute the operation warning method for an electric power operation site according to any one of claims 1 to 7.

Images