Safe distance monitoring and intelligent early warning method and system for live working personnel

Abstract

The invention discloses a safe distance monitoring and intelligent early warning method and system for live working personnel, and relates to the technical field of electric power working safety, and the method comprises the steps: collecting voltage data and image data of a working region in real time, carrying out the preprocessing, extracting the features of the image data through a depth estimation network, and generating a depth map; the depth map is converted into three-dimensional coordinates, and a three-dimensional scene model of live working is generated in combination with an RTK module; constructing a dynamic safety distance formula based on the electric field data and the three-dimensional scene model to calculate the safety distance between the operator and the electrified body; and performing early warning according to the safety distance and performing dynamic path optimization on the operating personnel. Image data features are extracted through a depth estimation network to generate a depth map, and a dynamically updated three-dimensional scene model is constructed in combination with an RTK module, so that the problem of incomplete spatial information in the traditional technology is effectively solved, the safety distance between an operator and an electrified body is accurately calculated based on a dynamic safety distance formula, and the real-time performance of safety assessment is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power operation safety, and particularly relates to a safety distance monitoring and intelligent early warning method and system for live working personnel. Background Art

[0002] In recent years, with the large-scale construction and intelligent upgrading of the power system, live working on high-voltage transmission lines and substations has become one of the important technical means to ensure the stable operation of the power grid. Live working effectively improves the power supply reliability and social and economic benefits of the power grid by avoiding power outage operations. The live working environment is complex and the electric field intensity is high. A little carelessness may cause serious safety hazards to the operating personnel. Therefore, how to accurately monitor the distance between the operating personnel and the live body, provide real-time early warning and dynamic path optimization has become a research hotspot in this field. At present, the monitoring of the safety distance mainly relies on the data collection and analysis based on a single sensor, such as a voltage sensor, a laser rangefinder, etc., or uses a relatively basic static model to judge the safe range. These methods can meet the basic safety requirements under specific conditions, but with the increasing complexity of the live working scenarios, their limitations are gradually emerging.

[0003] There are multiple deficiencies in the existing technologies for safety distance monitoring and early warning. In traditional methods, the judgment based on the static safety distance threshold is difficult to adapt to the dynamic environmental changes and difficult to adjust the safety area of the operating personnel in real time. The single-sensor mode has the problem of single-dimensional data collection and is difficult to comprehensively reflect the spatial information in the operating environment. Relying solely on the voltage sensor is difficult to provide the three-dimensional spatial distribution information between the live body and the operating personnel, while the laser rangefinder is limited by line-of-sight obstruction and environmental light interference, and its accuracy is difficult to guarantee in practical applications. Most of the existing early warning systems are passive, only triggering an alarm when the threshold is exceeded, lacking the functions of active path optimization and dynamic adjustment, and having poor adaptability to complex operating scenarios. How to construct a high-precision, safe and reliable monitoring and early warning system for live working personnel by integrating multi-source data, combining deep learning algorithms and three-dimensional modeling technologies is the key point and difficulty in the current technological development. Summary of the Invention

[0004] In view of the above existing problems, the present invention provides a safety distance monitoring and intelligent early warning method for live working personnel to solve the problem that the existing safety distance monitoring technology is difficult to effectively meet the requirements of accurate monitoring and early warning in dynamic working scenarios, especially the significant deficiencies in the real-time calculation of the safety distance and path optimization under the influence of multiple variables.

[0005] To solve the above technical problems, a safety distance monitoring and intelligent early warning method for live working personnel is proposed, including,

[0006] Real-time collect voltage data and image data of the operation area and preprocess them. Extract the features of the image data through a depth estimation network to generate a depth map; convert the depth map into three-dimensional coordinates and combine with the RTK module to generate a three-dimensional scene model of live working; construct a dynamic safety distance formula based on the electric field data and the three-dimensional scene model to calculate the safety distance between the operator and the live body; give an alarm according to the safety distance and perform dynamic path optimization for the operator; store all data in a relational database and manage it.

[0007] As a preferred solution of a safety distance monitoring and intelligent warning method for live working personnel described in the present invention, wherein: the real-time collection of voltage data and image data of the operation area and preprocessing thereof include,

[0008] Use an electric field intensity sensor and a voltage sensor and install them on the live body in the operation area to collect electric field data in real time. Arrange a monocular camera at the monitoring point and set the collection rate, collect the image of the operation area and transmit it to the central monitoring system;

[0009] The electric field data refers to the voltage value of each live body in the operation area;

[0010] The preprocessing includes enhancing the image contrast and equalizing the brightness of the image data, removing abnormal values and missing data in the electric field data, filling the missing data using linear interpolation method and performing standardization.

[0011] As a preferred solution of a safety distance monitoring and intelligent warning method for live working personnel described in the present invention, wherein: the generation of the depth map includes,

[0012] The depth estimation network includes an encoder and a decoder. The encoder adopts the DenseNet-169 network structure. Input the image data into the depth estimation network. Through the single-layer convolution operation of each layer of convolution in the encoder, use the output feature of each layer as a feature map for layer-by-layer transmission and processing. The decoder gradually restores the resolution of the feature map through convolution layers and bilinear upsampling operations and generates a predicted depth map. The value of each pixel point represents the predicted depth value;

[0013] Use depth difference loss, gradient smoothing loss and structural similarity loss as loss functions. The depth difference loss is:

[0014]

[0015] wherein, L1 represents the loss value of the depth difference loss, represents the predicted depth value of the pixel point p in the predicted depth map, D p represents the true depth value of the pixel point p in the true depth map;

[0016] Introduce an adaptive weight, and dynamically adjust the weight ω according to the prediction error of each pixel point p ;

[0017] The gradient smoothing loss is:

[0018]

[0019] where L2 represents the loss value of the gradient smoothing loss, and g x and g y respectively represent the gradients of the predicted depth map in the x and y directions, and respectively represent the gradient smoothing loss of pixel point p in the x direction and the gradient smoothing loss in the y direction;

[0020] The structural similarity loss is:

[0021]

[0022] where L3 represents the loss value of the structural similarity loss, represents the structural similarity index;

[0023] Construct the final loss function as:

[0024] L s = ω p (L1 + L2 + L3)

[0025] where L s represents the loss value of the final loss function;

[0026] Minimize the final loss function based on the regularization strategy, update the parameters of the depth estimation network, and repeat the processes of forward propagation, loss calculation, backpropagation, and parameter update until the loss function converges;

[0027] Use an existing publicly available depth estimation dataset to perform offline training on the depth estimation network, use the transfer learning strategy to accelerate the training process, and adopt the cosine annealing learning rate strategy to dynamically adjust the learning rate during training;

[0028] In the operation scenario, input the image captured by the monocular camera in real time into the trained depth estimation network to generate the depth map of the current scenario.

[0029] As a preferred solution of a safety distance monitoring and intelligent warning method for live working personnel according to the present invention, wherein: the three-dimensional scene model includes,

[0030] Define the internal parameter matrix and external parameter matrix of the camera coordinate system, and perform point-by-point processing on the coordinates and depth values of each pixel point in the depth map;

[0031] The point-by-point processing is to use the internal parameter of the camera to calculate the three-dimensional coordinates of each pixel point in the depth map in the camera coordinate system and combine them into a point cloud data structure, and convert the points in the camera coordinate system into points in the world coordinate system through the external parameter matrix. Each point is represented as the three-dimensional space coordinates in the world coordinate system;

[0032] Install high-precision RTK modules on the hands, shoulders, waists, knees and feet of the operator's shielding suit. The RTK modules communicate bidirectionally with the reference station, and transmit the high-precision positioning information of the reference station to the operator's RTK modules through differential signals, providing the three-dimensional space coordinates of the operator's hands, shoulders, waists, knees and feet in real time, and connecting to the central monitoring system through wireless network, and selecting a wireless communication frequency band for data transmission;

[0033] All three-dimensional space coordinates are converted into a standard point cloud format and integrated into a point cloud data set;

[0034] For the point cloud data generated at different time points, align them through the iterative closest point algorithm to generate a complete three-dimensional scene model.

[0035] As a preferred solution of a safety distance monitoring and intelligent warning method for live working personnel according to the present invention, wherein: the safety distance includes,

[0036] Convert the three-dimensional space coordinates of each live conductor in the three-dimensional scene model into cylindrical coordinates;

[0037] The conversion of the three-dimensional space coordinates to cylindrical coordinates includes obtaining the three-dimensional space coordinates, length and charge density of the center position of the live conductor, taking the three-dimensional space coordinates of the center position of the live conductor as the central axis, calculating the shortest radial distance r from the three-dimensional space coordinates of the operator to the central axis and obtaining the vertical height z of the three-dimensional space coordinates corresponding to the radial distance e , and taking the radial distance and the vertical height as the cylindrical coordinates;

[0038] Based on the cylindrical coordinates, use the equivalent charge method to calculate the electric field potential value generated by each live conductor as:

[0039]

[0040] Among them, φ i (r,z e ) represents the electric field potential value generated by the i-th live conductor in the cylindrical coordinates (r,z e ), τ represents the linear charge density, L represents the length of the live conductor, z s represents the integration variable along the central axis direction, ε represents the permittivity of air, and d represents the infinitesimal variable of the integration variable z s ;

[0041] The electric field strength value at the location of the operator is calculated through the electric potential values generated by each charged body as follows:

[0042]

[0043] where E i represents the electric field strength value of the i-th charged body, represents the negative value of the spatial rate of change of the electric potential value;

[0044] The electric field strength values of all charged bodies are superimposed to obtain the total electric field strength value of the operation area. The formula is:

[0045]

[0046] where E o represents the total electric field strength value of the operation area, n represents the number of all charged bodies in the operation area, and i is the variable index;

[0047] Taking the operation target position of the operator as the target point, partial derivatives are calculated in the x, y, and z directions of the target point and the electric field gradient is synthesized

[0048] A dynamic safety distance formula is constructed to calculate the safety distance between the operator and the charged body as:

[0049]

[0050] where d a represents the safety distance between the operator and the charged body, V represents the average voltage of all charged bodies in the operation area, and k1, k2, and k3 represent weights.

[0051] As a preferred solution of a safety distance monitoring and intelligent warning method for live working personnel according to the present invention, wherein: the dynamic path optimization includes,

[0052] Select an AR head-mounted display device and wear it on the head of the operator. Start the AR head-mounted display device and connect it to the central monitoring system through a wireless network. Initialize the device of the AR head-mounted display device, and the AR head-mounted display device displays the safety distance of the charged bodies around the operator;

[0053] Real-time calculate the Euclidean distance d b between the operator and the charged body a and compare it with the safety distance d

[0054] If d b ≥d a , it means that the current state of the operator is safe and no warning is required;

[0055] If d b <da , it indicates that the current operator's status is unsafe, triggering the warning mechanism;

[0056] The warning mechanism refers to using the vibration and voice of the AR headset device to prompt the operator to pay attention to safety and perform dynamic path optimization for the current operator;

[0057] The dynamic path optimization refers to setting the starting point of the path as the current three-dimensional coordinates of the operator, dividing the operation area into three-dimensional grid cells and defining a cost function, and using the D* algorithm for initial path planning;

[0058] The three-dimensional grid cells include the distance from the live body and the distance from the obstacle;

[0059] The cost function is defined as the smaller the distance between the operator and the live body, the higher the cost, and the smaller the distance between the operator and the obstacle, the higher the cost;

[0060] The initial path planning using the D* algorithm refers to searching for the optimal path from the starting point to the target point. The priority of each path node is determined according to the current path cost and the heuristic estimated cost. Nodes are gradually expanded from the starting point, and the path with the minimum cost is retrieved and a sequence of path points is generated;

[0061] The current path cost is the cumulative cost from the starting point to the current node;

[0062] The heuristic estimated cost is the estimated cost from the current node to the target point;

[0063] Real-time detection of newly added obstacles and changes in obstacle positions, updating the affected three-dimensional grid cells to an impassable state. If there are three-dimensional grid cells in the current path in an impassable state, the path in the affected area is re-planned locally;

[0064] The optimal path is displayed in the operator's AR headset device to provide real-time navigation prompts.

[0065] As a preferred solution of the safety distance monitoring and intelligent warning method for live working personnel described in the present invention, wherein: the storing all data in a relational database and managing it includes selecting a relational database to store and manage data and analysis results, designing the database table structure to store different types of data, setting up a regular backup task to back up all data in the database, performing permission management on database users and encrypting the storage of data.

[0066] Another object of the present invention is to provide a safety distance monitoring and intelligent warning system for live working personnel. The present invention solves the technical problems of safety distance monitoring, warning and data management in live working. By collecting electric field and image data in real time, generating a depth map and a three-dimensional scene model, dynamically calculating the safety distance between the working personnel and the energized body, and performing intelligent warning and dynamic path optimization accordingly, the safety of live working is improved. At the same time, all data is stored in a database for management, thereby enhancing the overall safety and efficiency of live working.

[0067] As a preferred embodiment of a safety distance monitoring and intelligent warning system for live working personnel according to the present invention, it is characterized in that it includes a data acquisition module, a depth map generation module, a model construction module, a dynamic calculation module, a classification warning module, and a data storage module;

[0068] The data acquisition module is used to collect electric field data and image data of the working area in real time and perform preprocessing;

[0069] The depth map generation module is used to extract the features of the image data through a depth estimation network to generate a depth map;

[0070] The model construction module is used to convert the depth map into three-dimensional coordinates and generate a three-dimensional scene model of live working in combination with the RTK module;

[0071] The dynamic calculation module is used to construct a dynamic safety distance formula based on the electric field data and the three-dimensional scene model to calculate the safety distance between the working personnel and the energized body;

[0072] The classification warning module is used to give warnings according to the safety distance and perform dynamic path optimization on the working personnel;

[0073] The data storage module is used to store all data in a relational database and manage it.

[0074] A computer device includes a memory and a processor. The memory stores a computer program. It is characterized in that when the processor executes the computer program, the steps of a safety distance monitoring and intelligent warning method for live working personnel as described above are implemented.

[0075] A computer-readable storage medium stores a computer program thereon. It is characterized in that when the computer program is executed by a processor, the steps of a safety distance monitoring and intelligent warning method for live working personnel as described above are implemented.

[0076] Advantages of the present invention: The present invention extracts the features of image data through a depth estimation network to generate a depth map, and constructs a dynamically updated three-dimensional scene model in combination with the RTK module, effectively solving the problem of incomplete spatial information in the traditional technology. Based on the electric field data and the dynamic safety distance formula, the safety distance between the operator and the energized body is accurately calculated, improving the real-time performance and adaptability of the safety assessment. Combining intelligent warning and dynamic path optimization technologies, the present invention provides the operator with the ability of real-time visual navigation and dynamic path adjustment, reducing the risk of entering high-risk areas during the operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0078] Figure 1 FIG. is the overall flowchart of a method for monitoring the safety distance and intelligent warning of live working personnel provided by an embodiment of the present invention.

[0079] Figure 2 FIG. is the system scheme module diagram of a system for monitoring the safety distance and intelligent warning of live working personnel provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0080] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the following will describe the specific embodiments of the present invention in detail with reference to the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0081] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0082] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation manner of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it an embodiment that is mutually exclusive with other embodiments alone or selectively.

[0083] The present invention will be described in detail with reference to the schematic diagrams. When describing the embodiments of the present invention, for the convenience of explanation, the cross-sectional views showing the device structure will be enlarged locally not in accordance with the general scale, and the schematic diagrams are only examples and should not limit the scope of protection of the present invention herein. In addition, in actual production, three-dimensional spatial dimensions including length, width, and depth should be included.

[0084] Meanwhile, in the description of the present invention, it should be noted that the orientation or positional relationship indicated by terms such as "upper, lower, inner, and outer" is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus cannot be understood as a limitation to the present invention. In addition, the terms "first, second, or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0085] Unless otherwise clearly defined and limited in the present invention, the terms "installation, connection, and coupling" should be understood in a broad sense. For example: it can be a fixed connection, a detachable connection, or an integral connection; it can also be a mechanical connection, an electrical connection, or a direct connection, and can also be indirectly connected through an intermediate medium, or can be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0086] Example 1, referring to Figure 1 , which is the first embodiment of the present invention. This embodiment provides a method for monitoring the safety distance and intelligent early warning of live working personnel, including:

[0087] S1: Real-time collect the voltage data and image data of the operation area and perform preprocessing, and extract the features of the image data through a depth estimation network to generate a depth map.

[0088] Specifically, real-time collecting the electric field data and image data of the operation area and performing preprocessing includes:

[0089] Use electric field intensity sensors and voltage sensors and install them on the live conductors in the operation area to collect electric field data in real time. Arrange a monocular camera at the monitoring point and set the collection rate, collect the images of the operation area and transmit them to the central monitoring system;

[0090] The voltage data refers to the voltage value of each live conductor in the operation area;

[0091] The live conductors include overhead transmission lines, busbars, low-voltage cables, medium-voltage distribution conductors, etc.;

[0092] The preprocessing includes enhancing the image contrast and equalizing the brightness of the image data, removing the abnormal values and missing data of the electric field data, filling the missing data using the linear interpolation method and performing normalization.

[0093] The above method significantly improves the quality and consistency of data by collecting real-time electric field intensity data and high-resolution image data in the operation area and preprocessing them. This processing method can provide more accurate and reliable input data for subsequent depth estimation and safety distance calculation, while enhancing the system's adaptability to complex operation scenarios and improving the accuracy and robustness of overall monitoring and early warning.

[0094] Furthermore, extracting features from the image data through a depth estimation network to generate a depth map includes:

[0095] The depth estimation network includes an encoder and a decoder. The encoder adopts the DenseNet-169 network structure. The image data is input into the depth estimation network. Through the single-layer convolution operation of each layer of convolution in the encoder, the output feature of each layer is used as a feature map for layer-by-layer transmission and processing. It is set that the first Q layers of convolution capture the edge and contour information of the image data as shallow features, and the remaining layers of convolution extract the scene semantics and context information of the image as deep features. The value of Q is adjusted according to the design logic of different network architectures. For example, when the number of network layers is small (5-10 layers), the shallow layer can be defined as the first 2-3 layers, and skip connections are used to directly transfer the shallow features to the decoder for fusion with the deep features. The decoder gradually restores the resolution of the feature map through convolutional layers and bilinear upsampling operations and generates a predicted depth map. The value of each pixel represents the predicted depth value;

[0096] The context information refers to the spatial position relationship between each pixel and its surrounding pixels. For example,

[0097] Pixels of the sky will appear coherently above the image and have relatively smooth color changes, and pixels of the road usually present continuous shapes and directions;

[0098] Using depth difference loss, gradient smoothing loss, and structural similarity loss as loss functions to optimize the depth estimation network, so that the predicted depth map not only has a value close to the true depth map numerically, but also performs well in local details (such as edges and textures) and global consistency. The depth difference loss formula is:

[0099]

[0100] where, L1 represents the loss value of the depth difference loss, represents the predicted depth value of pixel p in the predicted depth map, D p represents the true depth value of pixel p in the true depth map, which is obtained through labeled data or sensor acquisition;

[0101] The depth difference loss directly measures the difference between the predicted depth value and the true depth value. Minimizing this difference can make the depth value predicted by the model closer to the true value, thereby improving the accuracy of depth estimation;

[0102] Introduce an adaptive weight, and dynamically adjust the weight ω according to the prediction error of each pixel point p , and the formula is:

[0103]

[0104] where ω p represents the weight of pixel point p, γ is used to control the dynamic change range of the weight. The greater the difference between the predicted depth value and the true depth value of each pixel point, the greater the weight, highlighting the loss contribution of the key area. When the predicted depth value is close to the true depth value, the weight decreases;

[0105] The gradient smoothing loss formula is:

[0106]

[0107] where L2 represents the loss value of the gradient smoothing loss, g x and g y respectively represent the gradients of the predicted depth map in the x and y directions, used to constrain the smoothness of the predicted depth map, and respectively represent the gradient smoothing loss of pixel point p in the x direction and the gradient smoothing loss in the y direction, and the expression is:

[0108]

[0109] where is the partial derivative, and respectively represent the partial derivatives of the true depth map and the predicted depth map along the horizontal direction, reflecting the rate of change of the depth value along the horizontal direction;

[0110] and respectively represent the partial derivatives of the true depth map and the predicted depth map along the vertical direction, representing the rate of change of the depth value along the vertical direction;

[0111] This term constrains the local change of the depth value, avoiding depth jumps caused by noise;

[0112] The gradient smoothing loss constrains the gradient change of the depth map, avoiding unnatural "jumps" or noise in the predicted depth map. By smoothing the depth map, the model can make the boundary transition of complex scenes more natural;

[0113] The structural similarity loss formula is:

[0114]

[0115] Among them, L3 represents the loss value of the structural similarity loss, represents the structural similarity index, which measures the texture and detail consistency between the predicted depth map and the ground truth depth map. The formula is:

[0116]

[0117] where μ D and respectively represent the mean depth in the ground truth depth map and the predicted depth map, and respectively represent the variance of the depth values in the ground truth depth map and the predicted depth map, represents the covariance of the depth values in the ground truth depth map and the predicted depth map, and c1 and c2 respectively represent constants used to stabilize the calculation;

[0118] The structural similarity loss maximizes the structural similarity between the predicted depth map and the ground truth depth map, retains key details such as edges and textures in the depth map, and ensures that the model accurately identifies important features in complex scenes;

[0119] Construct the final loss function. The deviation between the prediction result and the ground truth target is measured by minimizing the final loss function. The formula is:

[0120] L s = ω p (L1 + L2 + L3)

[0121] where L s represents the loss value of the final loss function;

[0122] Based on the regularization strategy, minimize the final loss function, update the parameters of the depth estimation network, and repeat the processes of forward propagation, loss calculation, backpropagation, and parameter update until the loss function converges;

[0123] Use an existing publicly available depth estimation dataset (such as KITTI) to perform offline training on the depth estimation network, enabling it to accurately learn the mapping relationship between pixel values and depths. Use the transfer learning strategy to accelerate the training process, and adopt the cosine annealing learning rate strategy to dynamically adjust the learning rate during training;

[0124] In the work scenario, input the images captured in real time by a monocular camera into the trained depth estimation network to generate the depth map of the current scene.

[0125] By introducing a DenseNet-based encoder-decoder network structure, skip connection mechanism, adaptive weights, and the joint optimization of depth difference loss, gradient smoothing loss, and structural similarity loss, the depth estimation network performs excellently in numerical accuracy, local detail preservation, and global consistency. Combining transfer learning strategies and dynamic adjustment of the cosine annealing learning rate effectively accelerates network training and avoids falling into local optima. In the operation scenario, this method can generate high-quality depth maps in real time, accurately capture key depth features and spatial relationships in complex scenes, thereby providing highly reliable and adaptable support for depth perception tasks.

[0126] S2: Convert the depth map into three-dimensional coordinates and combine with the RTK module to generate a three-dimensional scene model for live working.

[0127] Furthermore, converting the depth map into three-dimensional coordinates and combining with the RTK module to generate a three-dimensional scene model for live working includes:

[0128] Define the internal parameter matrix and external parameter matrix of the camera coordinate system, and perform point-by-point processing on the coordinates and depth values of each pixel point in the depth map;

[0129] The internal parameter matrix describes the projection relationship of the camera from the camera coordinate system to the pixel coordinate system, which is a mathematical representation of the internal optical characteristics and imaging geometry of the camera, and is used to project the three-dimensional coordinates in the camera coordinate system onto the two-dimensional pixel plane of the image;

[0130] The external parameter matrix describes the position and orientation of the camera in the world coordinate system, including a rotation matrix and a translation vector, representing the transformation relationship from the world coordinate system to the camera coordinate system, and is used to convert the three-dimensional coordinates from the world coordinate system to the camera coordinate system;

[0131] The point-by-point processing refers to using the internal parameter of the camera to calculate the three-dimensional coordinates of each pixel point in the depth map in the camera coordinate system and combining them into a point cloud data structure, and converting the points in the camera coordinate system to points in the world coordinate system through the external parameter matrix, and each point is represented as the three-dimensional spatial coordinates in the world coordinate system;

[0132] Install high-precision RTK modules (such as Trimble R12) on the hands, shoulders, waists, knees, and feet of the insulating clothing of the operator. The RTK module communicates bidirectionally with the reference station, and transmits the high-precision positioning information of the reference station to the operator's RTK module through differential signals, providing the three-dimensional spatial coordinates of the operator's hands, shoulders, waists, knees, and feet in real time, and connecting to the central monitoring system through a wireless network, and selecting a wireless communication frequency band (such as LORA) for data transmission;

[0133] The high-precision RTK module refers to a high-precision satellite positioning technology device. Based on global navigation satellite system positioning and using differential correction algorithms, it can achieve centimeter-level or even millimeter-level positioning accuracy.

[0134] All three-dimensional space coordinates are converted into the standard point cloud format and integrated into a point cloud data set.

[0135] For the point cloud data generated at different time points, they are aligned through the iterative closest point algorithm to generate a complete three-dimensional scene model.

[0136] It should be noted that the above method processes the depth map point by point by combining the internal parameter matrix and external parameter matrix of the camera, accurately converts two-dimensional pixel data into three-dimensional point cloud coordinates, and uses the high-precision RTK module to obtain the three-dimensional coordinates of the key parts of the operator in real time, realizing the high-precision dynamic modeling of the operator and the live working environment. The point cloud data is aligned through standardized integration and the iterative closest point algorithm to generate a complete three-dimensional scene model, providing support for precise positioning, safety analysis, and path planning in complex working environments. This method has high precision, real-time performance, and scalability, and can significantly improve the digital and intelligent levels of live working scenarios.

[0137] S3: Based on the electric field data and the three-dimensional scene model, construct a dynamic safety distance formula to calculate the safety distance between the operator and the live conductor.

[0138] Furthermore, specifically, constructing a dynamic safety distance formula based on the electric field data and the three-dimensional scene model to calculate the safety distance between the operator and the live conductor includes:

[0139] Convert the three-dimensional space coordinates of each live conductor in the three-dimensional scene model to cylindrical coordinates.

[0140] Converting the three-dimensional space coordinates to cylindrical coordinates includes obtaining the three-dimensional space coordinates, length, and charge density of the center position of the live conductor. Using the three-dimensional space coordinates of the center position of the live conductor as the central axis, calculate the shortest radial distance r from the three-dimensional space coordinates of the operator to the central axis and obtain the vertical height z of the three-dimensional space coordinates corresponding to this radial distance. e , and take the radial distance and the vertical height as the cylindrical coordinates.

[0141] Based on the cylindrical coordinates, use the equivalent charge method to calculate the electric field potential value generated by each live conductor. The formula is:

[0142]

[0143] Among them, φ i (r,z e ) represents the electric field potential value of the i-th live conductor in the cylindrical coordinates (r,z eThe electric field potential value generated therein, τ represents the linear charge density, which is obtained by calculating the ratio of the total charge of the charged body to the length of the charged body, L represents the length of the charged body, and z s represents the integration variable along the central axis direction, ε represents the dielectric constant of air, which is measured by an experimental method, and d represents the infinitesimal variable of the integration variable z s ;

[0144] The electric field strength value at the position of the operator is calculated through the electric field potential value generated by each charged body. The formula is:

[0145]

[0146] where, E i represents the electric field strength value of the i-th charged body, represents the negative value of the spatial variation rate of the electric field potential value;

[0147] The total electric field strength value of the operation area is obtained by superimposing the electric field strength values of all charged bodies. The formula is:

[0148]

[0149] where, E o represents the total electric field strength value of the operation area, and n represents the number of all charged bodies in the operation area;

[0150] Taking the operation target position of the operator as the target point, the partial derivatives are calculated in the x, y, and z directions of the target point and the electric field gradient is synthesized The partial derivative represents the change rate of the electric field strength in each direction and is usually calculated by the numerical difference method. The formula is:

[0151]

[0152] where, represents the electric field gradient of the target point;

[0153] A dynamic safety distance formula is constructed to calculate the safety distance between the operator and the charged body. The formula is:

[0154]

[0155] where, d a represents the safety distance between the operator and the charged body, V represents the average voltage of all charged bodies in the operation area, and k1, k2, and k3 represent weights, which are obtained from experimental data.

[0156] In summary, by converting the coordinates of charged bodies in the three-dimensional scene model into cylindrical coordinates and combining the equivalent charge method to accurately calculate the electric potential value, electric field strength value, and electric field gradient of the charged bodies, a dynamic safety distance calculation formula is established. This method can dynamically adjust the safety distance between the operator and the charged bodies according to the real-time electric field distribution in the working environment, thereby effectively improving the accuracy and real-time performance of safety monitoring in complex working scenarios. Through the weight parameters determined by experiments and the comprehensive evaluation based on electric field strength and voltage, the scientificity and applicability of the safety distance model are further enhanced, ensuring that the operator is always within the safe range of electrical operations.

[0157] S4: Issue a warning based on the safety distance and perform dynamic path optimization for the operator.

[0158] Furthermore, specifically, issuing a warning based on the safety distance and performing dynamic path optimization for the operator includes:

[0159] Select an AR headset device (such as Microsoft HoloLens 2) and wear it on the operator's head. Start the AR headset device and connect it to the central monitoring system through a wireless network. Initialize the device for the AR headset device, and the AR headset device displays the safety distance of the charged bodies around the operator.

[0160] Real-time calculate the Euclidean distance d between the operator and the charged bodies b and compare it with the safety distance d a ;

[0161] If d b ≥d a , it means that the current operator status is safe and no warning is required;

[0162] If d b <d a , it means that the current operator status is unsafe and the warning mechanism is triggered;

[0163] The warning mechanism refers to using the vibration and voice of the AR headset device to prompt the operator to pay attention to safety and perform dynamic path optimization for the current operator;

[0164] The dynamic path optimization refers to setting the starting point of the path as the current three-dimensional coordinates of the operator, dividing the working area into three-dimensional grid cells and defining a cost function, and using the D* algorithm for initial path planning;

[0165] The three-dimensional grid cells include the distance from the charged bodies and the distance from the obstacles;

[0166] The cost function is defined as the smaller the distance between the operator and the charged bodies, the higher the cost, and the smaller the distance between the operator and the obstacles, the higher the cost;

[0167] The initial path planning using the D* algorithm refers to searching for the optimal path from the starting point to the target point. The priority of each path node is determined according to the current path cost and the heuristic estimated cost. Nodes are gradually expanded from the starting point, the path with the minimum cost is retrieved, and a sequence of path points is generated.

[0168] The current path cost refers to the cumulative cost from the starting point to the current node.

[0169] The heuristic estimated cost refers to the estimated cost from the current node to the target point, and the estimated cost is determined by the direct distance.

[0170] Newly added obstacles and changes in obstacle positions are detected in real time, and the affected 3D grid cells are updated to an impassable state. If there are 3D grid cells in the current path that are in an impassable state, the path in the affected area is replanned locally.

[0171] The optimal path is displayed in the AR headset device of the operator to provide real-time navigation prompts.

[0172] By combining the AR headset device, real-time safety distance calculation, and dynamic path optimization technology, not only is the safety of the operator in the live wire environment improved, but also the optimal path is dynamically planned through the D* algorithm to avoid obstacle interference and ensure real-time path updates, enhancing work efficiency and safety guarantee. The visual navigation and warning mechanism of the AR device effectively reduces the operation risk and provides an intuitive and efficient safety guide.

[0173] S5: Store all data in a relational database and manage it.

[0174] It should also be noted that a relational database is selected to store and manage data and analysis results. The database table structure is designed to store different types of data, a regular backup task is set to back up all data in the database, user permissions for the database are managed, and the data is encrypted for storage.

[0175] Embodiment 2, the second embodiment of the present invention, which is different from the previous two embodiments in that:

[0176] If the above-mentioned functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs.

[0177] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in combination with these instruction execution systems, apparatuses, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in combination with an instruction execution system, apparatus, or device.

[0178] More specific examples (non-exhaustive list) of computer-readable media include the following: electrical connection parts with one or more wirings (electronic devices), portable computer disk cartridges (magnetic devices), random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memories), optical fiber devices, and portable compact disc read-only memories (CDROM). Additionally, a computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or, if necessary, other suitable processing, and then stored in a computer memory.

[0179] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits with logic gate circuits for implementing logic functions on data signals, application specific integrated circuits with suitable combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0180] Example 3, referring to Figure 2 , which is the third embodiment of the present invention. This embodiment provides a safety distance monitoring and intelligent early warning system for live working personnel, including a data acquisition module, a depth map generation module, a model construction module, a dynamic calculation module, a classification and early warning module, and a data storage module;

[0181] The data acquisition module is used to collect the electric field data and image data of the operation area in real time and perform preprocessing;

[0182] The depth map generation module is used to extract the features of the image data through a depth estimation network to generate a depth map;

[0183] The model construction module is used to convert the depth map into three-dimensional coordinates and combine with the RTK module to generate a three-dimensional scene model of live working;

[0184] The dynamic calculation module is used to construct a dynamic safety distance formula based on the electric field data and the three-dimensional scene model to calculate the safety distance between the operator and the energized body;

[0185] The classification and early warning module is used to give early warnings according to the safety distance and perform dynamic path optimization for the operators;

[0186] The data storage module is used to store all data in a relational database and manage it.

[0187] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not restrictive. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A safety distance monitoring and intelligent early warning method for live working personnel, characterized in that: Including, Real-time collect voltage data and image data of the operation area and perform preprocessing, and extract features of the image data through a depth estimation network to generate a depth map; Convert the depth map into three-dimensional coordinates and combine with the RTK module to generate a three-dimensional scene model for live working; Construct a dynamic safety distance formula based on the electric field data and the three-dimensional scene model to calculate the safety distance between the operator and the energized body; Give an alarm according to the safety distance and perform dynamic path optimization for the operator; Store all data in a relational database and manage it.

2. The safety distance monitoring and intelligent early warning method for live working personnel according to claim 1, characterized in that: The real-time collection of voltage data and image data of the operation area and preprocessing includes, Use electric field intensity sensors and voltage sensors and install them on the energized bodies in the operation area to collect electric field data in real time. Arrange a monocular camera at the monitoring point and set the collection rate, collect images of the operation area and transmit them to the central monitoring system; The electric field data refers to the voltage value of each energized body in the operation area; The preprocessing includes enhancing the image contrast and equalizing the brightness of the image data, removing abnormal values and missing data from the electric field data, filling in the missing data using linear interpolation method and normalizing.

3. The safety distance monitoring and intelligent early warning method for live working personnel according to claim 2, characterized in that: The generation of the depth map includes, The depth estimation network includes an encoder and a decoder. The encoder adopts the DenseNet-169 network structure. Input the image data into the depth estimation network. Through the single-layer convolution operation of each layer of convolution in the encoder, use each layer's output feature as a feature map for layer-by-layer transmission and processing. The decoder gradually restores the resolution of the feature map through convolution layers and bilinear upsampling operations and generates a predicted depth map. The value of each pixel point represents the predicted depth value; Use depth difference loss, gradient smoothing loss and structural similarity loss as loss functions. The depth difference loss is: Among them, L1 represents the loss value of the depth difference loss, represents the predicted depth value of the pixel point p in the predicted depth map, D p represents the true depth value of the pixel point p in the true depth map; Introduce an adaptive weight to dynamically adjust the weight ω according to the prediction error of each pixel point p ; The gradient smoothing loss is: Among them, L2 represents the loss value of the gradient smoothing loss, and g x and g y respectively represent the gradients of the predicted depth map in the x and y directions, and respectively represent the gradient smoothing loss of pixel point p in the x direction and the gradient smoothing loss in the y direction; The structural similarity loss is: Among them, L3 represents the loss value of the structural similarity loss, represents the structural similarity index; Construct the final loss function as: L s = ω p (L1 + L2 + L3) Among them, L s represents the loss value of the final loss function; Based on the regularization strategy, minimize the final loss function, update the parameters of the depth estimation network, and repeat the process of forward propagation, loss calculation, backpropagation and parameter update until the loss function converges; Use the existing publicly available depth estimation dataset to perform offline training on the depth estimation network, use the transfer learning strategy to accelerate the training process, and adopt the cosine annealing learning rate strategy to dynamically adjust the learning rate during training; In the operation scenario, input the image captured by the monocular camera in real time into the trained depth estimation network to generate the depth map of the current scene.

4. The safety distance monitoring and intelligent early warning method for live working personnel according to claim 3, characterized in that: The three-dimensional scene model includes, Define the internal parameter matrix and external parameter matrix of the camera coordinate system, and perform point-by-point processing on the coordinates and depth values of each pixel point in the depth map; The point-by-point processing is to use the internal parameter of the camera to calculate the three-dimensional coordinates of each pixel point in the depth map in the camera coordinate system and combine them into a point cloud data structure, and convert the points in the camera coordinate system into points in the world coordinate system through the external parameter matrix. Each point is represented as a three-dimensional space coordinate in the world coordinate system; Install high-precision RTK modules on the hands, shoulders, waist, knees, and feet of the operator's shielding suit. The RTK modules communicate bidirectionally with the reference station, and transmit the high-precision positioning information of the reference station to the operator's RTK modules through differential signals, providing the three-dimensional spatial coordinates of the operator's hands, shoulders, waist, knees, and feet in real time, and connecting to the central monitoring system through a wireless network to select a wireless communication frequency band for data transmission; All three-dimensional spatial coordinates are converted into the standard point cloud format and integrated into a point cloud data set; For the point cloud data generated at different time points, align them through the iterative closest point algorithm to generate a complete three-dimensional scene model.

5. The safety distance monitoring and intelligent warning method for live working personnel according to claim 4, characterized in that: The safety distance includes, Convert the three-dimensional spatial coordinates of each charged body in the three-dimensional scene model into cylindrical coordinates; The conversion of the three-dimensional space coordinates to cylindrical coordinates includes obtaining the three-dimensional space coordinates, length, and charge density of the center position of the charged body, taking the three-dimensional space coordinates of the center position of the charged body as the central axis, calculating the shortest radial distance r from the three-dimensional space coordinates of the operator to the central axis, and obtaining the vertical height z of the three-dimensional space coordinates corresponding to the radial distance e , and taking the radial distance and the vertical height as the cylindrical coordinates; Based on the cylindrical coordinates, use the equivalent charge method to calculate the electric field potential value generated by each charged body as: Among them, φ i (r, z e ) represents the electric field potential value generated by the i-th charged body in cylindrical coordinates (r, z e ), τ represents the linear charge density, L represents the length of the charged body, z s represents the integration variable along the central axis direction, ε represents the dielectric constant of air, and d represents the infinitesimal variable of the integration variable z s ; Calculate the electric field strength value at the location of the operator through the electric field potential value generated by each charged body as: Among them, E i represents the electric field strength value of the i-th charged body, represents the negative value of the spatial change rate of the electric field potential value; Superimpose the electric field strength values of all charged bodies to obtain the total electric field strength value of the operation area. The formula is: Among them, E o represents the total electric field strength value of the operation area, n represents the number of all charged bodies in the operation area, and i is a variable index; Taking the operation target position of the operator as the target point, calculate the partial derivatives in the x, y, and z directions of the target point respectively and synthesize the electric field gradient Construct a dynamic safety distance formula to calculate the safety distance between the operator and the charged body as: Among them, d a represents the safety distance between the operator and the energized body, V represents the average voltage of all energized bodies in the operation area, and k1, k2, and k3 represent weights.

6. The safety distance monitoring and intelligent early warning method for live working personnel according to claim 5, characterized in that: The dynamic path optimization includes, Select an AR headset device and wear it on the operator's head. Start the AR headset device and connect it to the central monitoring system through a wireless network. Initialize the device of the AR headset device, and the AR headset device displays the safety distance of the charged bodies around the operator; Calculate the Euclidean distance d between the real-time computing operator and the live body b and compare it with the safety distance d a for comparison; If d b ≥ d a , it means that the current operator's status is safe and no warning is required; If d b <d a , it means that the current operator's status is unsafe, triggering the warning mechanism; The early warning mechanism refers to using the vibration and voice of the AR headset device to prompt the operator to pay attention to safety and perform dynamic path optimization on the current operator; The dynamic path optimization refers to setting the starting point of the path as the current three-dimensional coordinates of the operator, dividing the operation area into three-dimensional grid cells and defining a cost function, and using the D* algorithm for initial path planning; The three-dimensional grid cells include the distance from the charged body and the distance from the obstacle; The cost function is defined as the smaller the distance between the operator and the charged body, the higher the cost, and the smaller the distance between the operator and the obstacle, the higher the cost; The use of the D* algorithm for initial path planning refers to searching for the optimal path from the starting point to the target point. The priority of each path node is determined according to the current path cost and the heuristic estimated cost. Expand the nodes step by step from the starting point, retrieve the path with the minimum cost and generate a sequence of path points; The current path cost is the cumulative cost from the starting point to the current node; The heuristic estimated cost is the estimated cost from the current node to the target point; Real-time detect newly added obstacles and changes in the positions of obstacles, update the affected three-dimensional grid cells to the impassable state. If there are three-dimensional grid cells in the current path in the impassable state, then re-locally plan the path of the affected area; Display the optimal path in the operator's AR headset device to provide real-time navigation prompts.

7. The safety distance monitoring and intelligent early warning method for live working personnel according to claim 6, characterized in that: The storing and managing all data in a relational database includes: selecting a relational database to store and manage data and analysis results, designing the database table structure to store different types of data, setting up a regular backup task to back up all data in the database, performing permission management on database users, and storing the data in an encrypted manner.

8. A system adopting a safety distance monitoring and intelligent early warning method for live working personnel as described in any one of claims 1 to 7, characterized in that: It includes a data acquisition module, a depth map generation module, a model construction module, a dynamic calculation module, a classification and warning module, and a data storage module; The data acquisition module is used to collect the electric field data and image data of the operation area in real time and perform preprocessing; The depth map generation module is used to extract the features of the image data through a depth estimation network to generate a depth map; The model construction module is used to convert the depth map into three-dimensional coordinates and generate a three-dimensional scene model of live working in combination with the RTK module; The dynamic calculation module is used to construct a dynamic safety distance formula based on the electric field data and the three-dimensional scene model to calculate the safety distance between the operator and the energized body; The classification and warning module is used to give warnings according to the safety distance and perform dynamic path optimization on the operator; The data storage module is used to store all data in a relational database and manage it.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of a method for monitoring the safety distance and intelligent warning of live working personnel according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of a method for monitoring the safety distance and intelligent warning of live working personnel according to any one of claims 1 to 7.

Images