Power grid digital twin technology analysis method and system based on AI image recognition

Abstract

The present application relates to the technical field of power grid data processing, in particular to a power grid digital twin technology analysis method and system based on AI image recognition, comprising the following steps: collecting visible light images and ultraviolet images of insulator strings through a multi-spectral camera carried by a drone, and simultaneously acquiring line leakage current waveform data; generating a multi-scale contamination feature matrix; calculating the equivalent resistivity of the contamination conductive path; reconstructing the surface electric field distribution of the digital twin based on the equivalent resistivity, simulating the field strength distortion gradient under different humidity conditions; when the field strength distortion gradient reaches the preset percentage of the air breakdown field strength, a graded early warning signal is generated; the present application simulates the electric field distortion under different humidity conditions of the contamination layer, generates a humidity-field strength mapping relationship, and can further determine whether the insulation performance of electrical equipment is affected.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power grid data processing, and particularly relates to a power grid digital twin technology analysis method and system based on AI image recognition. BACKGROUND

[0002] In power grid equipment, insulators are key electrical components whose performance directly affects the safety and stability of the power system. The main function of insulators is to prevent current leakage and ensure the normal transmission of electricity. However, the surface of insulators is easily affected by environmental factors, especially the accumulation of dirt. Dirt such as dust, salt, and plant residues can adhere to the surface of insulators. Over time, these dirt layers gradually thicken, increasing the conductivity of the insulator surface and affecting its electrical performance. In humid environments, the conductivity of the dirt layer can be significantly exacerbated by increased moisture absorption. This change can lead to dangerous electrical breakdown and flashover.

[0003] Traditional insulator contamination state detection relies on regular patrols and manual visual inspection, which is limited by the frequency and accuracy of manual inspection, making it difficult to accurately grasp the real-time state of insulators. In addition, although image analysis techniques can be used to detect the appearance and thickness of the dirt layer in some cases, they often fail to fully consider the changes in the electrical performance of the dirt layer. Traditional image analysis techniques mainly focus on visible surface features, ignoring the dynamic impact of the dirt layer on the electrical properties of the insulator. SUMMARY

[0004] The present application provides a power grid digital twin technology analysis method and system based on AI image recognition.

[0005] The power grid digital twin technology analysis method based on AI image recognition includes the following steps:

[0006] S1: Collect visible light images and ultraviolet images of the insulator string by mounting a multispectral camera on a drone, and simultaneously obtain line leakage current waveform data;

[0007] S2: Construct a dual-channel deep residual network to extract the contamination texture features of the visible light images and the discharge spot distribution features of the ultraviolet images, respectively, and generate a multi-scale contamination feature matrix;

[0008] S3: Establish a dynamic contamination-current coupling model to align the multi-scale contamination feature matrix with the leakage current harmonic components in the time domain, and calculate the equivalent resistivity of the contamination conduction path;

[0009] S4: Reconstruct the surface electric field distribution of the digital twin of the insulator string based on the equivalent resistivity, and simulate the field intensity distortion gradient under different humidity conditions;

[0010] S5: When the field distortion gradient reaches a preset percentage of the air breakdown field strength, a graded early warning signal is generated.

[0011] Optionally, the S1 specifically comprises:

[0012] S11: A multispectral imaging device is integrated on the unmanned aerial vehicle, including a visible light camera (400-700nm) and a solar blind ultraviolet camera (240-280nm), which respectively collect visible light images and ultraviolet images of the insulator string. The visible light image resolution is set to 20 million pixels, and the ultraviolet image resolution is set to 1 million pixels;

[0013] S12: The unmanned aerial vehicle inspection path is planned to ensure that each piece of insulator is photographed at least from three different angles. The frame rate of the visible light and ultraviolet images is synchronized to 10fps, and the spatial registration error is less than 0.5 pixels;

[0014] S13: A wireless leakage current sensor is installed at the steel cap of the insulator string, and the sampling frequency is set to 10kHz. The leakage current waveform data is transmitted in real time to the unmanned aerial vehicle through the LoRa communication module.

[0015] Optionally, the S2 specifically comprises:

[0016] S21: A visible light channel network is designed, and a modified ResNet-34 architecture is adopted. A spatial attention module is embedded after the third residual block to output a pollution texture feature map, which includes a pollution thickness distribution (quantized value of 0-100μm) and a material classification probability (dust / salt crystallization / bird droppings);

[0017] S22: An ultraviolet channel network is constructed based on a deformable convolution to construct a U-Net structure. The distribution density, area ratio and morphological complexity of the discharge light spot are extracted through a light spot detection layer to generate a discharge activity index matrix;

[0018] S23: In the feature fusion stage, the fourth residual block feature map (size HxWx512) of the visible light channel and the decoder output (HxWx256) of the ultraviolet channel are cross-modal aligned. A bidirectional gate mechanism is used to realize feature interaction to generate a multiscale pollution feature matrix;

[0019] S24: It also includes extracting multiscale features through a pyramid pooling layer to output feature descriptors including four scales (1x1, 2x2, 4x4, global mean), which constitute a multiscale pollution feature matrix with a dimension of (HxWx1024).

[0020] Optionally, the S3 specifically comprises:

[0021] S31: Obtain leakage current waveform data, process the leakage current waveform data using fast Fourier transform, extract each harmonic component in the signal, and extract the 3rd, 5th and 7th harmonic components respectively;

[0022] S32: Use the time synchronization module to time-align the multi-scale contamination feature matrix and the harmonic component, and limit the search range to within ±10 milliseconds, which can ensure accurate alignment within a short time window and thus reduce errors caused by time synchronization;

[0023] S33: Construct a nonlinear relationship between the contamination feature and the harmonic current component, establish a nonlinear dynamic contamination-current coupling model by using the multi-scale contamination features extracted by the pyramid pooling network, and combine the weighted sum of the multi-scale contamination features with the corresponding harmonic component through an exponential function to model the influence of contamination on current. The weight parameter is dynamically learned according to the training process to adjust the influence degree of each scale feature. A harmonic attenuation coefficient is introduced to control the nonlinear influence of the contamination feature on the harmonic, and a noise term is added to consider errors and uncertainties in actual measurement;

[0024] S34: Evaluate the degree of influence of contamination on current by calculating the equivalent resistivity, which is obtained by combining the square sum of the amplitudes of multiple harmonic components with the effective value of the phase voltage. The value of the equivalent resistivity is used to judge the severity of the contamination on the surface of the insulator.

[0025] Optionally, the dynamic contamination-current coupling model is represented as:

[0026] wherein, n represents the harmonic order, represents the weighted sum of the pyramid-pooled feature vectors of the first scale, is a learnable weight parameter;

[0027] is a harmonic attenuation coefficient, represents the amplitude of the nth harmonic current at time , is the noise term of the nth harmonic, which considers possible errors or uncontrollable factors in the experiment.

[0028] Optionally, the equivalent resistivity of the contamination conduction path is :

[0029] wherein, is a geometric factor of the contamination layer, ​​is a phase voltage effective value, is a harmonic component amplitude.

[0030] Optionally, the S4 specifically comprises:

[0031] S41, constructing a three-dimensional finite element model and mapping conductivity: constructing a three-dimensional finite element model of the insulator according to the geometric parameters of the insulator, for describing the spatial form of the insulator and providing a basis for electric field strength calculation, and converting the equivalent resistivity value into the conductivity of the pollution layer in the three-dimensional finite element model, the conductivity being the inverse of the resistivity and reflecting the conductivity performance of the pollution layer material;

[0032] S42, introducing a humidity correction factor and dynamically adjusting the conductivity: considering the influence of humidity on the conductivity of the pollution layer, introducing a humidity correction factor, and dynamically adjusting the conductivity of the pollution layer according to the relative humidity of the environment;

[0033] S43, solving the electric field distribution by Poisson equation: using Poisson equation to solve the electric field distribution on the surface of the insulator, obtaining the potential distribution by solving the equation, and then calculating the electric field strength, and setting the boundary conditions in the calculation, including applying the phase voltage at the high-voltage end and setting the ground potential to zero, and through this equation, calculating the surface electric field distribution according to the change of the conductivity and reflecting the change of the electric field distortion;

[0034] S44, calculating the field strength distortion gradient: based on the electric field distribution, calculating the field strength distortion gradient , the field strength distortion gradient indicating the degree of change of the electric field strength, reflecting the local non-uniformity of the electric field, and the calculation of the field strength distortion gradient including taking the maximum value by taking the partial derivative of the electric field component (change in X, Y, Z direction) to obtain the gradient of the field strength distortion;

[0035] S45, simulating the field strength distortion gradient under different humidity conditions: simulating the field strength distortion gradient under different humidity conditions according to the humidity gradient, generating a mapping relationship surface between humidity and field strength by respectively simulating the field strength distortion gradient under different humidity conditions, and showing the influence of humidity change on the field strength distortion.

[0036] Optionally, the field strength distortion gradient is represented as: , indicating the electric field strength, i.e. the field strength, are respectively three components of the field strength, respectively indicating the components of the electric field in the x-axis, y-axis and z-axis directions.

[0037] Optionally, the preset percentage in the S5 is 45%, and the air breakdown field strength reference value is set, and the critical threshold is calculated.

[0038] S52: mapping the field strength distortion gradient to the surface of the digital twin of the insulator string mapping to the surface of the digital twin of the insulator string according to the spatial coordinates, generating a normalized risk coefficient matrix : ;

[0039] S53: dividing the early warning level according to the risk coefficient matrix.

[0040] The power grid digital twin technology analysis system based on AI image recognition is used to realize the power grid digital twin technology analysis method based on AI image recognition, and includes the following modules:

[0041] Image acquisition and data fusion: through the multi-spectral camera carried by the unmanned aerial vehicle, multi-dimensional image data including visible light image and ultraviolet image are collected, and line leakage current waveform data is synchronously acquired;

[0042] Contamination feature extraction and analysis: using a double-channel deep residual network, the contamination texture features and discharge light spot distribution features in the multi-dimensional image data are extracted respectively, a multi-scale contamination feature matrix is generated, and the distribution and state of the contamination layer are reflected;

[0043] Dynamic contamination current coupling model: through the establishment of a dynamic contamination-current coupling model, the multi-scale contamination features and leakage current harmonic components are time-domain aligned, the equivalent resistivity of the contamination conduction path is calculated, and the electrical characteristics of the contamination layer are quantified;

[0044] Digital twin reconstruction and field strength simulation: based on the equivalent resistivity, the digital twin of the insulator string is reconstructed, and the field strength distortion gradient under different humidity conditions is simulated;

[0045] Hierarchical early warning mechanism: when the field strength distortion gradient reaches the preset percentage of the set air breakdown field strength, a hierarchical early warning signal is automatically generated.

[0046] The beneficial effects of the present application are:

[0047] The present application, by comprehensively utilizing the harmonic analysis of leakage current and multi-scale feature extraction technology, combined with the environmental humidity correction factor, can dynamically evaluate the conductivity change of the contamination layer, break through the limitation of traditional single measurement technology, not only can real-time monitor the electrical characteristics of the contamination layer, but also can adjust the evaluation result according to the actual environmental conditions (such as humidity change), improve the accuracy and real-time of the contamination layer state evaluation, so as to provide more reliable data support for the operation safety of the power system.

[0048] The present application can realize field strength distortion gradient simulation by combining the calculated pollution layer conductivity and the influence of environmental humidity, through the simulation of electric field distortion under different humidity conditions of the pollution layer, generating a humidity-field strength mapping relationship, and then determining whether the insulation performance of the electrical equipment is affected, triggering a graded early warning when the field strength distortion gradient reaches the set risk threshold, effectively predicting potential flashover risks in advance, and improving the safety protection capability of power equipment. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only illustrate the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0050] Fig. 1 The present application is an analysis method flowchart.

[0051] Fig. 2 The present application is a field strength distortion gradient calculation and simulation diagram. DETAILED DESCRIPTION

[0052] The present application will be described in detail below in combination with the drawings and specific embodiments. It should be noted that in order to make the embodiments more detailed, the following embodiments are the best, preferred embodiments, and other alternative ways can also be used by those skilled in the art to implement them; and the drawings are only used to more specifically describe the embodiments, and are not intended to specifically limit the present application.

[0053] It should be noted that in the specification, "one embodiment", "embodiment", "exemplary embodiment", "some embodiments" and the like indicate that the described embodiments can include a specific feature, structure or property, but not necessarily every embodiment includes the specific feature, structure or property. In addition, when a specific feature, structure or property is described in combination with an embodiment, it should be within the knowledge of those skilled in the art to realize this feature, structure or property in combination with other embodiments (whether or not explicitly described).

[0054] Generally, the terms can be understood at least in part from the usage in context. For example, depending at least in part on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular or can be used to describe combinations of features, structures, or characteristics. In addition, the term "based on" can be understood as not necessarily intended to convey a set of exclusive factors, but can instead, depending at least in part on the context, allow for the existence of other factors not necessarily explicitly described.

[0055] As Figs. 1-2 shown, the power grid digital twin technology analysis method based on AI image recognition includes the following steps:

[0056] S1: Collect visible light images and ultraviolet images of insulator strings through unmanned aerial vehicles equipped with multispectral cameras, and simultaneously obtain line leakage current waveform data;

[0057] S2: Construct a dual-channel deep residual network to extract the contamination texture features of visible light images and the discharge light spot distribution features of ultraviolet images, respectively, and generate a multi-scale contamination feature matrix;

[0058] S3: Establish a dynamic contamination-current coupling model, align the multi-scale contamination feature matrix with the leakage current harmonic components in the time domain, and calculate the equivalent resistivity of the contamination conduction path;

[0059] S4: Reconstruct the surface electric field distribution of the digital twin of the insulator string based on the equivalent resistivity, and simulate the field intensity distortion gradient under different humidity conditions;

[0060] S5: When the field intensity distortion gradient reaches the preset percentage of the air breakdown field intensity, a graded warning signal is generated.

[0061] S1 specifically includes:

[0062] S11: Integrate multispectral imaging equipment on the unmanned aerial vehicle, including a visible light camera (400-700 nm) and a solar blind ultraviolet camera (240-280 nm), to collect visible light images and ultraviolet images of insulator strings respectively. Set the visible light image resolution to 20 million pixels and the ultraviolet image resolution to 1 million pixels;

[0063] S12: Plan the unmanned aerial vehicle inspection path to ensure that each piece of insulator is photographed from at least three different angles. The frame rate of the visible light and ultraviolet images is synchronized to 10 fps, and the spatial registration error is less than 0.5 pixels;

[0064] S13: Install a wireless leakage current sensor at the steel cap of the insulator string, set the sampling frequency to 10 kHz, and transmit the leakage current waveform data in real time to the unmanned aerial vehicle through the LoRa communication module;

[0065] Embed a time synchronization module in the unmanned aerial vehicle flight control system to ensure that the timestamps of the multispectral images and the leakage current waveform are aligned. The collected multispectral images and leakage current waveform data are transmitted in real time to the ground station through a 5G network to construct a spatiotemporally aligned multi-modal data set.

[0066] The visible light camera uses a global shutter CMOS sensor with a resolution of 20 million pixels and a dynamic range of ≥70 dB, supporting HDR mode to handle strong light / backlight scenarios.

[0067] The solar blind ultraviolet camera selects AlGaN focal plane array, spectral response 240-280nm, quantum efficiency ≥25%, equipped with narrow band filter to suppress background light interference.

[0068] The FPGA is used to realize the hardware synchronization of the dual cameras.

[0069] The unmanned aerial vehicle inspection path: flight speed 2m / s, and keep a safe distance of 2-3m from the insulator string.

[0070] The shooting angle: each insulator piece is shot from three angles of 0° (front), 45° (oblique side), and 90° (front side), ensuring full coverage.

[0071] The wireless leakage current sensor uses Rogowski coil type current sensor, range 0-100mA, accuracy ±0.5%.

[0072] The wireless transmission uses LoRa modulation, bandwidth 125kHz, transmission distance ≥2km, packet loss rate <0.1%.

[0073] S2 specifically comprises:

[0074] S21: design a visible light channel network, use an improved ResNet-34 architecture, embed a spatial attention module after the third residual block, output a contamination texture feature map, including contamination thickness distribution (0-100μm quantization value) and material classification probability (dust / salt crystallization / bird droppings);

[0075] 1. The spatial attention module in the visible light channel design is calculated as follows:

[0076] Input: is the feature map output by the third residual block, where is the spatial size, is the number of channels, is the spatial height, is the spatial width;

[0077] average pooling and maximum pooling :

[0078] ;

[0079] Concatenate the pooling results: , denotes concatenation;

[0080] Convolution operation: where is the convolution kernel, outputting a single-channel feature map;

[0081] Sigmoid activation: where, is the Sigmoid function, output spatial attention weight matrix ;

[0082] Output weighted feature map: where represents element-wise multiplication to apply the spatial attention weight matrix to the original input feature map, resulting in a weighted feature map.

[0083] The output layer uses a multi-task head design:

[0084] Regression branch: predicts the thickness of contamination (L1 Loss);

[0085] Classification branch: Softmax output material category (cross-entropy loss);

[0086] Contamination thickness regression branch: where is the predicted value of the thickness of contamination, and the loss function is L1Loss: is the true thickness label;

[0087] Material classification branch: where, is the material classification probability (dust, salt crystals, birds), and the loss function is cross-entropy loss, is the true material label.

[0088] S22: Construct the ultraviolet channel network, build a U-Net structure based on deformable convolution, extract the distribution density, area proportion and morphological complexity features of the discharge light spot through the light spot detection layer, and generate a discharge activity index matrix;

[0089] The core structure design of the ultraviolet channel is:

[0090] Encoder: 4 layers of deformable convolution (Deformable Conv), each layer with a step of 2;

[0091] Decoder: Transposed convolution + jump connection, restored to the original resolution;

[0092] Light spot detection layer: uses a morphologically inspired ring-shaped convolution kernel (radius 3 pixels, thickness 1 pixel) to enhance the light spot edge detection capability;

[0093] The light spot detection layer uses a morphologically inspired ring-shaped convolution kernel (radius 3 pixels, thickness 1 pixel) to enhance the light spot edge detection capability, and the output matrix contains the light spot density, area proportion and morphological parameters.

[0094] In the deformable convolution calculation, the input , is the feature map of the ultraviolet image input, and the deformable convolution is expressed as: , is the output position coordinate, is the convolution kernel weight, is the conventional convolution kernel offset, is the learnable offset.

[0095] S23: In the feature fusion stage, the fourth residual block feature map (size HxWx512) of the visible light channel is cross-modal aligned with the decoder output (HxWx256) of the ultraviolet channel, feature interaction is realized by using a bidirectional gate mechanism, and a multi-scale contamination feature matrix is generated;

[0096] The bidirectional gate mechanism realizes feature interaction as follows:

[0097] 1. Visible light gate calculation: , represents that the visible light feature and the ultraviolet feature are spliced together and input into convolution, then the Sigmoid activation function is used to obtain the gate weight ;

[0098] 2. Ultraviolet gate calculation: , similarly, the ultraviolet feature and the visible light feature are spliced, subjected to convolution and Sigmoid activation function, and the gate weight is obtained;

[0099] 3. Feature fusion: , finally, the gate weight and are respectively multiplied with the visible light feature and the ultraviolet feature , and the results are added to obtain the final fusion feature ;

[0100] S24: It also includes extracting multi-scale features through a pyramid pooling layer, outputting feature descriptors including four scales (1x1, 2x2, 4x4, global mean), and constituting a multi-scale contamination feature matrix with a dimension of (HxWx1024).

[0101] 1x1 scale: providing a global perspective for evaluating the overall state of the insulator string.

[0102] 2x2 scale: provides a regional perspective, used to identify areas of contamination accumulation and discharge hotspots.

[0103] 4x4 scale: provides a local perspective, used to analyze the contamination details of individual insulator pieces.

[0104] Global mean scale: provides a statistical perspective, used to quantify the statistical properties of overall contamination and discharge.

[0105] S3 specifically includes:

[0106] S31: Obtain leakage current waveform data, which contains the variation information of current in time domain, process the leakage current waveform data using fast Fourier transform, extract the 3rd, 5th and 7th harmonic components in the signal, these harmonic components represent different frequency components of current waveform, which can reflect the distortion degree of current, and further reveal the influence of contamination characteristics on current;

[0107] S32: In order to effectively match the multi-scale contamination feature matrix and harmonic component data in time, we need to perform time domain alignment, specifically, use the time synchronization module to align the multi-scale contamination feature matrix and the harmonic component in time, the search range is limited within ±10 milliseconds, which can ensure accurate alignment within a short time window, thereby reducing the error caused by time synchronization;

[0108] S33: Construct the nonlinear relationship between contamination characteristics and harmonic current components, establish a nonlinear dynamic contamination-current coupling model by using the multi-scale contamination features extracted by the pyramid pooling network, after weighting and summing the multi-scale contamination features, combine the exponential function with the corresponding harmonic components to model the influence of contamination on current, the weight parameter is dynamically learned according to the training process, adjusting the influence degree of each scale feature, introduce harmonic attenuation coefficient to control the nonlinear influence of contamination features on harmonic, and add noise term to consider the error and uncertainty in actual measurement;

[0109] S34: Evaluate the degree of contamination influence on current by calculating the equivalent resistivity, the equivalent resistivity is obtained by squaring and summing the amplitudes of multiple harmonic components and combining with the effective value of phase voltage, the value of equivalent resistivity is used to judge the severity of insulator surface contamination, if its value is lower than the normal level, it may indicate that the conductivity of the contamination layer increases, there is potential electrical risk.

[0110] Perform fast Fourier transform on the obtained leakage current waveform data to extract the 3rd, 5th, and 7th harmonic component amplitudes and total harmonic distortion rate THD;

[0111] The dynamic contamination-current coupling model is expressed as:

[0112] wherein, represents the harmonic order, represents the weighted sum of the pyramid-pooling first scale feature vector, is a learnable weight parameter;

[0113] is a harmonic attenuation coefficient, represents the amplitude of the first harmonic current over time , is a noise term for the first harmonic considering possible errors or uncontrollable factors in the experiment;

[0114] = 3, i.e., the third harmonic, = 0.8, = 1.1;

[0115] = 5, i.e., the fifth harmonic, = 0.6, = 0.9;

[0116] = 7, i.e., the seventh harmonic, = 0.4, = 0.7.

[0117] Equivalent resistivity of the contamination conductive path :

[0118] wherein, is a contamination layer geometry factor, reflecting the factor of the influence of the geometric shape of the contamination layer on the equivalent resistivity, related to the insulator shed structure, linking the macroscopic geometric characteristics of the contamination layer with the microscopic resistivity, the greater the value, the more significant the influence of the contamination layer on the electrical performance of the insulator, is the effective value of the phase voltage, is the amplitude of the harmonic component (the amplitudes of the third, fifth, and seventh harmonics, respectively), the effective value of which is calculated as follows: wherein, is the function of the change of the voltage over time, is the voltage cycle time, represents a small increment of time.

[0119] The value corresponds to the insulator type table:

[0120]

[0121] A higher THD value indicates more harmonic components in the leakage current, which means a more severe degree of contamination on the insulator surface or more active partial discharge activity. During the model training phase, sample data with higher contamination levels can be selected based on the THD value to ensure the model's generalization ability under extreme conditions.

[0122] S4 specifically includes:

[0123] S41. Constructing a 3D finite element model and mapping conductivity: A 3D finite element model of the insulator is constructed based on its geometric parameters to describe its spatial morphology and provide a basis for electric field strength calculation. In the 3D finite element model, the equivalent resistivity value is converted into the conductivity of the contamination layer. Conductivity is the reciprocal of resistivity and reflects the electrical conductivity of the contamination layer material. This conversion process lays the physical foundation for subsequent simulation of electric field distribution. The 3D finite element model is a discretized mathematical representation of the insulator for numerical calculations (such as electric field distribution, stress analysis, etc.).

[0124] S42, Introducing a humidity correction factor and dynamically adjusting conductivity: Considering the influence of humidity on the conductivity of the contaminant layer, a humidity correction factor is introduced to dynamically adjust the conductivity of the contaminant layer according to the relative humidity of the environment. This correction factor can nonlinearly correct the conductivity according to the humidity change, thereby simulating the conductivity change of the contaminant layer under different humidity conditions. This step reflects the significant influence of humidity on the electric field distribution on the surface of the insulator, making the model closer to the actual situation.

[0125] S43, Solving for electric field distribution using Poisson equation: The electric field distribution on the surface of the insulator is solved using the Poisson equation. By solving this equation, the potential distribution is obtained, and then the electric field strength is calculated. Boundary conditions are set during the calculation, including applying a phase voltage at the high-voltage end and setting the potential at the grounding end to zero. Through this equation, the surface electric field distribution is calculated based on the change in conductivity, and the change in electric field distortion is reflected.

[0126] S44, Calculate the electric field distortion gradient: Based on the electric field distribution, calculate the electric field distortion gradient. Field strength distortion gradient This indicates the degree of drastic change in electric field intensity, reflects the local non-uniformity of the electric field, and helps to understand the distribution of electric field intensity at different locations on the insulator surface, thus providing a basis for the analysis of the insulation performance of the contamination layer. The specific calculation method involves calculating the field strength distortion gradient by taking the partial derivatives of the electric field components (changes in the X, Y, and Z directions), finding the maximum value, and obtaining the gradient of the field strength distortion.

[0127] S45, simulate field strength distortion gradient under different humidity: simulate field strength distortion gradient under different humidity according to humidity gradient, generate a mapping relationship surface between humidity and field strength, show the influence of humidity change on field strength distortion.

[0128] According to the insulator geometric parameters (shed angle, spacing, diameter), a three-dimensional finite element model is constructed, and the equivalent resistivity is mapped to the conductivity of the pollution layer .

[0129] The three-dimensional finite element model is constructed for geometric modeling, as follows:

[0130] Use CAD software or finite element software to create a three-dimensional geometric model of the insulator;

[0131] Input the key geometric parameters of the insulator:

[0132] Shed angle (e.g. 20°);

[0133] Shed spacing (e.g. 50mm);

[0134] Shed diameter (e.g. 300mm);

[0135] Meshing: divide the geometric model into finite element meshes, which can use tetrahedral elements;

[0136] Set the grid density: increase the grid in the pollution layer and the electric field concentration area (shed edge), and appropriately sparse in other areas.

[0137] Material property definition:

[0138] Define the dielectric constant of the insulator material;

[0139] Define the conductivity of the pollution layer.

[0140] The humidity correction factor is represented as , adjust the conductivity , obtain the adjusted conductivity :

[0141] , wherein is the relative humidity of the environment, is the reference humidity, , are the material moisture absorption characteristic coefficients, , represents the degree of influence of humidity on conductivity change, , determines the degree of nonlinearity of humidity influence;

[0142] The surface electric field distribution is solved by Poisson equation, represented as: wherein, represents the surface electric field gradient, the boundary condition is that the phase voltage is applied at the high-voltage end , and the potential at the ground end is 0;

[0143] field strength distortion gradient is represented as: , represents the electric field strength, i.e., the field strength, are three components of the field strength, respectively representing the components of the electric field in the x-axis, y-axis, and z-axis directions.

[0144] The field strength distortion gradient is simulated according to the humidity gradients 50%, 70%, and 90% respectively, to obtain a humidity-field strength mapping relationship surface,

[0145] The 50%, 70%, and 90% are respectively brought into the humidity correction factor , i.e. , , , and then the pollution layer conductivity is adjusted according to the humidity correction factor , and based on the adjusted pollution layer conductivity , the electric field strength in three directions is calculated by solving the Poisson equation through software, and the field strength distortion gradient values under different humidities are plotted as a three-dimensional surface graph;

[0146] The three typical humidity points 50%, 70%, and 90% are selected, covering dry, humid, and saturated states, and the humidity-field strength mapping relationship surface is a three-dimensional surface graph, with the horizontal axis being humidity, the vertical axis being position coordinate, and the vertical axis being field strength distortion gradient value.

[0147] The preset percentage in S5 is 45%, and in the power system, the design and testing of power equipment such as insulators usually take 30 kV / cm as the reference value of air breakdown field strength, which helps to ensure the safety of the equipment in operation, and the air breakdown field strength reference value is set, and the critical threshold is calculated;

[0148] S52: Map the field strength distortion gradient to the digital twin surface of the insulator string according to the spatial coordinates (map ∇E from the finite element grid to the digital twin surface vertex, use the nearest neighbor interpolation method, display on the digital twin, and render the heat map on the digital twin surface), to generate a normalized risk coefficient matrix : ;

[0149] S53: According to the risk coefficient matrix, the warning level is divided into three levels:

[0150] Yellow warning Mark the risk area coordinates and increase the monitoring frequency to 1 per minute;

[0151] Orange warning Generate a cleaning path planning graph and push it to the operation and maintenance terminal;

[0152] Red warning : Trigger the emergency power-off command and start the insulator self-explosion protection mechanism;

[0153] The heat map is rendered in the HSV color space, and the hue gradually changes from green to red , and the saturation is positively correlated with the risk coefficient. The hierarchical warning signal and the heat map are transmitted in real time to the dispatching center through the LoRaWAN network, and the response delay is less than 200ms.

[0154] The power grid digital twin technology analysis system based on AI image recognition is used to realize the above-mentioned analysis method, which includes the following modules:

[0155] Image acquisition and data fusion: a multi-spectral camera is carried by a drone to collect multi-dimensional image data including visible light images and ultraviolet images, and line leakage current waveform data is synchronously acquired;

[0156] Contamination feature extraction and analysis: a dual-channel deep residual network is used to extract contamination texture features and discharge light spot distribution features from multi-dimensional image data respectively, generate a multi-scale contamination feature matrix, and reflect the distribution and state of the contamination layer;

[0157] Dynamic contamination current coupling model: through the establishment of a dynamic contamination-current coupling model, the multi-scale contamination features and leakage current harmonic components are time-domain aligned, the equivalent resistivity of the contamination conduction path is calculated, and the electrical characteristics of the contamination layer are quantified;

[0158] Digital twin reconstruction and field strength simulation: based on the equivalent resistivity, the digital twin of the insulator string is reconstructed, and the field strength distortion gradient under different humidity conditions is simulated;

[0159] Hierarchical warning mechanism: when the field strength distortion gradient reaches a preset percentage of the set air breakdown field strength, a hierarchical warning signal is automatically generated.

[0160] The present application encompasses any substitutions, modifications, equivalent methods and schemes made on the essence and scope of the present application. In order for the public to have a thorough understanding of the present application, specific details are described in the following preferred embodiments of the present application, and the present application can also be fully understood without the description of these details by those skilled in the art. In addition, in order to avoid unnecessary confusion to the essence of the present application, well-known methods, processes, procedures, elements and circuits, etc. are not described in detail.

[0161] The above merely describes the preferred embodiments of the present application, and it should be pointed out that those skilled in the art can make several improvements and refinements without departing from the principles of the present application, and these improvements and refinements should also be considered as falling within the protection scope of the present application.

Claims

1. A power grid digital twin technology analysis method based on AI image recognition, characterized in that, Includes the following steps: S1: Acquire visible and ultraviolet images of insulator strings using a drone equipped with a multispectral camera, and simultaneously obtain line leakage current waveform data; S2: Construct a dual-channel deep residual network to extract the dirt texture features of the visible light image and the discharge spot distribution features of the ultraviolet image, respectively, and generate a multi-scale dirt feature matrix; S3: Establish a dynamic pollution-current coupling model, align the multi-scale pollution feature matrix with the leakage current harmonic components in the time domain, and calculate the equivalent resistivity of the pollution conductive path. S4: Reconstructing the surface electric field distribution of a digital twin based on equivalent resistivity to simulate the field strength distortion gradient under different humidity conditions; specifically including: S41, Construct a three-dimensional finite element model and map conductivity: Construct a three-dimensional finite element model of the insulator based on the geometric parameters of the insulator to describe the spatial morphology of the insulator and provide a basis for electric field strength calculation. In the three-dimensional finite element model, the equivalent resistivity value is converted into the conductivity of the contamination layer. The conductivity is the reciprocal of the resistivity and reflects the conductivity of the contamination layer material. S42, Introduce a humidity correction factor and dynamically adjust conductivity: Considering the influence of humidity on the conductivity of the dirt layer, a humidity correction factor is introduced to dynamically adjust the conductivity of the dirt layer according to the relative humidity of the environment. S43, Solving the electric field distribution using the Poisson equation: The electric field distribution on the surface of the insulator is solved using the Poisson equation. By solving the equation, the potential distribution is obtained, and then the electric field strength is calculated. Boundary conditions are set during the calculation, including applying a phase voltage at the high-voltage end and setting the potential at the grounding end to zero. The surface electric field distribution is calculated based on the change in conductivity. S44, Calculate the electric field distortion gradient: Based on the electric field distribution, calculate the electric field distortion gradient. Field strength distortion gradient It represents the degree of drastic change in electric field intensity and reflects the local non-uniformity of the electric field. Calculating the field intensity distortion gradient involves taking the partial derivatives of the electric field components, taking the maximum value, and obtaining the gradient of the field intensity distortion. S45, Simulate field strength distortion gradient under different humidity conditions: Simulate field strength distortion gradient under different humidity conditions based on humidity gradient. By simulating the field strength distortion gradient under different humidity conditions, a mapping relationship surface between humidity and field strength is generated to show the influence of humidity change on field strength distortion. S5: When the field strength distortion gradient reaches a preset percentage of the air breakdown field strength, a graded early warning signal is generated.

2. The power grid digital twin technology analysis method based on AI image recognition according to claim 1, characterized in that, S1 specifically includes: S11: Integrates multispectral imaging equipment on the drone, including a visible light camera and a solar-blind ultraviolet camera, to acquire visible light and ultraviolet images of the insulator string, respectively; S12: Plan the drone inspection route to ensure that each insulator is photographed from at least 3 different angles; S13: Install a wireless leakage current sensor at the steel cap of the insulator string, set the sampling frequency to 10kHz, and transmit the leakage current waveform data to the drone in real time through the LoRa communication module.

3. The power grid digital twin technology analysis method based on AI image recognition according to claim 1, characterized in that, S2 specifically includes: S21: Design a visible light channel network, adopt an improved ResNet-34 architecture, embed a spatial attention module after the third residual block, and output a dirt texture feature map, including dirt thickness distribution and material classification probability. S22: Construct an ultraviolet channel network, build a U-Net structure based on deformable convolution, and extract the distribution density, area ratio and morphological complexity features of the discharge spot through the spot detection layer to generate a discharge activity index matrix; S23: In the feature fusion stage, the fourth residual block feature map of the visible light channel is aligned across modes with the decoder output of the ultraviolet channel. A bidirectional gating mechanism is used to realize feature interaction and generate a multi-scale contamination feature matrix. S24: It also includes extracting multi-scale features through pyramid pooling layers, outputting feature descriptors that include four scales.

4. The power grid digital twin technology analysis method based on AI image recognition according to claim 3, characterized in that, S3 specifically includes: S31: Obtain leakage current waveform data, process the leakage current waveform data using fast Fourier transform, extract each harmonic component in the signal, and extract the 3rd, 5th and 7th harmonic components respectively. S32: Using the time synchronization module, the multi-scale pollution feature matrix and harmonic components are time-aligned, and the search range is limited to ±10 milliseconds; S33: Construct a nonlinear relationship between pollution features and harmonic current components. By using multi-scale pollution features extracted by a pyramid pooling network, a nonlinear dynamic pollution-current coupling model is established. The dynamic pollution-current coupling model combines the multi-scale pollution features with the corresponding harmonic components through an exponential function to model the impact of pollution on current. The weight parameters are dynamically learned according to the training process to adjust the degree of influence of each scale feature. A harmonic attenuation coefficient is introduced to control the nonlinear impact of pollution features on harmonics. S34: The impact of pollution on current is assessed by calculating the equivalent resistivity. The equivalent resistivity is obtained by summing the squares of the amplitudes of multiple harmonic components and combining them with the effective value of the phase voltage. The value of the equivalent resistivity is used to determine the severity of pollution on the insulator surface.

5. The power grid digital twin technology analysis method based on AI image recognition according to claim 4, characterized in that, The dynamic pollution-current coupling model is expressed as follows: ,in, , indicating the harmonic order, Indicates the pyramid pooling method The weighted sum of eigenvectors at each scale. These are learnable weight parameters; Harmonic attenuation coefficient, Indicates time The first Second harmonic current amplitude For the first The noise term of the subharmonics.

6. The power grid digital twin technology analysis method based on AI image recognition according to claim 5, characterized in that, The equivalent resistivity : ,in, It is the geometric factor of the filth layer. It is the effective value of the phase voltage. It is the amplitude of the harmonic components.

7. The power grid digital twin technology analysis method based on AI image recognition according to claim 1, characterized in that, The field strength distortion gradient Represented as: , This represents the electric field strength, or field power. These are the three components of the electric field strength, representing the components of the electric field along the x-axis, y-axis, and z-axis, respectively.

8. The power grid digital twin technology analysis method based on AI image recognition according to claim 7, characterized in that, The preset percentage in S5 is 45%, which sets the air breakdown field strength reference value. Calculate the critical threshold ; S52: Transform the field strength distortion gradient By mapping spatial coordinates onto the surface of the digital twin, a normalized risk coefficient matrix is ​​generated. : ; S53: Classify early warning levels based on the risk coefficient matrix.

9. A power grid digital twin technology analysis system based on AI image recognition, used to implement the power grid digital twin technology analysis method based on AI image recognition as described in any one of claims 1-8, characterized in that, Includes the following modules: Image acquisition and data fusion: Using a drone equipped with a multispectral camera, multi-dimensional image data including visible light and ultraviolet images are acquired, and the leakage current waveform data of the line is acquired simultaneously; Fouling Feature Extraction and Analysis: A dual-channel deep residual network is used to extract fouling texture features and discharge spot distribution features from multi-dimensional image data, respectively, to generate a multi-scale fouling feature matrix that reflects the distribution and state of the fouling layer. Dynamic pollution-current coupling model: By establishing a dynamic pollution-current coupling model, the multi-scale pollution characteristics and leakage current harmonic components are aligned in the time domain to calculate the equivalent resistivity of the pollution conductive path and quantify the electrical characteristics of the pollution layer. Digital twin reconstruction and field strength simulation: Based on equivalent resistivity, the digital twin of the insulator string is reconstructed, and the field strength distortion gradient under different humidity conditions is simulated; Graded early warning mechanism: When the field strength distortion gradient reaches the preset percentage of the air breakdown field strength, a graded early warning signal is automatically generated.

Images