Power distribution overhead line icing thickness detection method, system and device and storage medium

Abstract

The invention discloses a power distribution overhead line icing thickness detection method, system and device and a storage medium, and relates to the field of icing thickness detection.The method comprises the steps that multi-source detection data of a power distribution overhead line is collected, self-adaptive interpolation completion is conducted based on data distribution characteristics, for example, bilinear interpolation is conducted on a two-dimensional image, linear interpolation is conducted on sequence data, and the detection result is obtained; complementing point cloud data in combination with information such as wind direction; performing multi-source icing thickness detection on the detection data set, and calculating a detection sequence based on image pixel gradient, point cloud intensity and the like; performing multi-source fusion by adopting Bayesian inference to obtain an accurate detection value and a confidence interval; predicting an icing trend in combination with historical data, and performing real-time monitoring and early warning; according to the invention, image edges and ice layer boundaries can be protected, so that interpolation better conforms to physical phenomena, and the detection precision is improved; icing thickness features are described from different aspects, posterior probability distribution can be updated by combining prior and real-time information, the icing thickness is accurately estimated, uncertainty is quantified, and powerful support is provided for decision making.

Description

Technical Field

[0001] The present invention relates to the field of ice thickness detection, and in particular to a method, system, device and storage medium for detecting ice thickness of distribution overhead lines. Background Art

[0002] Overhead distribution lines are significantly affected by the natural environment, particularly snow and ice. Ice and snow loads (icing) on ​​overhead lines can reduce their power transmission capacity and even cause line breakage, equipment damage, and outages. Therefore, accurately monitoring and assessing the thickness of ice on distribution overhead lines, especially during adverse weather conditions, is crucial to ensuring the safe and stable operation of power systems.

[0003] Existing ice detection technologies are mainly divided into the following categories: manual inspection: manually climbing poles to check or using telescopes to observe the ice situation, but this method is time-consuming, labor-intensive, inaccurate, and difficult to carry out on a large scale; power equipment monitoring: some power equipment (such as lightning arresters, grounding wires, etc.) can monitor a certain amount of ice coverage, but generally cannot accurately measure the thickness of the ice layer; remote sensing technology: using remote sensing equipment such as drones and satellites to collect data, and inferring the ice situation through image recognition technology, but due to factors such as weather and line of sight, the real-time and accuracy of remote sensing technology may be insufficient. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is that the existing ice detection technology (manual detection, power equipment monitoring, remote sensing technology) has the problems of being time-consuming, labor-intensive, inaccurate, difficult to carry out on a large scale, unable to accurately measure the thickness of the ice layer, and lacking real-time and accuracy.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a method for detecting ice thickness on a distribution overhead line, comprising:

[0008] Collect multi-source detection data of distribution overhead lines and perform adaptive interpolation and completion processing to obtain a detection data set;

[0009] Perform multi-source ice thickness detection on the detection data set to obtain a multi-source ice thickness detection sequence;

[0010] Multi-source fusion inference is performed on the multi-source ice thickness detection sequence to obtain the accurate detection value and confidence interval of the ice thickness of the distribution overhead line;

[0011] Combined with the historical data series of ice thickness on distribution overhead lines, the accurate detection value of ice thickness on distribution overhead lines is predicted to obtain the icing trend of distribution overhead lines.

[0012] Based on the icing trend of the distribution overhead lines, real-time monitoring and early warning of the icing status of the distribution overhead lines are carried out.

[0013] As a preferred solution for detecting ice thickness on distribution overhead lines:

[0014] The adaptive interpolation and completion processing is performed to obtain a detection data set including:

[0015] Count the number of pixels num1 in the horizontal direction and num2 in the vertical direction of the image detection data. If num1 is less than the preset row threshold and num2 is less than the preset column threshold, use bilinear interpolation to interpolate and fill in the missing pixels so that the number of pixels in the horizontal direction and the number of pixels in the vertical direction of the image detection data after interpolation and filling reach the preset row threshold and column threshold respectively.

[0016] This preferred technical solution has the beneficial effect of achieving an appropriate resolution for image detection data by counting the number of pixels in the horizontal and vertical directions and using bilinear interpolation to fill in missing pixels when the number of pixels falls below a preset threshold. This facilitates subsequent accurate analysis of the image data and improves the accuracy of image-based ice thickness detection, as the appropriate number of pixels more clearly demonstrates the characteristics and distribution of the ice layer.

[0017] As a preferred solution for detecting ice thickness on distribution overhead lines:

[0018] The adaptive interpolation and completion process to obtain the detection data set further includes:

[0019] The point cloud detection data is locally adaptively interpolated and supplemented using a local adaptive weighting method based on the environmental detection data to obtain the point cloud intensity at all three-dimensional point cloud coordinate positions;

[0020] Linear interpolation is used to interpolate and fill in the missing values ​​in the environmental detection data and electrical parameter detection data.

[0021] The beneficial effects of this preferred technical solution include: Adaptive interpolation and completion of point cloud data using a locally adaptive weighted approach based on environmental monitoring data fully accounts for the impact of environmental factors on point cloud intensity, ensuring that point cloud data has reasonable intensity values ​​at each 3D point cloud coordinate location, providing a more accurate data foundation for subsequent ice thickness measurements based on point cloud data. Furthermore, linear interpolation is used to complete missing values ​​in environmental and electrical parameter data, ensuring the integrity of both data types and preventing missing data from impacting the accuracy and reliability of ice thickness measurements.

[0022] As a preferred solution for detecting ice thickness on distribution overhead lines:

[0023] The performing multi-source ice thickness detection on the detection data set to obtain a multi-source ice thickness detection sequence includes:

[0024] The gradient value of each pixel in the image detection data set is calculated to form a gradient matrix that characterizes the distribution of the ice layer. The residual convolution module and the fully connected layer are used to perform residual convolution and mapping on the gradient matrix in sequence to obtain the image ice thickness detection value.

[0025] The beneficial effect of this preferred technical solution is that by calculating the gradient value of each pixel in the image detection data to form a gradient matrix, and then processing it using the residual convolution module and fully connected layers, it can fully explore the characteristic information of the ice layer in the image. The residual convolution module can effectively solve the gradient vanishing problem in deep neural networks, enabling the network to learn more complex ice layer characteristics, thereby accurately obtaining the ice thickness detection value of the image, and providing reliable image data detection results for multi-source fusion inference.

[0026] As a preferred solution for detecting ice thickness on distribution overhead lines:

[0027] The performing multi-source ice thickness detection on the detection data set to obtain a multi-source ice thickness detection sequence further includes:

[0028] Extract the point cloud detection data of the detection data set and calculate the point cloud ice thickness detection value;

[0029] Extract the environmental detection data from the detection data set and the environmental structural features and electrical structural features of the electrical parameter detection data, respectively construct an environmental ice thickness detection function and an electrical ice thickness detection function to perform ice thickness detection, and obtain the environmental ice thickness detection value and the electrical ice thickness detection value;

[0030] The environmental ice thickness detection function takes the environmental structural characteristics as input and the environmental ice thickness detection value as output. The electrical ice thickness detection function takes the electrical structural characteristics as input and the electrical ice thickness detection value as output.

[0031] As a preferred solution for detecting ice thickness on distribution overhead lines:

[0032] The above method combines the historical data series of ice thickness of the distribution overhead line to predict the accurate detection value of the ice thickness of the distribution overhead line, and obtains the ice trend of the distribution overhead line, including:

[0033] The historical data sequence of the ice thickness of the distribution overhead lines is a sequence composed of historical precise detection values. The precise detection values ​​of the ice thickness of the distribution overhead lines are added to the end of the historical data sequence. The time series prediction algorithm is used to predict the ice thickness of the distribution overhead lines in the future. The ice trend of the distribution overhead lines is also generated. The ice trend is the rate of change of the ice thickness in the future.

[0034] This preferred technical solution has the beneficial effect of appending accurate measurements of ice thickness on distribution overhead lines to the end of a historical data sequence and performing predictions using a time series prediction algorithm. This fully leverages historical data and current measurements, taking into account the temporal variation of ice thickness, to accurately predict the ice thickness of distribution overhead lines at future times and generate an ice trend. This ice trend reflects the rate of change in ice thickness, providing an important basis for subsequent real-time monitoring and early warning, helping to prepare countermeasures in advance.

[0035] As a preferred solution for detecting ice thickness on distribution overhead lines:

[0036] The real-time monitoring and early warning of the icing status of the distribution overhead lines based on the icing trend of the distribution overhead lines includes:

[0037] Based on the icing trend of the distribution overhead lines and combined with preset thresholds, real-time monitoring and early warning of the icing status of the distribution overhead lines are carried out; the preset thresholds include detection thresholds and trend thresholds. If the boundary value of the confidence interval exceeds the detection threshold or the icing trend exceeds the trend threshold, an early warning of the icing status of the distribution overhead lines is issued.

[0038] In a second aspect, an embodiment of the present invention provides a system for detecting ice thickness on distribution overhead lines, comprising:

[0039] The detection data set acquisition module is used to collect multi-source detection data of distribution overhead lines and perform adaptive interpolation and completion processing to obtain the detection data set;

[0040] A detection sequence acquisition module is used to perform multi-source ice thickness detection on the detection data set to obtain a multi-source ice thickness detection sequence;

[0041] The multi-source fusion inference module is used to perform multi-source fusion inference on the multi-source ice thickness detection sequence to obtain the accurate detection value and confidence interval of the ice thickness of the distribution overhead line;

[0042] The prediction module is used to combine the historical data series of ice thickness of distribution overhead lines to accurately predict the detection value of ice thickness of distribution overhead lines and obtain the ice trend of distribution overhead lines;

[0043] The monitoring and early warning module is used to monitor and warn the icing status of the distribution overhead lines in real time based on the icing trend of the distribution overhead lines.

[0044] In a third aspect, an embodiment of the present invention provides an electronic device, including:

[0045] memory and processor;

[0046] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the one or more programs are executed by the one or more processors, the one or more processors implement the method for detecting the ice thickness of distribution overhead lines as described in any embodiment of the present invention.

[0047] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the method for detecting ice thickness on distribution overhead lines.

[0048] Beneficial effects of the present invention: The present invention collects multi-source detection data in the distribution overhead line area, performs adaptive interpolation and completion based on the distribution characteristics of the data, adopts bilinear interpolation to interpolate and complete the two-dimensional image data, adopts linear interpolation to adaptively interpolate and complete the sequence data, and adopts a combination of wind direction contribution information and other information to interpolate and complete the point cloud detection data. By setting the dynamic standard deviation, the value is related to the distribution characteristics of the three-dimensional point cloud coordinates. In the sparse point cloud area, the dynamic standard deviation is larger, thereby increasing the interpolation smoothness and reducing the interpolation error. In the dense point cloud area, the dynamic standard deviation is smaller, the interpolation is more accurate, and more details are retained. Gradient information is introduced to optimize the local adaptive weight, so that the smoothing is reduced in the point cloud area with a larger gradient and increased in the point cloud area with a smaller gradient, thereby protecting the image edge and the ice layer boundary and improving the ice thickness. The detection accuracy is improved. The contribution of interpolation in the main wind direction is enhanced based on wind direction contribution information, making the interpolation more consistent with the real physical phenomenon and improving the detection accuracy. The relationship between multi-source detection data and ice formation is utilized. Based on image pixel gradient, point cloud intensity, mapping value of environmental detection data, and the difference between electrical parameter detection data and standard value, a multi-source ice thickness detection sequence is calculated to describe the ice thickness characteristics of the distribution overhead line area from different aspects. The Bayesian inference method is used for multi-source inference fusion. By combining prior information with current real-time observation information, the Bayesian inference method can update the posterior probability distribution of ice thickness after each measurement, thereby providing a more accurate estimate of ice thickness. Not only can the most likely value of ice thickness be obtained, but also uncertainty can be quantified and described, providing strong support for subsequent decision-making and actions. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0050] Figure 1 It is an overall flow chart of the method for detecting ice thickness of distribution overhead lines described in the present invention. DETAILED DESCRIPTION

[0051] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0052] Example 1, with reference to Figure 1 , which is the first embodiment of the present invention, provides a method for detecting ice thickness on a distribution overhead line, comprising:

[0053] S1: Collect multi-source detection data of distribution overhead lines and perform adaptive interpolation and completion processing to obtain a detection data set;

[0054] S2: Perform multi-source ice thickness detection on the detection data set to obtain a multi-source ice thickness detection sequence;

[0055] S3: Perform multi-source fusion inference on the multi-source ice thickness detection sequence to obtain the accurate detection value and confidence interval of the ice thickness of the distribution overhead line;

[0056] S4: Combined with the historical data series of ice thickness of distribution overhead lines, the accurate detection value of ice thickness of distribution overhead lines is predicted to obtain the ice trend of distribution overhead lines;

[0057] S5: Based on the icing trend of the distribution overhead lines, real-time monitoring and early warning of the icing status of the distribution overhead lines are carried out.

[0058] It should be noted that through steps S1-S5, the whole process management from data collection to ice thickness detection, evaluation, prediction and early warning is realized, which provides a strong guarantee for the safe operation of distribution overhead lines.

[0059] Example 2, reference Figure 1 , which is an embodiment of the present invention, provides a method for detecting ice thickness on a distribution overhead line based on the previous embodiment, comprising:

[0060] In this embodiment, collecting multi-source detection data of the distribution overhead line in the above step S1 includes:

[0061] Multi-source detection data includes image detection data, point cloud detection data, environmental detection data, and electrical parameter detection data;

[0062] The image detection data is the image of the distribution overhead line area, the point cloud detection data is the three-dimensional point cloud data of the distribution overhead line area, the environmental detection data is the environmental parameter detection sequence of the distribution overhead line area, where the environmental parameters include temperature, humidity, wind speed and air pressure, and the electrical parameter detection data is the electrical parameter detection sequence of the distribution overhead line area, where the electrical parameters include resistance, capacitance, current and voltage. The representation of multi-source detection data is K, which is expressed as:

[0063] K=(K1,K2,K3,K4)

[0064]

[0065] Among them, K1, K2, K3, and K4 are image detection data, point cloud detection data, environmental detection data, and electrical parameter detection data respectively;

[0066] The image detection data K1 is a grayscale image, and the pixel value of the pixel in the image detection data K1 is the grayscale value;

[0067] The point cloud detection data K2 consists of three-dimensional point cloud data, which includes three-dimensional point cloud coordinates and point cloud intensity;

[0068] Environmental detection data K3 is in the form of environmental parameter detection sequence, Represents count environmental detection data values ​​in the environmental detection data K3, which include temperature, humidity, wind speed, and air pressure. represents the environmental detection data value at the nth detection moment in the environmental detection data K3, n∈[1,count], count represents the total number of detection moments;

[0069] Indicates count electrical parameter detection data values ​​in electrical parameter detection sequence K4. The electrical parameter detection data values ​​include resistance, capacitance, current and voltage. Indicates the electrical parameter detection data value at the nth detection time in the electrical parameter detection data K4.

[0070] Specifically, a fixed bracket is deployed on the overhead distribution line, and a high-definition camera is installed on the fixed bracket to regularly capture images of the overhead distribution line as image detection data;

[0071] A laser radar is installed on a fixed bracket and periodically sends laser pulses to scan the distribution overhead line and its surrounding environment. The laser radar emits laser pulses to the surface of the distribution overhead line area and receives the return signal, measuring the distance to the surface, thereby obtaining the position of the reflection point in three-dimensional space, forming a three-dimensional point cloud coordinate, and using the signal strength of the return signal as the point cloud intensity corresponding to the three-dimensional point cloud coordinate;

[0072] Environmental sensors are deployed near the overhead distribution lines to regularly collect environmental detection data; electrical parameter sensors are deployed on the conductors of the overhead distribution lines to regularly collect electrical parameter detection data.

[0073] In another possible implementation, image detection data can be collected using an infrared thermal imaging camera. Specifically, a FLIR A65 series infrared thermal imager is installed on the distribution overhead line tower. Its spectral response range is 7.5-13μm and its temperature sensitivity is ≤0.05°C. When the conductor is covered with ice, due to the difference in emissivity between the ice layer and the conductor material (approximately 0.96 for ice and approximately 0.2 for aluminum conductor), the thermal imager can capture a clear temperature gradient distribution. The thermal image is processed through HSV color space conversion, and the chromaticity H component is extracted as an ice layer identification feature. Its typical value range is 200-240 (blue), corresponding to the ice-covered area.

[0074] In this embodiment, the adaptive interpolation and completion processing is performed in step S1 above, and the detection data set obtained includes:

[0075] Count the number of pixels num1 in the horizontal direction and num2 in the vertical direction of the image detection data K1. If num1 is less than the preset row threshold and num2 is less than the preset column threshold, use bilinear interpolation to interpolate and fill in the missing pixels so that the number of pixels in the horizontal direction and the number of pixels in the vertical direction of the image detection data K1 after interpolation and filling reach the preset row threshold and column threshold respectively. The formula for bilinear interpolation is:

[0076]

[0077] Among them, I(g) represents the bilinear interpolation complement value of pixel position g, and I0(g) represents the pixel position closest to pixel position g. The pixel value at position g is g = (g1, g2), where g1 represents the horizontal position of pixel position g, g2 represents the vertical position of pixel position g, and the pixel position The pixel value of is not a bilinear interpolation complement value; is the horizontal offset value of pixel position g, is the vertical offset value of pixel position g.

[0078] Furthermore, Represents the average horizontal distance between the pixel position g and the four nearest pixel positions that do not require interpolation. Represents the average vertical distance between the pixel position g and the four nearest pixel positions that do not require interpolation.

[0079] The point cloud detection data K2 is locally adaptively interpolated and completed using a local adaptive weighting method based on the environmental detection data to obtain the point cloud intensity at all three-dimensional point cloud coordinate positions, which is expressed as:

[0080]

[0081] Where V represents the mean wind speed in the environmental detection data of the detection dataset, which is in the form of a three-dimensional vector and describes the wind speed in the three-dimensional direction; ||·||1 represents the L1 norm; K(loc) represents the interpolation completion result of the point cloud intensity at the three-dimensional point cloud coordinate loc, and Ω loc Represents the adjacent three-dimensional point cloud coordinate set of the three-dimensional point cloud coordinate loc, K(i) represents the adjacent three-dimensional point cloud coordinate set Ω loc The point cloud intensity at coordinate i in the 3D point cloud is, Represents the adjacent 3D point cloud coordinate set Ω loc The local adaptive weight of the three-dimensional point cloud coordinate i in grad i It represents the gradient value of the corresponding pixel of the 3D point cloud coordinate i in the image detection data. The camera imaging model is used to establish the correspondence between the 3D point cloud coordinates and the pixel position. ρ represents the gradient influence factor. represents the dynamic standard deviation, which corresponds to the mean of the point cloud intensity differences between the 3D point cloud coordinate i and all the neighboring 3D point cloud coordinates of the 3D point cloud coordinate i. The neighboring 3D point cloud coordinates of the 3D point cloud coordinate i are: the 3D point cloud coordinates whose distance to the 3D point cloud coordinate i is less than the preset distance threshold; Represents the adjacent 3D point cloud coordinate set Ω loc The wind direction contribution information of the three-dimensional point cloud coordinate i in ε represents the wind direction control parameter.

[0082] Linear interpolation is used to interpolate and complete the missing values ​​in the environmental detection data K3 and the electrical parameter detection data K4, which can be expressed as:

[0083]

[0084] Among them, S time Indicates the missing value of the detection time, time1 and time2 indicate the adjacent detection time of the detection time, and there is no missing value in the adjacent detection time. time1 ,S time2 Indicates the data value of the proximity detection time time1, time2.

[0085] It should be noted that by setting the dynamic standard deviation, this value is related to the distribution characteristics of the three-dimensional point cloud coordinates. In the sparse point cloud area, the dynamic standard deviation is larger, thereby increasing the interpolation smoothness and reducing the interpolation error. In the dense point cloud area, the dynamic standard deviation is smaller, the interpolation is more accurate, and more details are retained. Gradient information is introduced to optimize the local adaptive weights, so that smoothing is reduced in the point cloud area with a larger gradient and smoothing is increased in the point cloud area with a smaller gradient, thereby protecting the image edge and ice layer boundary, improving the ice thickness detection accuracy, and enhancing the contribution of interpolation in the main wind direction based on the wind direction contribution information, so that the interpolation is more in line with the real physical phenomenon and improves the detection accuracy.

[0086] In another possible implementation, the system uses edge-aware infill technology for missing image data during adaptive interpolation. This technology first identifies the direction of the wire edge and then prioritizes filling pixels along that edge. In flat areas, a weighted average of the eight surrounding pixels is used, while in high-gradient areas, only a linear combination of the three nearest neighbor pixels is used. Testing has shown that the infilled image maintains over 95% edge clarity.

[0087] In another possible implementation, when environmental sensor data is missing during adaptive interpolation, the system automatically retrieves the last 24 hours of historical data, combines it with real-time forecast information from the meteorological station, and generates a fill-in value using a time series prediction algorithm. This algorithm prioritizes the diurnal cyclical variations in temperature and humidity, and maintains a fill-in error within ±5% of the actual value.

[0088] In this embodiment, in step S2, multi-source ice thickness detection is performed on the detection data set to obtain a multi-source ice thickness detection sequence, which includes:

[0089] The gradient value of each pixel in the image detection data set is calculated to form a gradient matrix that represents the distribution of the ice layer. The residual convolution module and the fully connected layer are used to perform residual convolution and mapping on the gradient matrix in sequence to obtain the image ice thickness detection value F1. The higher the image ice thickness detection value F1, the wider the ice distribution range in the distribution overhead line area.

[0090] Extract the point cloud detection data of the detection data set and calculate the point cloud ice thickness detection value F2, which is expressed as:

[0091]

[0092] weight j =exp(-dis j )

[0093] Among them, E j Represents the point cloud intensity of the jth group of 3D point cloud data in the point cloud detection data of the detection dataset, j∈[1,num], num represents the total number of 3D point cloud data in the point cloud detection data of the detection dataset, dis j It represents the distance from the 3D point cloud coordinates of the jth group of 3D point cloud data to the center of the distribution overhead line area, E ref -E j Indicates the point cloud intensity difference of the jth group of 3D point cloud data, weight j Indicates the point cloud intensity difference E ref -E j The weight of E refRepresents the average point cloud intensity of distribution overhead lines without ice coverage.

[0094] The higher the point cloud ice thickness detection value F2 is, the higher the ice density in the distribution overhead line area is.

[0095] Extract the environmental detection data from the detection data set and the environmental structural features and electrical structural features of the electrical parameter detection data, respectively construct an environmental ice thickness detection function and an electrical ice thickness detection function to perform ice thickness detection, and obtain an environmental ice thickness detection value F3 and an electrical ice thickness detection value F4;

[0096] The environmental structure feature is the mean vector of the environmental detection data of the detection dataset and the standard deviation vector The temperature, humidity, wind speed and air pressure values ​​of the environment test data set are represented in turn. The temperature standard deviation, humidity standard deviation, wind speed standard deviation, and air pressure standard deviation of the environmental detection data of the detection data set are represented in turn. The electrical structure feature is the mean of the electrical parameter detection data of the detection data set. and standard deviation The resistance mean, capacitance mean, current mean and voltage mean of the electrical parameter detection data of the detection data set are represented in turn. The resistance standard deviation, capacitance standard deviation, current standard deviation and voltage standard deviation of the electrical parameter detection data of the detection data set are represented in sequence.

[0097] The environmental ice thickness detection function takes the environmental structural characteristics as input and the environmental ice thickness detection value as output. The electrical ice thickness detection function takes the electrical structural characteristics as input and the electrical ice thickness detection value as output.

[0098] The expression of the environmental ice thickness detection function is:

[0099]

[0100] Where f3(·) represents the environmental ice thickness detection function, represents the environment mapping vector, b1 represents the environment bias; ||·||2 represents the L2 norm, and F3=f3(μ1,σ1).

[0101] The expression of the electrical ice thickness detection function is:

[0102]

[0103] Where f4(·) represents the electrical ice thickness detection function, z represents the nonlinear parameter, Represents the electrical mapping vector, F4=f4(μ2,σ2), and standard represents the standard value of the electrical parameter.

[0104] Construct a multi-source ice thickness detection sequence F, which can be expressed as:

[0105] F=(F1,F2,F3,F4)

[0106] Among them, F1, F2, F3, and F4 are the image ice thickness detection value, point cloud ice thickness detection value, environmental ice thickness detection value, and electrical ice thickness detection value, respectively.

[0107] In this embodiment, in step S3, the multi-source fusion inference is performed on the multi-source ice thickness detection sequence to obtain the accurate detection value and confidence interval of the ice thickness of the distribution overhead line, including:

[0108] Multi-source fusion is performed on the multi-source ice thickness detection sequence to obtain the ice thickness observation value G;

[0109] Specifically, the prior distribution of the ice thickness of the distribution overhead line is obtained. The Bayesian inference method is used to update the prior distribution in combination with the observation value G to obtain the accurate detection value and confidence interval of the ice thickness of the distribution overhead line. The prior distribution is the mean mean and variance δ of the historical accurate detection values ​​of the ice thickness of the distribution overhead line. The inference formula for the accurate detection value and confidence interval of the ice thickness of the distribution overhead line is:

[0110]

[0111] Among them, R represents the precise detection value, θ represents the pre-measured observation error variance; the confidence interval is The confidence level is 95%.

[0112] The multi-source fusion formula of the multi-source ice thickness detection sequence is:

[0113]

[0114] Among them, w d represents the fusion coefficient of the dth type of ice thickness detection value, where the 1st to 4th types of ice thickness detection values ​​are image ice thickness detection value, point cloud ice thickness detection value, environmental ice thickness detection value, and electrical ice thickness detection value, respectively; H d represents the information content of the dth ice thickness detection value, represents the information control factor of the d-th ice thickness detection value; grad_mean represents the gradient mean of the image detection data after interpolation and completion processing, grad_max represents the gradient maximum of the image detection data after interpolation and completion processing; num represents the total number of 3D point cloud data in the point cloud detection data after interpolation and completion processing; β represents the temperature variation coefficient, Tem represents the critical temperature for ice formation, Represents the mean temperature of the environmental detection data after interpolation and completion processing, They represent the mean resistance value and the mean capacitance value of the electrical parameter detection data after interpolation and completion processing, ω1 represents the standard resistance value, and ω2 represents the standard capacitance value.

[0115] Another possible implementation involves using a credibility-weighted fusion approach for multi-source inference. The system assigns dynamic weights to each type of detection data: image data is weighted based on its clarity score (0-100), point cloud data is weighted proportional to its point density, and environmental data is weighted based on sensor health. The final fusion result is a weighted average of the data sources, and re-collection is automatically triggered when the credibility of a particular data type falls below a threshold.

[0116] In this embodiment, in step S4, the historical data series of ice thickness of the distribution overhead line is combined to predict the accurate detection value of the ice thickness of the distribution overhead line, and the icing trend of the distribution overhead line is obtained, including:

[0117] The historical data sequence of ice thickness of distribution overhead lines is a sequence composed of historical accurate detection values. The accurate detection values ​​of ice thickness of distribution overhead lines are added to the end of the historical data sequence. The time series prediction algorithm is used to predict the ice thickness of distribution overhead lines in the future and generate the ice trend of distribution overhead lines. The ice trend is the rate of change of ice thickness in the future. The calculation formula of ice trend is (R * -R) / R, where R * Indicates the ice thickness of the distribution overhead lines at a future time.

[0118] Exemplarily, the time series prediction algorithm may be an LSTM model algorithm.

[0119] In another possible implementation, when predicting ice cover trends, a time-series model of ice growth can be established to automatically identify three typical growth patterns: linear growth (0.2-0.5 mm per hour), exponential growth (doubling every three hours), and step-like growth (sudden changes during cold waves). By matching current data with a library of historical patterns, thickness changes can be predicted over the next six hours.

[0120] In this embodiment, the real-time monitoring and early warning of the icing status of the distribution overhead line based on the icing trend of the distribution overhead line in step S5 includes:

[0121] Based on the icing trend of the distribution overhead lines and combined with preset thresholds, real-time monitoring and early warning of the icing status of the distribution overhead lines are carried out; the preset thresholds include detection thresholds and trend thresholds. If the boundary value of the confidence interval exceeds the detection threshold or the icing trend exceeds the trend threshold, an early warning of the icing status of the distribution overhead lines is issued.

[0122] In another possible implementation, a three-level warning system can be established during real-time monitoring and early warning: primary warning (blue): when the ice thickness reaches 30% of the design value, a text message notification is sent; intermediate warning (yellow): when the thickness reaches 50% of the design value, the automatic de-icing device is activated; advanced warning (red): when the thickness reaches 80% of the design value, the dispatching center is linked to adjust the power grid operation mode.

[0123] Example 3. The above is a schematic scheme of the method for detecting ice thickness on distribution overhead lines according to this embodiment. It should be noted that the technical scheme of the system for detecting ice thickness on distribution overhead lines and the technical scheme of the method for detecting ice thickness on distribution overhead lines are based on the same concept. For details not described in detail in the technical scheme of the system for detecting ice thickness on distribution overhead lines in this embodiment, please refer to the description of the technical scheme of the method for detecting ice thickness on distribution overhead lines.

[0124] This embodiment further provides a system for detecting ice thickness on distribution overhead lines, comprising:

[0125] The detection data set acquisition module is used to collect multi-source detection data of distribution overhead lines and perform adaptive interpolation and completion processing to obtain the detection data set;

[0126] A detection sequence acquisition module is used to perform multi-source ice thickness detection on the detection data set to obtain a multi-source ice thickness detection sequence;

[0127] The multi-source fusion inference module is used to perform multi-source fusion inference on the multi-source ice thickness detection sequence to obtain the accurate detection value and confidence interval of the ice thickness of the distribution overhead line;

[0128] The prediction module is used to combine the historical data series of ice thickness of distribution overhead lines to accurately predict the detection value of ice thickness of distribution overhead lines and obtain the ice trend of distribution overhead lines;

[0129] The monitoring and early warning module is used to monitor and warn the icing status of the distribution overhead lines in real time based on the icing trend of the distribution overhead lines.

[0130] This embodiment further provides an electronic device applicable to the method for detecting ice thickness of distribution overhead lines, including:

[0131] Memory and processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the method for detecting ice thickness on distribution overhead lines as proposed in the above embodiment.

[0132] This embodiment further provides a storage medium storing a computer program, which, when executed by a processor, implements the method for detecting ice thickness of distribution overhead lines proposed in the above embodiment.

[0133] The storage medium proposed in this embodiment and the method for detecting ice thickness of distribution overhead lines proposed in the above embodiment belong to the same inventive concept. For technical details not described in detail in this embodiment, please refer to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0134] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for detecting ice thickness of distribution overhead lines, characterized in that: include: Collect multi-source detection data of distribution overhead lines and perform adaptive interpolation and completion processing to obtain a detection data set; Perform multi-source ice thickness detection on the detection data set to obtain a multi-source ice thickness detection sequence; Multi-source fusion inference is performed on the multi-source ice thickness detection sequence to obtain the accurate detection value and confidence interval of the ice thickness of the distribution overhead line; Combined with the historical data series of ice thickness on distribution overhead lines, the accurate detection value of ice thickness on distribution overhead lines is predicted to obtain the icing trend of distribution overhead lines. Based on the icing trend of the distribution overhead lines, real-time monitoring and early warning of the icing status of the distribution overhead lines are carried out.

2. The method for detecting ice thickness of distribution overhead lines according to claim 1, wherein: The adaptive interpolation and completion processing is performed to obtain a detection data set including: Count the number of pixels num1 in the horizontal direction and num2 in the vertical direction of the image detection data. If num1 is less than the preset row threshold and num2 is less than the preset column threshold, use bilinear interpolation to interpolate and fill in the missing pixels so that the number of pixels in the horizontal direction and the number of pixels in the vertical direction of the image detection data after interpolation and filling reach the preset row threshold and column threshold respectively.

3. The method for detecting ice thickness of distribution overhead lines according to claim 2, wherein: The adaptive interpolation and completion process to obtain the detection data set further includes: The point cloud detection data is locally adaptively interpolated and supplemented using a local adaptive weighting method based on the environmental detection data to obtain the point cloud intensity at all three-dimensional point cloud coordinate positions; Linear interpolation is used to interpolate and fill in the missing values ​​in the environmental detection data and electrical parameter detection data.

4. The method for detecting ice thickness of distribution overhead lines according to claim 3, wherein: The performing multi-source ice thickness detection on the detection data set to obtain a multi-source ice thickness detection sequence includes: The gradient value of each pixel in the image detection data set is calculated to form a gradient matrix that characterizes the distribution of the ice layer. The residual convolution module and the fully connected layer are used to perform residual convolution and mapping on the gradient matrix in sequence to obtain the image ice thickness detection value.

5. The method for detecting ice thickness of distribution overhead lines according to claim 4, wherein: The performing multi-source ice thickness detection on the detection data set to obtain a multi-source ice thickness detection sequence further includes: Extract the point cloud detection data of the detection data set and calculate the point cloud ice thickness detection value; Extract the environmental detection data from the detection data set and the environmental structural features and electrical structural features of the electrical parameter detection data, respectively construct an environmental ice thickness detection function and an electrical ice thickness detection function to perform ice thickness detection, and obtain the environmental ice thickness detection value and the electrical ice thickness detection value; The environmental ice thickness detection function takes the environmental structural characteristics as input and the environmental ice thickness detection value as output. The electrical ice thickness detection function takes the electrical structural characteristics as input and the electrical ice thickness detection value as output.

6. The method for detecting ice thickness of distribution overhead lines according to claim 5, characterized in that: The above method combines the historical data series of ice thickness of the distribution overhead line to predict the accurate detection value of the ice thickness of the distribution overhead line, and obtains the ice trend of the distribution overhead line, including: The historical data sequence of the ice thickness of the distribution overhead lines is a sequence composed of historical precise detection values. The precise detection values ​​of the ice thickness of the distribution overhead lines are added to the end of the historical data sequence. The time series prediction algorithm is used to predict the ice thickness of the distribution overhead lines in the future. The ice trend of the distribution overhead lines is also generated. The ice trend is the rate of change of the ice thickness in the future.

7. The method for detecting ice thickness of distribution overhead lines according to claim 6, wherein: The real-time monitoring and early warning of the icing status of the distribution overhead lines based on the icing trend of the distribution overhead lines includes: Based on the icing trend of the distribution overhead lines and combined with preset thresholds, real-time monitoring and early warning of the icing status of the distribution overhead lines are carried out; the preset thresholds include detection thresholds and trend thresholds. If the boundary value of the confidence interval exceeds the detection threshold or the icing trend exceeds the trend threshold, an early warning of the icing status of the distribution overhead lines is issued.

8. A system for detecting ice thickness on distribution overhead lines, applying the method according to any one of claims 1 to 7, characterized in that: include: The detection data set acquisition module is used to collect multi-source detection data of distribution overhead lines and perform adaptive interpolation and completion processing to obtain the detection data set; A detection sequence acquisition module is used to perform multi-source ice thickness detection on the detection data set to obtain a multi-source ice thickness detection sequence; The multi-source fusion inference module is used to perform multi-source fusion inference on the multi-source ice thickness detection sequence to obtain the accurate detection value and confidence interval of the ice thickness of the distribution overhead line; The prediction module is used to combine the historical data series of ice thickness of distribution overhead lines to accurately predict the detection value of ice thickness of distribution overhead lines and obtain the ice trend of distribution overhead lines; The monitoring and early warning module is used to monitor and warn the icing status of the distribution overhead lines in real time based on the icing trend of the distribution overhead lines.

9. An electronic device comprising: memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions, when executed by a processor, implement the steps of the method according to any one of claims 1 to 7.

Images