Water accumulation detection method and device of transformer substation, storage medium and electronic equipment

Abstract

The invention discloses an accumulated water detection method and device of a transformer substation, a storage medium and electronic equipment. The method comprises the following steps: acquiring a current visible light image and a current infrared image of a target area where a transformer substation is located; based on the current visible light image, determining a ponding area detection result of a ponding area in the target area; based on the current infrared image, determining a ponding depth detection result of the ponding area; based on the ponding area detection result and the ponding depth detection result, a ponding detection result of the target area is obtained, and the ponding detection result is at least used for indicating whether ponding abnormity exists in the target area or not. According to the invention, the technical problem of inaccurate detection results of the accumulated water depth and the accumulated water area of the area where the transformer substation is located in the prior art is solved.

Description

Technical Field

[0001] This application relates to the field of smart grids, and more specifically, to a method, apparatus, storage medium, and electronic equipment for detecting water accumulation in substations. Background Art

[0002] Substations have densely packed equipment and complex terrain. Water accumulation is affected by obstructions, reflections, and shadows, making it difficult for sensors to provide comprehensive coverage and accurate water detection, especially in concealed areas. Current technologies using conventional visual algorithms for substation water detection are susceptible to environmental interference, have a high false detection rate, and struggle to handle water accumulation in complex backgrounds. Furthermore, determining the water area in these technologies relies primarily on manual calibration or 3D modeling, which is inefficient and fails to meet real-time detection requirements. This results in inaccurate detection results for water depth and area in substation areas.

[0003] To address the above-mentioned problems, no effective solutions have been proposed so far. Summary of the Invention

[0004] This application provides a method, device, storage medium, and electronic device for detecting water accumulation in substations, so as to at least solve the technical problem of inaccurate detection results of water depth and water area in the substation area in related technologies.

[0005] According to one aspect of the embodiments of this application, a method for detecting water accumulation in a substation is provided, comprising: acquiring a current visible light image and a current infrared image of a target area where the substation is located; determining a water accumulation area detection result in the target area based on the current visible light image; determining a water accumulation depth detection result in the target area based on the current infrared image; and obtaining a water accumulation detection result for the target area based on the water accumulation area detection result and the water accumulation depth detection result, wherein the water accumulation detection result is at least used to indicate whether there is an abnormal water accumulation in the target area.

[0006] According to another aspect of the embodiments of this application, a water accumulation detection device for a substation is provided, comprising: an image acquisition module for acquiring a current visible light image and a current infrared image of a target area where the substation is located; a water accumulation area determination module for determining a water accumulation area detection result of a water accumulation area in the target area based on the current visible light image; a water accumulation depth determination module for determining a water accumulation depth detection result of a water accumulation area based on the current infrared image; and a water accumulation detection result determination module for obtaining a water accumulation detection result of the target area based on the water accumulation area detection result and the water accumulation depth detection result, wherein the water accumulation detection result is at least used to indicate whether there is an abnormal water accumulation in the target area.

[0007] According to another aspect of the embodiments of this application, a non-volatile storage medium is provided, which stores multiple instructions adapted for a substation water accumulation detection method to be loaded by a processor and executed at any one of them.

[0008] According to another aspect of the embodiments of this application, an electronic device is provided, including: one or more processors and a memory, the memory being used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement any one of the substation water accumulation detection methods.

[0009] In this embodiment, current visible light and infrared images of the target area where the substation is located are acquired; based on the current visible light image, the water area detection result of the water accumulation region in the target area is determined; based on the current infrared image, the water depth detection result of the water accumulation region is determined; based on the water area detection result and the water depth detection result, the water accumulation detection result of the target area is obtained, wherein the water accumulation detection result is at least used to indicate whether there is an abnormal water accumulation in the target area. This achieves the goal of determining the water area detection result of the water accumulation region in the substation area based on the visible light image, and determining the water depth detection result of the water accumulation region in the substation area based on the infrared image, and accurately judging whether there is an abnormal water accumulation in the substation area based on the water area detection result and the water depth detection result. This improves the accuracy of the water depth and water area detection results for the substation area, thereby solving the technical problem of inaccurate water depth and water area detection results in related technologies. Attached Figure Description

[0010] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0011] Figure 1 This is a flowchart of an optional method for detecting water accumulation in a substation according to an embodiment of this application;

[0012] Figure 2 This is a structural block diagram of an optional water accumulation detection method for substations provided according to an embodiment of this application;

[0013] Figure 3 This is a schematic diagram of an optional water accumulation detection device for a substation according to an embodiment of this application. Detailed Implementation

[0014] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0015] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0016] According to an embodiment of this application, a method embodiment for detecting water accumulation in a substation is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0017] Figure 1 This is a flowchart of an optional substation water accumulation detection method provided according to an embodiment of this application, such as... Figure 1 As shown, the method includes the following steps:

[0018] Step S102: Acquire the current visible light image and the current infrared image of the target area where the substation is located;

[0019] It is understandable that data acquisition equipment is used to capture images of the target area where the substation is located, obtaining current visible light and infrared images. A wide-angle camera can be used to capture the current visible light image, and an infrared thermal imager can be used to capture the current infrared image. By acquiring these images, rich image information about the target area where the substation is located can be obtained, providing a data foundation for subsequent water accumulation detection.

[0020] Optionally, a wide-angle camera can be used to acquire visible light images of the substation, while an infrared thermal imager can be used to acquire infrared images. The wide-angle camera is a 4K resolution camera (meaning a resolution of approximately 4000 horizontal pixels) with a horizontal field of view ≥120°. The wide-angle camera is deployed in the substation's power distribution room, cable trench entrance, and on top of the transformer base, covering a radius ≥15 meters. The infrared thermal imager uses an uncooled infrared sensor with a wavelength of 8-14 μm (micrometers) and a temperature sensitivity ≤0.05℃, and is coaxially mounted with the visible light camera to penetrate rain and fog to detect the thermal radiation characteristics of accumulated water.

[0021] Optionally, when deploying wide-angle cameras and infrared thermal imagers, the latitude and longitude information and elevation data of the camera installation point (in the WGS84 coordinate system) can be recorded using an RTK (Real-Time Kinematic) positioning device.

[0022] Optionally, since the water accumulation detection result of the water accumulation area needs to be determined jointly based on information from both visible light and infrared images, a spatiotemporal synchronization mechanism is required to synchronize and align the acquired visible light and infrared images in both time and space. For temporal synchronization, the camera and RTK positioning device can achieve millisecond-level time synchronization via the NTP (Network Time Protocol), with each frame of image appended with a timestamp (accurate to milliseconds) and corresponding geographic coordinates. For spatial synchronization, the ORB-SLAM (OrientedFAST and Rotated BRIEF-Simultaneous Localization and Mapping) feature point matching algorithm can be used to align the infrared and visible light images, and RANSAC (Random Sample Consensus) can be used to eliminate mismatched points, ensuring that the matching error is ≤2 pixels.

[0023] Step S104: Based on the current visible light image, determine the water area detection result of the water accumulation area in the target area;

[0024] It is understandable that analyzing and processing the acquired visible light images helps determine the water area detection results for the water accumulation zone within the target area where the substation is located. By performing in-depth analysis of the current visible light images, the accuracy of the water area detection results for the substation area can be improved, enabling high-precision, real-time detection of water accumulation areas in substations under complex environments.

[0025] In one optional embodiment, when there are multiple water accumulation areas, the water accumulation area detection result of the water accumulation area in the target area is determined based on the current visible light image. This includes: using a target water accumulation area prediction model based on the current visible light image to determine the pixel vertex coordinates corresponding to each of the multiple water accumulation areas, wherein the target water accumulation area prediction model has pre-learned the correspondence between the visible light image of the target area and the pixel vertex coordinates of the water accumulation areas, and the pixel vertex coordinates are used to describe the vertex position of the water accumulation area in the image coordinate system; determining the pixel area corresponding to each of the multiple water accumulation areas based on the pixel vertex coordinates; and performing coordinate system transformation on the pixel areas corresponding to each of the multiple water accumulation areas to obtain the water accumulation area detection result corresponding to each of the multiple water accumulation areas in the geographic coordinate system.

[0026] It is understandable that if there are multiple water accumulation areas in the target area, it is necessary to determine the water area detection results corresponding to each of these areas. Based on the current visible light image, a target water accumulation area prediction model (e.g., the CBAM-YOLOv8 model) is used, which has pre-learned the correspondence between the visible light image of the target area and the pixel vertex coordinates of the water accumulation areas, to determine the pixel vertex coordinates corresponding to each of the multiple water accumulation areas. These pixel vertex coordinates represent the vertex positions of the water accumulation areas in the image coordinate system. Based on the pixel vertex coordinates corresponding to each of the multiple water accumulation areas, the pixel areas corresponding to each of the multiple water accumulation areas are calculated. The pixel areas corresponding to each of the multiple water accumulation areas are then transformed to obtain the water area detection results corresponding to each of the multiple water accumulation areas in the geographic coordinate system. Through accurate vertex coordinate prediction, efficient pixel area calculation, and accurate geographic coordinate system transformation, rapid and accurate detection of water accumulation areas in substations can be achieved, improving the safety and emergency response efficiency of substations.

[0027] Optionally, the CBAM-YOLOv8 model can be used as the prediction model for the aforementioned target water accumulation area to determine the water accumulation area detection results of the substation. The CBAM-YOLOv8 model refers to embedding the CBAM (Convolutional Block Attention Module) attention module after the C3 (Cross Stage Partial) layer (Backbone network, used for feature extraction) and the FPN (Feature Pyramid Network) layer (Neck network, used for feature fusion) of the YOLOv8 (You Only Look Once version 8) model's network architecture. This sequentially performs channel attention weight calculation and spatial attention mask generation, thereby enhancing the water accumulation texture features and suppressing reflective interference from substation equipment.

[0028] Optionally, a CBAM module is connected in series at the output of the C3 module (layers 2, 4, and 6) in the Backbone layer, and the channel attention compression ratio is set to 8:1 (i.e., the dimension of the fully connected layer is reduced from C to C / 8). Channel attention weights M c (F) can be determined in the following way:

[0029] M c (F)=σ(W1(W0(F avg )))

[0030] Among them, F avg W0 and W1 represent the global average pooling result, W0 and W1 represent the weight matrices of the fully connected layer, and σ() represents the activation function.

[0031] Optionally, a spatial attention module is inserted before the P3 and P4 feature map outputs of the Neck layer FPN module, and a 7×7 convolutional kernel is used to extract local texture features. A dynamic threshold activation mechanism can be used to adjust the feature activation threshold. This mechanism suppresses false activation of reflective areas (mistakenly identified as water accumulation areas) in substation equipment, reducing the false detection rate by 12.7%. The feature activation threshold is based on the spatial attention weights M. s (F) Dynamic adjustment is performed, and the feature activation threshold T can be determined in the following way:

[0032]

[0033] Where F represents the feature map obtained after feature extraction from the historical visible light image, and F contains the original pixel features of the historical visible light image.

[0034] Optionally, the original FPN is replaced with a bidirectional feature pyramid network, connecting layers P3 (80×80), P4 (40×40), and P5 (20×20) across scales, and image features are fused using learnable weights. For small areas of water accumulation, high-resolution features from layer P3 are preferentially used, and the weight w of the feature map from layer i in the fusion process... i It can be determined in the following way:

[0035]

[0036] Where γ = 2.0 represents a hyperparameter used to adjust the influence of image features from each layer on the calculation of IoU (Intersection over Union). IoU represents the intersection-over-union ratio of anchor boxes and ground truth boxes. Anchor boxes are a series of predefined bounding boxes set on historical visible light images to predict the location of potential water accumulation areas, while ground truth boxes are bounding boxes labeled on training images that accurately describe the location of water accumulation areas. i and j represent the number of layers in the model. w i This represents the weight of the i-th layer feature map during the fusion process.

[0037] Optionally, GIoU Loss (Generalized Intersection over Union Loss) can be used to determine the model's loss function and optimize the bounding box regression process. The loss function L... GIoU It can be determined in the following way:

[0038]

[0039] Where B represents the predicted bounding box, B gt C represents the ground truth bounding box, and C represents the smallest closed bounding box between the predicted box and the ground truth box.

[0040] In an optional embodiment, before determining the pixel vertex coordinates corresponding to multiple water accumulation areas based on the current visible light image and using a target water accumulation area prediction model, the method further includes: acquiring historical visible light images of the target area; performing data augmentation processing on the historical visible light images to obtain the target visible light image, wherein the data augmentation processing includes at least one of the following: substation equipment reflection simulation, dynamic rain and fog interference simulation, and small-area water accumulation area enhancement processing, where a small-area water accumulation area refers to a water accumulation area with an area smaller than a preset area threshold; and training a pre-determined initial water accumulation area prediction model based on the target visible light image to obtain the target water accumulation area prediction model.

[0041] It is understandable that historical visible light images of the target area where the substation is located are acquired. To enhance the identification capability of the target water accumulation area prediction model in environments with dense substation equipment, reflective interference, and dynamic rain and fog interference, as well as its ability to identify small water accumulation areas, data augmentation processing is performed on the acquired historical visible light images to obtain the target visible light image. The aforementioned data augmentation processing includes at least one of the following: substation equipment reflective simulation, dynamic rain and fog interference simulation, and small water accumulation area enhancement processing. The target visible light image is used to train a pre-determined initial water accumulation area prediction model to obtain the target water accumulation area prediction model. By training the model using the target visible light image obtained after the above data augmentation processing, the water accumulation area prediction model's ability to identify water accumulation in various complex environments, as well as its ability to identify small water accumulation areas, can be improved. The impact of interference conditions such as equipment density, reflective interference, and rain and fog obstruction on the detection results can be reduced, improving the accuracy of water accumulation area detection results, ensuring the safe operation of the substation, and enhancing the timely early warning and response capabilities to the substation's internal flooding risks.

[0042] Optionally, substation equipment reflection simulation can be performed as follows: First, a reflection texture library of 12 types of substation equipment surface materials (galvanized steel, ceramic insulators, silicone rubber bushings, etc.) is generated using CycleGAN (Cycle-Consistent Adversarial Networks), containing reflection templates under different illumination angles (0°-180°). Second, reflection phenomena in the actual environment are simulated to improve the model's ability to handle this type of interference. Next, a Poisson fusion overlay method is used to overlay the reflection templates onto historical visible light images with a transparency α = 0.4. The fusion process can be implemented as follows:

[0043] l out =α·l reflection +(1-α)·l original

[0044] Where α∈[0.3,0.5], α can be dynamically adjusted to simulate the actual reflective intensity. out This represents the image output after fusing a historical visible light image with a reflective template. original Represents historical visible light images, l reflection This represents a reflective image (i.e., a reflective template).

[0045] Optionally, dynamic rain and fog interference simulation can be performed through raindrop particle generation and rain / fog blurring. For raindrop particle generation, the raindrop trajectory can be simulated based on the NVIDIA Flow physics engine (a fluid dynamics simulation tool mainly used to generate realistic fluid effects such as water, smoke, and fire), and the raindrop density λ can be set to 0.2 drops / pixel. 2 (This means there are expected to be 0.2 raindrop particles per square pixel area, corresponding to moderate rain intensity). The raindrop particle size follows a normal distribution (mean 3 pixels, standard deviation 0.5), where droplets represent raindrops and pixel... 2 This represents the area of ​​a unit pixel. For rain and fog blurring, a Gaussian blur kernel (3×3, representing the size of the Gaussian blur kernel; ε=1.5, representing the standard deviation of the Gaussian blur kernel) is added to the historical visible light image, and salt-and-pepper noise (noise density 0.1) is superimposed to simulate the rain and fog scattering effect. The core of Gaussian blurring lies in using a Gaussian function as weights to perform a weighted average on each pixel of the image and its surrounding pixels, used to reduce image noise, smooth image edges, or generate certain visual effects.

[0046] Optionally, Mosaic four-image stitching and CutMix enhancement can be used to enhance small water accumulation areas. Mosaic four-image stitching involves randomly selecting four historical visible light images, stitching them together in a 2×2 grid, and forcibly including at least two small water accumulation areas (pixel area < 50×50). CutMix enhancement involves cropping small water accumulation areas from other historical visible light images and pasting them into the current historical visible light image at random positions and rotation angles (±15°), with an occlusion ratio ≤ 30%.

[0047] In one optional embodiment, the pixel areas corresponding to multiple water accumulation areas are transformed into coordinate systems to obtain water accumulation area detection results corresponding to multiple water accumulation areas in the geographic coordinate system. This includes: determining the extrinsic parameter matrix of the current visible light image acquisition device, wherein the extrinsic parameter matrix is ​​used to establish the geometric relationship between the image coordinate system and the geographic coordinate system; determining the affine transformation matrix based on the extrinsic parameter matrix, wherein the affine transformation matrix is ​​used to map points in the image coordinate system to the geographic coordinate system; and determining the water accumulation area detection results corresponding to multiple water accumulation areas based on the pixel areas corresponding to multiple water accumulation areas and the affine transformation matrix.

[0048] It is understood that the extrinsic parameter matrix used by the current visible light image acquisition device (e.g., a wide-angle camera) to establish the geometric relationship between the image coordinate system and the geographic coordinate system is determined. Based on the extrinsic parameter matrix, an affine transformation matrix is ​​determined to map points in the image coordinate system to the geographic coordinate system. Using the affine transformation matrix, the pixel areas corresponding to the multiple water accumulation areas obtained above are converted into water accumulation area detection results for the multiple water accumulation areas in the geographic coordinate system. By determining the extrinsic parameter matrix and the affine transformation matrix, the pixel areas of water accumulation areas in the image coordinate system can be accurately located in the geographic coordinate system, providing water accumulation area detection results for the water accumulation areas in the geographic coordinate system, thus enhancing the accuracy of the water accumulation area determination results.

[0049] Optionally, before performing pixel-to-geographic coordinate transformation, calibration boards need to be deployed first. The calibration boards can be deployed as follows: four sets of 1m×1m rectangular calibration boards are arranged on the substation ground, made of high-reflectivity aluminum plates (reflectivity ≥85%), with a spacing error ≤1cm; the corner coordinates of the calibration boards are obtained by measuring with a total station, with an accuracy of ±0.5mm.

[0050] Optionally, to convert pixel coordinates to geographic coordinates, the Perspective-n-Point (PnP) algorithm can be used to determine the extrinsic parameter matrix [R|t] of the wide-angle camera acquiring visible light images, where R represents the rotation matrix and t represents the translation vector, which are key parameters describing the position and orientation of the wide-angle camera in three-dimensional space. An affine transformation matrix H is constructed based on the extrinsic parameter matrix [R|t], thereby converting the pixel coordinates (u,v) to geographic coordinates (X,Y,Z). The geographic coordinates (X,Y,Z) can be determined as follows:

[0051]

[0052] Alternatively, the Levenberg-Marquardt algorithm can be used to optimize the parameters of the affine transformation matrix so that the reprojection error (the difference between the theoretical image point position and the observed actual image point position) is ≤1.5 pixels.

[0053] Optionally, to calculate the water accumulation area, the coordinates of n pixel vertices {(x1, y1), ..., (x...} of the water accumulation area polygon are extracted from the CBAM-YOLOv8 detection results. n ,y n Based on the above pixel vertex coordinates, calculate the pixel area S of the water accumulation region. pixel S pixel It can be determined in the following way:

[0054]

[0055] Optionally, the pixel area is converted into the water accumulation area detection result S in the geographic coordinate system through an affine transformation matrix H. real S real It can be determined in the following way:

[0056] S real =S pixel det(H) -1

[0057] Optionally, based on the above water accumulation area detection results, if the water accumulation area growth rate is ≥10% / minute for three consecutive frames, it is marked as a substation having a risk of waterlogging. When a risk of waterlogging is detected in the substation, a waterlogging risk warning can be triggered, and the SCADA system (Supervisory Control and Data Acquisition) can be linked to control the start and stop of the drainage pumps.

[0058] Step S106: Based on the current infrared image, determine the water depth detection result of the water accumulation area;

[0059] It is understandable that analyzing the acquired infrared images helps determine the water depth in the target area where the substation is located. By performing depth analysis on the current infrared images to determine the water depth in the target area where the substation is located, and combining this with the water area detection from the visible light images, the accuracy and reliability of the substation water accumulation detection results can be improved, enabling precise assessment and timely early warning of substation flooding risks.

[0060] In one optional embodiment, when there are multiple water accumulation areas, the water depth detection result of the water accumulation areas is determined based on the current infrared image, including: determining the current temperature distribution corresponding to each of the multiple water accumulation areas based on the current infrared image, wherein the current temperature distribution is used to describe the temperature status of multiple temperature measurement points in the water accumulation areas; performing difference processing on the actual temperature distribution corresponding to each of the multiple water accumulation areas and the corresponding current temperature distribution to obtain the temperature difference distribution corresponding to each of the multiple areas, wherein the actual temperature distribution is measured when there is no water accumulation; and determining the water depth detection result corresponding to each of the multiple water accumulation areas based on the temperature difference distribution corresponding to each of the multiple water accumulation areas.

[0061] It is understandable that if there are multiple waterlogged areas in the target area, it is necessary to determine the water area detection results for each of these areas. The current infrared image is analyzed to determine the current temperature distribution for each waterlogged area, representing the temperature conditions at multiple temperature measurement points within that area. The difference between the actual temperature distribution measured in the absence of water and the corresponding current temperature distribution is then calculated to obtain the temperature difference distribution for each area. Based on this temperature difference distribution, the water depth detection results for each waterlogged area are determined. By combining the current temperature distribution obtained from the current infrared image with the actual temperature distribution obtained under a preset water-free condition, the water depth in multiple waterlogged areas can be accurately detected, improving the accuracy of water depth detection results and the safe operation level of the substation.

[0062] Step S108: Based on the water accumulation area detection results and water accumulation depth detection results, obtain the water accumulation detection results of the target area, wherein the water accumulation detection results are used at least to indicate whether there is an abnormal water accumulation in the target area.

[0063] It is understandable that the water accumulation area and depth detection results are used to determine whether there is abnormal water accumulation in the target area. By integrating the water accumulation area and depth detection results, the water accumulation detection results become more comprehensive, thus accurately reflecting the degree of impact of water accumulation on substation safety.

[0064] In one optional embodiment, the water accumulation detection result of the target area is obtained based on the water accumulation area detection result and the water accumulation depth detection result, including: determining the target water accumulation level of the water accumulation area based on the water accumulation area detection result, the water accumulation depth detection result, a preset water accumulation area threshold, and a preset water accumulation depth threshold; if the target water accumulation level is greater than the preset level, the water accumulation detection result is determined to indicate that there is an abnormal water accumulation in the target area; or if the target water accumulation level is less than or equal to the preset level, the water accumulation detection result is determined to indicate that there is no abnormal water accumulation in the target area.

[0065] It is understandable that the target water accumulation level of the water-accumulated area is determined based on the water accumulation area detection results and a pre-set water accumulation area threshold, as well as the water accumulation depth detection results and a pre-set water accumulation depth threshold. If the target water accumulation level is greater than the preset level, it indicates that there is an abnormal water accumulation in the target area where the substation is located; if the target water accumulation level is less than or equal to the preset level, it indicates that there is no abnormal water accumulation in the target area where the substation is located. Combining the water accumulation status from both the water accumulation area and water accumulation depth dimensions allows for a more comprehensive assessment of the substation's water accumulation level, improving the accuracy of the water accumulation level determination results. Furthermore, by adaptively adjusting the water accumulation area and water accumulation depth thresholds, the applicability and effectiveness of substation water accumulation early warning can be improved, ensuring the safe operation of the power system.

[0066] Optionally, a tiered early warning mechanism can be used for substation flooding warnings. Different flooding levels are assigned corresponding flooding depth and flooding area thresholds. Based on the obtained flooding area and depth detection results, combined with the corresponding flooding depth and area thresholds, the target flooding level of the substation is determined, and flooding warnings are issued based on the target flooding level. For example, flooding levels are divided into warning levels and danger levels. If the flooding area is ≥5㎡ and the depth is ≥10cm (estimated from infrared images), the flooding level is considered a warning level, triggering a yellow warning and sending an SMS to substation maintenance personnel. If the flooding area is ≥10㎡ and the depth is ≥20cm, the current flooding level in the target area is considered a danger level, triggering a red warning. The substation will automatically start drainage pumps and activate the SCADA system to cut off power to the high-risk area.

[0067] Optionally, when conducting tiered early warnings, the substation's SCADA system can be linked to the OPC protocol (OLE for Process Control, an open platform communication protocol) to issue early warnings and push early warning information for water accumulation in the substation, and trigger drainage equipment to carry out drainage treatment.

[0068] As an optional embodiment, the water accumulation detection result of the target area is obtained based on the water accumulation area detection result and the water accumulation depth detection result, including: determining a first weight value corresponding to the water accumulation area detection result and a second weight value corresponding to the water accumulation depth detection result; determining the water accumulation score of the water accumulation area based on the first weight value and the second weight value; determining the target score interval of the water accumulation score from multiple score intervals, wherein the multiple score intervals correspond one-to-one with multiple water accumulation levels; determining the water accumulation level corresponding to the target score interval as the target water accumulation level of the water accumulation area; if the target water accumulation level is greater than a preset level, determining that the water accumulation detection result indicates that there is an abnormal water accumulation in the target area; or if the target water accumulation level is less than or equal to the preset level, determining that there is no abnormal water accumulation in the target area.

[0069] Optionally, the water accumulation detection results for the target area can be determined not only directly based on the water accumulation area detection results, water accumulation depth detection results, water accumulation area thresholds, and water accumulation depth thresholds, but also using a weighted calculation method. First, a first weight value corresponding to the water accumulation area detection results and a second weight value corresponding to the water accumulation depth detection results are determined. Then, a weighted calculation is performed based on the water accumulation area detection results, the first weight value, the water accumulation depth detection results, and the second weight value to obtain a water accumulation score for the water accumulation area. This score is compared with multiple scoring intervals to determine the target scoring interval to which the water accumulation score belongs, and the water accumulation level corresponding to this target scoring interval is determined as the target water accumulation level for the water accumulation area. If the target water accumulation level is greater than a preset level, it indicates that there is an abnormal water accumulation in the target area where the substation is located; if the target water accumulation level is less than or equal to the preset level, it indicates that there is no abnormal water accumulation in the target area where the substation is located. By quantifying the water accumulation area and depth and combining them with corresponding weight values ​​for comprehensive scoring, accurate quantification and classification of the water accumulation status of the substation can be achieved, thereby accurately identifying whether there is an abnormal water accumulation in the substation.

[0070] Through the above steps S102 to S108, the water area detection result of the water accumulation area in the substation area can be determined based on the visible light image, and the water depth detection result of the water accumulation area in the substation area can be determined based on the infrared image. The purpose of judging whether there is water accumulation anomaly in the substation area is achieved by using the water area detection result and the water depth detection result. This achieves the technical effect of improving the accuracy of the water depth and water area detection results in the substation area, thereby solving the technical problem of inaccurate water depth and water area detection results in the related technology.

[0071] Based on the above embodiments and optional embodiments, this application proposes an optional implementation method for water accumulation detection in substations. By acquiring visible light and infrared images of the substation, water accumulation detection is performed on the area where the substation is located, and water accumulation warnings are issued based on the detection results. This method achieves high-precision identification of water accumulation areas of various sizes and shapes within the substation area under complex backgrounds and automatically associates geographic coordinates for precise location of the water accumulation areas. Simultaneously, it enables rapid and automatic calculation of water accumulation area and depth to provide real-time water accumulation warnings and flood risk assessments for the substation, providing reliable assurance for the safe operation of the substation.

[0072] The substation water accumulation detection method in this embodiment is essentially a substation water accumulation identification method based on the CBAM-YOLOv8 model. This method solves the technical challenges of low detection accuracy for small-area water accumulation in complex backgrounds and reliance on manual calibration for area calculation by integrating multimodal data augmentation, attention mechanism optimization, and geographic information mapping techniques. This enables real-time detection and automated early warning of substation flooding risks. The method includes several parts: model construction, data augmentation, coordinate mapping calibration, water accumulation detection and calculation, and tiered early warning. Specifically:

[0073] The CBAM-YOLOv8 model was used to determine the water accumulation area of ​​the substation. The model construction involved embedding the CBAM attention module after the Backbone layer C3 module and the Neck layer FPN module feature pyramid network of the YOLOv8 model network architecture. Channel attention weight calculation and spatial attention mask generation were performed sequentially to enhance the water accumulation texture features and suppress the reflection interference of substation equipment.

[0074] Data augmentation refers to the dynamic enhancement processing of visible light images of substations, including simulating reflections on the metal surfaces of substation equipment (i.e., substation equipment reflection simulation), generating synthetic noise from rain and fog obstructions (i.e., dynamic rain and fog interference simulation), and Mosaic (a data augmentation method based on image mixing) four-image stitching data augmentation (i.e., enhancement processing of small water accumulation areas).

[0075] Coordinate mapping calibration refers to establishing an affine transformation matrix from image pixel coordinates to the geographic coordinate system based on pre-deployed substation ground calibration board or lidar point cloud data.

[0076] Water accumulation detection and calculation refers to using a trained CBAM-YOLOv8 model (i.e., a target water accumulation area prediction model) to output the bounding box of the water accumulation area, including the pixel vertex coordinates of the water accumulation area. The pixel area of ​​the water accumulation area is calculated using a polygon vertex integral algorithm, and then mapped to the water accumulation area detection result in a geographic coordinate system using an affine transformation matrix.

[0077] Tiered early warning refers to the use of pre-set water depth and water area thresholds to link the substation's SCADA system with the OPC protocol to issue water accumulation warnings and push warning information to the substation, and trigger drainage equipment to carry out drainage treatment.

[0078] Figure 2 This is a structural block diagram of an optional substation water accumulation detection method provided according to an embodiment of this application, such as... Figure 2 As shown, the steps of this method include:

[0079] Step S1, Data Acquisition and Enhancement.

[0080] Step S11: Multi-source data acquisition.

[0081] Wide-angle cameras are used to acquire visible light images of the substation, while infrared thermal imagers are used to acquire infrared images. The wide-angle cameras are 4K resolution cameras with a horizontal field of view ≥120° (meaning a resolution of approximately 4000 horizontal pixels). These wide-angle cameras are deployed in the substation's power distribution room, cable trench entrance, and on top of the transformer base, covering a radius ≥15 meters. The infrared thermal imager uses an uncooled infrared sensor with a wavelength of 8-14μm (micrometers) and a temperature sensitivity ≤0.05℃. It is coaxially mounted with the visible light camera to detect the thermal radiation characteristics of accumulated water through rain and fog.

[0082] When deploying wide-angle cameras and infrared thermal imagers, the latitude, longitude, and elevation data of the camera installation points are recorded using an RTK positioning device.

[0083] Since the water accumulation detection results of the water accumulation area need to be determined by combining information from visible light images and infrared images, a spatiotemporal synchronization mechanism is required to synchronize and align the acquired visible light images and infrared images in time and space.

[0084] For time synchronization, the camera and RTK positioning device achieve millisecond-level time synchronization through the NTP protocol, and each frame of image is appended with a timestamp (accurate to milliseconds) and the corresponding geographic coordinates.

[0085] For spatial synchronization alignment, the ORB-SLAM feature point matching algorithm is used to align the infrared image and the visible light image, and RANSAC is used to remove mismatched points to ensure that the matching error is ≤2 pixels.

[0086] Step S12, data augmentation processing.

[0087] The historical visible light image of the substation area is obtained by step S11. The historical visible light image is then augmented by substation equipment reflection simulation, dynamic rain and fog interference simulation, and small-area water accumulation area enhancement processing.

[0088] For simulating reflections on substation equipment, CycleGAN was first used to generate a reflection texture library for 12 types of substation equipment surface materials (galvanized steel, ceramic insulators, silicone rubber bushings, etc.), including reflection templates under different lighting angles (0°-180°). Secondly, reflection phenomena in the actual environment were simulated to improve the model's ability to handle this type of interference.

[0089] Next, a Poisson fusion overlay method is used to overlay the reflective template onto the historical visible light image with a transparency α = 0.4. The fusion process is the same as in the above embodiment and will not be described again here.

[0090] For dynamic rain and fog interference simulation, including raindrop particle generation and rain and fog blurring processing.

[0091] For raindrop particle generation, the raindrop trajectory is simulated using the NVIDIA Flow physics engine (a fluid dynamics simulation tool mainly used to generate realistic fluid effects such as water, smoke, and fire). The raindrop density λ can be set to 0.2 drops / pixel. 2 (This means there are expected to be 0.2 raindrop particles per square pixel area, corresponding to moderate rain intensity). The raindrop particle size follows a normal distribution (mean 3 pixels, standard deviation 0.5), where droplets represent raindrops and pixel... 2 This represents the area per unit pixel.

[0092] For rain and fog blurring, a Gaussian blur kernel (3×3, representing the size of the Gaussian blur kernel; ε=1.5, representing the standard deviation of the Gaussian blur kernel) is added to the historical visible light image, and salt-and-pepper noise (noise density 0.1) is superimposed to simulate the rain and fog scattering effect. The core of Gaussian blur is to use a Gaussian function as weights to perform a weighted average of each pixel in the image and its surrounding pixels, which is used to reduce image noise, smooth image edges, or generate certain visual effects.

[0093] Mosaic four-image stitching and CutMix enhancement were used to enhance small water accumulation areas. Mosaic four-image stitching involves randomly selecting four historical visible light images, stitching them together in a 2×2 grid, and ensuring at least two small water accumulation areas (pixel area < 50×50). CutMix enhancement involves cropping small water accumulation areas from other historical visible light images and pasting them into the current historical visible light image at random positions and rotation angles (±15°), with an occlusion ratio ≤ 30%.

[0094] Step S2, CBAM-YOLOv8 model improvement scheme.

[0095] The CBAM-YOLOv8 model was trained using historical visible light images. This trained model was then used as a prediction model for water accumulation areas in substations. Improvements to the CBAM-YOLOv8 model included optimized attention mechanism design and enhanced small target detection techniques.

[0096] A CBAM module is connected in series at the output of the C3 module (layers 2, 4, and 6) in the backbone layer, with the channel attention compression ratio set to 8:1 (i.e., the fully connected layer dimension is reduced from C to C / 8). The channel attention weights M... c The method for determining (F) is the same as in the above embodiments, and will not be repeated here.

[0097] A spatial attention module is inserted before the P3 and P4 feature map outputs of the FPN module in the Neck layer, and a 7×7 convolutional kernel is used to extract local texture features.

[0098] A dynamic threshold activation mechanism is used to adjust the feature activation threshold. This mechanism suppresses false activation of reflective areas (mistakenly identified as water accumulation areas) on substation equipment, reducing the false detection rate by 12.7%. The feature activation threshold is based on the spatial attention weight M. s (F) Dynamic adjustment is performed. The method for determining the feature activation threshold T is the same as in the above embodiments, and will not be repeated here.

[0099] The original FPN is replaced with a bidirectional feature pyramid network, connecting layers P3 (80×80), P4 (40×40), and P5 (20×20) across scales, and image features are fused using learnable weights. High-resolution features from layer P3 are prioritized for small water accumulation areas. The weight w of the feature map in the i-th layer during the fusion process... i The method for determining the value is the same as in the above embodiments, and will not be repeated here.

[0100] GIoU Loss is used to determine the model's loss function and optimize the bounding box regression process. The loss function L... GIoU The method for determining the bounding box localization error (IoU error) is the same as in the above embodiments, and will not be repeated here. Experiments show that GIoULoss reduces the bounding box localization error (IoU error) of small water accumulation areas by 18.3%.

[0101] Step S3: Calculation and geographic mapping of water accumulation area.

[0102] The calculation and geographic mapping of water accumulation area includes pixel-to-geographic coordinate transformation and dynamic water accumulation area calculation process.

[0103] Before performing pixel-to-geographic coordinate conversion, calibration boards need to be deployed first. The deployment process is as follows: four sets of 1m×1m rectangular calibration boards are placed on the substation ground. The material is high-reflectivity aluminum plate (reflectivity ≥85%), and the spacing error is ≤1cm. The corner coordinates of the calibration boards are obtained by measuring with a total station with an accuracy of ±0.5mm.

[0104] To convert pixel coordinates to geographic coordinates, the PnP algorithm is used to determine the extrinsic parameter matrix [R|t] of the wide-angle camera acquiring visible light images, where R represents the rotation matrix and t represents the translation vector, which are key parameters describing the position and orientation of the wide-angle camera in three-dimensional space. An affine transformation matrix H is constructed based on the extrinsic parameter matrix [R|t], thereby converting the pixel coordinates (u,v) to geographic coordinates (X,Y,Z). The determination of the geographic coordinates (X,Y,Z) is the same as in the previous embodiment and will not be repeated here.

[0105] The parameters of the affine transformation matrix are optimized using the Levenberg-Marquardt algorithm so that the reprojection error (the difference between the theoretical image point position and the observed actual image point position) is ≤1.5 pixels.

[0106] To calculate the area of ​​the water accumulation, the coordinates of n pixel vertices {(x1, y1), ..., (x...} of the polygon of the water accumulation region are extracted from the CBAM-YOLOv8 detection results. n ,y n Based on the above pixel vertex coordinates, calculate the pixel area S of the water accumulation region. pixel S pixel The method for determining the value is the same as in the above embodiments, and will not be repeated here.

[0107] The pixel area is converted into the water accumulation area detection result S in the geographic coordinate system through the affine transformation matrix H. real S real The method for determining the value is the same as in the above embodiments, and will not be repeated here.

[0108] Based on the above water accumulation area detection results, if the water accumulation area growth rate is ≥10% / minute for 3 consecutive frames, it is marked as a trend of waterlogging risk in the substation, triggering a waterlogging risk warning and linking the SCADA system to control the start and stop of the drainage pumps.

[0109] Step S4: Water accumulation warning for the substation.

[0110] A tiered early warning mechanism is adopted for substation flooding warnings. Different flooding levels have corresponding flooding depth and flooding area thresholds. Based on the obtained flooding area and depth detection results, combined with the corresponding flooding depth and area thresholds, the flooding level of the substation is determined, and flooding warnings are issued accordingly. In this embodiment, flooding levels are divided into warning levels and danger levels. If the flooding area is ≥5㎡ and the depth is ≥10cm (estimated from infrared images), the flooding level is considered a warning level, triggering a yellow warning and sending an SMS to substation maintenance personnel. If the flooding area is ≥10㎡ and the depth is ≥20cm, the flooding level is considered a danger level, triggering a red warning. The substation will automatically start drainage pumps and activate the SCADA system to cut off power to high-risk areas.

[0111] The above-mentioned optional implementation methods achieve at least the following effects: By embedding the CBAM module into the YOLOv8 model, the accuracy of water accumulation detection results is significantly improved, the model's adaptability to complex substation environments is enhanced, and the calculation efficiency for water accumulation area is increased; through accurate vertex coordinate prediction, efficient pixel area calculation, and accurate geographic coordinate system transformation, rapid and accurate detection of substation water accumulation area can be achieved, improving substation safety and emergency response efficiency; by performing depth analysis on the current infrared image to determine the water accumulation depth detection results in the target area where the substation is located, combined with water accumulation area detection from visible light images, the accuracy and reliability of substation water accumulation detection results can be improved, enabling accurate assessment and timely early warning of substation flooding risk.

[0112] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0113] This embodiment also provides a substation water accumulation detection device, which is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the terms "module" and "device" can refer to a combination of software and / or hardware that performs a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, hardware implementations, or a combination of software and hardware, are also possible and contemplated.

[0114] According to an embodiment of this application, an embodiment of an apparatus for implementing a method for detecting water accumulation in a substation is also provided. Figure 3 This is a schematic diagram of a water accumulation detection device for a substation according to an embodiment of this application, as shown below. Figure 3 As shown, the water accumulation detection device for the substation includes an image acquisition module 302, a water accumulation area determination module 304, a water accumulation depth determination module 306, and a water accumulation detection result determination module 308. The device will be described below.

[0115] Image acquisition module 302 is used to acquire the current visible light image and the current infrared image of the target area where the substation is located;

[0116] The water accumulation area determination module 304 is connected to the image acquisition module 302 and is used to determine the water accumulation area detection result of the water accumulation area in the target area based on the current visible light image.

[0117] The water depth determination module 306 is connected to the water area determination module 304 and is used to determine the water depth detection result of the water area based on the current infrared image.

[0118] The water accumulation detection result determination module 308 is connected to the water accumulation depth determination module 306 and is used to obtain the water accumulation detection result of the target area based on the water accumulation area detection result and the water accumulation depth detection result. The water accumulation detection result is used to indicate at least whether there is an abnormal water accumulation in the target area.

[0119] In a substation water accumulation detection device provided in this application embodiment, an image acquisition module 302 is set up to acquire the current visible light image and the current infrared image of the target area where the substation is located; a water accumulation area determination module 304 is connected to the image acquisition module 302 and is used to determine the water accumulation area detection result of the water accumulation area in the target area based on the current visible light image; a water accumulation depth determination module 306 is connected to the water accumulation area determination module 304 and is used to determine the water accumulation depth detection result of the water accumulation area based on the current infrared image; a water accumulation detection result determination module 308 is connected to the water accumulation depth determination module 306 and is used to obtain the water accumulation detection result of the target area based on the water accumulation area detection result and the water accumulation depth detection result, wherein the water accumulation detection result is at least used to indicate whether there is an abnormal water accumulation in the target area. This invention aims to determine the water accumulation area of ​​a substation based on visible light images and the water depth of a substation based on infrared images. It then uses these water accumulation area and depth results to determine whether there is any abnormal water accumulation in the substation area. This improves the accuracy of water depth and area detection results for substation areas, thereby solving the technical problem of inaccurate water depth and area detection results in related technologies.

[0120] It should be noted that the above modules can be implemented by software or hardware. For example, for the latter, it can be implemented in the following ways: the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0121] It should be noted that the image acquisition module 302, the water accumulation area determination module 304, the water accumulation depth determination module 306, and the water accumulation detection result determination module 308 mentioned above correspond to steps S102 to S108 in the embodiments. The instances and application scenarios implemented by the above modules and corresponding steps are the same, but are not limited to the content disclosed in the above embodiments. It should be noted that the above modules, as part of the device, can run in a computer terminal.

[0122] It should be noted that the optional or preferred implementation methods of this embodiment can be found in the relevant descriptions in the embodiments, and will not be repeated here.

[0123] The aforementioned substation water accumulation detection device may also include a processor and a memory. The image acquisition module 302, water accumulation area determination module 304, water accumulation depth determination module 306, and water accumulation detection result determination module 308 are all stored as program units in the memory. The processor executes the aforementioned program units stored in the memory to realize the corresponding functions.

[0124] The processor contains a core that retrieves the corresponding program unit from memory. One or more cores may be configured. Memory may include non-persistent memory in computer-readable media, random access memory (RAM), and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory includes at least one memory chip.

[0125] This application provides a non-volatile storage medium storing a program that, when executed by a processor, implements a method for detecting water accumulation in a substation.

[0126] This application provides an electronic device including a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it performs the following steps: acquiring a current visible light image and a current infrared image of the target area where the substation is located; determining the water area detection result of the water accumulation region in the target area based on the current visible light image; determining the water depth detection result of the water accumulation region based on the current infrared image; and obtaining the water accumulation detection result of the target area based on the water area detection result and the water depth detection result. The water accumulation detection result at least indicates whether there is an abnormal water accumulation in the target area. The device described herein may be a server, PC, etc.

[0127] This application also provides a computer program product, which, when executed on a data processing device, is suitable for executing an initialization program having the following method steps: acquiring a current visible light image and a current infrared image of the target area where the substation is located; determining the water area detection result of the water accumulation area in the target area based on the current visible light image; determining the water depth detection result of the water accumulation area based on the current infrared image; and obtaining the water accumulation detection result of the target area based on the water area detection result and the water depth detection result, wherein the water accumulation detection result is at least used to indicate whether there is an abnormal water accumulation in the target area.

[0128] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0129] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0130] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0131] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0132] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0133] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0134] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0135] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0136] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0137] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for detecting water accumulation in a substation, characterized in that, include: Acquire current visible light and infrared images of the target area where the substation is located; Based on the current visible light image, determine the water area detection result of the water accumulation area in the target region; Based on the current infrared image, determine the water depth detection result of the water accumulation area; Based on the water accumulation area detection results and the water accumulation depth detection results, the water accumulation detection results of the target area are obtained, wherein the water accumulation detection results are at least used to indicate whether there is an abnormal water accumulation in the target area.

2. The method according to claim 1, characterized in that, When there are multiple water accumulation areas, the step of determining the water accumulation area detection result of the water accumulation area in the target area based on the current visible light image includes: Based on the current visible light image, a target water accumulation area prediction model is used to determine the pixel vertex coordinates corresponding to multiple water accumulation areas. The target water accumulation area prediction model learns in advance the correspondence between the visible light image of the target area and the pixel vertex coordinates of the water accumulation area. The pixel vertex coordinates are used to describe the vertex position of the water accumulation area in the image coordinate system. Based on the pixel vertex coordinates corresponding to the multiple water accumulation areas, the pixel area corresponding to each of the multiple water accumulation areas is determined. The pixel areas corresponding to the multiple water accumulation areas are transformed into coordinate systems to obtain the water accumulation area detection results corresponding to the multiple water accumulation areas in the geographic coordinate system.

3. The method according to claim 2, characterized in that, Before determining the pixel vertex coordinates corresponding to multiple water accumulation areas based on the current visible light image and using a target water accumulation area prediction model, the method further includes: Acquire historical visible light images of the target area; The historical visible light image is subjected to data enhancement processing to obtain the target visible light image. The data enhancement processing includes at least one of the following: substation equipment reflection simulation, dynamic rain and fog interference simulation, and small area water accumulation area enhancement processing. The small area water accumulation area refers to a water accumulation area with an area smaller than a preset area threshold. Based on the target visible light image, a pre-determined initial water accumulation area prediction model is trained to obtain the target water accumulation area prediction model.

4. The method according to claim 2, characterized in that, The step of transforming the pixel areas corresponding to the multiple water accumulation areas into coordinate systems to obtain the water accumulation area detection results corresponding to the multiple water accumulation areas in the geographic coordinate system includes: Determine the extrinsic parameter matrix of the current visible light image acquisition device, wherein the extrinsic parameter matrix is ​​used to establish the geometric relationship between the image coordinate system and the geographic coordinate system; Based on the extrinsic parameter matrix, an affine transformation matrix is ​​determined, wherein the affine transformation matrix is ​​used to map points in the image coordinate system to the geographic coordinate system; Based on the pixel area corresponding to each of the multiple water accumulation areas and the affine transformation matrix, the water accumulation area detection result corresponding to each of the multiple water accumulation areas is determined.

5. The method according to claim 1, wherein When there are multiple water accumulation areas, determining the water depth detection result of each water accumulation area based on the current infrared image includes: Based on the current infrared image, the current temperature distribution corresponding to multiple water accumulation areas is determined, wherein the current temperature distribution is used to describe the temperature status of multiple temperature measurement points in the water accumulation area; The actual temperature distribution corresponding to each of the multiple waterlogged areas is compared with the corresponding current temperature distribution to obtain the temperature difference distribution corresponding to each of the multiple areas. The actual temperature distribution is measured under the condition that there is no waterlogging. Based on the temperature difference distribution corresponding to the multiple water accumulation areas, the water depth detection results corresponding to the multiple water accumulation areas are determined.

6. The method according to any one of claims 1 to 5, characterized in that, The process of obtaining the water accumulation detection results for the target area based on the water accumulation area detection results and the water accumulation depth detection results includes: Based on the water accumulation area detection results, the water accumulation depth detection results, the preset water accumulation area threshold, and the preset water accumulation depth threshold, the target water accumulation level of the water accumulation area is determined. If the target water accumulation level is greater than a preset level, the water accumulation detection result is determined to indicate that there is an abnormal water accumulation in the target area; or If the target water accumulation level is less than or equal to the preset level, the water accumulation detection result is determined to be that there is no abnormal water accumulation in the target area.

7. The method according to claim 1, characterized in that, The process of obtaining the water accumulation detection results for the target area based on the water accumulation area detection results and the water accumulation depth detection results includes: Determine a first weight value corresponding to the water accumulation area detection result and a second weight value corresponding to the water accumulation depth detection result; Based on the first weight value and the second weight value, the water accumulation score of the water accumulation area is determined; The target scoring interval for the water accumulation score is determined from multiple scoring intervals, wherein the multiple scoring intervals correspond one-to-one with multiple water accumulation levels; The water accumulation level corresponding to the target scoring interval is determined as the target water accumulation level of the water accumulation area; If the target water accumulation level is greater than a preset level, the water accumulation detection result is determined to indicate that there is an abnormal water accumulation in the target area; or If the target water accumulation level is less than or equal to the preset level, the water accumulation detection result is determined to be that there is no abnormal water accumulation in the target area.

8. A water accumulation detection device for a substation, characterized in that, include: The image acquisition module is used to acquire the current visible light image and the current infrared image of the target area where the substation is located; The water accumulation area determination module is used to determine the water accumulation area detection result of the water accumulation area in the target area based on the current visible light image; The water depth determination module is used to determine the water depth detection result of the water accumulation area based on the current infrared image; The water accumulation detection result determination module is used to obtain the water accumulation detection result of the target area based on the water accumulation area detection result and the water accumulation depth detection result, wherein the water accumulation detection result is used to at least indicate whether there is an abnormal water accumulation in the target area.

9. A non-volatile storage medium, characterized in that, The non-volatile storage medium stores multiple instructions, which are adapted to be loaded by a processor and executed by the substation water accumulation detection method according to any one of claims 1 to 7.

10. An electronic device, characterized in that, include: One or more processors and a memory, the memory being used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the substation water accumulation detection method according to any one of claims 1 to 7.

Images