Suspension insulator deicing robot control method and system

Abstract

The invention discloses a suspension insulator deicing robot control method and system, and relates to the technical field of deicing robots, and the method comprises the steps: selecting a deicing strategy according to the detected icing state of an insulator string and an ice bridge risk judgment result; under the selected deicing strategy, according to the icing state and the operation state of the robot, the propelling pressure, the propelling speed, the propelling distance and the lifting adjustment amount used for controlling operation of the robot are set; and the extrusion force applied to the insulator string by the robot in the operation process, the operation area image and the actual deicing time are obtained, so that the operation parameters of the robot are fed back and adjusted. The icing state of the insulator is automatically identified, and the operation strategy is optimized according to different icing conditions. And through high-precision positioning, force control and visual feedback technologies, high efficiency, precision and safety of deicing operation are ensured.

Description

Technical Field

[0001] This invention relates to the field of de-icing robot technology, and in particular to a control method and system for a suspended insulator de-icing robot. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] With the widespread application of power in the industry, especially in high-voltage transmission lines, suspension insulators frequently encounter frost accumulation under severe weather conditions, forming thick layers of ice. Severe icing not only affects the electrical performance of the insulators but can also lead to insulator breakage or equipment failure, seriously impacting the stability and safety of the power system.

[0004] Current de-icing methods often employ fixed strategies, such as constant pressure and uniform speed. This can easily lead to excessive compression that damages insulators or incomplete de-icing, leaving residual ice layers that still pose safety hazards. Moreover, focusing solely on whether the ice layer is removed does not consider the ice bridges formed between insulator strings (ice layers connecting adjacent insulators). Blindly applying pressure may cause the impact force when the ice bridge breaks to damage the insulator enamel surface, or cause the ice bridge to collapse, leading to conductor swaying and increasing the risk of wire breakage.

[0005] Secondly, during the de-icing process, the robot's propulsion mechanism may experience mechanical wear and environmental vibrations, causing the actual extrusion pressure to deviate from the set value. Robot positioning errors may also cause the robot to deviate from the center line of the insulator during operation. This can easily lead to uncontrolled extrusion pressure, damage to the insulator, incomplete de-icing, robot deviation, or de-icing time exceeding the limit, thus affecting the overall de-icing progress of the transmission line. Summary of the Invention To address the aforementioned issues, this invention proposes a control method and system for a suspended insulator de-icing robot. This system automatically identifies the icing state of the insulators, optimizes the operational strategy based on different icing conditions, and ensures the efficiency, accuracy, and safety of the de-icing operation through high-precision positioning, force control, and visual feedback technologies.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a control method for a suspended insulator de-icing robot, comprising: Based on the detected icing status of the insulator strings and the assessment of ice bridge risk, a de-icing strategy is selected. Under the selected de-icing strategy, the propulsion pressure, propulsion speed, propulsion distance, and lifting adjustment amount are set according to the icing state and robot operating state to control the robot's operation. The robot acquires data on the squeezing force applied to the insulator string, images of the work area, and the actual de-icing time during the operation, and uses this data to adjust the robot's operating parameters accordingly. Among them, when the pressure deviation between the extrusion pressure and the propulsion pressure is not less than the set pressure threshold, the amount of retraction of the robot propulsion is determined according to the pressure deviation. Extract the pixel deviation of the insulator string centerline in the working area image. When the pixel deviation is not less than the set pixel threshold, determine the lateral compensation amount based on the pixel deviation to adjust the horizontal movement of the robot. When the actual de-icing time is not less than the expected de-icing time, the enhanced extrusion pressure is determined based on the propulsion pressure, and the propulsion speed and propulsion distance are updated according to the enhanced extrusion pressure.

[0007] As an alternative implementation method, the process of selecting a de-icing strategy includes: The severity of icing is determined by comparing the icing thickness with a preset first thickness threshold and a preset second thickness threshold, and by combining the comparison of the robot motor load change over a set time period with a preset load threshold; wherein the first thickness threshold is less than the second thickness threshold. When the icing severity is mild and the ice bridge risk assessment result indicates the presence of bridging ice floes, a light-load compression strategy is selected. When the icing severity is moderate and no ice bridge risk is detected, the standard compression strategy is selected. When the icing severity is severe icing, or when the pressure feedback of a set number of consecutive compressions fails to achieve the desired de-icing effect, a powerful de-icing strategy is activated. When the ice bridge risk assessment result indicates that there are ice floes exceeding the predetermined bridging risk, a bridging risk handling strategy is adopted.

[0008] As an alternative implementation method, the process of determining the severity of icing includes: When the ice thickness is less than the first thickness threshold and the load change is less than the load threshold, it is judged as light icing; When the icing thickness is not less than the first thickness threshold and less than the second thickness threshold, or when the load change is greater than or equal to the load threshold, it is judged as moderate icing. When the ice thickness is not less than the second thickness threshold, or when the collected target area image contains continuous ice bodies exceeding the set area, it is determined to be heavily iced.

[0009] As an alternative implementation method, the process of setting the propulsion pressure, propulsion speed, propulsion distance, and lifting adjustment amount includes: The propulsion pressure is derived from the base pressure F0, icing thickness H, and motor load L: F target =F0+k h ·H+k l·L; Based on real-time pressure deviation e F = F target –F real The propulsion speed v is obtained as: v = v0 – k f ·e F ; The advancing distance D is obtained based on the initial distance D0 and the ice thickness H: D = D0 + k d ·H; The elevation adjustment Δh is obtained based on the visual center deviation Δpos and the IMU attitude deviation Δθ: Δh = k c ·Δpos+k θ ·Δθ; where k h k l k c k θ k f k d All are coefficients, F real This represents the actual extrusion pressure.

[0010] As an alternative implementation, the retraction amount Δd is Δd=k ret ·e F , where k ret e is the shrinkage gain coefficient. F This is the pressure deviation; the number of pulses of the forward thrust mechanism servo motor is reduced according to the retraction amount to achieve retraction, thereby reducing the thrust.

[0011] As an alternative implementation, when the absolute value of the pixel deviation is not less than a set pixel threshold, it is determined that the positional deviation between the working tool and the insulator exceeds the limit. The lateral compensation amount determined based on the pixel deviation Δu is Δx: Δx = k u ·Δu·S d , where k u S is the pixel compensation coefficient. d The depth scale factor represents the displacement that maps pixel deviations to actual space, thereby adjusting the robot's horizontal movement.

[0012] As an alternative implementation method, the desired de-icing time T is determined based on the current ice thickness, propulsion speed, and pressure application time. exp ; In pushing pressure F target The enhanced extrusion pressure F was determined based on this. enh For: F enh =F target +k enh · (T real -T exp ), where k enh To enhance the intensity coefficient, Treal This refers to the actual de-icing time.

[0013] As an alternative implementation method, the process of detecting the icing status of insulator strings and the results of ice bridge risk assessment includes: A pre-trained insulator string detection model is used to identify insulator string images under icing conditions, and to locate the position and icing status of the insulator strings. The pre-trained insulator disc recognition model is used to identify two adjacent insulator discs in an image. If there are multiple insulator discs in the image field of view, the two insulator discs in the center of the image are used as the recognition result. Using a pre-trained icicle detection algorithm model, it is determined whether there is still icicle between the two identified insulators. The ratio of the pixel length of the icicle to the pixel length of the distance between the two insulators is calculated. Based on the comparison of the ratio and a predetermined threshold, it is determined whether there is bridging icicle or icicle with a risk exceeding the predetermined bridging threshold.

[0014] As an alternative implementation, the method further includes: using a two-layer wireless communication architecture to realize remote communication between the robot and the ground control system; specifically including: LoRa communication as a long-distance communication link between the robot's lifting module and de-icing module; and SBUS protocol for multi-channel real-time control between the ground control system and the robot to control various actions of the robot.

[0015] Secondly, the present invention provides a control system for a suspended insulator de-icing robot, comprising: The selection module is configured to select a de-icing strategy based on the detected icing status of the insulator strings and the results of the ice bridge risk assessment. The control module is configured to set the propulsion pressure, propulsion speed, propulsion distance, and lifting adjustment amount for controlling the robot's operation based on the icing state and robot operating state under the selected de-icing strategy. The feedback module is configured to acquire the squeezing force applied by the robot to the insulator string, the image of the work area, and the actual de-icing time during the operation, so as to adjust the robot's operation parameters accordingly. Among them, when the pressure deviation between the extrusion pressure and the propulsion pressure is not less than the set pressure threshold, the amount of retraction of the robot propulsion is determined according to the pressure deviation. Extract the pixel deviation of the insulator string centerline in the working area image. When the pixel deviation is not less than the set pixel threshold, determine the lateral compensation amount based on the pixel deviation to adjust the horizontal movement of the robot. When the actual de-icing time is not less than the expected de-icing time, the enhanced extrusion pressure is determined based on the propulsion pressure, and the propulsion speed and propulsion distance are updated according to the enhanced extrusion pressure.

[0016] Thirdly, the present invention provides an electronic device including a memory and a processor, and computer instructions stored in the memory and running on the processor, wherein the computer instructions, when executed by the processor, perform the method described in the first aspect.

[0017] Fourthly, the present invention provides a computer-readable storage medium for storing computer instructions, which, when executed by a processor, perform the method described in the first aspect.

[0018] Fifthly, the present invention provides a computer program product, including a computer program that, when executed by a processor, implements the method described in the first aspect.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention innovatively proposes a dynamic matching technology for de-icing strategies of a suspended insulator de-icing robot, and develops a control system for the suspended insulator de-icing robot. By accurately judging the icing state of the insulator string and the risk of ice bridges, the de-icing strategy is selected, achieving a high degree of matching between the de-icing method and the actual icing situation. Simultaneously, under the determined de-icing strategy, key parameters such as propulsion pressure, propulsion speed, propulsion distance, and lifting adjustment are precisely set according to the icing state and robot operating status, laying the foundation for efficient robot operation. Furthermore, through force control and visual feedback technology, feedback adjustment of operating parameters is achieved, solving the problems of inaccurate matching of de-icing strategies for suspended insulators in complex high-altitude environments, and the difficulty in balancing operational efficiency and safety. This enables automated, safe, and efficient de-icing operations in complex high-altitude environments, improving the accuracy of de-icing strategy matching, the accuracy of operating parameter control, and the safety of high-altitude operations.

[0020] This invention innovatively proposes a feedback adjustment operation technology for a suspended insulator de-icing robot. By acquiring information such as the squeezing force applied to the insulator string, images of the work area, and the actual de-icing time during the robot's operation in real time, the robot's operating parameters are adjusted accordingly. This solves the problems of poor stability, insufficient adaptability, and easy damage to transmission lines by de-icing robots in complex and variable environments. It achieves adaptive and stable operation of the robot for different icing conditions and working environments, reduces the risk of additional damage to transmission lines and the cost of manual intervention, and improves the reliability, environmental adaptability, and operational safety of de-icing operations. This invention innovatively provides an intelligent identification method for insulator position and icing status. By fusing RGB color images and depth information, the fused information is processed using an insulator string position detection model. Furthermore, during the training of the insulator string position detection model, various types of image data are introduced, such as uniced steel caps or core rods, lightly iced steel caps or core rods, and heavily iced steel caps or core rods, as well as image data categorized by different icing statuses, such as slightly iced insulator strings, moderately iced insulator strings, and heavily iced insulator strings. This significantly improves the accuracy of insulator string position detection and filters out the influence of sky background, distant targets, etc., effectively reducing the impact of complex field environments and severe weather conditions.

[0021] This invention innovatively provides an icicle detection algorithm that determines whether there is still icicle between two identified insulators, calculates the ratio of the pixel length of the icicle to the pixel length of the distance between the two insulators, and determines whether there is bridging icicle or icicle with a risk exceeding a predetermined bridging threshold based on the ratio. This assists the robot in performing secondary de-icing, achieving pixel-level segmentation of insulator icicles and guiding the robot to thoroughly remove high-risk icicles.

[0022] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0024] Figure 1 This is a flowchart of the control method for the suspension insulator de-icing robot provided in Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of a two-layer wireless communication architecture provided in Embodiment 1 of the present invention; Figure 3 This is a schematic diagram of the system functional architecture provided in Embodiment 1 of the present invention. Detailed Implementation

[0025] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0026] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0027] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form as well. Furthermore, it should be understood that the terms “comprising” and “including”, and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or apparatus that includes 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.

[0028] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0029] Example 1 like Figure 1 As shown, this embodiment provides a control method for a suspension insulator de-icing robot, which specifically includes the following process: Based on the detected icing status of the insulator strings and the assessment of ice bridge risk, a de-icing strategy is selected. Under the selected de-icing strategy, the propulsion pressure, propulsion speed, propulsion distance, and lifting adjustment amount are set according to the icing state and robot operating state to control the robot's operation. The robot acquires data on the squeezing force applied to the insulator string, images of the work area, and the actual de-icing time during the operation, and uses this data to adjust the robot's operating parameters accordingly. Among them, when the pressure deviation between the extrusion pressure and the propulsion pressure is not less than the set pressure threshold, the amount of retraction of the robot propulsion is determined according to the pressure deviation. Extract the pixel deviation of the insulator string centerline in the working area image. When the pixel deviation is not less than the set pixel threshold, determine the lateral compensation amount based on the pixel deviation to adjust the horizontal movement of the robot. When the actual de-icing time is not less than the expected de-icing time, the enhanced extrusion pressure is determined based on the propulsion pressure, and the propulsion speed and propulsion distance are updated according to the enhanced extrusion pressure.

[0030] In this embodiment, the process of detecting the icing status of the insulator string and the result of the ice bridge risk assessment includes: A pre-trained insulator string detection model is used to identify images of insulator strings under icing conditions, locate the position and icing status of the insulator strings, and assist the robot in adjusting its posture. During the training process, RGB three-channel visible light image information and depth image information are fused together. The pre-trained insulator disc recognition model is used to identify two adjacent insulator discs in an image. If there are multiple insulator discs in the image field of view, the two insulator discs in the center of the image are used as the recognition result. Using a pre-trained icicle detection algorithm model, it is determined whether there is still icicle between the two identified insulators. The ratio of the pixel length of the icicle to the pixel length of the distance between the two insulators is calculated. Based on the comparison of the ratio and a predetermined threshold, it is determined whether there is bridging icicle or icicle with a risk exceeding the predetermined bridging threshold.

[0031] Specifically: RGB images and depth maps of the target area of ​​the insulator under different icing conditions are simultaneously acquired by an RGBD camera or a binocular camera mounted on a suspended insulator de-icing robot (hereinafter referred to as the robot). The depth map is then distorted using camera intrinsic parameters and aligned to the RGB image coordinate system. Subsequently, based on the RGB three-channel image, the depth map is directly stitched together as the fourth channel to construct an RGBD four-channel fused image. The fused four-channel image can simultaneously provide texture information and spatial geometric information.

[0032] Therefore, based on the fused images, the constructed insulator string detection model was trained. Specifically, insulator string images under different icing conditions were labeled and classified into lightly iced, moderately iced, and severely iced insulator strings, constructing a sample library of insulator strings under different icing conditions. The YOLOv11 algorithm was then used to train the insulator string detection model. Finally, the trained insulator string detection model was used to identify insulator string images under icing conditions, locating the position and icing state of the insulator strings. Simultaneously, image ranging software was used to calculate the icing thickness.

[0033] Next, images containing the two insulator regions are acquired, and the trained model is used to detect the two insulators. If there are multiple insulator pieces in the image field of view, the two insulators in the center of the image are used as the identification results. Then, the ice detection algorithm model is used to determine whether there is still ice between the two insulators, and the ratio of the pixel length of the ice to the pixel length of the distance between the two insulators is calculated. A threshold is set to determine whether there is still bridging ice or ice with a significant risk of bridging.

[0034] In this embodiment, the YOLOv11 deep learning model is used to improve the accuracy and robustness of ice detection and localization. To further improve recognition accuracy and speed, the upsampling method used by YOLOv11 is improved. YOLOv11 uses a combination of standard upsampling and 3x3 convolution for feature matching and enhancement, which has low computational efficiency, limiting its deployment on edge devices. Furthermore, the simple upsampling method can easily lead to feature blurring, affecting the feature fusion effect.

[0035] This embodiment improves the upsampling method and designs an upsampling-feature enhancement-channel matching processing flow: In terms of upsampling, bilinear interpolation with a scaling factor of 2 is used to expand the feature map to twice its original scale, and the primary matching skips the next level of feature resolution. In terms of feature enhancement, 3x3 depthwise convolution (DWC) is first used to extract local features, and then batch normalization (BN) and ReLU activation are performed to introduce non-linear features, thereby efficiently enhancing the feature map without significantly increasing computational overhead. In terms of channel matching, a 1x1 point convolution is adopted to reduce the number of channels and adjust the number of channels to be consistent with the next level of jump features. The above method enhances the information fusion between different levels and stages, which is more conducive to target detection and segmentation.

[0036] The specific mathematical formula can be expressed as: .

[0037] If the recognition model fails to recognize the insulator due to factors such as insulator disc occlusion during verification, the Canny edge detection algorithm is used to extract the edge of the image by utilizing the arc feature of the insulator disc edge. The least squares method or RANSAC algorithm is then used to fit the arc of the insulator edge to detect the edge of the insulator disc, which serves as the location information of the insulator disc.

[0038] For the ice detection algorithm model, the image is first preprocessed. In addition to commonly used enhancements such as rotation, contrast transformation, and affine transformation, Gaussian blur is added to the image to simulate the image defocusing caused by slight shaking of the robot gimbal. Salt and pepper noise or Gaussian noise is added to simulate the quality loss that may occur during image and video transmission.

[0039] In terms of semantic segmentation networks, the overall approach adopts U-Net v2, an improved version of the U-Net semantic segmentation network model. U-Net utilizes skip connections to connect between the encoder and decoder at each level, but it is insufficient to effectively integrate low-level and high-level features. U-Net v2 better integrates features from different levels through a new skip connection design, enhancing the fusion of semantic information in low-level features, while using finer details to optimize high-level features. It consists of three parts: an encoder, an SDI semantic and detail injection module, and a decoder.

[0040] First, a deep neural network encoder is used to extract multi-level features from the input image, from the bottom layer to the top layer. The encoder can adopt network structures such as ResNet, DenseNet, and MobileNet. For an input image I, the encoder extracts multi-scale M-level features. Let i represent the i-th feature, where... Extracted features The output will be fed into the next module for further optimization. Based on the M-level feature maps generated by the encoder, the semantic and detail injection module applies spatial attention and channel attention mechanisms to the features at each level, enabling the features to integrate local spatial information and global channel information, as shown in the following formula: = ( ).

[0041] in This represents the feature map after processing at the i-th level. This represents the spatial attention parameters in the i-th level. This represents the channel attention parameters in the i-th level. Specific spatial channel attention mechanisms can employ attention structures such as SE, ECA, and CBAM, followed by 1x1 convolution to reduce... The number of channels in the feature map is used to obtain the feature map. ; The resolution of feature maps at different levels is different, and they need to be adjusted to the target resolution (based on the i-th level feature map). resolution Based on the baseline, multi-level feature resolution alignment is performed, followed by fusion. Let the adjusted features be... For features of other levels j The adjustment rules are as follows: when : Downsampling using adaptive average pooling D, the formula is: ; when Using the identity mapping I, the features remain unchanged, and the formula is: ; when Upsampling via bilinear interpolation U is achieved using the following formula: .

[0042] After resolution adjustment, a 3x3 convolution is used to smooth each feature, removing noise introduced by interpolation or pooling while preserving feature details. The formula is as follows: ,in This represents a 3x3 convolution operation on the j-th feature at the i-th baseline. Appropriate padding is applied during convolution to ensure the output resolution matches the input resolution. Then, all aligned and smoothed features are integrated through element-wise multiplication, ensuring that each layer's features simultaneously contain both semantic and detailed information from each layer. For the i-th layer, the fusion formula is: The final output is the fused features. .

[0043] The decoder gradually restores the resolution through steps such as upsampling, skip fusion, and convolution, eventually restoring it to the original image resolution. It then maps the resolution to the number of segmentation categories using 1×1 convolution and normalizes the probability of each pixel using the Softmax function to obtain the output segmentation map.

[0044] To address the interference from complex background information in the field, edge detection feature maps are fused with U-Net v2 to enhance the accuracy of target edge processing and achieve more refined segmentation.

[0045] This embodiment provides two fusion methods. The first method is to use edge detection operators such as Canny and Laplacian to extract the edges of the original input image, and obtain a single-channel image after grayscale conversion. The edge single-channel image information is then stitched with the original RGB three-channel image to form a four-channel image. Then, the U-Net v2 network is used for feature extraction, fusion and other processing to output the final ice segmentation result.

[0046] The second approach is to refer to the residual network structure and introduce the edge detection operator into the residual structure. The original feature map, the feature map after multi-layer convolution operation, and the edge detection feature map are added and fused to form a new feature map for the current stage, which serves as the input for the next stage. The designed edge residual network structure can be introduced into the shallow structure of the first few layers of the network to enhance the model's ability to perceive edge information.

[0047] In some embodiments, the specific edge residual network structure branches are as follows: 1) Assuming the input feature map is x, the first branch performs an identity mapping on x, i.e. =x; 2) The second branch uses edge operators to extract edge features from x, obtaining... = Edge operator operation (x); 3) The third branch performs double convolution, normalization, activation function, and other operations on x to obtain... =BN(Conv(Relu(BN(Conv(x))))); 4) Output , , Feature maps are unified to the same scale, then summed and fused, and hyperparameters are designed. , , Weighted fusion It automatically learns the importance of different feature branches.

[0048] For two adjacent insulators in the operation, the midline point of the bottom edge of the rotating rectangle of the upper insulator piece and the midline point of the rotating rectangle of the lower insulator piece, as identified by the model, are selected as the basis for calculating the spacing. The vertical distance between the two points is taken as the pixel spacing distance of the insulator pieces. For each ice ridge region segmented by the ice ridge detection algorithm, the vertical coordinates of the region perpendicular to the midline point of the two insulators are used as a constraint. The coordinates of the top and bottom edges of each ice ridge region are calculated, and the vertical distance between the two points is taken as the pixel length of the ice ridge. When the ratio of the pixel length of the ice ridge to the pixel spacing distance between the two insulators is greater than the threshold t (for example, t is 0.6), it is determined that the ice ridge still has a bridging risk.

[0049] During the de-icing process, the robot's ability to identify the steel caps of porcelain insulators and the core rods of composite insulators, as well as their icing status, can guide the de-icing robot to determine whether it has completed the de-icing work on two adjacent insulators. In particular, since the steel caps are made of a hard material, if the robot cannot determine whether the de-icing operation has been completed, it may easily damage the de-icing device.

[0050] This embodiment employs a model training approach to construct a training dataset, including un-iced steel caps or core rods, lightly iced steel caps or core rods, and heavily iced steel caps or core rods. Collected data samples are labeled with rectangular bounding boxes. The Ylov11 algorithm is used to train an insulator steel cap or core rod icing state recognition model to detect the position of the insulator steel cap or core rod during operation. Using depth information from an RGBD camera, and with the center point of the detected steel cap or core rod rectangular bounding box as a reference point, the distance between the de-icing device and the steel cap or core rod is determined to guide the robot in effective and safe operation.

[0051] In this embodiment, a de-icing strategy is selected based on the detected icing state of the insulator string and the icing bridge risk assessment result. Under the selected de-icing strategy, the propulsion pressure, propulsion speed, propulsion distance, and lifting adjustment amount for controlling the robot operation are set according to the icing state and the robot's operating state.

[0052] Specifically, it includes: The severity of icing is determined by comparing the icing thickness H with the preset first and second thickness thresholds, and by combining the comparison of the load change of the robot motor load over a set time period with the preset load threshold. The first thickness threshold is less than the second thickness threshold. For example, the first thickness threshold is set to 5mm and the second thickness threshold is set to 15mm.

[0053] Specifically: when the ice thickness H is less than the first thickness threshold and the load change is less than the load threshold (considered as no abnormal change in motor load L), it is judged as light icing; When the icing thickness H is not less than the first thickness threshold and less than the second thickness threshold, or when the load change is greater than or equal to the load threshold, it is judged as moderate icing. When the ice thickness H is not less than the second thickness threshold, or when the collected target area image contains a continuous ice body exceeding the set area (which can be detected by the trained image recognition model), it is determined to be heavily iced.

[0054] Meanwhile, based on the results of the ice bridge risk assessment, ice floes that are likely to bridge are considered to have a bridging trend, and ice floes that exceed the predetermined bridging risk are considered to be high-risk bridging situations.

[0055] Based on this, the following de-icing strategy is selected: (1) When the severity of icing is mild and the ice bridge risk assessment result indicates a bridging trend, a light-load squeezing strategy is selected to quickly remove ice with low thrust and high speed. (2) When the icing severity is moderate and no ice bridge risk is detected, the standard squeezing strategy is selected, and the operation is carried out with moderate thrust and standard advance distance; (3) When the icing severity is severe, or when the feedback of three consecutive compressions fails to achieve the desired de-icing effect, a powerful de-icing strategy is initiated to break the ice with higher thrust and segmented propulsion. (4) When the ice bridge risk assessment result is a high-risk bridging situation, the bridging risk handling strategy should be called first, the speed should be reduced and the maximum thrust should be limited, and the safety should be ensured by gradually breaking the ice bridge. (5) In addition, if risk factors such as low ambient temperature, high wind speed, motor overload or abnormal posture are detected, the system will automatically enter the safety pause strategy, stop the operation and maintain the current position until the environment is restored or the operation is confirmed by a person before it can continue.

[0056] By using the above-mentioned graded judgment and rule-based strategy switching methods, the optimal de-icing strategy can be automatically selected under different icing and environmental conditions, thereby improving the stability and safety of de-icing operations.

[0057] Finally, under the selected de-icing strategy, the propulsion pressure, propulsion speed, propulsion distance, and lifting adjustment amount are set to control the robot's operation, so as to realize de-icing operations for different icing scenarios.

[0058] Specifically: (1) The propulsion pressure is derived from the base pressure F0 combined with the icing thickness H and the motor load L: F target =F0+k h ·H+k l ·L, where kh k l This is the empirical gain coefficient.

[0059] (2) The propulsion speed v is based on the real-time pressure deviation e F = F target –F real Dynamic adjustment: v = v0 – k f ·e F To ensure that the propulsion speed is automatically reduced when the pressure is too high, thus avoiding impact on the insulator, F real k represents the actual extrusion pressure. f is a coefficient.

[0060] (3) The advance distance D is set as D=D0+k based on the initial distance D0 and the icing thickness H. d ·H,k d is a coefficient.

[0061] (4) The elevation adjustment amount Δh is calculated based on the visual center deviation Δpos and the IMU attitude deviation Δθ: Δh=k c ·Δpos+k θ ·Δθ,k c and k θ is a coefficient.

[0062] Therefore, for the extrusion motor, the output pushing pressure is controlled by adjusting the drive current; for the forward pushing mechanism, the pushing speed and pushing distance are controlled by adjusting the pulse frequency of the servo motor; for the lifting mechanism, the lifting adjustment is finely adjusted by the stepping amount of the brake motor; and for the horizontal compensation mechanism, the slide is controlled to perform lateral movement according to the lateral compensation command. These actuators continuously correct pressure, displacement, and attitude under closed-loop feedback, ensuring that the robot's de-icing tool always remains in a safe and effective working state.

[0063] In this embodiment, during robot operation, high-precision lifting and lowering can be achieved through an electric telescopic rod and a brake motor, ensuring the robot can flexibly adjust to the target working position. Simultaneously, horizontal movement of the robot is supported; through precise control, the robot can move freely in suspended environments to complete de-icing tasks. Throughout this process, real-time feedback of sensor data ensures the robot's accuracy in every action, adapting to complex working scenarios.

[0064] In this embodiment, force control and visual feedback are used to ensure the protection of insulators and the accuracy of de-icing actions during operation by adjusting the pressure and posture of the de-icing tool in real time, thus avoiding damage.

[0065] The robot uses a built-in pressure sensor to monitor the squeezing force applied to the insulator by the gripper in real time, and uses a vision module to continuously detect the position of the insulator's centerline to identify the deviation of the de-icing tool.

[0066] Specifically: (1) When an abnormal increase in extrusion pressure is detected, the retraction action is automatically executed and the thrust is reduced.

[0067] (2) When visual detection shows a deviation in the center line of the insulator, the lateral compensation amount is calculated based on the deviation amount, and the horizontal moving mechanism is controlled to perform lateral compensation movement. (3) When the ice block does not fall off within the expected time, a secondary de-icing action with increased squeezing force is automatically triggered.

[0068] (4) If a dangerous condition is detected, such as abnormal insulator stress, motor stall, or abnormal tool posture, stop, retraction, and alarm protection actions shall be executed immediately.

[0069] More specifically: the calculation process for the above closed-loop regulation is as follows.

[0070] (1) Real-time reading of the measured extrusion force F from the pressure sensor real and the set propulsion pressure F target Compare the results, when F is satisfied real -F target ≥ ΔF th When this occurs, it is considered an abnormal increase in extrusion pressure, where ΔF th A pressure threshold is set to characterize the permissible deviation range from the normal operating pressure.

[0071] At this point, based on the pressure deviation e F =F real -F target Calculate the retraction amount Δd: Δd = k ret ·e F , where k ret This is the shrinkage gain coefficient.

[0072] Therefore, the number of pulses of the forward push mechanism servo motor is reduced according to the amount of retraction to achieve retraction, thereby automatically reducing the thrust and correcting the target current of the compression motor to prevent damage to the insulator.

[0073] This adjustment is used to control the pressure applied by the robot's de-icing tool, ensuring that appropriate pressure is applied to the insulator during de-icing operations to avoid damage to the target.

[0074] (2) Using the trained image recognition model, extract the pixel deviation Δu of the insulator string centerline in the image of the working area, and compare it with the set pixel threshold Δu. th When comparing, if |Δu|≥Δuth At that time, the positional deviation between the judgment tool and the insulator exceeded the limit.

[0075] At this point, the lateral compensation amount Δx is calculated based on the pixel deviation and image depth information: Δx = k u ·Δu·S d , where k u S is the pixel compensation coefficient. d Δx is the depth scale factor, representing the displacement that maps pixel deviation to actual space. Therefore, the stroke of the horizontal movement mechanism is adjusted according to Δx to realign the de-icing tool back to the center of the insulator.

[0076] (3) Establish the expected ice removal time model T based on the current ice thickness H, the propulsion speed v and the pressure application time t. exp When the actual de-icing time T real ≥T exp +ΔT th If the de-icing operation is deemed ineffective and a second de-icing operation with increased pressure is required, then the de-icing process must be performed.

[0077] Increase extrusion pressure F enh For: F enh =F target +k enh · (T real -T exp ), where k enh To enhance the intensity coefficient.

[0078] F enh As a new target pressure, the propulsion speed and distance are recalculated until the ice layer breaks off or the safety limit is reached.

[0079] Through the calculation process of pressure anomaly detection, visual deviation correction, and de-icing effect judgment described above, the extrusion pressure, tool position, and operation rhythm can be adjusted in real time in complex environments to achieve safe, reliable, and continuous de-icing operations.

[0080] As an alternative implementation method, real-time adjustments can be made based on environmental factors, such as optimizing the de-icing operation by considering temperature and humidity. The specific adjustment formula is as follows: Environmental compensation amount = k3 × (Ambient temperature change + Humidity change). When a temperature change or excessively high humidity is detected, the de-icing strategy is adjusted according to the set environmental compensation amount. For example, in low-temperature environments, it is necessary to increase the thrust to overcome the hardness of the ice layer; in high-humidity environments, it is necessary to adjust the speed and pressure of the de-icing tools to avoid affecting the operation results due to water droplets or melting ice.

[0081] As an alternative implementation method, the lifting height can also be finely adjusted in real time based on the position information fed back by the encoder of the upgrade module and the attitude angle of the IMU.

[0082] The position deviation calculation formula is Δpos = target position - measured position. This adjustment amount is used to control the robot's movement position. When a position deviation is detected, the robot's propulsion speed and direction are adjusted to ensure that the robot moves to the target position.

[0083] The formula for calculating attitude error is Δθ = visual recognition angle - attitude angle measured by IMU. This adjustment is used to control the robot's attitude and direction, ensuring that the robot's attitude is stable during operation and avoiding deviation from the predetermined trajectory.

[0084] The above-calculated adjustment amounts dynamically adjust the robot's movement, posture, and pressure based on the robot's actual working state and environmental conditions to cope with different working environments and operational needs, ensuring the robot's accuracy and stability in de-icing operations.

[0085] The control method for the suspension insulator de-icing robot provided in this embodiment mainly includes three core steps. First, based on the detected icing state of the insulator string and the assessment of ice bridge risk, the most suitable de-icing strategy is selected from multiple preset strategies to adapt to different icing conditions. Next, under the determined de-icing strategy, key parameters such as the propulsion pressure, propulsion speed, propulsion distance, and lifting adjustment amount are precisely set according to the icing state and the robot's operating status, laying the foundation for efficient robot operation. Finally, during the robot's operation, information such as the squeezing pressure applied by the robot to the insulator string, images of the working area, and actual de-icing time are acquired in real time. Based on this, the robot's operating parameters are adjusted to ensure the smooth progress of the de-icing operation. In practical applications, the icing conditions of insulator strings are complex and varied, and the risk of ice bridges also differs. This embodiment designs a de-icing strategy selection technology. When the detected ice thickness is thin and relatively uniform, and the risk of ice bridges is low, a gentler mechanical scraping de-icing strategy can be selected. However, when the ice thickness is large and there is a serious risk of ice bridges, a de-icing strategy with higher impact force is adopted. By selecting the de-icing strategy according to the actual situation, de-icing efficiency can be improved and de-icing costs can be reduced. Secondly, the settings for propulsion pressure, propulsion speed, propulsion distance, and lifting adjustment are closely related to the icing condition and the robot's operating status. If the ice is thick and hard, to ensure effective de-icing, it is necessary to increase the propulsion pressure, propulsion speed, and appropriately increase the propulsion distance to ensure thorough removal of the ice layer. However, the propulsion pressure and speed should not be excessive, otherwise, it may damage the insulators. Regarding the robot's operating status, if the robot wobbles or becomes unstable during movement, the lifting adjustment needs to be adjusted promptly to maintain the robot's balance and ensure the accuracy of the de-icing operation.

[0086] Finally, the feedback adjustment mechanism plays a crucial role in the robot's operation. Pressure sensors monitor the real-time compressive force applied by the robot to the insulator string. When the pressure deviation between the compressive force and the pre-set propulsion pressure is not less than a set pressure threshold, it indicates that the current operating parameters may need adjustment. At this point, the amount of robot retraction is determined based on the pressure deviation to restore the compressive force to an appropriate range. Image acquisition equipment captures images of the work area, and the pixel deviation of the insulator string centerline in the image is extracted. When the pixel deviation is not less than a set pixel threshold, it indicates a deviation in the robot's horizontal position. Lateral compensation is determined based on the pixel deviation to adjust the robot's horizontal movement and ensure the accuracy of the de-icing operation. Actual de-icing time is also an important feedback indicator. When the actual de-icing time is not less than the expected de-icing time, it indicates low de-icing efficiency. In this case, the compressive force is increased based on the propulsion pressure, and the propulsion speed and distance are updated accordingly to improve de-icing efficiency.

[0087] In this embodiment, a self-check and recovery mechanism is also provided throughout the robot's operation. This includes three stages: startup self-check, operation self-check, and anomaly recovery.

[0088] Specifically: After the robot is powered on, it enters the startup self-test process, which performs a comprehensive self-test on key modules such as the motor drive unit, encoder feedback, upper / lower limit of the telescopic rod, wireless communication link, battery voltage and temperature, and image sensor status. If an abnormality is detected, it enters the prohibited operation mode and prompts for manual handling.

[0089] During operation, the system continuously monitors the operating status of motor temperature, communication quality, pressure fluctuations, number of consecutive visual recognition failures, and telescopic pole load to determine in real time whether there are any potential faults.

[0090] When an anomaly occurs, the anomaly recovery process is initiated: In the case of communication interruption, the system automatically attempts to send a handshake packet and reconnect the LoRa communication link, and continues to execute the current operation steps after recovery; in the case of motor overload, the system reduces the operating load and resumes the operation after the temperature recovers; in the case of visual loss, the system automatically re-identifies the insulator position and returns to the safe position point; in the case of lifting error, the system performs a limit switch re-zeroing operation.

[0091] The aforementioned self-checking and recovery mechanisms can ensure the robot's continuous and stable operation in complex environments, thereby improving overall operational reliability.

[0092] In this embodiment, as Figure 2 As shown, a two-layer wireless communication architecture is used to achieve stable remote communication between the robot and the ground control system: Among them, LoRa communication serves as a long-distance communication link between the lifting and de-icing modules. The LoRa protocol features low power consumption and long distance, ensuring communication stability in high-altitude and strong electromagnetic environments.

[0093] The SBUS protocol is used for multi-channel real-time control between the remote controller and the robot, ensuring that the operator can accurately control the robot's various actions.

[0094] In this embodiment, to further improve the communication stability and security of the dual-layer wireless communication architecture under high-altitude, strong electromagnetic interference and severe weather conditions, the communication module incorporates anti-interference enhancement mechanisms, link adaptive mechanisms and redundancy protection mechanisms on both the LoRa and SBUS links.

[0095] Specifically: LoRa communication links utilize spread spectrum frequency hopping technology to automatically switch operating frequencies to avoid interference bands when they detect a decrease in signal-to-noise ratio, an increase in bit error rate, or increased external electromagnetic interference, thereby improving the link's anti-interference capability. Simultaneously, the spreading factor is adjusted in real-time based on parameters such as RSSI, SNR, and bit error rate, adaptively switching within the SF7 to SF12 range: increasing the spreading factor to enhance demodulation capability when the electromagnetic environment is harsh, and decreasing the spreading factor to improve communication efficiency when the environment is stable.

[0096] For forward error correction, LoRa communication employs variable error correction coding and automatically upgrades the error correction level when an increase in erroneous frames is detected, thereby significantly enhancing the data integrity of the link. In the event of abnormal data transmission, a tiered retransmission strategy is implemented, including rapid retransmission, retransmission after increasing transmit power, and reconnection after switching to a backup frequency, to ensure that the link can still restore communication in high-interference environments. Furthermore, LoRa communication enables AES-128 encryption and two-way authentication mechanisms to guarantee data security and attack resistance during long-distance communication.

[0097] For the SBUS link, to ensure low latency and high reliability even under strong electromagnetic interference, the hardware employs a metal braided shielding layer and a differential drive structure to reduce the impact of external interference on the signal. An LC filter network and a TVS diode surge protection circuit are added at the communication interface to suppress the effects of high-frequency noise and transient voltage on signal stability. At the protocol and parsing level, SBUS data frames utilize a real-time verification mechanism. Upon detecting a verification failure, a redundant frame mechanism automatically selects an available control quantity from the most recent valid data frame to prevent erroneous robot actions caused by brief interference. If the SBUS signal is continuously lost for a certain period, a safety mode is activated, automatically stopping horizontal movement, retrieving the de-icing gripper to a safe position, and locking the lifting mechanism to ensure the robot's safe operation in the event of communication failure.

[0098] Regarding the two-layer communication collaboration, this embodiment clarifies the functional division of LoRa and SBUS within the system. The LoRa link primarily undertakes long-distance communication tasks, including task command issuance, icing identification result feedback, equipment operating status monitoring, and abnormal information uploading, suitable for task-level scheduling and status monitoring during operation. The SBUS link is mainly used for short-range real-time motion control, enabling operators to precisely control actions such as lifting, horizontal movement, and gripper opening and closing, meeting the operational requirements of high real-time performance and low latency.

[0099] To enable coordinated operation of the two links, the system employs an internal communication scheduling mechanism: when the SBUS is connected, it prioritizes real-time actions, while LoRa focuses on status feedback; when the SBUS is not connected or experiences an interruption, the system automatically switches to LoRa control mode; when both links are available simultaneously, LoRa handles task-level control and status monitoring, while the SBUS handles action-level control, ensuring coordinated operation of the two links under different operating modes. Furthermore, in the event of a communication link anomaly, automatic redundancy switching between LoRa and SBUS is implemented. If the SBUS loses its signal, LoRa takes over control; if LoRa loses its signal, it enters a safe mode while retaining the local controllability of the SBUS, significantly improving the overall robustness and reliability of the communication system.

[0100] Through the aforementioned anti-interference enhancement mechanism, link adaptive adjustment strategy, and collaborative and redundant switching design of dual-layer communication, this embodiment can provide stable, reliable, and secure communication capabilities in strong electromagnetic, high-altitude, and complex environments, providing key assurance for the continuous operation of the suspended insulator de-icing robot in complex environments.

[0101] This embodiment also includes power management, responsible for monitoring battery status and ensuring a stable power supply. This includes: a battery management system that monitors battery charge, current, voltage, and temperature to ensure safe battery operation; and power distribution and charging management that intelligently adjusts battery charging strategies and rationally allocates power to ensure stable operation of each module. Efficient battery management extends the robot's runtime and improves operational reliability.

[0102] Specifically, the intelligent charging strategies for batteries include: phased charging strategy, temperature compensation strategy, load priority strategy, and battery health status management strategy.

[0103] Among them, the phased charging strategy adopts a constant current-constant voltage (CC-CV) two-stage charging method to improve efficiency and delay battery aging; the temperature compensation strategy automatically adjusts the charging current according to the battery temperature to avoid the risk of overcharging in high or low temperature environments; the load priority strategy dynamically allocates available current to each module according to the robot's current working status; and the health status management strategy automatically adjusts the maximum rechargeable amount according to parameters such as battery cycle count and voltage fluctuation to extend battery life.

[0104] The selection and adjustment of the charging strategy are based on the following rules: when the battery level is below the first threshold, the constant current fast charging mode is entered; when the battery level reaches the second threshold, the constant voltage fine-tuning mode is switched; when the battery temperature exceeds the set upper limit, the charging power is reduced or charging is paused; when the robot performs high-load de-icing operations, the power supply to non-critical modules is reduced, and priority is given to ensuring the lifting and squeezing power modules.

[0105] The power distribution module assigns different power supply priorities based on the importance of each module's function: the lifting motor and pressure control module are at level one, the vision recognition module at level two, the communication module at level three, and the auxiliary lighting and indicator module at level four. When the battery load approaches saturation, the power supply to lower-priority modules is reduced via PWM modulation, with the duty cycle dynamically adjusted to free up more power to support critical modules. By monitoring parameters such as current, voltage, and temperature in real time, the output power is dynamically allocated to ensure the continuous and stable operation of key functional units and improve the overall reliability of the machine.

[0106] This embodiment proposes a mechanically and electrically integrated lifting and de-icing coordinated control technology. The lifting module uses a brake motor in conjunction with a telescopic insulating rod, which can automatically lock and precisely control the lifting height in the event of a power outage. The de-icing module employs a squeezing gripper and a horizontal centering and pushing mechanism, achieving flexible de-icing through force control and visual feedback. This ensures precision and flexible control during the de-icing process, avoiding damage to the insulators. This technology achieves efficient de-icing without damaging the insulators, significantly improving the automation level and safety of live-line de-icing operations.

[0107] This embodiment proposes a dual-layer wireless communication architecture, employing the LoRa protocol for long-distance communication and the SBUS protocol for real-time control. The LoRa protocol, serving as the remote communication link between the lifting and de-icing modules, provides reliable remote control in strong electromagnetic and high-altitude environments, ensuring stability under these conditions. The SBUS protocol provides multi-channel real-time control between the robot and the remote controller. This dual-layer communication architecture provides the robot with highly reliable, low-latency communication, significantly improving the stability, accuracy, and reliability of communication during de-icing operations, meeting the real-time control requirements of de-icing operations.

[0108] The embodiment proposed as follows Figure 3The control system of the suspended insulator de-icing robot shown adopts a modular design concept. The whole machine consists of multiple functional modules, including fall protection, lifting, de-icing, power management, communication, and overall machine control. Each functional module operates independently and can be interconnected and mutually inspected through a unified protocol. It also has intelligent self-testing, self-recovery, and remote reset functions. Each module has a built-in independent control unit and communication interface, ensuring that the system can automatically detect and resume operation in the event of abnormalities, such as motor failure, communication interruption, or power failure, thereby improving the robot's safety and continuous operation capability.

[0109] The suspension insulator de-icing robot control system and method proposed in this embodiment significantly improve the operating efficiency, safety, and intelligence level of the de-icing robot by combining a dual-layer communication architecture, modular design, self-recovery function, and electromechanical integrated lifting and de-icing coordinated control technology. It is suitable for high-altitude operations in complex environments, and can maintain high efficiency and stable performance, especially under strong electromagnetic, high-altitude, and severe weather conditions.

[0110] Example 2 This embodiment provides a control system for a suspended insulator de-icing robot, including: The selection module is configured to select a de-icing strategy based on the detected icing status of the insulator strings and the results of the ice bridge risk assessment. The control module is configured to set the propulsion pressure, propulsion speed, propulsion distance, and lifting adjustment amount for controlling the robot's operation based on the icing state and robot operating state under the selected de-icing strategy. The feedback module is configured to acquire the squeezing force applied by the robot to the insulator string, the image of the work area, and the actual de-icing time during the operation, so as to adjust the robot's operation parameters accordingly. Among them, when the pressure deviation between the extrusion pressure and the propulsion pressure is not less than the set pressure threshold, the amount of retraction of the robot propulsion is determined according to the pressure deviation. Extract the pixel deviation of the insulator string centerline in the working area image. When the pixel deviation is not less than the set pixel threshold, determine the lateral compensation amount based on the pixel deviation to adjust the horizontal movement of the robot. When the actual de-icing time is not less than the expected de-icing time, the enhanced extrusion pressure is determined based on the propulsion pressure, and the propulsion speed and propulsion distance are updated according to the enhanced extrusion pressure.

[0111] It should be noted that the above modules correspond to the steps described in Embodiment 1, and the examples and application scenarios implemented by the above modules and the corresponding steps are the same, but are not limited to the content disclosed in Embodiment 1. It should also be noted that the above modules, as part of the system, can be executed in a computer system such as a set of computer-executable instructions.

[0112] In further embodiments, the following is also provided: An electronic device includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor, wherein the computer instructions, when executed by the processor, perform the method described in Embodiment 1. For brevity, further details are omitted here.

[0113] It should be understood that in this embodiment, the processor can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc.

[0114] Memory may include read-only memory and random access memory, and provides instructions and data to the processor. A portion of memory may also include non-volatile random access memory. For example, memory may also store information about the device type.

[0115] A computer-readable storage medium for storing computer instructions, which, when executed by a processor, perform the method described in Embodiment 1.

[0116] The method in Example 1 can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor. The software modules can reside in readily available storage media in the field, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method. To avoid repetition, a detailed description is not provided here.

[0117] A computer program product includes a computer program that, when executed by a processor, implements the method described in Embodiment 1.

[0118] The present invention also provides at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. The computer program product includes computer-executable instructions, such as instructions included in program modules, which execute in a device on a target real or virtual processor to perform the processes / methods described above. Typically, program modules include routines, programs, libraries, objects, classes, components, data structures, etc., that perform specific tasks or implement specific abstract data types. In various embodiments, the functionality of program modules can be combined or divided among program modules as needed. The machine-executable instructions for the program modules can execute within a local or distributed device. In a distributed device, the program modules can reside in both local and remote storage media.

[0119] The computer program code used to implement the methods of the present invention may be written in one or more programming languages. This computer program code may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the computer or other programmable data processing device, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a computer, partially on a computer, as a stand-alone software package, partially on a computer and partially on a remote computer, or entirely on a remote computer or server.

[0120] In the context of this invention, computer program code or related data may be carried by any suitable carrier to enable a device, apparatus, or processor to perform the various processes and operations described above. Examples of carriers include signals, computer-readable media, and the like. Examples of signals may include electrical, optical, radio, sound, or other forms of propagation signals, such as carrier waves, infrared signals, etc.

[0121] Those skilled in the art will recognize that the units and algorithm steps described in connection with the various examples of this embodiment can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention.

[0122] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A control method for a suspended insulator de-icing robot, characterized in that, include: Based on the detected icing status of the insulator strings and the assessment of ice bridge risk, a de-icing strategy is selected. Under the selected de-icing strategy, the propulsion pressure, propulsion speed, propulsion distance, and lifting adjustment amount are set according to the icing state and robot operating state to control the robot's operation. The robot acquires data on the squeezing force applied to the insulator string, images of the work area, and the actual de-icing time during the operation, and uses this data to adjust the robot's operating parameters accordingly. Among them, when the pressure deviation between the extrusion pressure and the propulsion pressure is not less than the set pressure threshold, the amount of retraction of the robot propulsion is determined according to the pressure deviation. Extract the pixel deviation of the insulator string centerline in the working area image. When the pixel deviation is not less than the set pixel threshold, determine the lateral compensation amount based on the pixel deviation to adjust the horizontal movement of the robot. When the actual de-icing time is not less than the expected de-icing time, the enhanced extrusion pressure is determined based on the propulsion pressure, and the propulsion speed and propulsion distance are updated according to the enhanced extrusion pressure.

2. The control method for a suspended insulator de-icing robot as described in claim 1, characterized in that, The process of selecting a de-icing strategy includes: The severity of icing is determined by comparing the icing thickness with a preset first thickness threshold and a preset second thickness threshold, and by combining the comparison of the robot motor load change over a set time period with a preset load threshold; wherein the first thickness threshold is less than the second thickness threshold. When the icing severity is mild and the ice bridge risk assessment result indicates the presence of bridging ice floes, a light-load compression strategy is selected. When the icing severity is moderate and no risk of ice bridges is detected, the standard extrusion strategy is selected. When the icing severity is severe icing, or when the pressure feedback of a set number of consecutive cycles fails to achieve the desired de-icing effect, a powerful de-icing strategy is activated. When the ice bridge risk assessment result indicates that there are ice floes exceeding the predetermined bridging risk, a bridging risk handling strategy is adopted.

3. The control method for a suspended insulator de-icing robot as described in claim 2, characterized in that, The process of determining the severity of icing includes: When the ice thickness is less than the first thickness threshold and the load change is less than the load threshold, it is judged as light icing; When the ice thickness is not less than the first thickness threshold and less than the second thickness threshold, or when the load change is greater than or equal to the load threshold, it is judged as moderate icing. When the ice thickness is not less than the second thickness threshold, or when the collected target area image contains continuous ice bodies exceeding the set area, it is determined to be heavily iced.

4. The control method for a suspended insulator de-icing robot as described in claim 1, characterized in that, The process of setting propulsion pressure, propulsion speed, propulsion distance, and elevation adjustment includes: The propulsion pressure is derived from the base pressure F0, icing thickness H, and motor load L: F target =F0+k h ·H+k l ·L; Based on real-time pressure deviation e F = F target –F real The propulsion speed v is obtained as: v = v0 – k f ·e F ; The advancing distance D is obtained based on the initial distance D0 and the ice thickness H: D = D0 + k d ·H; The elevation adjustment Δh is obtained based on the visual center deviation Δpos and the IMU attitude deviation Δθ: Δh = k c ·Δpos+k θ ·Δθ; where k h k l k c k θ k f k d All are coefficients, F real This represents the actual extrusion pressure.

5. The control method for a suspended insulator de-icing robot as described in claim 1, characterized in that, The amount of retraction Δd is Δd=k ret ·e F , where k ret e is the shrinkage gain coefficient. F This is the pressure deviation; the number of pulses of the forward thrust mechanism servo motor is reduced according to the retraction amount to achieve retraction, thereby reducing the thrust.

6. The control method for a suspended insulator de-icing robot as described in claim 1, characterized in that, When the absolute value of the pixel deviation is not less than the set pixel threshold, it is determined that the positional deviation between the working tool and the insulator exceeds the limit. The lateral compensation amount determined based on the pixel deviation Δu is Δx: Δx = k u ·Δu·S d , where k u S is the pixel compensation coefficient. d The depth scale factor represents the displacement that maps pixel deviations to actual space, thereby adjusting the robot's horizontal movement.

7. The control method for a suspended insulator de-icing robot as described in claim 1, characterized in that, The expected de-icing time T is determined based on the current ice thickness, advance speed, and pressure application time. exp ; In pushing pressure F target The enhanced extrusion pressure F was determined based on this. enh For: F enh =F target +k enh · (T real -T exp ), where k enh To enhance the intensity coefficient, T real This refers to the actual de-icing time.

8. The control method for a suspended insulator de-icing robot as described in claim 1, characterized in that, The process of detecting the icing status of insulator strings and assessing the risk of ice bridges includes: A pre-trained insulator string detection model is used to identify insulator string images under icing conditions, and to locate the position and icing status of the insulator strings. The pre-trained insulator disc recognition model is used to identify two adjacent insulator discs in an image. If there are multiple insulator discs in the image field of view, the two insulator discs in the center of the image are used as the recognition result. Using a pre-trained icicle detection algorithm model, it is determined whether there is still icicle between the two identified insulators. The ratio of the pixel length of the icicle to the pixel length of the distance between the two insulators is calculated. Based on the comparison of the ratio and a predetermined threshold, it is determined whether there is bridging icicle or icicle with a risk exceeding the predetermined bridging threshold.

9. The control method for a suspended insulator de-icing robot as described in claim 1, characterized in that, The method also includes: using a two-layer wireless communication architecture to realize remote communication between the robot and the ground control system; specifically including: LoRa communication as a long-distance communication link between the robot's lifting module and de-icing module; and SBUS protocol for multi-channel real-time control between the ground control system and the robot to control the robot's various actions.

10. A control system for a suspended insulator de-icing robot, characterized in that, include: The selection module is configured to select a de-icing strategy based on the detected icing status of the insulator strings and the results of the ice bridge risk assessment. The control module is configured to set the propulsion pressure, propulsion speed, propulsion distance, and lifting adjustment amount for controlling the robot's operation based on the icing state and robot operating state under the selected de-icing strategy. The feedback module is configured to acquire the squeezing force applied by the robot to the insulator string, the image of the work area, and the actual de-icing time during the operation, so as to adjust the robot's operation parameters accordingly. Among them, when the pressure deviation between the extrusion pressure and the propulsion pressure is not less than the set pressure threshold, the amount of retraction of the robot propulsion is determined according to the pressure deviation. Extract the pixel deviation of the insulator string centerline in the working area image. When the pixel deviation is not less than the set pixel threshold, determine the lateral compensation amount based on the pixel deviation to adjust the horizontal movement of the robot. When the actual de-icing time is not less than the expected de-icing time, the enhanced extrusion pressure is determined based on the propulsion pressure, and the propulsion speed and propulsion distance are updated according to the enhanced extrusion pressure.

11. An electronic device, characterized in that, It includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor, which, when executed by the processor, perform the method according to any one of claims 1-9.

12. A computer-readable storage medium, characterized in that, Used to store computer instructions, which, when executed by a processor, perform the method described in any one of claims 1-9.

13. A computer program product, characterized in that, Includes a computer program, which, when executed by a processor, implements the method described in any one of claims 1-9.

Images