A power grid de-icing method and system based on environmental perception

Abstract

The present application relates to the technical field of power grid deicing, and discloses a power grid deicing method and system based on environmental perception, which comprises the following steps: collecting environmental data such as temperature, humidity and wind speed, and state data such as conductor tension, vibration amplitude and echo signal to form an original monitoring data set, extracting icing boundary features after preprocessing, combining trends to generate icing sign determination information, and confirming icing risk through multi-parameter joint analysis; after the risk is confirmed, the initial estimated value of thickness is taken as a base point to correct the target thickness parameter in combination with real-time environmental and historical information, when the safety threshold is exceeded, the corresponding deicing mode is selected according to the thickness distribution and environmental conditions, and when the thickness of three consecutive time windows is all rising, early warning information is generated, and finally the data before and after deicing is stored and updated in a database. The present application realizes accurate identification and efficient processing of icing risk.

Description

Technical Field

[0001] This invention relates to the field of power grid de-icing technology, and more specifically, to a power grid de-icing method and system based on environmental perception. Background Technology

[0002] With the continuous expansion of the power grid and the long-term operation of transmission lines in complex environments, icing has become one of the main risks affecting the safe and stable operation of the power grid. Traditional manual inspection methods suffer from low monitoring efficiency, poor timeliness, and a high risk of missed detections and misjudgments, making it difficult to promptly detect and address icing hazards. Furthermore, transmission lines are significantly affected by environmental factors such as temperature, humidity, and wind speed during icy and snowy weather, resulting in a complex ice formation and melting process. Traditional monitoring methods based on single parameters often fail to accurately reflect the dynamic changes in icing. While some existing technologies have incorporated sensors, infrared cameras, and laser ranging, they still have shortcomings in data fusion, accuracy of icing assessment, and rationality of de-icing decisions. In addition, de-icing operations often lag behind icing formation, causing some lines to operate under risky conditions for extended periods, increasing the probability of failure and maintenance costs. Summary of the Invention

[0003] In view of this, the present invention proposes a power grid de-icing method and system based on environmental perception, aiming to solve the problem of how existing technologies can detect icing in a timely manner and take effective de-icing measures during the operation of transmission lines.

[0004] In one aspect, the present invention proposes a power grid de-icing method and system based on environmental perception, comprising: Temperature, humidity and wind speed are collected as environmental monitoring data, and conductor tension, vibration amplitude, echo signal delay and signal strength are collected as condition monitoring data. The environmental monitoring data and condition monitoring data are then combined to obtain the raw monitoring dataset. The original monitoring dataset is preprocessed to obtain a preprocessed monitoring dataset. The ice appearance images in the preprocessed monitoring dataset are compared with historical images to obtain the ice boundary change characteristics. Based on the change trend of environmental monitoring parameters in the preprocessed monitoring dataset, ice sign judgment information is generated. Based on the analysis of the difference between laser detection distance and actual distance, echo signal delay and signal strength deviation, and icing boundary change characteristics in the preprocessed monitoring dataset, when at least two of the following are simultaneously abnormal, the icing sign judgment information is confirmed as an icing risk judgment, and it is determined that the transmission line has an icing risk. After determining that there is a risk of icing on the transmission line, the initial estimate of the icing thickness in the preprocessed monitoring dataset is used as the correction base point. The first stage correction thickness is obtained based on the real-time environmental monitoring parameters. The second stage correction thickness is obtained based on the difference in boundary features between historical icing stripping information and icing appearance images. The first stage correction thickness and the second stage correction thickness are superimposed to obtain the corrected target icing thickness parameter. When the target ice thickness parameter exceeds the safety threshold, the de-icing method is selected based on the target ice thickness distribution and environmental monitoring parameters, and the surface of the transmission line is de-iced. The target ice thickness parameter is tracked continuously. When it shows an upward trend for three consecutive time windows, a de-icing warning is generated. After the de-icing process is completed, the images of the transmission lines before and after de-icing, along with the corresponding environmental monitoring data and condition monitoring data, are stored synchronously and the historical database is updated.

[0005] Furthermore, the process of preprocessing the original monitoring dataset to obtain the preprocessed monitoring dataset includes: The environmental monitoring data and condition monitoring data in the original monitoring dataset are aligned according to timestamps and organized according to a unified data format specification to obtain an initial data sequence with consistent labels. The environmental monitoring data and condition monitoring data in the initial data sequence are subjected to outlier removal and normalization processes to obtain standardized environmental monitoring parameters and standardized condition monitoring parameters. The ice-covered appearance images in the original monitoring dataset are subjected to feature extraction and enhancement processing. Ice boundary feature data are obtained based on image edge clarity and grayscale gradient information. The ice boundary feature data are associated with the corresponding timestamps and stored to obtain the preprocessed ice-covered image feature set. The standardized environmental monitoring parameters and standardized condition monitoring parameters are matched with the preprocessed icing image feature set to obtain the preprocessed monitoring dataset.

[0006] Furthermore, the process of generating information to determine signs of icing includes: Based on the standardized environmental monitoring parameters in the preprocessed monitoring dataset, the changing trends of temperature, humidity and wind speed are extracted. When the change amplitude within a continuous time window exceeds the set interval or the fluctuation frequency continues to increase, it is considered as obtaining the environmental characteristics of icing conditions, and thus obtaining the environmental trend characteristic sequence. Based on the standardized status monitoring parameters in the preprocessed monitoring dataset, the change trajectory of echo signal delay and echo signal intensity is extracted. When the delay continues to increase or the intensity continues to decrease, it is considered as a manifestation of dielectric coverage on the line surface, and a signal anomaly feature sequence is obtained. Based on the feature set of icing images in the preprocessed monitoring dataset, the changes in the expansion rate and edge sharpness of the image boundaries are extracted. When the boundary expansion rate shows an increasing trend or the edge sharpness decreases, it is determined to be a sign of ice adhesion or thickening, and an image difference feature sequence is obtained. The environmental trend feature sequence, signal anomaly feature sequence, and image difference feature sequence are fused together. When at least two types of feature sequences point to the possibility of icing, the presence of icing signs on the line is confirmed, and the confirmation result is output as icing sign judgment information.

[0007] Furthermore, after the information for determining icing signs is generated, the information for determining icing signs is cross-validated, including: The environmental monitoring parameters in the preprocessed monitoring dataset are matched with the transmission line condition monitoring parameters to verify consistency. When the temperature and humidity conditions are consistent with the trend of changes in line tension or vibration amplitude, the consistency verification is used as the basis for confirming the validity of the indication. The boundary features of the icing appearance image are compared with normal operation images in the historical database. When the direction of boundary expansion matches the icing growth trajectory in the historical record, the comparison result is used as the basis for confirming the reliability of the indication. When both the consistency verification and comparison results are valid, the icing sign determination information is confirmed as valid.

[0008] Furthermore, the process of assessing icing risk includes: After obtaining the information on signs of icing, the difference between the laser detection distance and the actual distance in the preprocessed monitoring dataset is compared. When the difference continues to increase and exceeds the set threshold, the difference is used as the quantitative basis for thickness accumulation to obtain the distance difference feature. Based on the analysis of echo signal delay and echo signal intensity deviation in the preprocessed monitoring dataset, when the delay value shows a continuous increase and the intensity value shows a continuous decrease, this joint feature is used as the basis for the propagation anomaly caused by the attached material, and the signal deviation feature is obtained. Trend tracking is performed based on the ice boundary feature data in the preprocessed monitoring dataset. When the boundary expansion rate is higher than the historical average or the boundary morphology changes irregularly, this trend is used as the basis for the rapid evolution of the ice layer, and boundary anomaly features are obtained. Distance difference features, signal deviation features, and boundary anomaly features are matched according to timestamps, and consistency is compared based on the direction and magnitude of change of each feature. When at least two of the three types of features show an abnormal trend toward increasing icing within the same time window, the two types of features are superimposed for confirmation to obtain the fusion analysis results. When the fusion analysis results are consistent with the icing sign determination information, the icing sign determination information is upgraded to icing risk determination, confirming that the transmission line is at risk of icing.

[0009] Furthermore, the process of correcting for icing thickness includes: After confirming that there is a risk of icing on the transmission line, the initial estimate of the icing thickness in the preprocessed monitoring dataset is used as the correction base point, and this correction base point is used as the reference value for subsequent phased adjustments. The initial estimate of the icing thickness is the reference value of the icing thickness obtained by comparing the difference between the laser detection distance and the actual distance, the boundary scale of the icing appearance image, and the signal strength attenuation in the preprocessed monitoring dataset. Based on real-time environmental monitoring parameters, the instantaneous changes in temperature, humidity and wind speed are extracted. When the temperature is decreasing, the humidity is increasing and the wind speed is lower than the critical wind speed for ice reduction obtained from the historical database, the instantaneous changes are used as a correction factor for thickness increase and superimposed with the correction base point to form the first stage of corrected thickness. Based on the difference in boundary features between historical icing stripping information and current icing appearance image, the boundary expansion rate and morphological consistency are extracted. When the current boundary expansion rate exceeds the historical reference range or the boundary morphology is inconsistent with the historical reference state, the difference in boundary features of the current icing appearance image is used as a compensation factor for thickness growth and superimposed on the first stage of corrected thickness to form the second stage of corrected thickness. The first-stage correction thickness and the second-stage correction thickness are compared together. When the correction directions of the two stages are consistent, they are directly superimposed to obtain the corrected target icing thickness parameter. When the correction directions of the two stages are different, the second-stage correction thickness is given priority, and the first-stage correction thickness is adjusted in magnitude before being superimposed to obtain the corrected target icing thickness parameter.

[0010] Furthermore, the de-icing process includes: After determining the corrected target icing thickness parameter, the parameter is compared with a safety threshold. When the safety threshold is exceeded, the de-icing judgment process is initiated. The safety threshold is based on the icing thickness records stored in the historical database. The icing thickness data corresponding to the abnormal line operation status detected during operation is extracted, the thickness data is classified and statistically analyzed, and the smallest critical thickness range is selected as the safety threshold. Based on the boundary features of the icing appearance image and the difference between the laser detection distance and the actual distance, the thickness distribution information of the target icing thickness in the line length direction and cross-sectional direction is obtained; When the thickness distribution information shows local concentration, vibration de-icing is selected based on wind direction conditions in the environmental monitoring parameters; when the thickness distribution information shows overall coverage, electric heating de-icing is selected based on environmental monitoring parameters; when the thickness distribution information shows irregular accumulation accompanied by abnormal signals, electric pulse de-icing is selected. When the thickness distribution information simultaneously exhibits two or more characteristics such as local concentration, overall coverage, or irregular accumulation, multiple methods among electric heating, vibration, and pulse methods are selected and executed in combination based on environmental monitoring parameters. During the execution process, the de-icing method is adjusted according to the dynamic changes of the corrected target icing thickness parameters.

[0011] Furthermore, the process of tracking and generating early warnings for changes in icing thickness includes: Based on the corrected target icing thickness parameters, records and comparisons are made within continuous time windows. The direction and magnitude of thickness parameter changes in adjacent time windows are correlated and analyzed to form a thickness change sequence. When the thickness change sequence shows an upward trend in three consecutive time windows, the trend is judged as a state of continuous ice layer growth. When the trend determination matches the conditions in the environmental monitoring parameters that the temperature remains low and the humidity remains high, a de-icing early warning message is generated.

[0012] Furthermore, the process of data storage and database updates before and after de-icing includes: Before performing de-icing, images of the iced surface, environmental monitoring data, and condition monitoring data are collected and stored as baseline information before de-icing. After the de-icing process is completed, images of the iced surface, environmental monitoring data, and condition monitoring data are collected again and stored as reference information after de-icing. The baseline information before de-icing and the comparison information after de-icing are matched with the same timestamp and line identifier to form a record of de-icing effect; Update the de-icing effect records to the historical database.

[0013] Compared with existing technologies, the advantages of this invention are as follows: By simultaneously collecting environmental monitoring data such as temperature, humidity, and wind speed, as well as status monitoring data such as conductor tension, vibration amplitude, echo signal delay, and signal strength, a raw monitoring dataset covering meteorological conditions and line operating status is formed. This avoids the one-sidedness caused by a single data source in existing technologies, providing a more comprehensive basis for icing assessment. Based on this, the raw monitoring dataset is preprocessed, and the preprocessed icing appearance image is compared with historical images to obtain boundary change characteristics. Simultaneously, icing sign judgment information is generated by combining the changing trends of environmental monitoring parameters. This allows for early identification of risks before icing causes serious impact, enabling monitoring and early warning of early icing conditions. Furthermore, by jointly analyzing the difference between laser detection distance and actual distance, echo signal delay and signal strength deviation, and icing boundary change characteristics, and setting at least two simultaneous anomalies as confirmation conditions, misjudgments caused by fluctuations in a single monitoring parameter are effectively avoided, making icing risk assessment more reliable. Upon confirming the risk of icing, this invention does not directly use a single thickness measurement. Instead, it uses an initial estimate of the icing thickness obtained through multi-feature comparison as a correction base point. This is then combined with real-time environmental monitoring parameters, historical icing stripping information, and boundary feature differences for phased correction, resulting in a corrected target icing thickness parameter. This phased correction mechanism improves the accuracy and stability of the thickness results. Furthermore, when the target icing thickness parameter exceeds a safety threshold, this invention can select the most suitable de-icing method based on thickness distribution characteristics and environmental monitoring parameters, achieving differentiated processing under different icing morphologies. This approach is more adaptable and flexible than existing technologies that rely on a single de-icing method. Simultaneously, this invention tracks the continuous changes in the target icing thickness parameter, generating a de-icing warning when it shows an upward trend for three consecutive time windows. This enables early response to rapid ice development, enhancing the safety protection capabilities of the line operation. Finally, after completing the de-icing process, this invention stores and updates the icing appearance images before and after de-icing, along with the corresponding environmental and status monitoring data, to a historical database, ensuring traceability of the judgment and correction process.

[0014] On the other hand, this application also provides an environmental perception-based power grid de-icing system for applying the above-mentioned environmental perception-based power grid de-icing method, comprising: The data acquisition module collects temperature, humidity and wind speed as environmental monitoring data, and collects conductor tension, vibration amplitude, echo signal delay and signal strength as condition monitoring data. It also combines the environmental monitoring data and the condition monitoring data to obtain the raw monitoring dataset. The data preprocessing module preprocesses the original monitoring dataset to obtain a preprocessed monitoring dataset. It compares the ice appearance images in the preprocessed monitoring dataset with historical images to obtain the ice boundary change characteristics. Based on the change trend of environmental monitoring parameters in the preprocessed monitoring dataset, it generates ice sign judgment information. The risk assessment module analyzes the difference between the laser detection distance and the actual distance, the echo signal delay and signal strength deviation, and the icing boundary change characteristics in the preprocessed monitoring dataset. When at least two of the laser detection distance and the actual distance, the echo signal delay and signal strength deviation, and the icing boundary change characteristics are abnormal at the same time, the icing sign assessment information is confirmed as an icing risk assessment, and it is determined that there is an icing risk in the transmission line. The thickness correction module, after determining that there is a risk of icing on the transmission line, uses the initial estimate of the icing thickness in the preprocessed monitoring dataset as the correction base point, obtains the first-stage corrected thickness based on the real-time environmental monitoring parameters, and then obtains the second-stage corrected thickness based on the boundary feature difference between historical icing stripping information and icing appearance images. The first-stage corrected thickness and the second-stage corrected thickness are superimposed to obtain the corrected target icing thickness parameters. The de-icing control module selects a de-icing method based on the target ice thickness distribution and environmental monitoring parameters when the target ice thickness parameter exceeds the safety threshold, and performs de-icing treatment on the surface of the transmission line. The early warning module tracks the continuous changes in the target ice thickness parameter. When the ice thickness shows an upward trend for three consecutive time windows, it generates a de-icing early warning message. The data update module, after completing the de-icing process, synchronously stores the images of the icy transmission lines before and after de-icing, along with the corresponding environmental monitoring data and status monitoring data, and updates the historical database.

[0015] It is understandable that the aforementioned environmental perception-based power grid de-icing system has the same beneficial effects, and will not be elaborated upon here. Attached Figure Description

[0016] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A flowchart illustrating a power grid de-icing method based on environmental perception, provided for an embodiment of the present invention; Figure 2 This is a structural block diagram of a power grid de-icing system based on environmental perception, provided for an embodiment of the present invention. Detailed Implementation

[0017] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey its scope to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features described herein can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0018] In some embodiments of this application, see Figure 1 As shown, a power grid de-icing method based on environmental perception includes: S100: Collects temperature, humidity and wind speed as environmental monitoring data, collects conductor tension, vibration amplitude, echo signal delay and signal strength as condition monitoring data, and combines environmental monitoring data and condition monitoring data to obtain the raw monitoring dataset; S200: The original monitoring dataset is preprocessed to obtain a preprocessed monitoring dataset. The ice appearance images in the preprocessed monitoring dataset are compared with historical images to obtain the ice boundary change characteristics. Based on the change trend of environmental monitoring parameters in the preprocessed monitoring dataset, ice sign judgment information is generated. S300: Based on the analysis of the difference between laser detection distance and actual distance, echo signal delay and signal strength deviation, and icing boundary change characteristics in the preprocessed monitoring dataset, when at least two of the differences between laser detection distance and actual distance, echo signal delay and signal strength deviation, and icing boundary change characteristics are abnormal at the same time, the icing sign judgment information is confirmed as the icing risk judgment, and it is determined that there is an icing risk in the transmission line. S400: After determining that there is a risk of icing on the transmission line, the initial estimate of the icing thickness in the preprocessed monitoring dataset is used as the correction base point. The first stage correction thickness is obtained based on the real-time environmental monitoring parameters. The second stage correction thickness is obtained based on the difference in boundary features between historical icing stripping information and icing appearance images. The first stage correction thickness and the second stage correction thickness are superimposed to obtain the corrected target icing thickness parameter. S500: When the target icing thickness parameter exceeds the safety threshold, the de-icing method is selected based on the target icing thickness distribution and environmental monitoring parameters, and the surface of the transmission line is de-iced. S600: Tracks the continuous changes in the target icing thickness parameter. When the icing thickness shows an upward trend for three consecutive time windows, it generates a de-icing warning. S700: After de-icing is completed, the images of the transmission line before and after de-icing, along with the corresponding environmental monitoring data and condition monitoring data, are stored synchronously and the historical database is updated.

[0019] Specifically, the method involves simultaneously collecting environmental monitoring data and condition monitoring data during the operation of the transmission line to form a raw monitoring dataset. Environmental monitoring data reflects the real-time status of the external environment in which the transmission line is located, while condition monitoring data reflects the response of the transmission line itself during operation. The raw monitoring dataset is preprocessed to obtain a preprocessed monitoring dataset. Then, the icing appearance images within this dataset are compared with historical images. Historical images are reference images collected and stored in a database when the transmission line is in a normal or known icing state. The comparison between the two datasets reveals the characteristics of icing boundary changes. Simultaneously, combined with the changing trends of environmental monitoring parameters in the preprocessed monitoring dataset, indicator information is generated to determine whether icing is likely to occur on the transmission line. Subsequently, a comprehensive analysis is performed based on the difference between the laser detection distance and the actual distance, echo signal delay, signal strength deviation, and icing boundary change characteristics. When at least two of these characteristics are simultaneously abnormal, the aforementioned indicator information is confirmed as a risk assessment, indicating that the transmission line already faces the risk of icing. After risk confirmation, an initial estimate of the icing thickness is extracted from the pre-processed monitoring dataset as a correction baseline. This initial estimate is reasonably limited by referencing historical icing stripping information, which records boundary and thickness changes of the transmission line during past icing stripping processes. Based on this initial estimate, phased corrections are performed in conjunction with real-time conditions to obtain the corrected target icing thickness parameter. When this parameter exceeds a set safety threshold, an appropriate method is selected to de-ic the transmission line surface based on the target icing thickness distribution and environmental conditions. The safety threshold is statistically set based on icing thickness records in the historical database. Simultaneously, continuous changes in the target icing thickness parameter are tracked. When it shows an upward trend for three consecutive time windows, a de-icing warning is generated. After the de-icing operation is completed, images of the iced appearance of the transmission line before and after de-icing, along with corresponding environmental monitoring data and status monitoring data, are stored and updated to the historical database, providing foundational data for subsequent risk assessment and thickness correction. For example, during the operation of a power transmission line in a mountainous area, environmental monitoring data showed a continuous decrease in temperature, an increase in humidity, and low wind speed. Status monitoring data showed increased echo signal delay and weakened signal strength. Simultaneously, comparison of the icing appearance image with historical images showed an expansion of the icing boundary. Analysis of this data revealed at least two abnormal trends, thus confirming the icing indications as a risk assessment. Subsequently, an initial estimate of the icing thickness was limited using historical icing removal information as a reference, and a corrected target icing thickness parameter was obtained by combining real-time monitoring conditions. When this parameter exceeded a safety threshold, an appropriate method was selected to de-ic the line. After the de-icing operation was completed, the images before and after de-icing, along with relevant monitoring data, were stored and updated to the historical database, providing a reference for subsequent risk assessment and thickness correction.

[0020] Understandably, the joint collection and analysis of environmental monitoring data and condition monitoring data, compared to methods relying on a single parameter, can avoid misjudgments caused by fluctuations in a single indicator, thereby improving the reliability of icing identification. By comparing icing appearance images with historical images, the changes in ice layer boundaries can be visually presented, providing a morphological basis for determining icing signs. Combining the difference between laser detection distance and actual distance, echo signal delay, signal strength deviation, and icing boundary change characteristics for joint analysis, and confirming risk only when at least two types of features are simultaneously abnormal, can reduce false alarm rates and improve the accuracy of risk assessment. Introducing historical icing stripping information during the thickness correction process ensures that the initial thickness... The estimated values ​​are closer to the actual situation, reducing the impact of single measurement errors. The corrected thickness parameters are obtained through step-by-step correction in the first and second stages, ensuring the scientific and rational timing of de-icing. Triggering de-icing when the thickness parameters exceed the safety threshold can prevent the line from operating at dangerous thicknesses for a long time. Generating early warning information when the thickness parameters maintain an upward trend within a continuous time window allows for early intervention during the risk accumulation stage. After de-icing is completed, the images and monitoring data before and after de-icing are stored and updated to the historical database, which can continuously improve the reference information needed for subsequent judgments and corrections, thereby forming a closed-loop monitoring and control mechanism and improving the timeliness, accuracy, and safety of power grid de-icing.

[0021] In some embodiments of this application, the process of preprocessing the original monitoring dataset to obtain a preprocessed monitoring dataset includes: The environmental monitoring data and condition monitoring data in the original monitoring dataset are aligned according to timestamps and organized according to a unified data format specification to obtain an initial data sequence with consistent labels. The environmental monitoring data and condition monitoring data in the initial data sequence are subjected to outlier removal and normalization processes to obtain standardized environmental monitoring parameters and standardized condition monitoring parameters. The ice-covered appearance images in the original monitoring dataset are subjected to feature extraction and enhancement processing. Ice boundary feature data are obtained based on image edge clarity and grayscale gradient information. The ice boundary feature data are associated with the corresponding timestamps and stored to obtain the preprocessed ice-covered image feature set. The standardized environmental monitoring parameters and standardized condition monitoring parameters are matched with the preprocessed icing image feature set to obtain the preprocessed monitoring dataset.

[0022] Specifically, in the preprocessing of the original monitoring dataset, the first step is to align the collected environmental monitoring data and condition monitoring data according to timestamps to ensure comparability of data from different sources within the same time window. After timestamp alignment, the data is organized according to a unified data format specification to form an initial data sequence with consistency markers. These consistency markers are used to identify the correspondence between data in terms of time and source. Subsequently, outlier removal and normalization are performed on the environmental monitoring data and condition monitoring data in the initial data sequence, removing extreme outliers caused by external interference and unifying parameters of different dimensions to a relatively standard range, thus obtaining standardized environmental monitoring parameters and standardized condition monitoring parameters. Simultaneously, feature extraction and enhancement processing are performed on the icing appearance images in the original monitoring dataset. By enhancing contrast and edge information, feature data reflecting the icing boundary morphology is obtained, and the boundary features are associated and stored with the corresponding timestamps to form a preprocessed icing image feature set. Finally, the standardized environmental monitoring parameters and condition monitoring parameters are matched with the preprocessed icing image feature set to ensure that data from different sources are correlated at the same point in time, thereby obtaining the preprocessed monitoring dataset. For example, during a single monitoring operation, the collected temperature, humidity, and wind speed data differed temporally from the conductor tension, vibration amplitude, and echo signal data. By aligning the data according to timestamps, parameters collected at different times were unified into the same time window. Then, the data was organized according to a standardized data format to form an initial data sequence with consistent labeling. Subsequently, outlier removal was performed on this sequence to eliminate extreme values ​​caused by transient interference. Normalization was then used to unify data of different dimensions to a comparable range, resulting in standardized environmental monitoring parameters and condition monitoring parameters. Simultaneously, feature extraction and enhancement processing were performed on synchronously collected images of icing appearance, highlighting boundary regions in the images to obtain icing boundary feature data. This data was then correlated with the corresponding timestamps to form a preprocessed image feature set. Finally, the standardized numerical data and the image feature set were matched to obtain the preprocessed monitoring dataset.

[0023] In some embodiments of this application, the process of generating icing sign determination information includes: Based on the standardized environmental monitoring parameters in the preprocessed monitoring dataset, the changing trends of temperature, humidity and wind speed are extracted. When the change amplitude within a continuous time window exceeds the set interval or the fluctuation frequency continues to increase, it is considered as obtaining the environmental characteristics of icing conditions, and thus obtaining the environmental trend characteristic sequence. Based on the standardized status monitoring parameters in the preprocessed monitoring dataset, the change trajectory of echo signal delay and echo signal intensity is extracted. When the delay continues to increase or the intensity continues to decrease, it is considered as a manifestation of dielectric coverage on the line surface, and a signal anomaly feature sequence is obtained. Based on the feature set of icing images in the preprocessed monitoring dataset, the changes in the expansion rate and edge sharpness of the image boundaries are extracted. When the boundary expansion rate shows an increasing trend or the edge sharpness decreases, it is determined to be a sign of ice adhesion or thickening, and an image difference feature sequence is obtained. The environmental trend feature sequence, signal anomaly feature sequence, and image difference feature sequence are fused together. When at least two types of feature sequences point to the possibility of icing, the presence of icing signs on the line is confirmed, and the confirmation result is output as icing sign judgment information.

[0024] Specifically, in generating information to determine icing signs, the standardized environmental monitoring parameters in the pre-processed monitoring dataset are first analyzed, extracting the trajectories of temperature, humidity, and wind speed changes over continuous time windows. When the temperature gradually decreases over multiple time windows, the humidity gradually increases over the same period, and the wind speed is insufficient to reduce the accumulated ice layer, this trend is integrated into an environmental trend feature sequence to characterize that external conditions are developing in a direction favorable to icing. Secondly, the standardized state monitoring parameters in the pre-processed monitoring dataset are processed, converting changes in echo signal delay and signal strength into trend information. When the echo signal delay gradually increases or the signal strength continuously decreases over adjacent time windows, this change is identified as a gradual thickening of the medium on the line surface, and a signal anomaly feature sequence is generated accordingly. Simultaneously, the icing image feature set in the pre-processed monitoring dataset is analyzed, tracking the expansion rate of image boundaries and changes in edge sharpness over multiple time windows. When the boundaries gradually expand outward or the edge blurring increases, this result is organized into an image difference feature sequence to characterize the continuous adhesion and thickening of the icing layer. Finally, the environmental trend feature sequence, signal anomaly feature sequence, and image difference feature sequence are fused according to a unified time window. When at least two of the three feature types simultaneously show an icing trend within the same time window, the presence of icing signs on the transmission line is confirmed, and this confirmation result is output as icing sign judgment information for subsequent risk assessment and thickness correction. For example, during a winter operation monitoring process, environmental monitoring data for three consecutive time windows showed a continuous decrease in temperature, an increase in humidity from 60% to 85% during the same period, and a persistently low wind speed. Simultaneously, status monitoring data showed a gradual increase in the echo signal delay value and a continuous decrease in signal strength. Within the same time window, comparison between the icing appearance image and historical images showed an expansion of the image boundary range and a significant decrease in edge clarity. After fusing the above data, at least two of the environmental trend feature sequence, signal anomaly feature sequence, and image difference feature sequence simultaneously point to an increasing icing trend. Therefore, the system confirms the presence of icing signs on the line and outputs icing sign judgment information.

[0025] In some embodiments of this application, after the icing sign determination information is formed, the icing sign determination information is cross-validated, including: The environmental monitoring parameters in the preprocessed monitoring dataset are matched with the transmission line condition monitoring parameters to verify consistency. When the temperature and humidity conditions are consistent with the trend of changes in line tension or vibration amplitude, the consistency verification is used as the basis for confirming the validity of the indication. The boundary features of the icing appearance image are compared with normal operation images in the historical database. When the direction of boundary expansion matches the icing growth trajectory in the historical record, the comparison result is used as the basis for confirming the reliability of the indication. When both the consistency verification and comparison results are valid, the icing sign determination information is confirmed as valid.

[0026] Specifically, after generating information to determine icing signs, cross-validation is required to avoid misjudgments due to a single feature. First, the environmental monitoring parameters in the preprocessed monitoring dataset are matched with the transmission line condition monitoring parameters, comparing the trends of temperature and humidity with the trends of conductor tension or vibration amplitude. When the external conditions of gradually decreasing temperature and gradually increasing humidity are consistent with the internal response of increased tension or intensified vibration amplitude, the correspondence is considered reasonable and used as the basis for confirming the validity of the sign. Second, the boundary features of the icing appearance images in the preprocessed monitoring dataset are compared with normal operation images stored in the historical database. Normal operation images refer to reference images collected and stored when the transmission line is not iced or is in a negligible icing state. Through comparison, when the boundary expansion direction of the icing appearance image matches the known icing growth trajectory in historical records, the comparison result can be used as the basis for confirming the reliability of the sign. Finally, when both the consistency verification and image comparison results point to an increasing icing trend, the icing sign determination information can be confirmed as a valid step.

[0027] In some embodiments of this application, the process of determining the risk of icing includes: After obtaining the information on signs of icing, the difference between the laser detection distance and the actual distance in the preprocessed monitoring dataset is compared. When the difference continues to increase and exceeds the set threshold, the difference is used as the quantitative basis for thickness accumulation to obtain the distance difference feature. Based on the analysis of echo signal delay and echo signal intensity deviation in the preprocessed monitoring dataset, when the delay value shows a continuous increase and the intensity value shows a continuous decrease, this joint feature is used as the basis for the propagation anomaly caused by the attached material, and the signal deviation feature is obtained. Trend tracking is performed based on the ice boundary feature data in the preprocessed monitoring dataset. When the boundary expansion rate is higher than the historical average or the boundary morphology changes irregularly, this trend is used as the basis for the rapid evolution of the ice layer, and boundary anomaly features are obtained. Distance difference features, signal deviation features, and boundary anomaly features are matched according to timestamps, and consistency is compared based on the direction and magnitude of change of each feature. When at least two of the three types of features show an abnormal trend toward increasing icing within the same time window, the two types of features are superimposed for confirmation to obtain the fusion analysis results. When the fusion analysis results are consistent with the icing sign determination information, the icing sign determination information is upgraded to icing risk determination, confirming that the transmission line is at risk of icing.

[0028] Specifically, in assessing icing risk, the laser detection distance is first compared. Laser detection distance refers to the one-way measurement distance calculated from the laser beam's path from the emission point to the conductor's outer surface and back. Actual distance refers to the reference distance determined based on the conductor's installation geometry and historical benchmark measurements under icing-free conditions. When the comparison shows that the difference between the detected distance and the actual distance gradually increases within a continuous time window and exceeds the minimum abnormal difference obtained from historical database statistics, this difference is considered an indirect quantitative indicator reflecting the ice accumulation trend, thus forming a distance difference characteristic. Secondly, the echo signal delay and signal strength in the pre-processed monitoring dataset are jointly analyzed. When the delay value continuously increases and the signal strength gradually decreases, it indicates that there are adhering substances on the line surface affecting electromagnetic wave propagation, thus forming a signal deviation characteristic. Simultaneously, the boundary feature data of the icing appearance image are dynamically tracked. When the boundary expansion rate exceeds the historical average or the boundary morphology shows irregular changes, it indicates that the icing layer is in a rapid evolution state; this result is extracted as a boundary anomaly characteristic. Subsequently, the distance difference feature, signal deviation feature, and boundary anomaly feature are matched according to their timestamps, allowing for comparison within the same time window, and consistency is compared based on the direction and magnitude of change. When at least two of the three types of features simultaneously show an increasing trend of icing within the same time window, the two types of features are superimposed for confirmation. Superposition confirmation refers to accumulating the number of features that meet the conditions within the same time window, and using feature changes with consistent trends as a common basis to form a fusion analysis result. Finally, when the fusion analysis result is consistent with the previously generated icing sign judgment information, the icing sign judgment information can be upgraded to an icing risk judgment, clearly indicating that the transmission line is in a state of icing risk. For example, during an operational monitoring process, the laser distance detection display showed that the measured distance between the outer surface of the conductor and the reference position gradually shortened within three adjacent time windows. The difference formed after comparing with the actual distance increased from 5 mm to 12 mm, exceeding the minimum abnormal difference of 8 mm obtained from the historical database statistics, thus generating a distance difference feature. At the same time, the echo signal delay increased from the original 0.8 microseconds to 1.1 microseconds within the same time period, and the signal strength decreased by about 12%, which was identified as a signal deviation feature. Analysis of the icing appearance images showed that the boundary expansion rate was higher than the historical average and exhibited irregular expansion directions, which were extracted as boundary anomaly features. After timestamp comparison, both the distance difference feature and the boundary anomaly feature pointed to an ice thickening trend within the same time window, meeting the overlay confirmation criteria and forming a fusion analysis result. When this fusion analysis result was consistent with the previously obtained icing sign assessment information, the icing sign assessment information was ultimately upgraded to an icing risk assessment, confirming that the transmission line was in a state of icing risk.

[0029] In some embodiments of this application, the process of correcting icing thickness includes: After confirming the risk of icing on the transmission line, the initial estimate of the icing thickness in the preprocessed monitoring dataset is used as the correction base point, and this correction base point is used as the reference value for subsequent phased adjustments. The initial estimate of the icing thickness is the reference value of the icing thickness obtained by comparing the difference between the laser detection distance and the actual distance, the boundary scale of the icing appearance image, and the signal strength attenuation in the preprocessed monitoring dataset. Based on real-time environmental monitoring parameters, the instantaneous changes in temperature, humidity and wind speed are extracted. When the temperature is decreasing, the humidity is increasing and the wind speed is lower than the critical wind speed for ice reduction obtained from the historical database, the instantaneous changes are used as a correction factor for thickness increase and superimposed with the correction base point to form the first stage of corrected thickness. Based on the difference in boundary features between historical icing stripping information and current icing appearance image, the boundary expansion rate and morphological consistency are extracted. When the current boundary expansion rate exceeds the historical reference range or the boundary morphology is inconsistent with the historical reference state, the difference in boundary features of the current icing appearance image is used as a compensation factor for thickness growth and superimposed on the first stage of corrected thickness to form the second stage of corrected thickness. The first-stage correction thickness and the second-stage correction thickness are compared together. When the correction directions of the two stages are consistent, they are directly superimposed to obtain the corrected target icing thickness parameter. When the correction directions of the two stages are different, the second-stage correction thickness is given priority, and the first-stage correction thickness is adjusted in magnitude before being superimposed to obtain the corrected target icing thickness parameter.

[0030] Specifically, in the process of correcting icing thickness, the initial estimate of icing thickness needs to be determined first. This initial estimate is not a single measurement result, but a reference value obtained by comparing multiple features. Specifically, the distance obtained by laser detection is compared with the reference physical distance in the uniced state; the difference reflects the degree to which the conductor surface has shifted outward due to ice coverage. Next, the boundary scale of the icing appearance image is compared with normal operation images stored in the historical database. When the boundary expansion exceeds the boundary reference range statistically obtained from the historical database, this expansion value is considered a direct indication of ice thickening. Simultaneously, the signal strength attenuation within a continuous time window is analyzed. By comparing the attenuation amplitude with typical icing scenarios in the historical database, this serves as indirect evidence of propagation anomalies caused by ice adhesion. When these three types of features are compared within the same time window, and all three results—a gradually increasing laser detection difference, a continuously expanding boundary scale, and a continuously decreasing signal strength—point to increasing ice thickness, it is considered a consistent trend indicating ice growth. At this point, a comprehensive judgment is made to obtain the icing thickness reference value, which is then used as the correction base. Subsequently, real-time changes in temperature, humidity, and wind speed are extracted based on real-time environmental monitoring parameters. When the temperature shows a decreasing trend, humidity shows an increasing trend, and wind speed is insufficient to reduce the ice layer, this trend is used as a thickness increase correction factor and superimposed on the correction base point to obtain the first-stage corrected thickness. Then, a difference comparison is performed between historical icing peeling information and the boundary features of the current icing appearance image. When the current boundary expansion rate exceeds the upper limit of the reference range stored in the historical database, or when the boundary morphology is inconsistent with the historical peeling state, this difference is used as a thickness increase compensation factor and superimposed on the first-stage corrected thickness to obtain the second-stage corrected thickness. Finally, the first-stage and second-stage corrected thicknesses are comprehensively compared. When their correction directions are consistent, the values ​​are directly added to form the corrected target icing thickness parameter. When their correction directions differ, the second-stage corrected thickness is prioritized, and the magnitude of the first-stage corrected thickness is adjusted before addition to obtain the final correction result. By using a reference range determined by a historical database as a benchmark, comparing multiple features to form reference values, and gradually adding and correcting them, errors caused by a single monitoring condition can be effectively avoided, thus improving the accuracy and stability of ice thickness determination.

[0031] In some embodiments of this application, the de-icing process includes: After determining the corrected target icing thickness parameter, the parameter is compared with the safety threshold. When the safety threshold is exceeded, the de-icing judgment process is initiated. The safety threshold is based on the icing thickness records stored in the historical database. The icing thickness data corresponding to the abnormal line operation status detected during operation is extracted, the thickness data is classified and statistically analyzed, and the smallest critical thickness range is selected as the safety threshold. Based on the boundary features of the icing appearance image and the difference between the laser detection distance and the actual distance, the thickness distribution information of the target icing thickness in the line length direction and cross-sectional direction is obtained; When the thickness distribution information shows local concentration, vibration de-icing is selected based on wind direction conditions in the environmental monitoring parameters; when the thickness distribution information shows overall coverage, electric heating de-icing is selected based on environmental monitoring parameters; when the thickness distribution information shows irregular accumulation accompanied by abnormal signals, electric pulse de-icing is selected. When the thickness distribution information simultaneously exhibits two or more characteristics such as local concentration, overall coverage, or irregular accumulation, multiple methods among electric heating, vibration, and pulse methods are selected and executed in combination based on environmental monitoring parameters. During the execution process, the de-icing method is adjusted according to the dynamic changes of the corrected target icing thickness parameters.

[0032] Specifically, during the de-icing process, the first step is to perform a threshold comparison on the corrected target icing thickness parameter, comparing it to a safety threshold. When the parameter value exceeds the safety threshold, the de-icing determination process is initiated. The safety threshold is not set arbitrarily, but rather is a critical range derived from icing thickness records stored in a historical database. Specifically, icing thickness data corresponding to abnormal operating conditions such as abnormal line tension, sudden changes in vibration amplitude, or sharp attenuation of echo signals caused by icing is extracted from a large amount of historical data. This data is then categorized and statistically analyzed, and the smallest critical thickness range is selected as the safety threshold under the current operating conditions, ensuring the objectivity and reliability of the determination. Subsequently, by extracting the scale of the boundary features of the icing appearance image and comparing it with the difference between the laser detection distance and the actual distance, the distribution trend of the target icing thickness along the transmission line length and the coverage status along the cross-section are obtained, forming thickness distribution information. The appropriate de-icing method is selected based on the distribution information: When the thickness distribution shows localized concentration, it indicates that the ice is mainly accumulated in certain local sections. In this case, vibration de-icing is selected based on wind direction conditions in the environmental monitoring parameters to quickly remove the local ice layer using vibration. When the thickness distribution shows overall coverage, it indicates that the line surface has been iced over a large area. In this case, electric heating de-icing is selected based on environmental monitoring parameters to increase the overall temperature and accelerate ice removal. When the thickness distribution shows irregular accumulation accompanied by signal anomalies, it indicates that the ice morphology is complex and there is media interference. In this case, electric pulse de-icing is selected to use instantaneous energy to break the ice layer interface. If the thickness distribution shows two or more of the characteristics of localized concentration, overall coverage, or irregular accumulation, multiple methods, including electric heating, vibration, and pulse methods, are selected and executed in combination based on environmental monitoring parameters. During the combined execution, the dynamic changes of the corrected target ice thickness parameters are tracked and adjusted in real time. For example, the pulse intensity is increased when the thickness descent rate is insufficient, and the vibration time is extended when local residual ice is not completely removed, thereby achieving efficient and controllable de-icing results.

[0033] In some embodiments of this application, the process of tracking and generating early warnings for changes in icing thickness includes: Based on the corrected target icing thickness parameters, records and comparisons are made within continuous time windows. The direction and magnitude of thickness parameter changes in adjacent time windows are correlated and analyzed to form a thickness change sequence. When the thickness change sequence shows an upward trend in three consecutive time windows, the trend is judged as a state of continuous ice layer growth. When the trend determination matches the conditions in the environmental monitoring parameters that the temperature remains low and the humidity remains high, a de-icing early warning message is generated.

[0034] Specifically, in the process of tracking and generating early warnings for changes in ice thickness, the first step is to continuously record the corrected target ice thickness parameters, compare the thickness values ​​within multiple time windows, and calculate the direction and magnitude of change between adjacent time windows to form a thickness change sequence. This thickness change sequence not only reflects the absolute change in value but also the continuous trend of change. When the analysis results show that the thickness change sequence shows a positive increase in three consecutive time windows, that is, the thickness parameter in each time window is greater than the value in the previous time window, this continuous trend is judged as a state of continuous ice growth. When the temperature parameter remains at a low level in consecutive time windows and the humidity parameter remains at a high level in the same time window, it indicates that the external meteorological conditions are consistent with the trend of increasing thickness. At this time, a de-icing early warning is generated and output to the operation management stage to indicate that de-icing may be necessary.

[0035] In some embodiments of this application, the process of data storage and database update before and after de-icing includes: Before performing de-icing, images of the iced surface, environmental monitoring data, and condition monitoring data are collected and stored as baseline information before de-icing. After the de-icing process is completed, images of the iced surface, environmental monitoring data, and condition monitoring data are collected again and stored as reference information after de-icing. The baseline information before de-icing and the comparison information after de-icing are matched with the same timestamp and line identifier to form a record of de-icing effect; Update the de-icing effect records to the historical database.

[0036] Specifically, in the process of data storage and database updates before and after de-icing, firstly, before the de-icing process begins, images of the iced appearance of the transmission line, along with corresponding environmental and status monitoring data, are collected. These data are bound to the same timestamp and line identifier and stored as pre-de-icing baseline information for subsequent comparison. After the de-icing process is completed, images of the iced appearance of the transmission line, environmental monitoring data, and status monitoring data are collected again and organized according to the same timestamp and line identifier, stored as post-de-icing comparison information. Subsequently, the pre-de-icing baseline information and the post-de-icing comparison information are matched to compare the differences in iced appearance images, changes in environmental parameters, and changes in status parameters before and after de-icing. The comparison results are summarized into a de-icing effect record. Finally, this de-icing effect record is updated to the historical database, so that the historical database not only stores monitoring data at each stage but also includes effect comparison information before and after each de-icing process, thus providing a reliable historical basis for subsequent icing trend analysis and de-icing strategy optimization.

[0037] In another preferred embodiment based on the above embodiments, see [reference] Figure 2 As shown, this embodiment provides a power grid de-icing system based on environmental perception, including: The data acquisition module collects temperature, humidity and wind speed as environmental monitoring data, and collects conductor tension, vibration amplitude, echo signal delay and signal strength as condition monitoring data. It also combines the environmental monitoring data and the condition monitoring data to obtain the raw monitoring dataset. The data preprocessing module preprocesses the original monitoring dataset to obtain a preprocessed monitoring dataset. It compares the ice appearance images in the preprocessed monitoring dataset with historical images to obtain the ice boundary change characteristics. Based on the change trend of environmental monitoring parameters in the preprocessed monitoring dataset, it generates ice sign judgment information. The risk assessment module analyzes the difference between the laser detection distance and the actual distance, the echo signal delay and signal strength deviation, and the icing boundary change characteristics in the preprocessed monitoring dataset. When at least two of the laser detection distance and the actual distance, the echo signal delay and signal strength deviation, and the icing boundary change characteristics are abnormal at the same time, the icing sign assessment information is confirmed as an icing risk assessment, and it is determined that there is an icing risk in the transmission line. The thickness correction module, after determining that there is a risk of icing on the transmission line, uses the initial estimate of the icing thickness in the preprocessed monitoring dataset as the correction base point, obtains the first-stage corrected thickness based on the real-time environmental monitoring parameters, and then obtains the second-stage corrected thickness based on the boundary feature difference between historical icing stripping information and icing appearance images. The first-stage corrected thickness and the second-stage corrected thickness are superimposed to obtain the corrected target icing thickness parameters. The de-icing control module selects a de-icing method based on the target ice thickness distribution and environmental monitoring parameters when the target ice thickness parameter exceeds the safety threshold, and performs de-icing treatment on the surface of the transmission line. The early warning module tracks the continuous changes in the target ice thickness parameter. When the ice thickness shows an upward trend for three consecutive time windows, it generates a de-icing early warning message. The data update module, after completing the de-icing process, synchronously stores the images of the icy transmission lines before and after de-icing, along with the corresponding environmental monitoring data and status monitoring data, and updates the historical database.

[0038] Understandably, by sequentially setting up interconnected functional modules for data collection, data preprocessing, risk assessment, thickness correction, de-icing control, early warning, and data updating, this invention enables dynamic monitoring, real-time assessment, and effective handling of icing conditions throughout the entire operation of transmission lines. The data acquisition module ensures complete collection of environmental parameters and line operating status, providing a foundation for subsequent analysis; the data preprocessing module standardizes and extracts features from raw monitoring data, making the data more comparable and usable; the risk assessment module reduces the risk of misjudgment due to single signal errors through multi-feature cross-analysis; the thickness correction module uses real-time meteorological conditions and historical stripping information to adjust the thickness results in stages, improving the accuracy of thickness assessment; the de-icing control module achieves differentiated handling of different icing morphologies through joint judgment of thickness distribution and environmental conditions; the early warning module generates early warning information in advance when the ice layer grows rapidly through continuous trend analysis, improving the foresight of protection; and the data updating module stores comparison information into a historical database after each de-icing process, ensuring the historical reference required for subsequent assessment and correction. Thus, the entire system constructs a closed-loop process covering data acquisition, analysis, judgment, execution, and feedback. This not only improves the real-time performance and reliability of power transmission line de-icing but also enables continuous optimization of judgment and control strategies through the accumulation of historical data.

[0039] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A power grid de-icing method based on environmental perception, characterized in that, include: Temperature, humidity and wind speed are collected as environmental monitoring data, and conductor tension, vibration amplitude, echo signal delay and signal strength are collected as condition monitoring data. The environmental monitoring data and condition monitoring data are then combined to obtain the raw monitoring dataset. The original monitoring dataset is preprocessed to obtain a preprocessed monitoring dataset. The ice appearance images in the preprocessed monitoring dataset are compared with historical images to obtain the ice boundary change characteristics. Based on the change trend of environmental monitoring parameters in the preprocessed monitoring dataset, ice sign judgment information is generated. Based on the analysis of the difference between laser detection distance and actual distance, echo signal delay and signal strength deviation, and icing boundary change characteristics in the preprocessed monitoring dataset, when at least two of the following are simultaneously abnormal, the icing sign judgment information is confirmed as an icing risk judgment, and it is determined that the transmission line has an icing risk. After determining that there is a risk of icing on the transmission line, the initial estimate of the icing thickness in the preprocessed monitoring dataset is used as the correction base point. The first stage correction thickness is obtained based on the real-time environmental monitoring parameters. The second stage correction thickness is obtained based on the difference in boundary features between historical icing stripping information and icing appearance images. The first stage correction thickness and the second stage correction thickness are superimposed to obtain the corrected target icing thickness parameter. When the target ice thickness parameter exceeds the safety threshold, the de-icing method is selected based on the target ice thickness distribution and environmental monitoring parameters, and the surface of the transmission line is de-iced. The target ice thickness parameter is tracked continuously. When it shows an upward trend for three consecutive time windows, a de-icing warning is generated. After the de-icing process is completed, the images of the transmission lines before and after de-icing, along with the corresponding environmental monitoring data and condition monitoring data, are stored synchronously and the historical database is updated.

2. The power grid de-icing method based on environmental perception according to claim 1, characterized in that, The process of preprocessing the original monitoring dataset to obtain the preprocessed monitoring dataset includes: The environmental monitoring data and condition monitoring data in the original monitoring dataset are aligned according to timestamps and organized according to a unified data format specification to obtain an initial data sequence with consistent labels. The environmental monitoring data and condition monitoring data in the initial data sequence are subjected to outlier removal and normalization processes to obtain standardized environmental monitoring parameters and standardized condition monitoring parameters. The ice-covered appearance images in the original monitoring dataset are subjected to feature extraction and enhancement processing. Ice boundary feature data are obtained based on image edge clarity and grayscale gradient information. The ice boundary feature data are associated with the corresponding timestamps and stored to obtain the preprocessed ice-covered image feature set. The standardized environmental monitoring parameters and standardized condition monitoring parameters are matched with the preprocessed icing image feature set to obtain the preprocessed monitoring dataset.

3. The power grid de-icing method based on environmental perception according to claim 2, characterized in that, The process of generating information to determine signs of icing includes: Based on the standardized environmental monitoring parameters in the preprocessed monitoring dataset, the changing trends of temperature, humidity and wind speed are extracted. When the change amplitude within a continuous time window exceeds the set interval or the fluctuation frequency continues to increase, it is considered as obtaining the environmental characteristics of icing conditions, and thus obtaining the environmental trend characteristic sequence. Based on the standardized status monitoring parameters in the preprocessed monitoring dataset, the change trajectory of echo signal delay and echo signal intensity is extracted. When the delay continues to increase or the intensity continues to decrease, it is considered as a manifestation of dielectric coverage on the line surface, and a signal anomaly feature sequence is obtained. Based on the feature set of icing images in the preprocessed monitoring dataset, the changes in the expansion rate and edge sharpness of the image boundaries are extracted. When the boundary expansion rate shows an increasing trend or the edge sharpness decreases, it is determined to be a sign of ice adhesion or thickening, and an image difference feature sequence is obtained. The environmental trend feature sequence, signal anomaly feature sequence, and image difference feature sequence are fused together. When at least two types of feature sequences point to the possibility of icing, the presence of icing signs on the line is confirmed, and the confirmation result is output as icing sign judgment information.

4. The power grid de-icing method based on environmental perception according to claim 3, characterized in that, After the information indicating signs of icing is generated, the information is cross-validated, including: The environmental monitoring parameters in the preprocessed monitoring dataset are matched with the transmission line condition monitoring parameters to verify consistency. When the temperature and humidity conditions are consistent with the trend of changes in line tension or vibration amplitude, the consistency verification is used as the basis for confirming the validity of the indication. The boundary features of the icing appearance image are compared with normal operation images in the historical database. When the direction of boundary expansion matches the icing growth trajectory in the historical record, the comparison result is used as the basis for confirming the reliability of the indication. When both the consistency verification and comparison results are valid, the icing sign determination information is confirmed as valid.

5. The power grid de-icing method based on environmental perception according to claim 4, characterized in that, The process of assessing icing risk includes: After obtaining the information on signs of icing, the difference between the laser detection distance and the actual distance in the preprocessed monitoring dataset is compared. When the difference continues to increase and exceeds the set threshold, the difference is used as the quantitative basis for thickness accumulation to obtain the distance difference feature. Based on the analysis of echo signal delay and echo signal intensity deviation in the preprocessed monitoring dataset, when the delay value shows a continuous increase and the intensity value shows a continuous decrease, this joint feature is used as the basis for the propagation anomaly caused by the attached material, and the signal deviation feature is obtained. Trend tracking is performed based on the ice boundary feature data in the preprocessed monitoring dataset. When the boundary expansion rate is higher than the historical average or the boundary morphology changes irregularly, this trend is used as the basis for the rapid evolution of the ice layer, and boundary anomaly features are obtained. Distance difference features, signal deviation features, and boundary anomaly features are matched according to timestamps, and consistency is compared based on the direction and magnitude of change of each feature. When at least two of the three types of features show an abnormal trend toward increasing icing within the same time window, the two types of features are superimposed for confirmation to obtain the fusion analysis results. When the fusion analysis results are consistent with the icing sign determination information, the icing sign determination information is upgraded to icing risk determination, confirming that the transmission line is at risk of icing.

6. The power grid de-icing method based on environmental perception according to claim 5, characterized in that, The process of correcting for icing thickness includes: After confirming that there is a risk of icing on the transmission line, the initial estimate of the icing thickness in the preprocessed monitoring dataset is used as the correction base point, and this correction base point is used as the reference value for subsequent phased adjustments. The initial estimate of the icing thickness is the reference value of the icing thickness obtained by comparing the difference between the laser detection distance and the actual distance, the boundary scale of the icing appearance image, and the signal strength attenuation in the preprocessed monitoring dataset. Based on real-time environmental monitoring parameters, the instantaneous changes in temperature, humidity and wind speed are extracted. When the temperature is decreasing, the humidity is increasing and the wind speed is lower than the critical wind speed for ice reduction obtained from the historical database, the instantaneous changes are used as a correction factor for thickness increase and superimposed with the correction base point to form the first stage of corrected thickness. Based on the difference in boundary features between historical icing stripping information and current icing appearance image, the boundary expansion rate and morphological consistency are extracted. When the current boundary expansion rate exceeds the historical reference range or the boundary morphology is inconsistent with the historical reference state, the difference in boundary features of the current icing appearance image is used as a compensation factor for thickness growth and superimposed on the first stage of corrected thickness to form the second stage of corrected thickness. The first-stage correction thickness and the second-stage correction thickness are compared together. When the correction directions of the two stages are consistent, they are directly superimposed to obtain the corrected target icing thickness parameter. When the correction directions of the two stages are different, the second-stage correction thickness is given priority, and the first-stage correction thickness is adjusted in magnitude before being superimposed to obtain the corrected target icing thickness parameter.

7. The power grid de-icing method based on environmental perception according to claim 6, characterized in that, The de-icing process includes: After determining the corrected target icing thickness parameter, the parameter is compared with a safety threshold. When the safety threshold is exceeded, the de-icing judgment process is initiated. The safety threshold is based on the icing thickness records stored in the historical database. The icing thickness data corresponding to the abnormal line operation status detected during operation is extracted, the thickness data is classified and statistically analyzed, and the smallest critical thickness range is selected as the safety threshold. Based on the boundary features of the icing appearance image and the difference between the laser detection distance and the actual distance, the thickness distribution information of the target icing thickness in the line length direction and cross-sectional direction is obtained; When the thickness distribution information shows local concentration, vibration de-icing is selected based on wind direction conditions in the environmental monitoring parameters; when the thickness distribution information shows overall coverage, electric heating de-icing is selected based on environmental monitoring parameters; when the thickness distribution information shows irregular accumulation accompanied by abnormal signals, electric pulse de-icing is selected. When the thickness distribution information simultaneously exhibits two or more characteristics such as local concentration, overall coverage, or irregular accumulation, multiple methods among electric heating, vibration, and pulse methods are selected and executed in combination based on environmental monitoring parameters. During the execution process, the de-icing method is adjusted according to the dynamic changes of the corrected target icing thickness parameters.

8. The power grid de-icing method based on environmental perception according to claim 7, characterized in that, The process of tracking and generating early warnings for changes in icing thickness includes: Based on the corrected target icing thickness parameters, records and comparisons are made within continuous time windows. The direction and magnitude of thickness parameter changes in adjacent time windows are correlated and analyzed to form a thickness change sequence. When the thickness change sequence shows an upward trend in three consecutive time windows, the trend is judged as a state of continuous ice layer growth. When the trend determination matches the conditions in the environmental monitoring parameters that the temperature remains low and the humidity remains high, a de-icing early warning message is generated.

9. The power grid de-icing method based on environmental perception according to claim 8, characterized in that, The process of data storage and database updates before and after de-icing includes: Before performing de-icing, images of the iced surface, environmental monitoring data, and condition monitoring data are collected and stored as baseline information before de-icing. After the de-icing process is completed, images of the iced surface, environmental monitoring data, and condition monitoring data are collected again and stored as reference information after de-icing. The baseline information before de-icing and the comparison information after de-icing are matched with the same timestamp and line identifier to form a record of de-icing effect; Update the de-icing effect records to the historical database.

10. An environmentally-aware power grid de-icing system, used to apply the environmentally-aware power grid de-icing method as described in any one of claims 1-9, characterized in that, include: The data acquisition module collects temperature, humidity and wind speed as environmental monitoring data, and collects conductor tension, vibration amplitude, echo signal delay and signal strength as condition monitoring data. It also combines the environmental monitoring data and the condition monitoring data to obtain the raw monitoring dataset. The data preprocessing module preprocesses the original monitoring dataset to obtain a preprocessed monitoring dataset. It compares the ice appearance images in the preprocessed monitoring dataset with historical images to obtain the ice boundary change characteristics. Based on the change trend of environmental monitoring parameters in the preprocessed monitoring dataset, it generates ice sign judgment information. The risk assessment module analyzes the difference between the laser detection distance and the actual distance, the echo signal delay and signal strength deviation, and the icing boundary change characteristics in the preprocessed monitoring dataset. When at least two of the laser detection distance and the actual distance, the echo signal delay and signal strength deviation, and the icing boundary change characteristics are abnormal at the same time, the icing sign assessment information is confirmed as an icing risk assessment, and it is determined that there is an icing risk in the transmission line. The thickness correction module, after determining that there is a risk of icing on the transmission line, uses the initial estimate of the icing thickness in the preprocessed monitoring dataset as the correction base point, obtains the first-stage corrected thickness based on the real-time environmental monitoring parameters, and then obtains the second-stage corrected thickness based on the boundary feature difference between historical icing stripping information and icing appearance images. The first-stage corrected thickness and the second-stage corrected thickness are superimposed to obtain the corrected target icing thickness parameters. The de-icing control module selects a de-icing method based on the target ice thickness distribution and environmental monitoring parameters when the target ice thickness parameter exceeds the safety threshold, and performs de-icing treatment on the surface of the transmission line. The early warning module tracks the continuous changes in the target ice thickness parameter. When the ice thickness shows an upward trend for three consecutive time windows, it generates a de-icing early warning message. The data update module, after completing the de-icing process, synchronously stores the images of the icy transmission lines before and after de-icing, along with the corresponding environmental monitoring data and status monitoring data, and updates the historical database.

Images