Abnormal Monitoring Method, Device, Equipment and Storage Medium of Transformer

Abstract

The present invention relates to the technical field of abnormal monitoring of transformers, and particularly relates to an abnormal monitoring method, device, equipment and storage medium for transformers. The method includes the following steps: obtaining multi-source state parameters of the transformer; performing time-series load change analysis on the multi-source state parameters of the transformer, and conducting in-depth correlation mining between features to generate a load-temperature change map; performing three-dimensional imaging scanning on the transformer and marking the transformer components; calculating the spatial positioning of the center points of the transformer components to obtain the precise position coordinates of multiple components; performing stereoscopic morphological point cloud modeling on the load-temperature change map based on the precise position coordinates of multiple components, and performing dynamic state attribute rendering, thereby constructing an operating state twin model; performing multi-period state trend prediction on the operating state twin model, and performing sequential serialization processing to construct a state trend prediction curve. The present invention realizes efficient and accurate abnormal monitoring of transformers.

Description

Technical Field

[0001] The present invention relates to the technical field of transformer anomaly monitoring, and particularly to an anomaly monitoring method, device, equipment and storage medium for a transformer. Background Art

[0002] As a key device in the power system, the stable and reliable operation of a transformer is of utmost importance. However, during long-term operation, the transformer is affected by various internal and external factors, such as load fluctuations, environmental condition changes, component aging, etc. These factors will all have a certain impact on the operating state of the transformer.

[0003] Traditional transformer condition monitoring methods mainly rely on regular manual inspections and fault diagnoses based on single monitoring indicators. This method has problems such as limited monitoring scope, low analysis accuracy, and low efficiency. With the increasing demand for intelligent management of power equipment, there is an urgent need for a new monitoring technology that can comprehensively perceive the operating state of a transformer and intelligently analyze its abnormal changes. Based on this, an intelligent transformer anomaly monitoring method is required. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention proposes an anomaly monitoring method, device, equipment and storage medium for a transformer to solve at least one of the above technical problems.

[0005] To achieve the above object, the present invention provides an anomaly monitoring method for a transformer, including the following steps:

[0006] Step S1: Obtain multi-source state parameters of the transformer; perform time-series load change analysis on the multi-source state parameters of the transformer, and conduct in-depth correlation mining between features to generate a load-temperature change map;

[0007] Step S2: Perform three-dimensional imaging scanning on the transformer, and mark the transformer components; calculate the spatial positioning of the center points of the transformer components to obtain the precise position coordinates of multiple components;

[0008] Step S3: Based on the precise position coordinates of multiple components, perform three-dimensional morphological point cloud modeling on the load-temperature change map, and perform dynamic state attribute rendering, thereby constructing an operating state twin model;

[0009] Step S4: Perform multi-period state trend prediction on the operating state twin model, and perform sequential serialization processing to construct a state trend prediction curve;

[0010] Step S5: Extract curve outlier mutation points based on the state trend prediction curve; perform abnormal trend positioning on the curve outlier mutation points, and mark abnormal state trend nodes;

[0011] Step S6: Analyze the parameters of components near the abnormal points of the abnormal state trend nodes, make an abnormal diagnosis decision, and construct a transformer abnormal state diagnosis strategy.

[0012] The present invention also provides an abnormal monitoring device for a transformer, including:

[0013] A deep association mining module, configured to obtain multi-source state parameters of the transformer; perform time-series load change analysis on the multi-source state parameters of the transformer, and conduct in-depth association mining between features to generate a load-temperature change map;

[0014] A spatial positioning module, configured to perform three-dimensional imaging scanning on the transformer and mark the transformer components; perform central point spatial positioning calculation on the transformer components to obtain accurate position coordinates of multiple components;

[0015] A three-dimensional modeling module, configured to perform three-dimensional morphological point cloud modeling on the load-temperature change map based on the accurate position coordinates of multiple components, and perform dynamic state attribute rendering, thereby constructing an operating state twin model;

[0016] A trend prediction module, configured to perform multi-period state trend prediction on the operating state twin model, and perform sequential serialization processing to construct a state trend prediction curve;

[0017] An abnormal trend positioning module, configured to extract curve outlier mutation points based on the state trend prediction curve; perform abnormal trend positioning on the curve outlier mutation points, and mark abnormal state trend nodes;

[0018] An abnormal diagnosis decision module, configured to analyze the parameters of components near the abnormal points of the abnormal state trend nodes, make an abnormal diagnosis decision, and construct a transformer abnormal state diagnosis strategy.

[0019] The present invention also provides a computer device, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the steps of the abnormal monitoring method for the transformer described in any one of the above are implemented.

[0020] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the abnormal monitoring method for the transformer described in any one of the above are implemented.

[0021] The beneficial effects of the present invention are specifically as follows: By analyzing the temporal load changes of multi-source state parameters, the operation conditions of the transformer under different loads are understood, providing basic data for anomaly detection. Through in-depth correlation mining between features, a load-temperature change map is generated to reveal the potential relationships between state parameters, helping to monitor the operation state of the transformer. Through three-dimensional imaging scanning and spatial positioning calculation, the position information of the transformer components is accurately obtained, providing accurate basic data for subsequent modeling and analysis. Marking the position coordinates of the components to establish a precise three-dimensional point cloud model of the transformer's shape provides visual support for anomaly monitoring. The three-dimensional point cloud modeling based on precise position coordinates more realistically reflects the operation state of the transformer, improving the accuracy of anomaly detection. The dynamic state attribute rendering monitors the operation state of the transformer in real time, promptly detecting abnormal situations. Conducting state trend prediction for the operation state twin model helps predict the future operation state of the transformer, taking preventive measures in advance to avoid problems. Constructing a state trend prediction curve visually shows the change trend of the transformer's state, providing a reference basis for anomaly monitoring. Extracting the outlier mutation points of the curve and locating the abnormal trend nodes quickly identify abnormal situations during the operation of the transformer, promptly handling problems. Marking the abnormal state trend nodes helps determine the time and location of the anomaly, providing a basis for further anomaly diagnosis. Analyzing the parameters of the components adjacent to the abnormal points of the abnormal state trend nodes helps find out the causes of the anomaly and propose corresponding diagnostic decisions. Constructing a transformer abnormal state diagnosis strategy systematically handles abnormal situations, improving the operation efficiency and reducing losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a schematic flow chart of the steps of an abnormal monitoring method for a transformer according to the present invention;

[0023] Figure 2 It is a schematic detailed implementation step flow chart of step S1;

[0024] Figure 3 It is a schematic detailed implementation step flow chart of step S2;

[0025] Figure 4 It is a schematic detailed implementation step flow chart of step S3. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0026] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0027] The embodiments of the present application provide an abnormal monitoring method, device, equipment and storage medium for a transformer. The execution subjects of the abnormal monitoring method, device, equipment and storage medium for the transformer include, but are not limited to, the following general computing nodes equipped with this system: mechanical equipment, data processing platforms, cloud server nodes, network uploading devices, etc. The data processing platform includes, but is not limited to, at least one of an audio and image management system, an information management system, and a cloud data management system.

[0028] In the embodiments of the present invention, refer to Figure 1 , which is a schematic diagram of the step flow of an abnormal monitoring method for a transformer according to the present invention. In this example, the steps of the method include:

[0029] Step S1: Obtain multi-source state parameters of the transformer; perform time-series load change analysis on the multi-source state parameters of the transformer, and conduct in-depth correlation mining between features to generate a load-temperature change map;

[0030] In this embodiment, through various sensor devices installed on the transformer, multi-dimensional state parameter data during the operation of the transformer are collected. These parameters include load current, winding temperature, cooling system status, vibration noise, etc., covering multiple fields such as electricity, heat, and machinery. For key parameters such as load current and winding temperature, time-series analysis is performed to extract their change rules. Using data mining methods, the correlation features between the load current and the winding temperature are deeply analyzed. By establishing a mathematical model, an analysis map that can reflect the load-temperature change relationship of the transformer is formed. Trend analysis, periodic analysis, etc. are performed on the collected time-series data such as load current and winding temperature to extract their change rules. Using time-series analysis models such as ARIMA models and Fourier analysis, the time evolution characteristics of these state parameters are mined. Using data mining methods such as correlation analysis and regression analysis, the internal correlation rules between the load current and the winding temperature are deeply explored, and a mathematical model that can reflect the relationship between the two is established, such as a linear regression model and a neural network model. The results of time-series analysis and feature correlation analysis are integrated together to form an analysis map that can comprehensively describe the load-temperature change relationship of the transformer. This map can not only reflect the corresponding relationship between the load and the temperature, but also show their change characteristics in the time dimension.

[0031] Step S2: Perform three-dimensional imaging scanning on the transformer and mark the transformer components; perform central point space positioning calculation on the transformer components to obtain accurate position coordinates of multiple components;

[0032] In this embodiment, a three-dimensional scanning device, such as a laser scanner, a structured light scanner, etc., is used to perform high-precision three-dimensional modeling on the overall shape of the transformer, obtain the point cloud data on the surface of the transformer, and construct a complete three-dimensional solid model. On the three-dimensional model, each component of the transformer, such as windings, iron cores, insulators, etc., is manually or automatically labeled and segmented to generate a three-dimensional model of the transformer marked with the position information of each component. Using computer vision and geometric modeling methods, the three-dimensional coordinates of the center point of each component are accurately calculated, and the point cloud data on the surface of the component is fitted and geometrically analyzed using a mathematical model to determine its accurate spatial position. Preprocessing operations such as denoising and filtering are performed on the three-dimensional point cloud data of the transformer obtained in the steps to improve the data quality. For each segmented component point cloud data block, the centroid coordinates are calculated using the geometric model fitting method, and the three-dimensional spatial position of the center point of the component is determined using techniques such as principal component analysis and least squares fitting. The calculated center point coordinates of each component are integrated into a complete three-dimensional position data of the transformer components, and the point cloud data is segmented and extracted at the component level using algorithms such as segmentation and clustering.

[0033] Step S3: Based on the accurate position coordinates of multiple components, perform three-dimensional shape point cloud modeling on the load-temperature change map and perform dynamic state attribute rendering, so as to construct an operating state twin model;

[0034] In this embodiment, a three-dimensional modeling software (such as Blender, 3ds Max or Unity) is used for point cloud modeling. The accurate position coordinates of the components and the load-temperature change map are imported into the modeling software. Corresponding three-dimensional point clouds are generated according to the component position coordinates to ensure the correct positions of the components in space. Detail processing is performed on the generated point clouds, such as denoising and resampling, to improve the quality and accuracy of the model. A suitable rendering engine (such as Unity, Unreal Engine or Three.js) is selected for dynamic attribute rendering. The data of the load-temperature change map is integrated with the point cloud model for rendering preparation. Determine the dynamic attributes to be rendered, such as temperature change, load status and running time, etc. Map these attributes to the corresponding components of the point cloud model, and use colors, brightness or animation effects to represent the state changes. Dynamically update the display status of the point cloud model according to the load and temperature change data to simulate the performance of the transformer under different working conditions. Set monitoring indicators to reflect the state changes of the transformer in real time, help with fault warning and performance evaluation. Integrate the rendered dynamic point cloud model with the actual operation data of the transformer to construct a complete operating state twin model. Verify the twin model to ensure that it accurately reflects the actual operating state of the transformer. Save and publish the constructed operating state twin model for subsequent monitoring and analysis.

[0035] Step S4: Perform multi-period state trend prediction on the operating state twin model, and conduct sequential serialization processing to construct a state trend prediction curve;

[0036] In this embodiment, technical means such as time series prediction and dynamic simulation are adopted to predict the state change trend within a certain period in the future. The prediction includes the development trends of multi-dimensional parameters such as load current, winding temperature, and cooling system state in the future time period, generating a piece of data that can describe the future evolution trend of the overall state of the transformer. Integrate and arrange them in chronological order to form a continuous state change sequence. Use methods such as curve fitting to connect these discrete state prediction points into a smooth trend prediction curve, forming a prediction curve that can intuitively reflect the future development trajectory of the overall state of the transformer. For key state parameters such as load current and winding temperature in the operating state twin model, methods such as time series analysis and machine learning are used to construct a prediction model for future state changes. Use these models to predict the future evolution trend of state parameters. Based on the future predicted values of each state parameter, integrate and arrange them in chronological order to form a complete sequence data of the overall state change of the transformer in the future time period. Use methods such as curve fitting to connect the discrete state prediction points into a smooth trend prediction curve, generating a prediction curve that can intuitively reflect the future development trajectory of the overall state of the transformer.

[0037] Step S5: Extract curve outlier mutation points based on the state trend prediction curve; Locate the abnormal trend of the curve outlier mutation points and mark the abnormal state trend nodes;

[0038] In this embodiment, methods such as outlier detection and mutation point analysis are adopted to identify the outlier points and mutation points on the curve. These outlier mutation points indicate that the transformer state has abnormal changes at certain time points. Combine with the operating state twin model to deeply analyze the state change trend of these abnormal points, determine the specific time, duration, change amplitude and other characteristics of the abnormal state occurrence, and mark these abnormal state trend nodes on the operating state twin model of the transformer.

[0039] Step S6: Analyze the parameters of the components adjacent to the abnormal points of the abnormal state trend nodes, and make an abnormal diagnosis decision to construct a transformer abnormal state diagnosis strategy.

[0040] In this embodiment, according to the structure of the transformer, components adjacent to the abnormal state trend node are identified. Generally, spatial position and functional similarity are considered. The state parameters of adjacent components related to the abnormal state node, such as temperature, pressure, load, etc., are collected. The state parameters of adjacent components are extracted from the real-time monitoring system or historical database, and the extracted parameters are organized into a structured format to ensure data consistency and integrity. A suitable analysis method, such as correlation analysis, regression analysis, or statistical test, is selected to evaluate the relationship between the abnormal state and the parameters of adjacent components, analyze the parameter changes of adjacent components, identify the parameter characteristics related to the abnormal state, formulate criteria and strategies for abnormal diagnosis, including how to evaluate the abnormal state and its impact on the operation of the transformer, determine key diagnostic indicators, such as the amplitude of temperature rise, abnormal pressure, etc. According to the abnormal state characteristics and adjacent parameters, a suitable abnormal diagnosis model (such as decision tree, support vector machine, or fuzzy logic system) is selected, and the diagnostic model is trained using historical data to improve its prediction accuracy. The trained model is applied to newly identified abnormal state trend nodes for abnormal diagnosis. According to the results output by the model, the nature and potential causes of the abnormal state are evaluated. According to the diagnostic results, corresponding countermeasures are formulated, including maintenance, inspection, or replacement of components, etc. An early warning mechanism for abnormal states is established to ensure timely response when an abnormality occurs. The diagnostic results and strategies are recorded in the database for subsequent analysis and reference. A feedback mechanism is established to continuously monitor the effects after implementation and adjust the diagnostic strategy according to the actual situation.

[0041] In this embodiment, refer to Figure 2 , which is a schematic diagram of the detailed implementation steps of step S1. In this embodiment, the detailed implementation steps of step S1 include:

[0042] Step S11: Real-time monitor the operating state of the transformer to obtain multi-source state parameters of the transformer;

[0043] Step S12: Calculate the load characteristics of the multi-source state parameters of the transformer to obtain the load characteristic data of the transformer;

[0044] Step S13: Conduct time-series load change analysis on the load characteristic data of the transformer to obtain time-series load change characteristics;

[0045] Step S14: Identify the temperature fluctuations of the multi-source state parameters of the transformer to obtain the temperature fluctuation data of the transformer;

[0046] Step S15: Fit the fluctuation distribution of the temperature fluctuation data of the transformer to generate a temperature fluctuation distribution field;

[0047] Step S16: Conduct in-depth correlation mining between features on the temperature fluctuation distribution field based on the time-series load change characteristics to generate a load-temperature change map.

[0048] In this embodiment, sensors are installed at key parts of the transformer (such as windings and oil tanks) to monitor state parameters such as current, voltage, temperature, and oil level, constructing a data acquisition system to collect data from each sensor in real time, ensuring the stability and real-time nature of data transmission. The multi-source state parameters collected are stored in a database for subsequent analysis. Determine load characteristic indicators such as rated load, actual load, and load factor. Use appropriate calculation methods (such as weighted average) to calculate the load characteristics of the multi-source state parameters. Organize the calculated load characteristic data into a structured format for subsequent analysis. Organize the load characteristic data into time-series data according to timestamps. Select suitable time-series analysis methods such as moving average and exponential smoothing to analyze the time-series data, identify the trends and periodic characteristics of load changes, and organize the time-series load change characteristics into a report for easy understanding and subsequent application. Extract temperature-related data from the multi-source state parameters to ensure data integrity. Determine the identification criteria for temperature fluctuations such as fluctuation amplitude and frequency. Use statistical methods (such as standard deviation calculation) to identify temperature fluctuations and record them as temperature fluctuation data. Organize the temperature fluctuation data into a structured format for subsequent analysis. Select a suitable distribution fitting method such as normal distribution and Gamma distribution. Use statistical tools (such as the SciPy library in Python) to fit the temperature fluctuation data and generate a probability distribution model of temperature fluctuations. Generate a temperature fluctuation distribution field according to the fitting results to represent the temperature fluctuation characteristics under different conditions. Organize the results of the temperature fluctuation distribution field into charts or data sets for subsequent analysis. Select a suitable association analysis method such as correlation analysis, regression analysis, or machine learning models. Integrate the time-series load change characteristics with the data of the temperature fluctuation distribution field for in-depth analysis. Perform in-depth association mining to identify the relationship between load changes and temperature fluctuations and generate a load-temperature change map. Visualize the generated load-temperature change map to help understand the dynamic relationship between load and temperature.

[0049] In this embodiment, refer to Figure 3 , which is a schematic diagram of the detailed implementation steps of step S2. In this embodiment, the detailed implementation steps of step S2 include:

[0050] Step S21: Perform three-dimensional imaging scanning on the transformer to extract the three-dimensional image of the transformer;

[0051] Step S22: Perform multi-directional cutting on the three-dimensional image of the transformer to obtain transformer images in multiple directions;

[0052] Step S23: Identify components at different parts of the transformer images in multiple directions and mark the transformer components;

[0053] Step S24: Perform central point spatial positioning calculation on the transformer components to obtain the central position coordinates of multiple components;

[0054] Step S25: Perform three-dimensional position correction calculation on the central position coordinates of multiple components based on the transformer images in multiple orientations to obtain the accurate position coordinates of multiple components.

[0055] In this embodiment, use a three-dimensional imaging device (such as a laser scanner or a structured light scanner) for preparation, ensure correct calibration of the device, conduct a comprehensive scan of the transformer, collect three-dimensional point cloud data, record the external and internal structures of the transformer, use software (such as MeshLab or AutoCAD) to process the original point cloud data into a three-dimensional image, generate a three-dimensional model of the transformer, determine the cutting orientation and cutting plane to ensure coverage of all key parts of the transformer, use three-dimensional modeling software to cut the three-dimensional image of the transformer to generate transformer images in multiple orientations, save the multiple orientation images obtained by cutting in different file formats (such as STL, OBJ) for subsequent processing, determine the identification criteria for components, such as shape, size, and material properties, select a suitable image processing algorithm (such as edge detection, morphological processing) for component identification, analyze the images in multiple orientations, identify different components of the transformer (such as windings, oil tanks, insulators), and mark them, determine the calculation method for the geometric center point of each component, usually using centroid calculation, calculate the coordinates of each marked component to obtain its spatial center point position coordinates, organize the calculated central position coordinates of the components into structured data for convenient subsequent correction, select a suitable correction method, such as rigid body transformation or similarity transformation, to ensure coordinate consistency, correct the central position coordinates of the components based on the image data in multiple orientations, and correct the positioning error caused by the perspective change, organize the corrected accurate position coordinates of the components into a data set, and record the final position of each component.

[0056] In this embodiment, refer to Figure 4 as the schematic diagram of the detailed implementation steps of step S3. In this embodiment, the detailed implementation steps of step S3 include:

[0057] Step S31: Perform state feature positioning inference on the load-temperature change map based on the accurate position coordinates of multiple components to obtain the matching positions of each associated feature;

[0058] Step S32: Conduct spatial layout analysis on the transformer components based on the accurate position coordinates of multiple components to generate component spatial layout data;

[0059] Step S33: Mine the three-dimensional spatial topological relationship of the component spatial layout data to obtain the component three-dimensional topological structure diagram;

[0060] Step S34: Perform stereoscopic morphological point cloud modeling on the three-dimensional topological structure diagram of the components to generate a three-dimensional point cloud model of the transformer;

[0061] Step S35: Use the matching positions of each associated feature to perform dynamic state attribute rendering on the three-dimensional point cloud model of the transformer with the load-temperature change map, thereby constructing an operating state twin model.

[0062] In this embodiment, key features in the load-temperature change map are identified, such as temperature peaks, load peaks, etc. A suitable matching algorithm (such as nearest neighbor matching or interpolation method) is selected to associate the map features with the component positions. For each identified feature, its matching position in the component space is inferred and recorded as the state feature localization result. A suitable spatial analysis method, such as clustering analysis or geometric shape analysis, is selected to organize the precise position coordinates of multiple components into a data set for preparing spatial layout analysis. The spatial positions of the components are analyzed to identify the relative position relationships between the components. The analysis results are organized into component spatial layout data to describe the layout features of the components. The topological relationship criteria between the components, such as adjacency relationships and connection relationships, are determined. A suitable topological mining algorithm (such as graph theory algorithm) is selected to analyze the relationships between the components. A three-dimensional topological structure diagram is generated according to the layout data to represent the connections and relative positions between the components. The generated three-dimensional topological structure diagram is saved for subsequent visualization and analysis. Three-dimensional modeling software (such as CloudCompare or MeshLab) is used for point cloud modeling. According to the three-dimensional topological structure diagram, the spatial layout of the components is transformed into point cloud data to generate a three-dimensional point cloud model of the transformer. The point cloud model is optimized in detail, such as denoising, resampling, etc., to improve the model quality. The generated three-dimensional point cloud model of the transformer is saved in a usable format (such as PLY, OBJ). A suitable rendering tool (such as Unity, Unreal Engine) is selected for dynamic attribute rendering. The state features at the matching positions are integrated with the three-dimensional point cloud model for preparing rendering. According to the features of the load-temperature change map, dynamic state attribute rendering is performed on the three-dimensional point cloud model of the transformer to display the operating state. The rendered model is saved as a dynamic operating state twin model for real-time monitoring and analysis.

[0063] In this embodiment, step S4 includes the following steps:

[0064] Step S41: Perform state feature trend evolution on the operating state twin model to generate operating state trend feature data;

[0065] Step S42: Perform sliding window segmentation processing on the operating state trend feature data to obtain multiple trend feature segment data;

[0066] Step S43: performing deep time series evolution learning on multiple trend feature segment data to generate transformer trend feature evolution law;

[0067] Step S44: using the transformer trend feature evolution law to perform multi-period state trend prediction on the operation state twin model to generate state trend prediction data for multiple time periods;

[0068] Step S45: sequentially serialize the state trend prediction data of multiple time periods to construct a state trend prediction curve.

[0069] In this embodiment, according to the transformer operating state twin model, the dynamic change trend of key operating state characteristics such as load and temperature is analyzed, and the change law and development trend of these state characteristics in the time series are extracted. The transformer operating state twin model, the dynamic change trend of key operating state characteristics such as load and temperature is analyzed, and the change law and development trend of these state characteristics in the time series are extracted. A deep learning time series analysis model, such as a recurrent neural network, is used to perform deep time series learning on it, and the evolution law of the overall operating state of the transformer in the time dimension is excavated to form a mathematical model that can describe the development trend of the transformer state. Based on the evolution law obtained by learning, a state trend prediction model is constructed, and the constructed prediction model is used to perform multi-period state trend prediction on the operating state twin model, and state trend prediction data for multiple time periods is generated. A suitable serialization method (such as timestamp serialization) is selected to convert the state trend prediction data into a sequence, and a state trend prediction curve is drawn according to the serialized data. It is usually implemented using a chart tool (such as Matplotlib or Excel), and the constructed state trend prediction curve is visualized to ensure that the information is clear and easy to understand, and the trend prediction curve and related data are saved for subsequent reference and decision support.

[0070] In this embodiment, step S5 includes the following steps:

[0071] Step S51: Identify trend outlier mutation points on the state trend prediction curve and extract the curve outlier mutation points;

[0072] Step S52: Analyze the abnormal trend change of the outlier mutation point of the curve to obtain abnormal state trend data;

[0073] Step S53: locate the abnormal trend of the operating status twin model according to the abnormal status trend data and mark the abnormal status trend node.

[0074] In this embodiment, relevant data points are extracted from the state trend prediction curve to construct an analyzable numerical array. A suitable outlier detection algorithm is selected, such as the Z-score method, the IQR (interquartile range) method, or the LOF (local outlier factor) algorithm. The selected algorithm is applied to detect the curve data to identify outlier mutation points, that is, data points that significantly deviate from the normal trend. The identified outlier mutation points are organized into a data set, and their positions and values are recorded for subsequent analysis. The characteristics of the abnormal trend are determined, such as rapid rise, sharp drop, or increased fluctuation. A suitable analysis method is selected, such as time series analysis, trend analysis, or statistical tests, to conduct a detailed analysis of the outlier mutation points to identify the reasons and impacts behind them. The change characteristics of the abnormal state are recorded. The abnormal state trend data obtained from the analysis is organized into a structured format for subsequent use. The marking criteria for the abnormal trend nodes are determined, such as the abnormal amplitude, duration, or influence range. Based on the abnormal state trend data, the abnormal trend nodes in the running state twin model are marked to ensure easy identification. The marked abnormal nodes in the running state twin model are visualized to enhance visibility. The marked abnormal state trend nodes and their characteristics are recorded in the database for subsequent monitoring and analysis.

[0075] In this embodiment, step S6 includes the following steps:

[0076] Step S61: Identify the abnormal type of the abnormal state trend node to generate the node abnormal type;

[0077] Step S62: Visualize the abnormal trend of the abnormal state trend node to generate an abnormal state trend visualization model;

[0078] Step S63: Analyze the parameters of the adjacent components of the abnormal points in the abnormal state trend visualization model to obtain the adjacent component parameters;

[0079] Step S64: Conduct an abnormal state association analysis of the abnormal state trend node based on the adjacent component parameters to obtain the component abnormal association data;

[0080] Step S65: Make an abnormal diagnosis decision on the abnormal state trend visualization model based on the node abnormal type and the component abnormal association data to construct a transformer abnormal state diagnosis strategy.

[0081] In this embodiment, the types of different abnormal states are determined, such as overload, temperature anomaly, vibration anomaly, etc. Relevant feature data, such as amplitude change, duration, and frequency characteristics, are extracted from the abnormal state trend nodes. A suitable classification algorithm (such as decision tree, random forest, or support vector machine) is selected for abnormal type recognition. The model is trained and applied to the abnormal state trend nodes to generate the abnormal type labels for each node. A suitable visualization tool (such as Tableau, Matplotlib, or D3.js) is selected for data visualization. The data of the abnormal state trend nodes and their abnormal types are sorted out to ensure that the format is suitable for visualization. An abnormal state trend visualization model is generated in the selected tool to display the abnormal nodes and their characteristic attributes and optimize the visualization effect to ensure that the information is clear and easy to understand for analysis and decision-making. The components adjacent to the abnormal nodes are determined, usually based on spatial location or functional similarity. The status data of the adjacent components, such as parameters like temperature, load, and vibration, are collected from the monitoring system. A suitable analysis method (such as correlation analysis or regression analysis) is selected to analyze the parameters of the adjacent components. The parameters of the adjacent components are analyzed to identify the parameter characteristics related to the abnormal nodes. A suitable association analysis method (such as Pearson correlation or Cohen's d) is selected to evaluate the relationship between abnormal states. The abnormal state trend nodes and the parameters of their adjacent components are integrated into an analysis data set. The integrated data is analyzed to identify the correlation between the abnormal states and the adjacent components. The analysis results are sorted out as component-abnormal association data, and the relationship between each component and the abnormal node is recorded. The criteria and strategies for abnormal diagnosis are determined, including how to make decisions based on the abnormal type and association data. A suitable decision support algorithm (such as fuzzy logic, expert system, or machine learning model) is selected for abnormal diagnosis. The selected algorithm is applied to make an abnormal diagnosis decision based on the node abnormal type and component-abnormal association data. A transformer abnormal state diagnosis strategy is generated according to the diagnosis results, and the countermeasures and suggestions are recorded.

[0082] In this embodiment, the present invention also provides an abnormal monitoring device for a transformer, including:

[0083] A deep association mining module, configured to obtain multi-source state parameters of the transformer; perform time-series load change analysis on the multi-source state parameters of the transformer, and conduct in-depth association mining between features to generate a load-temperature change map;

[0084] A spatial positioning module, configured to perform three-dimensional imaging scanning on the transformer and mark the transformer components; perform central point spatial positioning calculation on the transformer components to obtain the precise position coordinates of multiple components;

[0085] A three-dimensional modeling module, configured to perform three-dimensional morphological point cloud modeling on the load-temperature change map based on the precise position coordinates of multiple components, and perform dynamic state attribute rendering, thereby constructing an operating state twin model;

[0086] A trend prediction module, which is used to perform multi-period state trend prediction on the operation state twin model, perform sequential serialization processing, and construct a state trend prediction curve;

[0087] An abnormal trend positioning module, which is used to extract curve outlier mutation points based on the state trend prediction curve; perform abnormal trend positioning on the curve outlier mutation points, and mark abnormal state trend nodes;

[0088] An abnormal diagnosis and decision-making module, which is used to analyze the parameters of components adjacent to the abnormal points for the abnormal state trend nodes, perform abnormal diagnosis and decision-making, and construct a transformer abnormal state diagnosis strategy.

[0089] The present invention also provides a computer device, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the abnormal monitoring method of the transformer described in any one of the above is implemented.

[0090] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the abnormal monitoring method of the transformer described in any one of the above are implemented.

[0091] Those skilled in the art will clearly understand that for the convenience and brevity of description, the specific working processes of the above-described systems, systems, and units refer to the corresponding processes in the foregoing method embodiments, and will not be described herein again.

[0092] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it is stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present application essentially, or the part that contributes to the prior art, or all or part of the technical solution, is embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present application. The foregoing storage medium includes: various media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc that store program codes.

[0093] Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, it is intended to cover all changes falling within the meaning and scope of the equivalent elements of the application documents within the present invention.

[0094] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the widest scope consistent with the principles and novel features invented herein.

Claims

1. An abnormal monitoring method for a transformer, characterized in that, It includes the following steps: Step S1: Obtain the multi-source state parameters of the transformer; conduct time-series load change analysis on the multi-source state parameters of the transformer, and perform in-depth correlation mining between features to generate a load-temperature change map; Step S2: Conduct three-dimensional imaging scanning on the transformer and mark the transformer components; perform central point spatial positioning calculation on the transformer components to obtain the precise position coordinates of multiple components; Step S3: Based on the precise position coordinates of multiple components, perform three-dimensional morphological point cloud modeling on the load-temperature change map and conduct dynamic state attribute rendering, thereby constructing an operating state twin model; Step S4: Conduct multi-period state trend prediction on the operating state twin model and perform sequential serialization processing to construct a state trend prediction curve; Step S5: Extract the curve outlier mutation points based on the state trend prediction curve; perform abnormal trend positioning on the curve outlier mutation points and mark the abnormal state trend nodes; Step S6: Conduct analysis on the parameters of the components adjacent to the abnormal points of the abnormal state trend nodes and make an abnormal diagnosis decision to construct a transformer abnormal state diagnosis strategy.

2. The abnormal monitoring method of the transformer according to claim 1, characterized in that, The specific steps of Step S1 are as follows: Step S11: Monitor the operating state of the transformer in real time and obtain the multi-source state parameters of the transformer; Step S12: Perform load characteristic calculation on the multi-source state parameters of the transformer to obtain the transformer load characteristic data; Step S13: Conduct time-series load change analysis on the transformer load characteristic data to obtain the time-series load change characteristics; Step S14: Identify the temperature fluctuations of the multi-source state parameters of the transformer to obtain the transformer temperature fluctuation data; Step S15: Fit the fluctuation distribution of the transformer temperature fluctuation data to generate a temperature fluctuation distribution field; Step S16: Based on the time-series load change characteristics, perform in-depth correlation mining between features on the temperature fluctuation distribution field to generate a load-temperature change map.

3. The abnormal monitoring method of the transformer according to claim 1, wherein, The specific steps of Step S2 are as follows: Step S21: Conduct three-dimensional imaging scanning on the transformer and extract the three-dimensional image of the transformer; Step S22: Perform multi-directional cutting processing on the three-dimensional image of the transformer to obtain transformer images in multiple directions; Step S23: Identify the components in different parts of the transformer images in multiple directions and mark the transformer components; Step S24: Perform central point spatial positioning calculation on the transformer components to obtain the central position coordinates of multiple components; Step S25: Perform three-dimensional position correction calculation on the central position coordinates of multiple components according to the transformer images in multiple directions to obtain the precise position coordinates of multiple components.

4. The abnormal monitoring method of the transformer according to claim 1, wherein The specific steps of Step S3 are as follows: Step S31: Based on the precise position coordinates of multiple components, conduct state feature positioning inference on the load-temperature change map to obtain the matching position of each associated feature; Step S32: Conduct spatial layout analysis on the transformer components according to the precise position coordinates of multiple components to generate component spatial layout data; Step S33: Conduct three-dimensional spatial topological relationship mining on the component spatial layout data to obtain a component three-dimensional topological structure diagram; Step S34: Perform three-dimensional morphological point cloud modeling on the component three-dimensional topological structure diagram to generate a transformer three-dimensional point cloud model; Step S35: Use the matching positions of each associated feature to perform dynamic state attribute rendering on the load-temperature change map for the three-dimensional point cloud model of the transformer, thereby constructing an operating state twin model.

5. The abnormal monitoring method of the transformer according to claim 1, characterized in that, The specific steps of Step S4 are as follows: Step S41: Perform state feature trend evolution on the operating state twin model to generate operating state trend feature data; Step S42: Perform sliding window segmentation processing on the operating state trend feature data to obtain multiple trend feature segment data; Step S43: Perform deep time series evolution learning on multiple trend feature segment data to generate the evolution law of the transformer trend features; Step S44: Use the evolution law of the transformer trend features to perform multi-period state trend prediction on the operating state twin model to generate state trend prediction data for multiple time periods; Step S45: Perform sequential serialization processing on the state trend prediction data for multiple time periods to construct a state trend prediction curve.

6. The abnormal monitoring method of the transformer according to claim 1, wherein, The specific steps of Step S5 are as follows: Step S51: Identify trend outlier mutation points on the state trend prediction curve and extract the curve outlier mutation points; Step S52: Perform abnormal trend change analysis on the curve outlier mutation points to obtain abnormal state trend data; Step S53: Locate the abnormal trend on the operating state twin model according to the abnormal state trend data and mark the abnormal state trend nodes.

7. The abnormal monitoring method of the transformer according to claim 1, wherein, The specific steps of Step S6 are as follows: Step S61: Identify the abnormal type of the abnormal state trend node to generate the node abnormal type; Step S62: Visualize the abnormal trend of the abnormal state trend node to generate an abnormal state trend visualization model; Step S63: Analyze the parameters of the adjacent components of the abnormal point on the abnormal state trend visualization model to obtain the adjacent component parameters; Step S64: Perform abnormal state correlation analysis on the abnormal state trend node based on the adjacent component parameters to obtain component abnormal correlation data; Step S65: Make an abnormal diagnosis decision on the abnormal state trend visualization model based on the node abnormal type and the component abnormal correlation data to construct a transformer abnormal state diagnosis strategy.

8. An abnormal monitoring device for a transformer, characterized in that, The device for executing the abnormal monitoring method of the transformer as described in claim 1 includes: A deep association mining module, configured to obtain multi-source state parameters of the transformer; perform time-series load change analysis on the multi-source state parameters of the transformer, and perform in-depth association mining between features to generate a load-temperature change map; A spatial positioning module, configured to perform three-dimensional imaging scanning on the transformer and mark the transformer components; perform central point spatial positioning calculation on the transformer components to obtain accurate position coordinates of multiple components; A three-dimensional modeling module, configured to perform three-dimensional morphological point cloud modeling on the load-temperature change map based on the accurate position coordinates of multiple components, and perform dynamic state attribute rendering, thereby constructing an operating state twin model; A trend prediction module, configured to perform multi-period state trend prediction on the operating state twin model, and perform sequential serialization processing to construct a state trend prediction curve; An abnormal trend positioning module, configured to extract curve outlier mutation points based on the state trend prediction curve; locate the abnormal trend of the curve outlier mutation points and mark the abnormal state trend nodes; An abnormal diagnosis decision-making module is used to analyze the parameters of components near the abnormal point for the abnormal state trend node, make an abnormal diagnosis decision, and construct a diagnosis strategy for the abnormal state of the transformer.

9. A computer device, comprising a memory and a processor, wherein a computer program is stored in the memory, characterized in that, When the processor executes the computer program, the steps of the abnormal monitoring method of the transformer according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, the steps of the abnormal monitoring method of the transformer according to any one of claims 1 to 7 are implemented.

Images