Line fault diagnosis method and device based on knowledge forest and multi-modal data

Abstract

The invention provides a line fault diagnosis method and device based on knowledge forest and multi-modal data, and relates to the technical field of power grid line fault diagnosis. The method comprises the following steps: acquiring multi-modal data of a to-be-detected power distribution network line, wherein the multi-modal data comprises operating parameters of a power distribution network, temperature data of the line and image data; inputting the multi-modal data into a pre-trained multi-modal basic model to obtain multi-modal features, the multi-modal features being key features in different modal data; fusing the multi-modal features to obtain comprehensive features; analyzing the comprehensive features by using a pre-constructed knowledge forest to obtain a fault analysis result; wherein the knowledge forest comprises line equipment information, fault conditions and line fault types, the fault analysis result comprises line fault types and fault positions, and the line fault types comprise line open circuit and line damage. The accuracy of line fault diagnosis can be improved.

Description

Technical Field

[0001] This invention relates to the field of power grid line fault diagnosis technology, and in particular to a line fault diagnosis method, apparatus and equipment based on knowledge forest and multimodal data. Background Technology

[0002] The power system is an indispensable infrastructure of modern society, and its stable operation is crucial for industrial production and residential life. Line faults can lead to power outages, affecting a wide area and causing significant economic losses. Timely and accurate diagnosis of line faults enables rapid location and repair of the fault point, reducing power outage time and economic losses.

[0003] With the continuous development and increasing complexity of the power system, the power system collects multi-source data from various aspects. This data contains a wealth of information and can reflect the operating status of the power grid from different perspectives.

[0004] However, due to the lack of effective means to integrate and analyze multi-source data, fault analysis and judgment often rely on the experience of dispatchers when line faults occur. Since dispatchers need to combine multi-source data analysis and calculation, it is difficult to diagnose power grid line faults in a timely manner, which in serious cases will threaten the safe and stable operation of the power grid.

[0005] Therefore, there is an urgent need for a line fault diagnosis method that can analyze multi-source data. Summary of the Invention

[0006] This invention provides a method and apparatus for diagnosing line faults based on knowledge forests and multimodal data, in order to solve the problem that in line fault diagnosis, the inability to effectively fuse and analyze multi-source data leads to inaccurate fault identification.

[0007] In a first aspect, embodiments of the present invention provide a line fault diagnosis method based on knowledge forest and multimodal data, including:

[0008] Acquire multimodal data of the distribution network lines to be tested. The multimodal data includes the operating parameters of the distribution network, the temperature data of the lines, and image data.

[0009] Multimodal data is input into a pre-trained multimodal base model to obtain multimodal features, which are key features in different modal data.

[0010] Multimodal features are fused to obtain comprehensive features;

[0011] The comprehensive features are analyzed using a pre-constructed knowledge forest to obtain fault analysis results. The knowledge forest includes line equipment information, fault conditions, and line fault types. The fault analysis results include line fault types and fault locations. Line fault types include line open circuits and line damage.

[0012] In one possible implementation, the operating parameters include the carrying capacity;

[0013] The comprehensive features are analyzed using a pre-constructed knowledge forest, including:

[0014] Determine the location of the line fault based on image data;

[0015] The cross-sectional area of ​​the line to be tested is calculated based on the current carrying capacity and temperature data corresponding to the location of the line fault.

[0016] The type of line fault is determined by the cross-sectional area being inspected.

[0017] In one possible implementation, determining the line fault type based on the detected cross-sectional area includes:

[0018] When the cross-sectional area detected is zero, the line fault type is determined to be a line open circuit.

[0019] When the detected cross-sectional area is smaller than the actual cross-sectional area but greater than the preset threshold ratio, the line fault type is determined to be line damage.

[0020] In one possible implementation, the formula for calculating the cross-sectional area is:

[0021]

[0022] ρ=ρ0×(1+α×(T-T0))

[0023] In the formula, S is the cross-sectional area of ​​the cable, ρ is the resistivity of the cable material, L is the length of the cable, I is the current carrying capacity of the cable, U is the rated voltage of the cable, k is the comprehensive coefficient, ρ0 is the resistivity at the reference temperature T0, α is the temperature coefficient of resistance, T is the actual operating temperature, and T0 is the reference temperature.

[0024] In one possible implementation, the image data is acquired via infrared imaging;

[0025] Determining the location of a line fault based on image data includes:

[0026] Image data preprocessing includes median filtering for noise reduction and histogram equalization enhancement.

[0027] Temperature features, grayscale features, and edge features are extracted from the preprocessed image data.

[0028] The extracted temperature features, grayscale features, and edge features are normalized.

[0029] The normalized temperature features, grayscale features, and edge features are input into the fault location judgment model to obtain the coordinates of the fault point.

[0030] In one possible implementation, the process of constructing a knowledge forest includes:

[0031] Acquire multi-source data from the power system, including operating parameters, temperature data, image data, and text data of line faults;

[0032] Knowledge extraction is performed on multi-source data. Knowledge extraction includes extracting entities, extracting relations, and extracting attributes. Entities include line equipment and line fault types, relations include the associations between entities, and attributes include the characteristics of entities.

[0033] The extracted entities, relationships, and attributes are merged, and subject faceted trees are constructed based on the merged entities, relationships, and attributes. A knowledge forest is then constructed based on multiple subject faceted trees.

[0034] In one possible implementation, the training process of the fault location determination model includes:

[0035] A dataset was constructed using historical fault-related data from a knowledge forest. This dataset includes equipment information, fault conditions, and line fault types from the knowledge forest. Equipment information includes operating parameters, temperature data, and image data. Fault conditions include ambient temperature, current carrying capacity, and cross-sectional area. Line fault types include line open circuits and line damage. Specifically, the dataset includes:

[0036] For line outage and line damage conditions, the data in the dataset is classified according to different temperature data, different cross-sectional area or different current carrying capacity, and sub-data sets corresponding to different temperature data, different cross-sectional area or different current carrying capacity are constructed.

[0037] Fault location determination models were obtained by training convolutional neural networks using subsets of the dataset.

[0038] In one possible implementation, multimodal features are fused to obtain comprehensive features, including:

[0039] Normalize each modal feature in the multimodal features;

[0040] The normalized modal features are concatenated in sequence to obtain the comprehensive features; the order is as follows: operating parameters, temperature data, and image data.

[0041] Secondly, embodiments of the present invention provide a line fault diagnosis device based on knowledge forest and multimodal data, comprising:

[0042] The multimodal data acquisition module is used to acquire multimodal data of the distribution network line to be tested. The multimodal data includes the operating parameters of the distribution network, the temperature data of the line, and image data.

[0043] The multimodal feature acquisition module is used to input multimodal data into a pre-trained multimodal base model to obtain multimodal features, which are key features in different modal data.

[0044] The feature fusion module is used to fuse multimodal features to obtain comprehensive features;

[0045] The fault analysis module is used to analyze the comprehensive features using a pre-built knowledge forest to obtain fault analysis results. The knowledge forest includes line equipment information, fault conditions, and line fault types. The fault analysis results include line fault types and fault locations. Line fault types include line open circuits and line damage.

[0046] Thirdly, embodiments of the present invention provide an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method described in the first aspect or any possible implementation thereof.

[0047] In this embodiment of the invention, data from multiple modalities reflects the operating status of distribution network lines from different perspectives, enabling a more comprehensive capture of fault information. The multimodal basic model can extract key features from data of different modalities. These key features are the most representative and discriminative parts of the data, effectively reducing data dimensionality and improving the efficiency of subsequent processing. By fusing multimodal features, a comprehensive feature is obtained, which can more comprehensively reflect the operating status of the line and provide richer information for subsequent analysis. The knowledge forest contains knowledge such as line equipment information, fault conditions, and line fault types, providing theoretical support and decision-making basis for fault diagnosis. By acquiring operating parameters and temperature data in real time, real-time monitoring of distribution network lines can be achieved, enabling timely detection of potential faults. Through the comprehensive utilization of multimodal data, the application of the multimodal basic model, feature fusion, and the knowledge forest, the accuracy and efficiency of line fault diagnosis are improved. Attached Figure Description

[0048] Figure 1 This is a flowchart illustrating the implementation of the line fault diagnosis method based on knowledge forest and multimodal data provided in this embodiment of the invention.

[0049] Figure 2 This is a schematic diagram of the structure of the line fault diagnosis device based on knowledge forest and multimodal data provided in an embodiment of the present invention;

[0050] Figure 3 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0051] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0052] See Figure 1 The document illustrates a flowchart of the implementation of the line fault diagnosis method based on knowledge forest and multimodal data provided in this embodiment of the invention, which is described in detail below:

[0053] Step 101: Obtain multimodal data of the distribution network line to be tested. The multimodal data includes the operating parameters of the distribution network, the temperature data of the line, and image data.

[0054] In this embodiment, the operating parameters of the distribution network include current, voltage, power factor, and frequency, reflecting the electrical characteristics of the distribution network during operation. For example, abnormal changes in current and voltage may indicate problems such as overload or short circuit in the line. Temperature sensors acquire line temperature information; abnormal temperature increases may indicate overheating, possibly caused by line aging, poor contact, or overload. Cameras (such as infrared imaging cameras) acquire line images, which can help visually identify physical damage, insulator breakage, foreign object entanglement, and other problems in the line.

[0055] Step 102: Input the multimodal data into the pre-trained multimodal base model to obtain multimodal features, which are the key features in different modal data.

[0056] In this embodiment, the operating parameters in the multimodal data reflect the electrical operating status of the power distribution network, the temperature data reflects the thermal status of the lines, and the image data reflects the physical status of the lines.

[0057] In this embodiment, the pre-trained multimodal base model is a model trained on a large amount of data, capable of handling multiple types of data (multimodal data). This model is a deep learning model, such as a convolutional neural network for image data, a recurrent neural network or a transformer for time series data, etc. The model's goal is to extract key features from data of different modalities. The model extracts the most representative and discriminative features from each modality of data. For example, it extracts current and voltage fluctuation features from operating parameters, temperature change rate from temperature data, and edge and texture features from image data. The key features extracted from different modalities of data can more comprehensively describe the operating status of the distribution network lines.

[0058] For example, suppose there is a power distribution network line, and the following data is collected: current 100A, voltage 10kV, power factor 0.9, and line temperature 60℃. Infrared imaging shows a significant overheated area at a certain point in the line. This data is input into a pre-trained multimodal baseline model, which extracts the following key features: current and voltage fluctuation characteristics; temperature change rate and absolute temperature value; and the location, size, and temperature gradient of the overheated area. These key features are used for subsequent fault diagnosis analysis.

[0059] Step 103: Fuse the multimodal features to obtain the comprehensive features.

[0060] In this embodiment, features extracted from different modalities are integrated to form a more comprehensive and representative feature representation, i.e., a composite feature. By fusing data features from different sources or types, the complexity and diversity of the data can be better captured, thereby improving the model's performance and generalization ability.

[0061] Step 104: Analyze the comprehensive features using a pre-constructed knowledge forest to obtain fault analysis results; wherein, the knowledge forest includes line equipment information, fault conditions and line fault types, and the fault analysis results include line fault types and fault locations, and line fault types include line open circuits and line damage.

[0062] In this embodiment, the process of analyzing comprehensive features using a knowledge forest integrates multimodal data and professional knowledge to achieve accurate diagnosis of distribution network line faults, including determining the type and location of the fault, so as to carry out targeted repair and maintenance work. The knowledge forest is a structured knowledge representation form containing various information related to the distribution network lines, specifically including: line equipment information such as the specifications, operating parameters, and maintenance records of cables, circuit breakers, insulators, etc. Fault conditions are various factors and situations that lead to line faults, such as overload, short circuit, lightning strike, and line aging. Two main fault types are clearly identified: line open circuit and line damage. A line open circuit usually refers to a complete break in the line at a certain point, causing current to be unable to flow; a line damage refers to a certain degree of damage to the line, such as insulation damage or partial breakage of the conductor, but not a complete disconnection.

[0063] In this embodiment, the comprehensive feature is a feature vector obtained by processing multimodal data (operating parameters, temperature data, and image data), which contains key information from different modes and can comprehensively reflect the operating status of the distribution network line.

[0064] In this embodiment, the parsing process refers to using comprehensive features and knowledge forest to match, reason, and analyze the comprehensive features based on knowledge such as line equipment information, fault conditions, and fault types from the knowledge forest.

[0065] In this embodiment, the fault type is determined by whether the fault occurring in the line is an open circuit or a broken line. For example, if the detected cross-sectional area is zero, it is determined to be an open circuit; if the detected cross-sectional area is smaller than the actual cross-sectional area but greater than a preset threshold ratio, it is determined to be a broken line.

[0066] In this embodiment, by analyzing information such as image data in the comprehensive features and combining it with line equipment information in the knowledge forest, the specific location of the fault in the distribution network line can be accurately located, such as a certain section of line or near a certain equipment.

[0067] In one possible implementation, the operating parameters include the carrying capacity;

[0068] The comprehensive features are analyzed using a pre-constructed knowledge forest, including:

[0069] Determine the location of the line fault based on image data;

[0070] The cross-sectional area of ​​the line to be tested is calculated based on the current carrying capacity and temperature data corresponding to the location of the line fault.

[0071] The type of line fault is determined by the cross-sectional area being inspected.

[0072] In this embodiment, image recognition technology (such as convolutional neural networks) is used to analyze image data to determine the specific location of the fault. After determining the fault location, the current carrying capacity and temperature data at that location are acquired. Specific formulas (such as the resistance method and voltage drop method) are used in conjunction with parameters such as current carrying capacity and temperature to calculate the cross-sectional area of ​​the line. The degree of line damage is determined based on the detected cross-sectional area, thus identifying the type of line fault. By combining the intuitiveness of image data with the accuracy of electrical parameters, this method can more comprehensively and accurately diagnose line faults, providing a scientific basis for the maintenance and repair of power systems.

[0073] In one possible implementation, determining the line fault type based on the detected cross-sectional area includes:

[0074] When the cross-sectional area detected is zero, the line fault type is determined to be a line open circuit.

[0075] When the detected cross-sectional area is smaller than the actual cross-sectional area but greater than the preset threshold ratio, the line fault type is determined to be line damage.

[0076] In this embodiment, if the calculated detection cross-sectional area is zero, it indicates that the line is completely broken at that location, and current cannot pass through, i.e., the line is open-circuited. If the cross-sectional area is smaller than the actual cross-sectional area and smaller than a preset threshold ratio, it indicates that the line has a certain degree of damage, such as insulation damage or partial breakage of the conductor, but not completely disconnected, i.e., the line is damaged. This application obtains the time during which the current line can operate stably based on the detection cross-sectional area, the preset threshold, and industry standard data for steady-state operation, so as to perform line maintenance within the stable operation time, thereby improving maintenance efficiency and quality.

[0077] For example, suppose during an inspection, infrared imaging reveals a significant overheated area in a section of the line. Further data on the current carrying capacity and temperature at that location are obtained, and the calculated cross-sectional area is zero. Combining this information, it can be accurately determined that an open circuit fault has occurred in that section of the line.

[0078] For example, suppose that in a power system, historical data analysis reveals that the cross-sectional area detected when a line is damaged is typically between 70% and 90% of the actual cross-sectional area. Referring to relevant industry standards, a preliminary threshold ratio is set at 80%. Then, in actual testing, pilot tests are conducted on select lines, creating faults of varying degrees, and their cross-sectional areas are measured. If the test results show that when the detected cross-sectional area is greater than 80% of the actual cross-sectional area, line damage can be accurately identified with a low false positive rate, then this preset threshold ratio can be applied to line fault diagnosis throughout the entire power system.

[0079] In one possible implementation, the formula for calculating the cross-sectional area is:

[0080]

[0081] ρ=ρ0×(1+α×(T-T0))

[0082] In the formula, S is the cross-sectional area of ​​the cable, ρ is the resistivity of the cable material, L is the length of the cable, I is the current carrying capacity of the cable, U is the rated voltage of the cable, k is the comprehensive coefficient, ρ0 is the resistivity at the reference temperature T0, α is the temperature coefficient of resistance, T is the actual operating temperature, and T0 is the reference temperature.

[0083] In this embodiment, the cross-sectional area of ​​the cable is calculated by measuring parameters such as current carrying capacity, voltage and temperature. If the cross-sectional area is greater than a preset threshold ratio but less than the actual cross-sectional area, it indicates that the line may be broken or damaged.

[0084] In one possible implementation, the image data is acquired via infrared imaging;

[0085] Determining the location of a line fault based on image data includes:

[0086] Image data preprocessing includes median filtering for noise reduction and histogram equalization enhancement.

[0087] Temperature features, grayscale features, and edge features are extracted from the preprocessed image data.

[0088] The extracted temperature features, grayscale features, and edge features are normalized.

[0089] The normalized temperature features, grayscale features, and edge features are input into the fault location judgment model to obtain the coordinates of the fault point.

[0090] In this embodiment, the extracted temperature features, grayscale features, and edge features are normalized to unify the range of these feature values ​​to a specific interval (e.g., [0, 1] or [-1, 1]). Normalization can employ methods such as min-max normalization or Z-score normalization. The fault location determination model is a pre-trained model used to determine the location of the fault point based on the input features. This model can be a machine learning-based model (e.g., support vector machine, random forest) or a deep learning model (e.g., convolutional neural network). The model outputs the coordinates of the fault point in the image, which can be used to accurately locate the fault point in actual power distribution lines.

[0091] In reality, circuit breaking is not instantaneous, or localized overheating may occur before the circuit breaks. For example, poor contact or arcing may occur before the circuit completely breaks, both of which can lead to localized temperature increases, resulting in overheated areas in infrared images. Furthermore, other parts of the circuit may still be energized and operating when a circuit breaks, and these parts may overheat due to overload or other reasons. Therefore, overheated areas associated with the circuit break can still be detected in infrared images.

[0092] In one possible implementation, the process of constructing a knowledge forest includes:

[0093] Acquire multi-source data from the power system, including operating parameters, temperature data, image data, and text data of line faults;

[0094] Knowledge extraction is performed on multi-source data. Knowledge extraction includes extracting entities, extracting relations, and extracting attributes. Entities include line equipment and line fault types, relations include the associations between entities, and attributes include the characteristics of entities.

[0095] The extracted entities, relationships, and attributes are merged, and subject faceted trees are constructed based on the merged entities, relationships, and attributes. A knowledge forest is then constructed based on multiple subject faceted trees.

[0096] In this embodiment, firstly, multi-source data of the power system is acquired, including: operating parameters such as current carrying capacity, voltage, and current; line temperature information acquired through sensors; images of line equipment acquired through infrared imaging; and textual data of line faults, such as fault reports and maintenance records.

[0097] Then, knowledge extraction is performed on the acquired multi-source data, including: Entity extraction: identifying and extracting key entities from the data, such as line equipment (cables, circuit breakers, etc.) and fault types (line breakage, line damage, etc.). Relationship extraction: determining the relationships between entities, such as the relationship between a certain fault type and a specific piece of equipment. Attribute extraction: extracting the features of entities, such as equipment operating parameters, temperature data, etc.

[0098] Secondly, the extracted entities, relationships, and attributes are merged to eliminate redundant and inconsistent information. For example, operating parameters, temperature data, and image data of the same device from different data sources are integrated to form a complete knowledge forest.

[0099] Furthermore, based on the fused knowledge, multiple subject-based faceted trees are constructed. Each tree revolves around a specific subject or theme, organizing knowledge from different facets. For example: Equipment faceted tree: with line equipment as the theme, facets include equipment type, operating parameters, maintenance records, etc. Fault faceted tree: with fault type as the theme, facets include fault causes, manifestations, detection methods, etc. Environmental faceted tree: with the operating environment as the theme, facets include temperature, humidity, load, etc.

[0100] Finally, the multiple subject-specific faceted trees are integrated into a knowledge forest, forming a structured and systematic knowledge system. Each subject-specific faceted tree functions as an independent knowledge unit within the knowledge forest, while simultaneously connecting with other trees through relationships to form an organic whole. The study examines how each subject-specific faceted tree organizes knowledge from different facets around a specific subject or theme, thereby constructing a structured and systematic knowledge forest that provides comprehensive and in-depth knowledge support for power system line fault diagnosis.

[0101] In one possible implementation, the training process of the fault location determination model includes:

[0102] A dataset was constructed using historical fault-related data from a knowledge forest. This dataset includes equipment information, fault conditions, and line fault types from the knowledge forest. Equipment information includes operating parameters, temperature data, and image data. Fault conditions include ambient temperature, current carrying capacity, and cross-sectional area. Line fault types include line open circuits and line damage. Specifically, the dataset includes:

[0103] For line outage and line damage conditions, the data in the dataset is classified according to different temperature data, different cross-sectional area or different current carrying capacity, and sub-data sets corresponding to different temperature data, different cross-sectional area or different current carrying capacity are constructed.

[0104] Fault location determination models were obtained by training convolutional neural networks using subsets of the dataset.

[0105] In this embodiment, two operating conditions, line open circuit and line damage, are classified separately to ensure that the model can learn the characteristics of these two fault types. Within each fault type, the data is further subdivided based on different temperature data, different cross-sectional areas, or different current carrying capacities, forming multiple subsets. This helps the model accurately determine the fault location under different environments.

[0106] In one possible implementation, multimodal features are fused to obtain comprehensive features, including:

[0107] Normalize each modal feature in the multimodal features;

[0108] The normalized modal features are concatenated in sequence to obtain the comprehensive features; the order is as follows: operating parameters, temperature data, and image data.

[0109] In this embodiment, the features of different modalities are arranged and combined in sequence, which can completely preserve the complementary information contained in each modality. This makes the fused comprehensive features more fully reflect the characteristics of the data, providing a rich information foundation for subsequent model training and fault diagnosis, and improving the performance and reliability of the fault diagnosis model.

[0110] In other possible implementations, the features of different modalities are linearly weighted and summed to obtain the comprehensive features.

[0111] In this embodiment, the computational complexity of weighted summation is low, but it requires that the feature dimensions of each modality be the same. The importance of certain modalities is emphasized by adjusting the weights.

[0112] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0113] The following are device embodiments of the present invention. For details not described in detail, please refer to the corresponding method embodiments described above.

[0114] Figure 2 The diagram shows a schematic of the line fault diagnosis device based on knowledge forest and multimodal data provided in an embodiment of the present invention. For ease of explanation, only the parts related to the embodiment of the present invention are shown, and are described in detail below:

[0115] like Figure 2 As shown, the line fault diagnosis device 2 based on knowledge forest and multimodal data includes:

[0116] The multimodal data acquisition module 21 is used to acquire multimodal data of the distribution network line to be tested. The multimodal data includes the operating parameters of the distribution network, the temperature data of the line, and image data. The operating parameters include current carrying capacity. The image data is acquired through infrared imaging.

[0117] The multimodal feature acquisition module 22 is used to input multimodal data into a pre-trained multimodal base model to obtain multimodal features, which are key features in different modal data.

[0118] Feature fusion module 23 is used to fuse multimodal features to obtain comprehensive features;

[0119] The fault analysis module 24 is used to analyze the comprehensive features using a pre-built knowledge forest to obtain fault analysis results. The knowledge forest includes line equipment information, fault conditions and line fault types. The fault analysis results include line fault types and fault locations. Line fault types include line open circuit and line damage.

[0120] In one possible implementation, the fault resolution module 24 can be used for:

[0121] Determine the location of the line fault based on image data;

[0122] The cross-sectional area of ​​the line to be tested is calculated based on the current carrying capacity and temperature data corresponding to the location of the line fault.

[0123] The type of line fault is determined by the cross-sectional area being inspected.

[0124] In one possible implementation, the fault resolution module 24 can be used for:

[0125] When the cross-sectional area detected is zero, the line fault type is determined to be a line open circuit.

[0126] When the detected cross-sectional area is smaller than the actual cross-sectional area but greater than the preset threshold ratio, the line fault type is determined to be line damage.

[0127] In one possible implementation, the fault resolution module 24 can be used for:

[0128] Image data preprocessing includes median filtering for noise reduction and histogram equalization enhancement.

[0129] Temperature features, grayscale features, and edge features are extracted from the preprocessed image data.

[0130] The extracted temperature features, grayscale features, and edge features are normalized.

[0131] The normalized temperature features, grayscale features, and edge features are input into the fault location judgment model to obtain the coordinates of the fault point.

[0132] In one possible implementation, the fault resolution module 24 can be used for:

[0133] Acquire multi-source data from the power system, including operating parameters, temperature data, image data, and text data of line faults;

[0134] Knowledge extraction is performed on multi-source data. Knowledge extraction includes extracting entities, extracting relations, and extracting attributes. Entities include line equipment and line fault types, relations include the associations between entities, and attributes include the characteristics of entities.

[0135] The extracted entities, relationships, and attributes are merged, and subject faceted trees are constructed based on the merged entities, relationships, and attributes. A knowledge forest is then constructed based on multiple subject faceted trees.

[0136] Figure 3 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. For example... Figure 3 As shown, the electronic device 3 of this embodiment includes a processor 30 and a memory 31. The memory 31 stores a computer program 32. When the processor 30 executes the computer program 32, it implements the steps in the various method embodiments described above. Alternatively, when the processor 30 executes the computer program 32, it implements the functions of each module / unit in the various device embodiments described above.

[0137] For example, computer program 32 may be divided into one or more modules / units, which are stored in memory 31 and executed by processor 30 to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of computer program 32 in electronic device 3.

[0138] Electronic device 3 may include, but is not limited to, processor 30 and memory 31. Those skilled in the art will understand that... Figure 3 This is merely an example of electronic device 3 and does not constitute a limitation on electronic device 3. It may include more or fewer components than shown, or combine certain components, or different components. For example, electronic device 3 may also include input / output devices, network access devices, buses, etc.

[0139] For the sake of simplicity and clarity, only the above-described functional modules / units are used as examples. In practical applications, the functions described above can be assigned to different functional modules / units as needed. These modules / units can be implemented in hardware, software, or a combination of both.

[0140] In the above embodiments, the descriptions of each embodiment have their own emphasis. Parts not detailed or described in a particular embodiment can be referred to in the relevant descriptions of other embodiments. Unless otherwise specified or in conflict with logic, the terminology and / or descriptions between different embodiments are consistent and can be referenced interchangeably. Technical features in different embodiments can be combined to form new embodiments based on their inherent logical relationships.

[0141] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for diagnosing line faults based on knowledge forest and multimodal data, characterized in that, include: Acquire multimodal data of the distribution network line to be tested, including the operating parameters of the distribution network, temperature data of the line, and image data; The multimodal data is input into a pre-trained multimodal base model to obtain multimodal features, which are key features in different modal data. Multimodal features are fused to obtain comprehensive features; The comprehensive features are analyzed using a pre-constructed knowledge forest to obtain fault analysis results; wherein, the knowledge forest includes line equipment information, fault conditions and line fault types, and the fault analysis results include line fault types and fault locations, and the line fault types include line open circuits and line damage.

2. The line fault diagnosis method based on knowledge forest and multimodal data according to claim 1, characterized in that, The operating parameters include the carrying capacity; The process of parsing the comprehensive features using a pre-constructed knowledge forest includes: Determine the location of the line fault based on the image data; The cross-sectional area of ​​the line to be tested is calculated based on the current carrying capacity and temperature data corresponding to the location of the line fault. The type of line fault is determined based on the cross-sectional area of ​​the detection.

3. The line fault diagnosis method based on knowledge forest and multimodal data according to claim 2, characterized in that, The step of determining the line fault type based on the detected cross-sectional area includes: When the detected cross-sectional area is zero, the line fault type is determined to be a line open circuit. When the detected cross-sectional area is smaller than the actual cross-sectional area but greater than a preset threshold ratio, the line fault type is determined to be line damage.

4. The line fault diagnosis method based on knowledge forest and multimodal data according to claim 3, characterized in that, The formula for calculating the cross-sectional area of ​​the detector is: ρ=ρ0×(1+α×(T-T0)) In the formula, S is the cross-sectional area of ​​the cable, ρ is the resistivity of the cable material, L is the length of the cable, I is the current carrying capacity of the cable, U is the rated voltage of the cable, k is the comprehensive coefficient, ρ0 is the resistivity at the reference temperature T0, α is the temperature coefficient of resistance, T is the actual operating temperature, and T0 is the reference temperature.

5. The line fault diagnosis method based on knowledge forest and multimodal data according to claim 2, characterized in that, The image data was acquired through infrared imaging; Determining the location of the line fault based on the image data includes: The image data is preprocessed; wherein the preprocessing includes median filtering for noise reduction and histogram equalization enhancement. Temperature features, grayscale features, and edge features are extracted from the preprocessed image data. The extracted temperature features, grayscale features, and edge features are normalized. The normalized temperature features, grayscale features, and edge features are input into the fault location judgment model to obtain the coordinates of the fault point.

6. The line fault diagnosis method based on knowledge forest and multimodal data according to claim 5, characterized in that, The training process of the fault location determination model includes: A dataset is constructed using data related to historical faults from a knowledge forest. This data includes equipment information, fault conditions, and line fault types from the knowledge forest. The equipment information includes operating parameters, temperature data, and image data. The fault conditions include ambient temperature, current carrying capacity, and cross-sectional area. The line fault types include line open circuits and line damage, specifically including: For line outage and line damage conditions respectively, the data in the dataset is classified according to different temperature data, different cross-sectional area or different current carrying capacity, and sub-data sets corresponding to different temperature data, different cross-sectional area or different current carrying capacity are constructed. The fault location determination model is obtained by training a convolutional neural network using the aforementioned subset dataset.

7. The line fault diagnosis method based on knowledge forest and multimodal data according to claim 1, characterized in that, The construction process of the knowledge forest includes: Acquire multi-source data from the power system, including operating parameters, temperature data, image data, and text data of line faults; Knowledge extraction is performed on the multi-source data. The knowledge extraction includes extracting entities, extracting relationships, and extracting attributes. The entities include line equipment and line fault types. The relationships include associations between entities. The attributes include features of entities. The extracted entities, relationships, and attributes are merged, and subject faceted trees are constructed based on the merged entities, relationships, and attributes. The knowledge forest is then constructed based on multiple subject faceted trees.

8. The line fault diagnosis method based on knowledge forest and multimodal data according to claim 1, characterized in that, The process of fusing multimodal features to obtain comprehensive features includes: Normalize each modal feature in the multimodal features; The normalized modal features are concatenated in sequence to obtain the comprehensive features; wherein, the sequence is operating parameters, temperature data, and image data.

9. A line fault diagnosis device based on knowledge forest and multimodal data, characterized in that, include: The multimodal data acquisition module is used to acquire multimodal data of the distribution network line to be tested. The multimodal data includes the operating parameters of the distribution network, the temperature data of the line, and image data. The multimodal feature acquisition module is used to input the multimodal data into a pre-trained multimodal base model to obtain multimodal features, wherein the multimodal features are key features in different modal data; The feature fusion module is used to fuse multimodal features to obtain comprehensive features; The fault analysis module is used to analyze the comprehensive features using a pre-built knowledge forest to obtain fault analysis results; wherein, the knowledge forest includes line equipment information, fault conditions and line fault types, and the fault analysis results include line fault types and fault locations, and the line fault types include line open circuits and line damage.

10. An electronic device, characterized in that, It includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method as described in any one of claims 1 to 8.

Images