High-voltage direct-current transformer substation risk visualization method based on label propagation algorithm

Abstract

The invention relates to a high-voltage direct-current transformer substation risk visualization method and device based on a label propagation algorithm and a medium, and the method comprises the steps: collecting the real-time operation data of eight types of devices, such as a transformer, a switch and a primary measurement device, and constructing a quintuple sample containing a device type, a subitem data type, a data format, a normal value range and a measurement value; constructing a substation state description graph based on the state vector, and calculating an edge weight between nodes through an exponential decay function; using a semi-supervised label propagation algorithm to initialize an 8 * 10-dimensional risk label matrix, and iteratively propagating an equipment security state label through a probability transfer matrix; calculating the geometric center of each category by adopting a confidence learning mechanism, and removing noise samples deviating from the center and exceeding a threshold multiple distance; and outputting the risk tags after batch standardization processing, and generating a three-level visual thermodynamic diagram of an equipment layer, a subitem data layer and a risk tag layer. According to the method, real-time risk perception of multi-dimensional data fusion is realized, and the power grid abnormity early warning precision and the operation and maintenance efficiency are improved.

Description

Technical Field

[0001] This invention belongs to the field of power grid safety early warning, and more specifically, relates to a method, equipment and storage medium for visualizing risks in high voltage DC substations based on a tag propagation algorithm. Background Technology

[0002] In ultra-high voltage direct current (UHVDC) substations, a large amount of structured and unstructured power grid data exists. In recent years, a large amount of image data from AC / DC power grids, inverter equipment, and lines has been integrated, enhancing grid visibility while placing higher demands on data processing, identification, and the mining of hidden information. Current technologies, such as traditional power grid inspections relying on manual threshold setting and independent equipment monitoring, cannot effectively capture the correlation between equipment states and the transmission paths of hidden risks. This results in inaccurate power image recognition, failing to meet the precision and real-time requirements of power grid safety early warning. Therefore, there is an urgent need for a method that can achieve multi-dimensional data fusion and improve the accuracy of risk identification to solve the challenge of safety situational awareness in UHVDC substations operating within a massive heterogeneous data environment. Summary of the Invention

[0003] To address the problem of inaccurate existing power image recognition methods, this invention utilizes convolutional neural networks to improve label recognition and image denoising methods. The main contents of the solution are as follows.

[0004] This invention provides a risk visualization method for high-voltage direct current (HVDC) substations based on a label propagation algorithm. The method includes the following steps: acquiring real-time operating data of various types of equipment in the substation, including at least transformers, switches, primary measuring devices, surge arresters, DC-side systems, valve cooling systems, AC / DC filtering devices, and infrared temperature measuring devices; constructing a substation state description graph based on the operating data, where nodes represent substation state vectors at different times, and edge weights are determined by an exponential decay function of the Euclidean distance between nodes; propagating risk labels to unlabeled nodes in the state description graph using a semi-supervised label propagation algorithm, and initializing a label matrix using pre-set labeled sample nodes; cleaning label noise using a confidence learning mechanism, calculating the geometric center of each sample category, and removing abnormal samples that deviate from the center; outputting risk labels for the operating data of each equipment sub-item, and generating a visualized risk distribution map based on the label levels.

[0005] In one embodiment, the data sample construction step includes: defining data information for ultra-high voltage substation samples, and performing data dimensional analysis and data sample construction based on the data information, in the following manner:

[0006] f = <D,T,S,V,N>

[0007] Where f is a set of substation status data, D is the equipment type, namely the eight equipment types including the above substation operation data; T is the sub-item operation data, namely the sub-item operation data of the above eight items; S is the data format type, including two types: switch status data (discrete type) and operation status data (continuous type); V is the normal operating range of this type of data; and N is the measurement value of the sub-item data.

[0008] In one embodiment, a substation status description diagram is constructed based on substation data samples, and the method is as follows:

[0009]

[0010] Where X is the set of substation states at m time points, and for the substation state x at the i-th time point... i Its state consists of n sub-item data measurements, and the edges between nodes represent the similarity between samples. The edge weights between nodes i and j are:

[0011]

[0012] Where α is a hyperparameter. i With x j Let be the modulus values ​​of nodes i and j.

[0013] In one embodiment, the method further includes: predicting the label information of unlabeled nodes from the label information of labeled nodes based on a semi-supervised learning method, wherein each node label is propagated to neighboring nodes according to similarity, and at each step of node propagation, each node updates its own label based on the labels of neighboring nodes.

[0014] In one embodiment, the specific label propagation algorithm steps are as follows:

[0015] Define a probability transition matrix P:

[0016]

[0017] Where Pij represents the probability of transitioning from node i to node j; Wij is the weight from i to j.

[0018] In one embodiment, tag noise mitigation employs confidence learning to clean the tags, and the algorithm steps are as follows:

[0019] Find the center sample of each category. For a certain category of samples X1 to Xn, calculate its geometric center X. mid as follows:

[0020]

[0021] Calculate each sample with Xmid The average distance d is:

[0022]

[0023] Define distance from X mid Samples whose distance exceeds three times the average value are considered outliers, and their existing sample labels are removed. After filtering out erroneous samples, the sample category weights are readjusted, and the system is retrained to finally obtain converged risk label recognition results for each type.

[0024] In one embodiment, the iterative process of the label propagation includes: performing a label propagation operation: F = P × F, where F is the label matrix; resetting the values ​​of labeled nodes to the initial label; repeating steps (a)-(b) until the label matrix converges.

[0025] In one embodiment, the risk level mapping is a three-layer structure: the equipment type layer, the equipment sub-item data layer, and the risk label layer, which are presented in the form of a heat map on the visualization interface.

[0026] It's important to note that the output process for anomaly risk assessment results involves batch standardization of neuron input values. Batch standardization uses certain normalization techniques to forcibly pull the distribution of the input value of any neuron in each layer of the neural network back to a standard normal distribution with a mean of 0 and a variance of 1. This forces the increasingly skewed distribution back to a more standard distribution, ensuring that the activation input values ​​fall within the region where the nonlinear function is highly sensitive to input. Therefore, small changes in the input will lead to significant changes in the loss function, meaning the gradient is increased, preventing the gradient vanishing problem. Furthermore, a larger gradient means faster learning convergence, significantly accelerating the training process. After training, the optimal hyperparameters are obtained, yielding the output results for all samples, indicating the safety risk level represented by each image.

[0027] Secondly, the present invention also provides a high-voltage direct current substation risk visualization device based on a tag propagation algorithm, comprising: one or more processors; and a memory storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement any of the high-voltage direct current substation risk visualization methods based on the tag propagation algorithm described above.

[0028] Thirdly, the present invention provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to execute any of the aforementioned high-voltage DC substation risk visualization methods based on the tag propagation algorithm.

[0029] Beneficial effects

[0030] This invention provides a risk visualization method for high-voltage direct current (HVDC) substations based on a tag propagation algorithm. The constructed automatic tag propagation and risk visualization system fundamentally solves the challenges of accuracy and real-time performance in safety situational awareness of UHVDC substations under massive heterogeneous data environments. Traditional power grid inspections rely on manual threshold setting and independent equipment monitoring, failing to effectively capture the state correlations between equipment and the transmission paths of hidden risks. This technology, through structured modeling of multi-type equipment operating data, achieves for the first time a unified mathematical representation of discrete switching signals and continuous operating parameters, laying the foundation for multi-source data fusion analysis. The tag propagation algorithm innovatively transforms the substation state description graph into an undirected graph network, using a similarity function to dynamically calculate the edge weights between nodes, enabling risk tags to automatically propagate along physically correlated equipment nodes—for example, when an anomaly occurs in the converter valve cooling system, the algorithm preferentially propagates the risk tag to the associated DC-side power nodes through high-weight connections, significantly improving the sensitivity for identifying small-sample abnormal operating conditions. The confidence learning mechanism introduces geometric center calculation in multi-dimensional space to replace the conventional mean method, effectively resisting false alarms caused by transient sensor interference. In practical deployments, the system can accurately filter false signals generated by electromagnetic interference, avoiding accidental triggering of shutdown commands. The visualization engine employs hierarchical heatmap rendering technology, allowing maintenance personnel to intuitively observe how abnormal equipment status is transmitted to related systems through current fluctuations. Long-term operational verification has shown that the system's accuracy in identifying core defects far surpasses traditional infrared inspection methods. The semi-supervised learning feature significantly reduces reliance on historical fault labeling data. When the system detects an abnormal combination of key parameters, it can pinpoint the root cause in a very short time, achieving an order-of-magnitude improvement in efficiency compared to manual diagnosis. This technology advances the fault warning window to the nascent stage of potential problems, providing decision support for preventative maintenance through visualized simulations of risk transmission paths. In extreme weather emergency responses, the system successfully warned of insulation failure risks caused by environmental factors, guiding maintenance personnel to activate protective devices in advance and preventing cascading grid accidents. Full lifecycle analysis shows that this technology can significantly reduce unplanned downtime, extend the service life of critical equipment, and comprehensively improve the power supply resilience and disaster response capabilities of the smart grid. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0032] Figure 1 A flowchart illustrating the risk visualization method for high-voltage direct current substations based on the tag propagation algorithm provided in this embodiment of the invention.

[0033] Figure 2 A schematic diagram of the structure of a high-voltage direct current substation risk visualization device based on the tag propagation algorithm provided in an embodiment of the present invention. Detailed Implementation

[0034] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0035] It should be noted that if the embodiments of the present invention involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of the components in a specific posture. If the specific posture changes, the directional indicators will also change accordingly.

[0036] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Thus, features defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the use of "and / or" or "and / or" throughout the text implies three parallel solutions. For example, "A and / or B" includes solution A, solution B, or a solution where both A and B are satisfied. Furthermore, the technical solutions of the various embodiments can be combined, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0037] To address the problem of inaccurate existing power image recognition methods, this invention utilizes convolutional neural networks to improve label recognition and image denoising methods. The main contents of the solution are as follows.

[0038] refer to Figure 1 This invention provides a risk visualization method for high-voltage direct current substations based on a tag propagation algorithm, characterized by the following steps:

[0039] S100. Obtain real-time operating data of various types of equipment in the substation, wherein the types of equipment include at least transformers, switches, primary measuring equipment, surge arresters, DC side systems, valve cooling systems, AC / DC filtering devices, and infrared temperature measuring devices.

[0040] S200. Construct a substation state description diagram based on the operating data, wherein the nodes of the diagram represent the substation state vectors at different times, and the weights of the edges are determined by the exponential decay function of the Euclidean distance between the nodes.

[0041] S300. Risk label propagation is performed on unlabeled nodes in the state description graph based on a semi-supervised label propagation algorithm, and the label matrix is ​​initialized using preset labeled sample nodes.

[0042] S400: A confidence learning mechanism is used to clean up label noise, calculate the geometric center of each sample category, and remove abnormal samples that deviate from the center.

[0043] S500 outputs risk labels for the operating data of each device sub-item and generates a visual risk distribution map based on the label level.

[0044] It should be noted that the algorithm for constructing the data sample is as follows: Define the data information for the UHV substation sample, including the following basic data types: Transformer operation data: Data monitoring the operation of the transformer, mainly including bushing pressure, tap changer operation count, tap changer pressure, oil temperature, wire temperature, oil level, core clamp current, tap position, etc. Switch operation data: Real-time data monitoring the insulation and operating structure of the switch, mainly including SF6 pressure, operation count, pressurization count, oil pressure or air pressure, etc. Primary measuring equipment data: Real-time data used to monitor the operation of primary equipment, mainly including drive current, operating temperature, etc. Surge arrester operation data: Real-time data monitoring the operation of surge arresters, mainly including operation count and leakage current, etc. DC side operation data: Real-time data monitoring the correctness of the high-voltage DC system control, mainly including DC voltage, current, power, firing angle, and tap position, etc. Valve cooling system operation data: Real-time data monitoring the cooling effect of water-cooled converter valves, mainly including pressure, flow rate, temperature, conductivity, water level, and other operating data. AC / DC filter operation data: Real-time data monitoring the normal operation of capacitors, mainly including capacitor unbalanced current and other operating data. Infrared thermography operation data: Real-time temperature operation data of equipment monitored by an infrared imager, including the temperature of voltage-induced heating bushings, voltage transformers, and current-induced heating switches, disconnectors, transformers, valves, and other equipment.

[0045] In one embodiment, the data sample construction step includes: defining data information for ultra-high voltage substation samples, and performing data dimensional analysis and data sample construction based on the data information, in the following manner:

[0046] f = <D,T,S,V,N>

[0047] Where f is a set of substation status data, D is the equipment type, namely the eight equipment types including the above substation operation data; T is the sub-item operation data, namely the sub-item operation data of the above eight items; S is the data format type, including two types: switch status data (discrete type) and operation status data (continuous type); V is the normal operating range of this type of data; and N is the measurement value of the sub-item data.

[0048] In one embodiment, a substation status description diagram is constructed based on substation data samples, and the method is as follows:

[0049]

[0050] Where X is the set of substation states at m time points, and for the substation state x at the i-th time point... i Its state consists of n sub-item data measurements, and the edges between nodes represent the similarity between samples. The edge weights between nodes i and j are:

[0051]

[0052] Where α is a hyperparameter. i With x j Let be the modulus values ​​of nodes i and j.

[0053] In one embodiment, the method further includes: predicting the label information of unlabeled nodes from the label information of labeled nodes based on a semi-supervised learning method, wherein each node label is propagated to neighboring nodes according to similarity, and at each step of node propagation, each node updates its own label based on the labels of neighboring nodes.

[0054] In one embodiment, the specific label propagation algorithm steps are as follows:

[0055] Define an NxN probability transition matrix P:

[0056]

[0057] Where Pij represents the probability of moving from node i to node j.

[0058] In one embodiment, tag noise mitigation employs confidence learning to clean the tags, and the algorithm steps are as follows:

[0059] Find the center sample of each category. For a certain category of samples X1 to Xn, calculate its geometric center X. mid as follows:

[0060]

[0061] Calculate each sample with X mid The average distance d is:

[0062]

[0063] Define distance from X mid Samples whose distance exceeds three times the average value are considered outliers, and their existing sample labels are removed. After filtering out erroneous samples, the sample category weights are readjusted, and the system is retrained according to claim 3 to finally obtain converged risk label recognition results for each type.

[0064] In one embodiment, the iterative process of the label propagation includes: performing a label propagation operation: F = P × F, where F is the label matrix; resetting the values ​​of labeled nodes to the initial label; repeating steps (a)-(b) until the label matrix converges.

[0065] In one embodiment, the risk level mapping is a three-layer structure: the equipment type layer, the equipment sub-item data layer, and the risk label layer, which are presented in the form of a heat map on the visualization interface.

[0066] It's important to note that the output process for anomaly risk assessment results involves batch standardization of neuron input values. Batch standardization uses certain normalization techniques to forcibly pull the distribution of the input value of any neuron in each layer of the neural network back to a standard normal distribution with a mean of 0 and a variance of 1. This forces an increasingly skewed distribution back to a more standard one, ensuring that the activation input values ​​fall within the region where the nonlinear function is highly sensitive to input. Therefore, small changes in the input will lead to significant changes in the loss function, essentially increasing the gradient and preventing the vanishing gradient problem. Furthermore, a larger gradient means faster learning convergence, significantly accelerating the training process. After training, the optimal hyperparameters are obtained, yielding the output results for all samples and indicating the safety risk level represented by each image.

[0067] In some embodiments, based on image detection results from a local ultra-high voltage substation, a convolutional neural network algorithm is used to identify equipment and defects within the image. Based on the obtained fusion detection model and the data to be detected in the image, the preliminary detection results are classified and analyzed to reduce the error of a single detection method, and the initial results of both detection methods are received. Next, the class of the detected sample is determined, and the corresponding detection model is adopted to correct the two detection results, resulting in the final detection results shown in Tables 1 and 2.

[0068] Table 1 shows the identification results for the devices.

[0069]

[0070] Table 1 shows the identification results for the devices.

[0071]

[0072] Table 2 shows the results of fault and defect identification.

[0073] The results above demonstrate that the proposed algorithm can accurately identify basic lines and equipment, with a significant improvement in power topology identification, particularly in complex environments. Furthermore, it can accurately identify power grid defects in various scenarios. This solution can save on normal power grid operation and maintenance costs and provide timely early warnings for faults of various levels.

[0074] Secondly, the present invention also provides a high-voltage direct current substation risk visualization device based on a tag propagation algorithm, comprising: one or more processors; and a memory storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement any of the high-voltage direct current substation risk visualization methods based on the tag propagation algorithm described above.

[0075] Thirdly, the present invention provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to execute any of the aforementioned high-voltage DC substation risk visualization methods based on the tag propagation algorithm.

[0076] Figure 2 This is a schematic diagram of the structure of a high-voltage direct current (HVDC) substation risk visualization device based on a tag propagation algorithm, provided in an embodiment of this application. The device includes a processor 31, a memory 32, a communication device 33, an input device 34, and an output device 35. The number of processors 31 in the tag propagation algorithm-based HVDC substation risk visualization device can be one or more, and the number of memories 32 in the long-term fire hazard testing and evaluation device for solid insulation materials of power equipment can also be one or more. The processor 31, memory 32, communication device 33, input device 34, and output device 35 of this tag propagation algorithm-based HVDC substation risk visualization device can be connected via a bus or other means.

[0077] The memory 32, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the tag propagation algorithm-based high-voltage DC substation risk visualization method in any embodiment of this application. The memory 32 may mainly include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a function; the data storage area may store data created based on the use of the device, etc. Furthermore, the memory 32 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to the device via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0078] The communication device 33 is used for data transmission.

[0079] The processor 31 executes various functional applications and data processing of the device by running software programs, instructions and modules stored in the memory 32, thereby realizing the above-mentioned risk visualization method for high voltage DC substations based on tag propagation algorithm.

[0080] Input device 34 can be used to receive input digital or character information, and to generate key signal inputs related to user settings and function control of the device. Output device 35 may include display devices such as a display screen.

[0081] The high-voltage direct current substation risk visualization device based on the tag propagation algorithm provided above can be used to execute the high-voltage direct current substation risk visualization method based on the tag propagation algorithm provided in the above embodiments, and has corresponding functions and beneficial effects.

[0082] This application also provides a storage medium containing computer-executable instructions. When executed by a computer processor, the computer-executable instructions are used to execute a risk visualization method for high-voltage DC substations based on a label propagation algorithm. The method includes the following steps: acquiring real-time operating data of various types of equipment in the substation, wherein the equipment types include at least transformers, switches, primary measuring devices, surge arresters, DC-side systems, valve cooling systems, AC / DC filtering devices, and infrared temperature measuring devices; constructing a substation state description graph based on the operating data, wherein the nodes of the graph represent substation state vectors at different times, and the edge weights are determined by an exponential decay function of the Euclidean distance between nodes; propagating risk labels to unlabeled nodes in the state description graph using a semi-supervised label propagation algorithm, and initializing a label matrix using preset labeled sample nodes; cleaning label noise using a confidence learning mechanism, calculating the geometric center of each sample category and removing abnormal samples that deviate from the center; outputting risk labels for the operating data of each equipment sub-item, and generating a visualized risk distribution map based on the label levels.

[0083] Storage medium – any type of memory device or storage apparatus. The term “storage medium” is intended to include: mounting media, such as CD-ROMs, floppy disks, or magnetic tape devices; computer system memory or random access memory, such as DRAM, DDR RAM, SRAM, EDO RAM, Rambus RAM, etc.; non-volatile memory, such as flash memory, magnetic media (e.g., hard disks or optical storage); registers or other similar types of memory elements, etc. Storage media may also include other types of memory or combinations thereof. Furthermore, storage media may reside in a first computer system in which a program is executed, or may reside in a different second computer system connected to the first computer system via a network (such as the Internet). The second computer system can provide program instructions to the first computer for execution. The term “storage medium” can include two or more storage media residing in different locations (e.g., in different computer systems connected via a network). Storage media may store program instructions (e.g., specifically implemented as a computer program) executable by one or more processors.

[0084] Of course, the computer-executable instructions provided in the embodiments of this application are not limited to the above-mentioned high-voltage DC substation risk visualization method based on tag propagation algorithm, but can also execute related operations in the high-voltage DC substation risk visualization method based on tag propagation algorithm provided in any embodiment of this application.

[0085] The high-voltage direct current substation risk visualization device based on the tag propagation algorithm provided in the above embodiments can execute the high-voltage direct current substation risk visualization method based on the tag propagation algorithm provided in any embodiment of this application. For technical details not described in detail in the above embodiments, please refer to the high-voltage direct current substation risk visualization method based on the tag propagation algorithm provided in any embodiment of this application.

[0086] This invention provides a risk visualization method for high-voltage direct current (HVDC) substations based on a tag propagation algorithm. The constructed automatic tag propagation and risk visualization system fundamentally solves the challenges of accuracy and real-time performance in safety situational awareness of UHVDC substations under massive heterogeneous data environments. Traditional power grid inspections rely on manual threshold setting and independent equipment monitoring, failing to effectively capture the state correlations between equipment and the transmission paths of hidden risks. This technology, through structured modeling of multi-type equipment operating data, achieves for the first time a unified mathematical representation of discrete switching signals and continuous operating parameters, laying the foundation for multi-source data fusion analysis. The tag propagation algorithm innovatively transforms the substation state description graph into an undirected graph network, using a similarity function to dynamically calculate the edge weights between nodes, enabling risk tags to automatically propagate along physically correlated equipment nodes—for example, when an anomaly occurs in the converter valve cooling system, the algorithm preferentially propagates the risk tag to the associated DC-side power nodes through high-weight connections, significantly improving the sensitivity for identifying small-sample abnormal operating conditions. The confidence learning mechanism introduces geometric center calculation in multi-dimensional space to replace the conventional mean method, effectively resisting false alarms caused by transient sensor interference. In practical deployments, the system can accurately filter false signals generated by electromagnetic interference, avoiding accidental triggering of shutdown commands. The visualization engine employs hierarchical heatmap rendering technology, allowing maintenance personnel to intuitively observe how abnormal equipment status is transmitted to related systems through current fluctuations. Long-term operational verification has shown that the system's accuracy in identifying core defects far surpasses traditional infrared inspection methods. The semi-supervised learning feature significantly reduces reliance on historical fault labeling data. When the system detects an abnormal combination of key parameters, it can pinpoint the root cause in a very short time, achieving an order-of-magnitude improvement in efficiency compared to manual diagnosis. This technology advances the fault warning window to the nascent stage of potential problems, providing decision support for preventative maintenance through visualized simulations of risk transmission paths. In extreme weather emergency responses, the system successfully warned of insulation failure risks caused by environmental factors, guiding maintenance personnel to activate protective devices in advance and preventing cascading grid accidents. Full lifecycle analysis shows that this technology can significantly reduce unplanned downtime, extend the service life of critical equipment, and comprehensively improve the power supply resilience and disaster response capabilities of the smart grid.

[0087] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural transformations made using the contents of the present invention's specification and drawings under the inventive concept of the present invention, or direct / indirect applications in other related technical fields, are included within the patent protection scope of the present invention.

Claims

1. A risk visualization method for high-voltage direct current substations based on a tag propagation algorithm, characterized in that, Includes the following steps: The system acquires real-time operating data of various types of equipment in the substation, including at least transformers, switches, primary measuring devices, surge arresters, DC side systems, valve cooling systems, AC / DC filtering devices, and infrared temperature measuring devices. A substation state description graph is constructed based on the aforementioned operational data, where the nodes of the graph represent the substation state vectors at different times, and the weights of the edges are determined by the exponential decay function of the Euclidean distance between the nodes. Risk labels are propagated for unlabeled nodes in the state description graph based on a semi-supervised label propagation algorithm, and the label matrix is ​​initialized using pre-set labeled sample nodes. A confidence learning mechanism is used to clean up label noise, calculate the geometric center of each sample category, and remove outlier samples that deviate from the center. Output risk labels for the operational data of each device sub-item, and generate a visual risk distribution map based on the label level.

2. The high-voltage direct current substation risk visualization method based on tag propagation algorithm according to claim 1, characterized in that, Also includes: Define the data information for ultra-high voltage substation samples, and based on this data information, perform data dimensional analysis and data sample construction in the following manner: f = <D,T,S,V,N> Where f is a set of substation status data, D is the equipment type, namely the eight equipment types including the above substation operation data; T is the sub-item operation data, namely the sub-item operation data of the above eight items; S is the data format type, including switch status data and operation status data; V is the normal operating range of this type of data; and N is the sub-item data measurement value.

3. The high-voltage direct current substation risk visualization method based on tag propagation algorithm according to claim 2, characterized in that, The method for constructing a substation status description diagram based on substation data samples is as follows: Where X is the set of substation states at m time points, and for the substation state x at the i-th time point... i Its state consists of n sub-item data measurements, and the edges between nodes represent the similarity between samples. The edge weights between nodes i and j are: Where α is a hyperparameter, x i With x j Let be the modulus values ​​of nodes i and j.

4. The high-voltage direct current substation risk visualization method based on tag propagation algorithm according to claim 2, characterized in that, Also includes: Based on a semi-supervised learning method, the label information of unlabeled nodes is predicted from the label information of labeled nodes. Each node's label is propagated to neighboring nodes according to similarity. At each step of the node propagation, each node updates its own label based on the labels of its neighboring nodes.

5. The high-voltage direct current substation risk visualization method based on tag propagation algorithm according to claim 1, characterized in that, The specific steps of the label propagation algorithm are as follows: Define a probability transition matrix P: Where P ij W represents the probability of transitioning from node i to node j; ij Let be the weights from i to j.

6. The high-voltage direct current substation risk visualization method based on tag propagation algorithm according to claim 1, characterized in that, Anti-label noise is addressed by using confidence learning to clean the labels. The algorithm steps are as follows: Find the center sample of each category, where X1 to Xn are the category samples, and calculate their geometric center X. mid as follows: Calculate each sample with X mid The average distance d is: Define distance from X mid Samples whose distance exceeds three times the average are considered outliers and are removed. mid Sample labels are already available; After filtering out erroneous samples, the sample category weights are readjusted and the system is retrained to finally obtain converged risk label recognition results for each type.

7. The high-voltage direct current substation risk visualization method based on tag propagation algorithm according to claim 1, characterized in that, The iterative process of tag propagation includes: Perform the label propagation operation: F = P × F, where F is the label matrix; reset the values ​​of labeled nodes to the initial label; repeat steps (a)-(b) until the label matrix converges.

8. The risk visualization method for high-voltage direct current substations based on the tag propagation algorithm according to claim 1, characterized in that, The risk level mapping is a three-layer structure, namely the equipment type layer, the equipment sub-item data layer, and the risk label layer, which are presented in the form of a heat map on the visualization interface.

9. A risk visualization device for high-voltage direct current substations based on a tag propagation algorithm, characterized in that, include: One or more processors; The memory stores one or more programs that, when executed by one or more processors, enable the one or more processors to implement the high-voltage direct current substation risk visualization method based on the tag propagation algorithm as described in any one of claims 1-8.

10. A storage medium containing computer-executable instructions, characterized in that, The computer-executable instructions, when executed by a computer processor, are used to perform the high-voltage direct current substation risk visualization method based on the tag propagation algorithm as described in any one of claims 1-8.

Images