Low-dimensional vector and power grid topology thought chain fusion switching operation risk identification method

Abstract

The application provides a low-dimensional vector and power grid topology chain fusion switching operation risk identification method, and relates to the technical field of power systems, and comprises the following steps: standardizing and preprocessing real-time operation data of a power grid, extracting core features through principal component analysis (PCA) to generate low-dimensional vector features; adopting a graph database to store the topological relationship between nodes and edges of the power grid and constructing a power grid topology model; taking a topological element corresponding to an abnormal low-dimensional feature in the low-dimensional vector features as a starting node, and deducing a risk propagation path along a topological chain; bidirectionally cross- verifying the topological deduction results of the low-dimensional vector features and the risk propagation path; and outputting a risk report containing a risk position, a level, a conduction path and a disposal measure in combination with the bidirectionally cross-verified results; the application guarantees the safety of the power grid, avoids causing systematic risks, improves operation efficiency, reduces operation delays, reduces equipment operation and maintenance costs, and builds a solid foundation for high-quality operation of the power grid and continuous and reliable power supply.

Description

Technical Field

[0001] This invention relates to the field of power system technology, and more specifically, to a method for identifying switching operation risks by integrating low-dimensional vectors and power grid topology thinking chains. Background Technology

[0002] In the stable operation of a power system, grid switching operations are a core component for adjusting grid operation modes and equipment maintenance. However, the risks associated with these operations directly affect the safe and stable operation of the grid and the reliable supply of electricity. Misjudgments or erroneous actions during these operations can lead to topological disturbances, causing power flow overload, bus voltage exceeding limits, and in severe cases, triggering cascading faults that result in equipment damage, localized power outages, or even large-scale grid collapse. This not only impacts industrial production and residential electricity use but can also lead to significant economic losses and social consequences.

[0003] With the continuous expansion of the power grid and the increasing penetration of new energy sources, switching operation scenarios are becoming more complex. Traditional risk management methods that rely on manual experience are no longer able to cope with challenges such as high-dimensional data redundancy and complex topological relationships, highlighting the problems of lagging risk identification and misjudgment or omission. Summary of the Invention

[0004] The purpose of this invention is to provide a method for identifying switching operation risks by integrating low-dimensional vectors and power grid topology thinking chains. This method can safeguard the power grid safety barrier, avoid systemic risks caused by operational errors, improve operational efficiency, reduce operational delays caused by untimely risk warnings, reduce equipment maintenance costs and power outage losses, and lay a solid foundation for high-quality power grid operation and continuous and reliable power supply.

[0005] The technical solution of this invention is as follows:

[0006] Firstly, this application provides a method for identifying switching operation risks by fusing low-dimensional vectors and power grid topology thinking chains, which includes the following steps:

[0007] S1. Standardize and preprocess the real-time operation data of the power grid, and extract core features through principal component analysis (PCA) to generate low-dimensional vector features.

[0008] S2. A power grid topology model is constructed by storing the topological relationships between power grid nodes and edges using a graph database.

[0009] S3. Based on the power grid topology model, the risk propagation path is derived along the topology thinking chain, starting with the topology element corresponding to the abnormal low-dimensional feature in the low-dimensional vector feature and taking the topology element corresponding to the abnormal low-dimensional feature in the low-dimensional vector feature as the starting node.

[0010] S4. Verify the topological derivation results of low-dimensional vector features and risk propagation paths through bidirectional cross-validation;

[0011] S5. Based on the results of two-way cross-validation, output a risk report that includes the risk location, level, transmission path, and response measures.

[0012] Further, step S1 includes:

[0013] Z-score standardization preprocessing is performed on real-time power grid operation data to eliminate dimensional differences; the calculation formula for Z-score standardization includes:

[0014]

[0015] In the formula, For the standardized data of the i-th sample on the j-th indicator, Let i be the time sample, and i be the i-th sample. Let be the value of the i-th sample on the j-th indicator;

[0016] The linear correlation between power grid indicators is quantified using covariance analysis on the preprocessed real-time power grid operation data; the formula for calculating the covariance matrix is ​​defined as follows:

[0017]

[0018] In the formula, Let the covariance be the sum of the j-th and k-th indicators. For the i-th sample, the standardized data is based on the k-th indicator;

[0019] Principal components are selected based on the linear correlation among power grid indicators and the 90% cumulative contribution rate criterion; whereby the cumulative contribution rate of the first k principal components is defined as:

[0020]

[0021] In the formula, k represents the first k principal components, and m is the number of monitoring indicators. For the k-th eigenvalue, This is the m-th eigenvalue;

[0022] Principal Component Analysis (PCA) is used to extract core features to generate low-dimensional vector features.

[0023] Furthermore, it also includes verifying the effectiveness of the low-dimensional vector through reconstruction error, the formula for calculating the reconstruction error is:

[0024]

[0025] In the formula, MSE represents the reconstruction error, and m is the number of monitoring indicators. Let i be the time sample, and i be the i-th sample. Let be the value of the i-th sample on the j-th indicator. This is a low-dimensional vector matrix of power grid data. It is the transpose of the projection matrix. Let j be the standard deviation of the j-th indicator. Let be the mean of the j-th indicator.

[0026] Furthermore, in step S2, the process of constructing the power grid topology model includes: using a graph database to store the topological relationships between power grid nodes and edges, and constructing a triplet rule base that supports automatic matching and updating, consisting of principal components, physical quantities, and topological elements.

[0027] Furthermore, in step S3, the process of deriving the risk propagation path along the topological thought chain includes:

[0028] The state of the edge is dynamically updated according to the switching operation steps;

[0029] Starting with the topological elements corresponding to the abnormal low-dimensional features in the low-dimensional vector features, the risk propagation path is derived according to the priority of bus, adjacent lines, associated transformers, and downstream loads.

[0030] Furthermore, the above priority determination method includes: assigning risk weights based on the degree of anomaly of low-dimensional vector features, and determining the priority based on the risk weights, with higher weights indicating higher priority.

[0031] Furthermore, in step S4, the above-mentioned bidirectional cross-validation process includes:

[0032] The authenticity of low-dimensional vector feature anomalies is verified by the topological derivation results of the risk propagation path;

[0033] The effectiveness of the topological derivation of the risk propagation path is verified by using low-dimensional vector features.

[0034] Therefore, by verifying the authenticity of low-dimensional vector feature anomalies, it can be determined whether the anomalies are caused by changes in topology. By verifying the validity of the topological derivation results of the risk propagation path, it can be determined whether the reasoning conclusions are supported by data features.

[0035] Secondly, this application provides an electronic device, comprising:

[0036] Memory, used to store one or more programs;

[0037] processor;

[0038] When one or more of the above programs are executed by the above processor, a method for identifying switching operation risks by fusing low-dimensional vectors and power grid topology thinking chains, as described in any of the first aspects above, is implemented.

[0039] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a method for identifying switching operation risks by fusing low-dimensional vectors and power grid topology thinking chains as described in any of the first aspects above.

[0040] Compared with the prior art, the present invention has at least the following advantages or beneficial effects:

[0041] (1) The present invention provides a method for identifying switching operation risks by integrating low-dimensional vectors and power grid topology thinking chain. The method extracts core features by using algorithms such as principal component analysis through low-dimensional vector calculation, effectively breaking through the dimensional bottleneck and providing efficient and accurate feature input for risk identification.

[0042] (2) This invention uses the power grid topology thinking chain fusion reasoning to dynamically track the topological changes before and after the operation, and uses the "node-edge" relationship as the core to deduce the risk path. The fusion of the two can realize the closed-loop identification of "data characteristics-structural changes-risk source tracing", which fundamentally improves the timeliness and accuracy of risk identification and avoids equipment damage, partial power outages or even systemic power grid accidents caused by misjudgment or omission.

[0043] (3) This invention combines the objectivity of mathematical algorithms with the logic of the power grid physical structure through deep coupling of low-dimensional vector calculation and topological thinking chain, laying the foundation for the automation and intelligence of subsequent scheduling decisions. The low-dimensional feature vector output by the system can be directly connected to the scheduling AI decision model, and the risk path deduced by the topological thinking chain can provide accurate basis for the optimization of the operation process, which meets the core requirements of the new power system for intelligent scheduling.

[0044] (4) The research content of this invention can significantly reduce the incidence of switching operation risks, reduce equipment maintenance costs and power outage losses caused by operational errors, and ensure the continuity of electricity supply for residential use and industrial production. In terms of economic benefits, low-dimensional vector calculation greatly improves data processing efficiency, shortens the time for risk identification, and reduces the cost of manual intervention. At the same time, the dynamic reasoning ability of topological thinking chain can optimize the operation process and reduce the time cost and resource consumption of scheduling operations. Attached Figure Description

[0045] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is a flowchart illustrating the steps of the switching operation risk identification method based on the fusion of low-dimensional vectors and power grid topology thinking chain of the present invention.

[0047] Figure 2 This is a schematic structural block diagram of an electronic device according to an embodiment of the present invention.

[0048] Icons: 101, memory; 102, processor; 103, communication interface. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0050] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0051] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0052] It should be noted that, in this document, the term "comprising" or any other variation thereof is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0053] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the various embodiments and features described below can be combined with each other.

[0054] Example 1

[0055] Please see Figure 1 , Figure 1 The diagram shows the steps of a switching operation risk identification method that integrates low-dimensional vectors and power grid topology thinking chain according to an embodiment of this application.

[0056] This application provides a method for identifying switching operation risks by fusing low-dimensional vectors and power grid topology thinking chains, which includes the following steps:

[0057] S1. Standardize and preprocess the real-time operation data of the power grid, and extract core features through principal component analysis (PCA) to generate low-dimensional vector features.

[0058] S2. A power grid topology model is constructed by storing the topological relationships between power grid nodes and edges using a graph database.

[0059] S3. Based on the power grid topology model, the risk propagation path is derived along the topology thinking chain, starting with the topology element corresponding to the abnormal low-dimensional feature in the low-dimensional vector feature and taking the topology element corresponding to the abnormal low-dimensional feature in the low-dimensional vector feature as the starting node.

[0060] S4. Verify the topological derivation results of low-dimensional vector features and risk propagation paths through bidirectional cross-validation;

[0061] S5. Based on the results of two-way cross-validation, output a risk report that includes the risk location, level, transmission path, and response measures.

[0062] In constructing the power grid topology model, this application integrates a low-dimensional vector embedded with BiLSTM triples containing knowledge representation. The core of this low-dimensional vector is to transform the symbolic knowledge of switching operations into a sequence that BiLSTM can process. Through bidirectional encoding, a distributed vector with both rule constraints and semantic associations is generated, transforming the symbolic knowledge of switching operations (including entity attributes, rule constraints, and structural associations) into an ordered sequence, laying the foundation for distributed mapping. The key is to preserve the logical structure of the symbolic knowledge. Taking a typical triple ("bus B2", "power outage operation", "grounding switch G1 closure") as an example, the sequence transformation needs to completely preserve the head entity attribute sequence, the tail entity attribute sequence, and the relation rule sequence. Integrating these three complete sequences into a logical structure containing symbolic knowledge can adapt to the time-series processing characteristics of BiLSTM and achieve lossless transformation of symbolic knowledge into sequence data.

[0063] Building upon this foundation, this application constructs a dedicated mapping dictionary based on symbolic knowledge classification, ensuring a correspondence between symbolic knowledge and vectors. Each entity symbol corresponds to a unique initial quantity, and vector initialization employs One-Hot encoding to convert to low-dimensional embedding. Relational symbol vectors reflect the directionality of operations, and rule symbols are highly correlated with associated operation symbols. During pre-training, the Vec model is used to train multi-dimensional vectors on several switching operation corpora, focusing on optimizing the distributed mapping of symbolic knowledge to ensure that vector associations are consistent with symbol rules. This includes vector constraints between rule symbols and operation symbols, vector constraints between attribute symbols and entity symbols, and vector constraints between reverse operation symbols. Through pre-encoding, each symbol element in the sequence is transformed into a multi-dimensional preliminary distributed vector, forming a pre-encoding matrix, completing the mapping from symbols to the preliminary distribution. At this point, the vectors already possess basic symbolic association semantics.

[0064] This application utilizes the bidirectional temporal modeling capabilities of BiLSTM to integrate the sequential dependencies of symbolic rules in the sequence, such as "operation → rule → state," into distributed vectors. This achieves a dual enhancement of "semantic association + rule constraint," which is the core step in knowledge representation from "preliminary distribution" to "enhanced distribution." BiLSTM, through forward and reverse encoding, imbues vectors with symbolic rules. For each sequence position t, bidirectional information is fused through vector concatenation to generate a multidimensional enhanced vector. Furthermore, core vectors are extracted based on the type of symbolic knowledge, realizing a distributed representation of entity knowledge vectors. Relational knowledge vector and state knowledge vector By incorporating rule constraints into the traditional translation distance model, a scoring function is constructed to implement symbolic rules constraining distributed vectors.

[0065]

[0066] in, For distributed vector constraints, R is a rule knowledge base built based on power safety regulations; It is a norm that measures the consistency of distributed semantics, corresponding to the SPO structure logic of symbolic knowledge; This is a rule penalty item; if the rule knowledge base is violated, then... A value of 0 indicates direct association with the symbolic rule base; The rule weights ensure that symbolic rules take precedence over semantic associations; when a rule conflicts with semantics, the rule penalty term dominates the scoring result. By comparing positive and negative samples, the rule compliance of the vectors is optimized to ensure consistency between the distributed representation and the symbolic rules, and the loss is calculated accordingly.

[0067]

[0068] in, For a set of compliant triples, For the set of illegal triples, The margin parameter helps prevent confusion between compliant and non-compliant triples; the trained multidimensional vectors satisfy the needs of distributed computation while strictly adhering to symbolization rules, achieving a unity of interpretability and computability of knowledge.

[0069] In a preferred embodiment, step S1 includes:

[0070] Z-score standardization preprocessing is performed on real-time power grid operation data to eliminate dimensional differences; the calculation formula for Z-score standardization includes:

[0071]

[0072] In the formula, For the standardized data of the i-th sample on the j-th indicator, Let i be the time sample, and i be the i-th sample. Let be the value of the i-th sample on the j-th indicator;

[0073] It should be noted that the difference in dimensions of different indicators in the power grid will cause PCA to favor indicators with larger values, such as power which is much greater than voltage. In such cases, Z-score standardization is needed to convert all indicators into dimensionless data with a mean of 0 and a standard deviation of 1. For example, the original active power data of a certain line in a certain operating power grid may have an average of 80-120MW. 100MW, standard deviation If the value is 10MW, then the original value of a certain sample is... =110KW can be standardized as: (110-100) / 10=1, that is, the active power value of this line in this sample is one standard deviation higher than the mean, and can be directly compared with the voltage standardized value;

[0074] The linear correlation between power grid indicators is quantified using covariance analysis on the preprocessed real-time power grid operation data; the formula for calculating the covariance matrix is ​​defined as follows:

[0075]

[0076] In the formula, Let the covariance be the sum of the j-th and k-th indicators. If the value is greater than 0, then these two indicators are positively correlated; for example, the larger the line current, the greater the active power. If the value is less than 0, then these two indicators are negatively correlated; for example, as bus voltage increases, reactive power consumption decreases. If the values ​​approach 0, then the two indicators have almost no linear correlation, such as the lack of correlation between transformer temperature and bus frequency. For the i-th sample, the standardized data is based on the k-th indicator;

[0077] It's important to note that the covariance matrix is ​​a core input of PCA, used to measure the degree of linear correlation between different power grid indicators. Many indicators in the power grid exhibit physical correlations, such as a positive correlation between line active power and bus current. The covariance matrix quantifies these correlations, facilitating subsequent principal component extraction. The covariance matrix of power grid data typically exhibits a block diagonalization characteristic, meaning that the covariance within the same type of indicator is large, while the covariance across different types of indicators is small. For example, suppose an operating state block includes indicators such as node voltage, line power flow, and bus frequency, with large covariances among these indicators; a health state block includes indicators such as equipment temperature and dissolved gas content in oil, also with large covariances among these indicators. However, the covariance between this operating state block and the equipment health block will be close to 0, indicating almost no correlation between the operating state and equipment health. This characteristic facilitates subsequent risk identification of corresponding operational or equipment risks.

[0078] Principal components are selected based on the linear correlation among power grid indicators and the 90% cumulative contribution rate criterion; whereby the cumulative contribution rate of the first k principal components is defined as:

[0079]

[0080] In the formula, k represents the first k principal components, and m is the number of monitoring indicators. For the k-th eigenvalue, This is the m-th eigenvalue;

[0081] Principal Component Analysis (PCA) is used to extract core features to generate low-dimensional vector features.

[0082] It should be noted that the core objective of the PCA involved in this application is to retain most of the power grid information with a small number of principal components. The top k principal components need to be selected through the cumulative contribution rate criterion to ensure that the low-dimensional vector is small enough to reduce the amount of computation while covering enough information to retain risk-related features in a relatively fully variable manner.

[0083] As a preferred implementation, after the PCA principal component calculation is completed, the effectiveness of the low-dimensional vector must be verified by the reconstruction error. The high-dimensional data is then inferred from the low-dimensional vector to verify whether the reconstructed data has a smaller error than the original data, thereby determining whether the low-dimensional vector retains the core features. The formula for calculating the reconstruction error is:

[0084]

[0085] In the formula, MSE represents the reconstruction error, and m is the number of monitoring indicators. Let i be the time sample, and i be the i-th sample. Let be the value of the i-th sample on the j-th indicator. This is a low-dimensional vector matrix of power grid data. It is the transpose of the projection matrix. Let j be the standard deviation of the j-th indicator. Let be the mean of the j-th indicator.

[0086] As a preferred implementation, step S2, the process of constructing the power grid topology model includes: using a graph database to store the topological relationships between power grid nodes and edges and a knowledge graph to construct a triplet rule base that supports automatic matching and updating, consisting of principal components, physical quantities and topological elements.

[0087] Among them, the triple rule base uses knowledge graph construction to transform each rule into a triple of entity (principal component)-relation (correspondence)-entity (topological element), which includes the mapping rule of "principal component-physical quantity-topological element", low-dimensional feature anomaly threshold, and topological reasoning priority rules (such as bus association taking precedence over load association).

[0088] It should be noted that the eigenvector coefficients used in PCA dimensionality reduction can clearly identify the corresponding power grid physical quantity for each low-dimensional principal component. These physical quantities are then mapped one-to-one with the nodes and branches in the power grid topology model, forming an association rule base.

[0089] As a preferred implementation, step S3, deriving the risk propagation path along the topological thought chain, includes:

[0090] The state of the edge is dynamically updated according to the switching operation steps;

[0091] Starting with the topological elements corresponding to the abnormal low-dimensional features in the low-dimensional vector features, the risk propagation path is derived according to the priority of bus, adjacent lines, associated transformers, and downstream loads.

[0092] During the switching operation, a low-dimensional vector for the current operation step is calculated in real time according to the operation process. The principal component values ​​are compared with the feature anomaly threshold. If a principal component value exceeds the threshold, the corresponding topological element is located through the association rule base and used as the starting node for topological thinking chain reasoning. If the principal component value in the current low-dimensional vector is greater than the feature anomaly threshold, the branch or node corresponding to this low-dimensional vector may have a safety risk, and the topological chain effect needs to be deduced.

[0093] It should be noted that the power grid topology thinking chain established in this application is essentially based on dynamic structural reasoning using graph theory. It abstracts the power grid as a topological graph of nodes (buses, equipment) and edges (lines, switches). Switching operations (such as switching on / off) change the state of the edges (connected / disconnected), thereby altering the connectivity of the topological graph. The risk transmission path must be derived from the aforementioned anchored starting node, following the topological connection chain.

[0094] Table 1 shows examples of principal component topological inference:

[0095] Low-dimensional principal components Corresponding physical quantity Related topology elements Feature anomaly threshold Principal component 1 Line L1 current, bus B1 power Branch L1, Node B1 >1.5 (Overload anomaly) Principal component 2 Bus B2 voltage, transformer T1 voltage Node B2, branch T1 <-1.2 (low pressure anomaly) Principal component 3 Transformer T2 temperature Branch T2 >2.0 (overheating abnormality)

[0096] Table 1

[0097] Dynamically update the topology graph: Based on the current switching operation steps, such as opening the L1 switch of the line, update the edge status of the power grid topology graph, mark the edge corresponding to line L1 as "disconnected", and generate the "topology graph after operation".

[0098] Constructing a topological reasoning chain: Starting with the initial node, the reasoning chain is built according to the priority of equipment associations, such as busbars, adjacent lines, associated transformers, and downstream loads. Each step of the reasoning needs to verify the likelihood of risk by incorporating low-dimensional features.

[0099] An example of an inference rule based on the starting node B1 is as follows:

[0100] Step 1: Analyze the direct associated branches of node B1—node B1 connects line L1 (which is now disconnected), line L2, and transformer T1; in the low-dimensional vector, "principal component 1 = 1.8 (power flow anomaly)," it is necessary to determine whether the power flow has shifted to L2 after "L1" is disconnected;

[0101] Step 2: Verify the operating characteristics of L2 - By checking whether the standardized value of the associated physical quantity (L2 current) of "principal component 1" in the low-dimensional vector is close to the threshold (e.g., the current standardized value of L2 current = 1.2, which is close to the abnormal threshold of 1.5), it can be determined that "L2 has a risk of power flow overload".

[0102] Step 3: Deducing the chain reaction of L2 overload - L2 is connected to bus B2. If L2 is overloaded, it is necessary to further verify the value of "principal component 2 (bus B2 voltage)" (e.g., the current principal component 2 = -1.1, close to the low voltage threshold of -1.2) to determine that "the B2 bus voltage is at risk of exceeding the limit".

[0103] Step 4: Termination of inference condition - When the inference reaches "downstream load node B3", and its associated low-dimensional principal components (such as principal component 4=0.5, which is far below the abnormal threshold) are not abnormal, the inference chain is terminated, forming a complete risk path: L1 tripping, power flow transfer to L2, L2 overload, and B2 bus voltage exceeding the limit.

[0104] After reasoning, the risk weights of the reasoning chain need to be quantified: For each step of the reasoning chain, such as L2 overload or B2 voltage limit violation, a risk weight is assigned based on the degree of anomaly of the low-dimensional features. For example, principal component 1 = 1.8 corresponds to a weight of 0.8, and principal component 2 = -1.1 corresponds to a weight of 0.6. The higher the weight, the higher the risk priority.

[0105] As a preferred implementation method, the priority is determined by: assigning risk weights based on the degree of anomaly of the low-dimensional vector features, and determining the priority based on the risk weights, with higher weights indicating higher priority.

[0106] In a preferred implementation, step S4, the two-way cross-validation process includes:

[0107] The authenticity of low-dimensional vector feature anomalies is verified by the topological derivation results of the risk propagation path;

[0108] The effectiveness of the topological derivation of the risk propagation path is verified by using low-dimensional vector features.

[0109] Therefore, by verifying the authenticity of low-dimensional vector feature anomalies, it can be determined whether the anomalies are caused by changes in topology. By verifying the validity of the topological derivation results of the risk propagation path, it can be determined whether the reasoning conclusions are supported by data features.

[0110] This application achieves precise transformation from power grid characteristic data to topology risk, ultimately improving the efficiency and accuracy of switching operation risk identification. It avoids misjudgments that can easily occur with single-dimensional judgments. For example, if a low-dimensional vector shows voltage anomalies but the topology has no weak points, it may be due to sensor fluctuations; or if topology inference indicates "overload risk" but the low-dimensional vector shows no anomalies, it may be due to biased inference assumptions or other factors. Two-way cross-validation is needed to ensure the accuracy of risk identification: Topology inference is used to verify low-dimensional features: If a low-dimensional vector shows an anomaly such as "principal component 2 = -1.3, voltage anomaly," the topology thinking chain must be used to verify whether the anomaly is caused by a change in the topology. If topology inference finds that bus B2 only connects L2 and T1, and L2 overload causes voltage drop, then the voltage anomaly is verified as a real risk caused by topology changes; if topology inference finds that bus B2 has no structural changes and the voltage anomaly is only caused by local load fluctuations, then the anomaly is judged to be a temporary disturbance, not an operational risk. Low-dimensional features are used to verify topology inference: If topology inference indicates a risk such as "T1 transformer overload," the low-dimensional vector must be used to verify whether the risk is supported by data features. For example, if "principal component 3 (T1 temperature) = 0.8 (normal)" and "T1 power principal component = 0.6 (normal)" in the low-dimensional vector, then the inference conclusion of "T1 overload" is overturned, eliminating misjudgment; if "principal component 3 = 1.9 (close to the abnormal threshold)" in the low-dimensional vector, then the conclusion of "T1 overload risk" is strengthened, and it is included in the high-priority risk. Combining the risk path and risk weight after cross-validation, the final switching operation risk report is generated, clarifying the risk location, risk level, transmission path, and disposal measures, and associating key evidence of low-dimensional features and topological structure, finally completing the risk identification of the switching operation ticket.

[0111] Example 2

[0112] Please see Figure 2 , Figure 2 This is a schematic structural block diagram of an electronic device provided in an embodiment of this application.

[0113] An electronic device includes a memory 101, a processor 102, and a communication interface 103. The memory 101, processor 102, and communication interface 103 are electrically connected directly or indirectly to enable data transmission or interaction. For example, these components can be electrically connected to each other via one or more communication buses or signal lines. The memory 101 can be used to store software programs and modules. The processor 102 executes the software programs and modules stored in the memory 101 to perform various functional applications and data processing. The communication interface 103 can be used for signaling or data communication with other node devices.

[0114] The memory 101 may be, but is not limited to, random access memory (RAM), read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), etc.

[0115] The processor 102 can be an integrated circuit chip with signal processing capabilities. The processor 102 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0116] It is understood that the structure shown in the figure is for illustrative purposes only. A method for identifying switching operation risks by fusing low-dimensional vectors and power grid topology may include more or fewer components than those shown in the figure, or have a different configuration. The components shown in the figure can be implemented in hardware, software, or a combination thereof.

[0117] In the embodiments provided in this application, it should be understood that the disclosed methods can also be implemented in other ways. The embodiments described above are merely illustrative. For example, the flowcharts or block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and computer program products according to various embodiments of this application. In this regard, each block in the flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0118] In addition, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0119] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0120] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0121] It will be apparent to those skilled in the art that this application is not limited to the details of the exemplary embodiments described above, and that this application can be implemented in other specific forms without departing from the spirit or essential characteristics of this application. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of this application is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within this application. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for identifying switching operation risks by fusing low-dimensional vectors and power grid topology thinking chains, characterized in that, Includes the following steps: S1. Standardize and preprocess the real-time operation data of the power grid, and extract core features through principal component analysis (PCA) to generate low-dimensional vector features. S2. A power grid topology model is constructed by storing the topological relationships between power grid nodes and edges using a graph database. S3. Based on the power grid topology model, the risk propagation path is derived along the topology thinking chain, starting with the topology element corresponding to the abnormal low-dimensional feature in the low-dimensional vector feature and taking the topology element corresponding to the abnormal low-dimensional feature in the low-dimensional vector feature as the starting node. S4. Verify the topological derivation results of low-dimensional vector features and risk propagation paths through bidirectional cross-validation; S5. Based on the results of two-way cross-validation, output a risk report that includes the risk location, level, transmission path, and response measures.

2. The method for identifying switching operation risks by fusing low-dimensional vectors and power grid topology thinking chains as described in claim 1, characterized in that, Step S1 includes: Z-score standardization preprocessing is performed on real-time power grid operation data to eliminate dimensional differences; the calculation formula for Z-score standardization includes: In the formula, For the standardized data of the i-th sample on the j-th indicator, Let i be the time sample, and i be the i-th sample. Let be the value of the i-th sample on the j-th indicator; The linear correlation between power grid indicators is quantified using covariance analysis on the preprocessed real-time power grid operation data; the formula for calculating the covariance matrix is ​​defined as follows: In the formula, Let the covariance be the sum of the j-th and k-th indicators. For the i-th sample, the standardized data is based on the k-th indicator; Principal components are selected based on the linear correlation among power grid indicators and the 90% cumulative contribution rate criterion; whereby the cumulative contribution rate of the first k principal components is defined as: In the formula, k represents the first k principal components, and m is the number of monitoring indicators. For the k-th eigenvalue, This is the m-th eigenvalue; Principal Component Analysis (PCA) is used to extract core features to generate low-dimensional vector features.

3. The method for identifying switching operation risks by fusing low-dimensional vectors and power grid topology thinking chains as described in claim 2, characterized in that, It also includes verifying the effectiveness of low-dimensional vectors through reconstruction error, the formula for which the reconstruction error is calculated is: In the formula, MSE represents the reconstruction error, and m is the number of monitoring indicators. Let i be the time sample, and i be the i-th sample. Let be the value of the i-th sample on the j-th indicator. This is a low-dimensional vector matrix of power grid data. It is the transpose of the projection matrix. Let j be the standard deviation of the j-th indicator. Let be the mean of the j-th indicator.

4. The method for identifying switching operation risks by fusing low-dimensional vectors and power grid topology thinking chains as described in claim 1, characterized in that, In step S2, the process of constructing the power grid topology model includes: using a graph database to store the topological relationships between power grid nodes and edges and a knowledge graph to construct a triplet rule base that supports automatic matching and updating, consisting of principal components, physical quantities and topological elements.

5. The method for identifying switching operation risks by fusing low-dimensional vectors and power grid topology thinking chains as described in claim 1, characterized in that, In step S3, the process of deriving the risk propagation path along the topological thought chain includes: The state of the edge is dynamically updated according to the switching operation steps; Starting with the topological elements corresponding to the abnormal low-dimensional features in the low-dimensional vector features, the risk propagation path is derived according to the priority of bus, adjacent lines, associated transformers, and downstream loads.

6. The method for identifying switching operation risks by fusing low-dimensional vectors and power grid topology thinking chains as described in claim 5, characterized in that, The priority determination method includes: assigning risk weights based on the anomaly degree of low-dimensional vector features, and determining the priority based on the risk weights, with higher weights indicating higher priority.

7. The method for identifying switching operation risks by fusing low-dimensional vectors and power grid topology thinking chains as described in claim 1, characterized in that, In step S4, the bidirectional cross-validation process includes: The authenticity of low-dimensional vector feature anomalies is verified by the topological derivation results of the risk propagation path; The effectiveness of the topological derivation of the risk propagation path is verified by using low-dimensional vector features.

8. An electronic device, characterized in that, include: Memory, used to store one or more programs; processor; When the processor executes the one or more programs, it implements the switching operation risk identification method that integrates low-dimensional vectors and power grid topology thinking chains as described in any one of claims 1-7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the switching operation risk identification method that integrates low-dimensional vectors and power grid topology thinking chains as described in any one of claims 1-7.

Images