Intelligent analysis method of power system and related device

Abstract

The invention discloses an intelligent analysis method of a power system and a related device, and the method comprises the steps: obtaining power grid multi-modal data of the power system, the power grid multi-modal data comprises power grid topological data, power grid static parameters and real-time measurement and power grid time sequence data; performing data feature extraction on the power grid multi-modal data to obtain topological representation, equipment measurement representation and time sequence dynamic representation, and performing preliminary fusion on the topological representation and the equipment measurement representation to obtain local fusion representation; fusing the time sequence dynamic representation and the local fusion representation to obtain a multi-dimensional feature, and applying a physical constraint to the multi-dimensional feature to obtain a final representation; and executing the final representation to obtain an analysis result of the electric power system, and the method and the related device can improve the analysis accuracy of the electric power system.

Description

Technical Field

[0001] This invention belongs to the field of power systems and relates to an intelligent analysis method and related devices for power systems. Background Technology

[0002] Currently, power system-level models rely heavily on modular, statically assumed models and offline integration. Power flow and state estimation typically require precise topology and parameters, making them ill-suited for frequent switching and equipment aging. Multi-source measurements suffer from gaps, asynchronicity, and heterogeneous dimensions, often relying on heuristic weighting or simple summarization, making it difficult to express device-level and network-level information in the same domain. This leads to problems with noise, low coverage, and N... 1. Insufficient robustness under perturbations and easy violation of feasible region constraints. Artificial intelligence generally suffers from limitations of "single-modal or spliced ​​fusion," ignoring structural priors and differences in measurement quality. Assemblies of measurements do not satisfy permutation invariance, and time-series modeling is insufficient for long-term dependencies and responses to sudden events. End-to-end training lacks physical consistency constraints and uncertainty characterization, easily generating infeasible solutions with weak interpretability. The model's generalization ability to scene transfer and distributed off-conditions is limited. In summary, existing technologies cannot meet the accuracy requirements for intelligent analysis of power systems. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the prior art and provide an intelligent analysis method and related apparatus for power systems, which can improve the accuracy of power system analysis.

[0004] To achieve the above objectives, this invention discloses an intelligent analysis method for power systems, comprising: Acquire multimodal data of the power grid of the power system, including power grid topology data, power grid static parameters and real-time measurements, and power grid time series data; Data features are extracted from the multimodal data of the power grid to obtain topology representation, equipment measurement representation, and time-series dynamic representation. The topology representation and equipment measurement representation are initially fused to obtain a local fusion representation. The temporal dynamic representation and the local fusion representation are fused to obtain multidimensional features. Physical constraints are then applied to the multidimensional features to obtain the final representation. By executing the final representation, the analysis results of the power system are obtained.

[0005] A further improvement to the intelligent analysis method for power systems described in this invention is as follows: Furthermore, before extracting data features from the multimodal data of the power grid, the method further includes: The multimodal data of the power grid is subjected to data cleaning, missing value imputation, outlier handling, resampling, slicing and alignment.

[0006] Furthermore, the process of extracting data features from the multimodal data of the power grid to obtain topological representation, equipment measurement representation, and time-series dynamic representation is as follows: The topology representation is obtained by extracting data features from the power grid topology data using a topology encoder. The static parameters and real-time measurements of the power grid are extracted by the equipment measurement encoder to obtain the equipment measurement characterization. The time-series dynamic representation is obtained by extracting data features from the power grid time-series data using a time-series encoder.

[0007] Furthermore, the process of initially fusing the topological representation and the device measurement representation to obtain a locally fused representation is as follows: The topological representation and the device measurement representation are bilinearly interactively fused to obtain a fusion result. The fusion result is then processed through a gating mechanism to obtain the local fusion representation.

[0008] Furthermore, the process of fusing the temporal dynamic representation and the local fusion representation to obtain multidimensional features is as follows: The temporal dynamic representation and the local fusion representation are fused through a bidirectional cross-attention mechanism to obtain multidimensional features.

[0009] Furthermore, the process of executing the final representation to obtain the analysis results of the power system is as follows: The final representation is executed through the task header to obtain the analysis results of the power system.

[0010] This invention discloses an intelligent analysis system for power systems, comprising: The acquisition module is used to acquire multimodal data of the power grid of the power system, including power grid topology data, power grid static parameters and real-time measurements, and power grid time series data. The preliminary fusion module is used to extract data features from the multimodal data of the power grid to obtain topology representation, equipment measurement representation and time-series dynamic representation, and to perform preliminary fusion of the topology representation and equipment measurement representation to obtain local fusion representation. The fusion module is used to fuse the temporal dynamic representation and the local fusion representation to obtain multi-dimensional features, and apply physical constraints to the multi-dimensional features to obtain the final representation; The execution module is used to execute the final representation to obtain the analysis results of the power system.

[0011] A further improvement to the intelligent analysis system for the power system described in this invention is as follows: Furthermore, it also includes: The preprocessing module is used to perform data cleaning, missing value imputation, outlier handling, resampling, slicing, and alignment on the power grid multimodal data.

[0012] Furthermore, the process of extracting data features from the multimodal data of the power grid to obtain topological representation, equipment measurement representation, and time-series dynamic representation is as follows: The topology representation is obtained by extracting data features from the power grid topology data using a topology encoder. The static parameters and real-time measurements of the power grid are extracted by the equipment measurement encoder to obtain the equipment measurement characterization. The time-series dynamic representation is obtained by extracting data features from the power grid time-series data using a time-series encoder.

[0013] Furthermore, the process of initially fusing the topological representation and the device measurement representation to obtain a locally fused representation is as follows: The topological representation and the device measurement representation are bilinearly interactively fused to obtain a fusion result. The fusion result is then processed through a gating mechanism to obtain the local fusion representation.

[0014] Furthermore, the process of fusing the temporal dynamic representation and the local fusion representation to obtain multidimensional features is as follows: The temporal dynamic representation and the local fusion representation are fused through a bidirectional cross-attention mechanism to obtain multidimensional features.

[0015] Furthermore, the process of executing the final representation to obtain the analysis results of the power system is as follows: The final representation is executed through the task header to obtain the analysis results of the power system.

[0016] This invention discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the intelligent analysis method for the power system.

[0017] The present invention discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the intelligent analysis method for the power system.

[0018] The present invention has the following beneficial effects: In practical operation, the intelligent analysis method and related devices for power systems described in this invention acquire multimodal data of the power grid, including grid topology data, grid static parameters and real-time measurements, and grid time series data. Data features are extracted from the multimodal data to obtain topology representation, equipment measurement representation, and time-series dynamic representation. The topology representation and equipment measurement representation are initially fused to obtain a local fusion representation. The time-series dynamic representation and the local fusion representation are then fused to obtain multidimensional features, avoiding the limitations of single-modality or simple splicing fusion. This also considers the structure and both sides, ensuring permutation invariance and meeting the requirements of time-series modeling for long-term dependencies and responses to sudden events. Physical constraints are then applied to the multidimensional features to obtain the final representation, which improves the model's generalization ability to scene migration and distributed external operating conditions to a limited extent. The final representation is then executed to obtain the analysis results of the power system, thereby improving the accuracy of power system analysis. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a system structure diagram of the present invention. Detailed Implementation

[0021] 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 some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] In the description of this invention, it should be understood that the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0023] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0024] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Additionally, the character " / " in this invention generally indicates that the preceding and following objects have an "or" relationship.

[0025] It should be understood that although terms such as first, second, third, etc., may be used in the embodiments of the present invention to describe the preset range, these preset ranges should not be limited to these terms. These terms are only used to distinguish the preset ranges from one another. For example, without departing from the scope of the embodiments of the present invention, the first preset range may also be referred to as the second preset range, and similarly, the second preset range may also be referred to as the first preset range.

[0026] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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 some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0028] The accompanying drawings illustrate various structural schematic diagrams according to embodiments disclosed in this invention. These drawings are not to scale, and some details have been enlarged for clarity, and some details may have been omitted. The shapes of the various regions and layers shown in the drawings, as well as their relative sizes and positional relationships, are merely exemplary and may deviate from reality due to manufacturing tolerances or technical limitations. Furthermore, those skilled in the art can design regions / layers with different shapes, sizes, and relative positions as needed.

[0029] Example 1 refer to Figure 1 The intelligent analysis method for power systems described in this invention includes the following steps: 1) Acquire multimodal power grid data of the power system, preprocess the multimodal power grid data, and obtain preprocessed multimodal power grid data; The power grid multimodal data includes power grid topology data, power grid static parameters and real-time measurements, and power grid time series data. The measurement data includes at least one of the second-level / minute-level measurement data provided by the SCADA system and the millisecond-level synchronous phasor data provided by the PMU system. The power grid topology data includes at least one of the following: power grid node connection relationships, circuit breaker / disconnector status, and equipment parameters.

[0030] The preprocessing process for the multimodal power grid data is as follows: data cleaning, missing value imputation, and outlier handling are performed on the multimodal power grid data. Then, data with different sampling frequencies are resampled and sliced ​​according to a preset time window. Spatial and version alignment is then performed to ensure that the power grid topology data, static parameters, real-time measurements, and time series data at the same time are semantically consistent. Finally, a standardized data batch is output.

[0031] 2) Extract data features from the preprocessed power grid multimodal data using a three-channel encoder to obtain topology representation, equipment measurement representation, and time-series dynamic representation. Perform preliminary fusion of the topology representation and equipment measurement representation to obtain a local fusion representation. The three-channel encoder is a topology encoder, a backup measurement encoder, and a time-series encoder. The specific process of step 2) is as follows: The topology representation is obtained by extracting data features from the preprocessed power grid topology data through a topology encoder. The device measurement characterization is obtained by extracting the characteristics of the preprocessed static parameters and real-time data of the power grid using the device measurement encoder. By extracting data features from the preprocessed power grid time series data using a time encoder, a time-series dynamic representation is obtained.

[0032] The topology representation and the device measurement representation are bilinearly interactively fused to obtain a fusion result. The fusion result is then processed through a gating mechanism to form the local fusion representation. The gating mechanism dynamically adjusts the intensity of each information path according to the current operating scenario and data quality, and finally forms a local fusion representation aligned to each power grid node. The local fusion representation serves as a candidate vector for query, key, and value in the subsequent attention fusion mechanism.

[0033] It should be noted that the topology encoder uses a graph structure as its carrier and utilizes a graph neural network (GNN) to aggregate information such as electrical distance, impedance, and operating status at nodes and edges to obtain a structural representation with global and local power grid semantics. The equipment measurement encoder uses a long short-term memory (LSTM) network to perform permutation-invariant modeling on the disordered measurement set of each device. It captures the correlation between channels by induced point aggregation and suppresses noise and missing data by projection of the physical feasible region and confidence gating based on q, outputting a robust measurement representation. The time-series encoder introduces a multi-head self-attention Transformer, which learns dynamic evolution and perturbation response with the help of relative position and exogenous variable embedding to form a time-series representation.

[0034] 3) The temporal dynamic representation and the local fusion representation are fused through a bidirectional cross-attention mechanism to obtain multidimensional features. Physical constraints are applied to the multidimensional features to obtain the final representation. First, bidirectional cross-attention computation is performed. One path uses temporal dynamic representation as the query and local fusion representation as the key and value to capture the dependence of dynamic processes on structure and parameters. The other path uses local fusion representation as the query and temporal dynamic representation as the key and value to mine temporal patterns under structural parameter constraints. Through this bidirectional interaction, cross-modal dependencies are modeled across the entire network to generate multidimensional features.

[0035] Then, the multidimensional features are regularized and corrected by incorporating physical consistency and feasible region projection. For example, power balance and voltage within a reasonable range are ensured. The final output is used for downstream tasks of different granularities, such as nodes, branches, and the entire network.

[0036] 4) Execute the final representation through the task header to obtain the analysis results of the power system.

[0037] Specifically, the analysis results of the power system include at least one of power system state estimation, power flow calculation, and source-load prediction. The task header also carries the deployment configuration of the model and can be customized according to actual application scenarios, thereby achieving an end-to-end engineering implementation closed loop from multimodal data input to specific task result output.

[0038] Step 4) involves inputting the final representation into the corresponding task-specific header network according to task requirements. For example, for a state estimation task, the task header maps the final representation to the optimal estimates of node voltage magnitude and phase angle; for a power flow calculation task, the task header maps the final representation to branch power flow distribution, offering greater flexibility and scalability.

[0039] This invention enhances the model's ability to extract electrical, temporal, and topological features through three encoders. The extracted features are highly representative and discriminative, and possess a certain degree of generalization ability. Furthermore, feature fusion unifies features from different dimensions into a single vector, enabling the trained model to correctly and accurately learn the fused electrical, temporal, and topological features. This accelerates the model's learning speed, allowing it to achieve the same results with fewer training epochs and less computational resources.

[0040] This invention features the following characteristics: More reliable features. Structural, measurement, and temporal information are not simply pieced together, but mutually constrained within the same coordinate system, coupled with physical verification, reducing situations where numerical values ​​appear correct but are impractical in engineering. More efficient training and deployment. Alignment and reliability management reduce invalid gradients, node-level fusion shortens the cross-modal learning path, and the model converges faster; during online runtime, only affected local updates are needed, keeping resource consumption controllable. Directly deployable. The process is consistent with existing network data formats, requiring no modifications to monitoring and scheduling systems for integration; a single fused feature set can support multiple downstream tasks, reducing maintenance costs.

[0041] Example 2 The specific operation process of this embodiment is as follows: 1) Acquire multimodal power grid data of the power system, preprocess the multimodal power grid data, and obtain preprocessed multimodal power grid data; Assume the power grid is a weighted directed graph. With N nodes and M branches, a unified computing platform acquires multi-source data from SCADA (Supervisory Control and Data Acquisition) systems, PMU (Phasor Measurement Unit), and operational records. The point set... Corresponding to busbars, aggregation points, transformer substations, etc., edge sets This corresponds to primary equipment such as lines, transformers, and switches.

[0042] In the data preprocessing stage, the first steps are to perform dimensional unification, anomaly suppression, and missing data completion. For continuous quantities, robust scaling or a log-normalization hybrid transformation is applied. For missing segments, time-domain interpolation is performed to generate quality scores q∈[0,1] to characterize reliability. Taking a single channel x as an example, robust scaling can be implemented using the following formula:

[0043] in, For robust scaling, This indicates that it is used for robust estimation of central tendency and dispersion.

[0044] Subsequently, spatiotemporal alignment of the multimodal sequences was performed, mapping different sampling frequency sequences to a unified sliding window {t}. L+1,…,t}, and establish an index relationship between devices, nodes, and branches, mapping device IDs to topology nodes or edge terminals; when one-to-many or many-to-one mappings occur, learnable weights are used to merge and synchronously output the alignment quality score qv at the node / device level.

[0045] 2) Data features are extracted from the preprocessed power grid multimodal data using a three-channel encoder to obtain topology representation, equipment measurement representation, and time-series dynamic representation. The topology representation and equipment measurement representation are then initially fused to obtain a local fusion representation. The power grid topology encoder takes the graph structure and its node / edge attributes as input, and uses a message-passing network with residuals to obtain the structural representation. Its core update is written as:

[0046] in, For the electrical field, Represents the node structure; For domain messages; Indicate the characteristics of the branch; Represents the graph update function, The number of convolutional layers in the graph.

[0047] The device measurement encoder adopts a two-stage structure of "LSTM + permutation-invariant aggregation": First, for each measurement channel k on each device (or node), the sequence within the most recent window is taken. After passing through LSTM, the channel time sequence summary is obtained as follows:

[0048] in, For nodes / devices The Each measuring point at time... The sampled values, For nodes / devices The Each measuring point at time... Window summary, For memory units.

[0049] Secondly, treating the summary sets of different channels as unordered sets, we introduce attention-based aggregation (PMA) with induced points to achieve permutation-invariant modeling of channel arrangement, resulting in the device measurement aggregation representation as follows:

[0050] in, For the number of channels, For nodes / devices The A window summary of each measurement point.

[0051] To inject the feasible domain prior, ranges are set for key derived quantities such as temperature rise, current carrying capacity, and power / reactive power. Let differentiability penalty be applied and one-step approximate projection be performed to obtain the physically consistent measurement embedding as follows:

[0052] in, To be The device-level synthesized vector is obtained by performing symmetric aggregation (sum / mean / attention) on the output of the single-channel LSTM. This is the inferred value of the physical quantity. This is the gradient projection step size; This indicates penalties for exceeding boundaries and projection.

[0053] By combining alignment quality scores to construct confidence gating, we can obtain the device measurement characterization. for:

[0054] in, To measure the reliability gate, For robust measurement characterization.

[0055] The temporal encoder takes multi-channel time window data from each node as input and uses a multi-head self-attention Transformer to learn long-range dependencies and perturbation responses. The standard attention calculation is as follows:

[0056] in, This represents the attention triplet. Indicates the scaling dimension.

[0057] Through multi-head parallel processing, feedforward processing, and residual normalization, the time series representation is obtained. At the node scale, to achieve deep coupling between structural priors and measurement evidence, a bilinear interaction and gating adjustment are constructed, resulting in a structure-measurement fusion representation. for:

[0058] in, For nodes The topology-side and measurement-side features, For bilinear interaction weights, For fusion and gating weights.

[0059] 3) The temporal dynamic representation and the local fusion representation are fused through a bidirectional cross-attention mechanism to obtain multidimensional features. Physical constraints are applied to the multidimensional features to obtain the final representation. To enable bidirectional constraints between static conditions and dynamic context, in and Constructing bidirectional cross-attention: using time series as the query and structure-measurement as the key to obtain conditional time series representation. Using structure-measurement as the query and time series as the key, a conditional structure representation is obtained. for:

[0060] in, Represents the cross-modal projection matrix; This indicates a conditional update.

[0061] The two results are combined into a multidimensional feature after residual and layer normalization:

[0062] in, Represents the linear integration weights. Representation layer normalization, This indicates the final fusion feature.

[0063] Regarding physical consistency and feasible region constraints, the residuals of the AC power flow equations are used as the core of soft constraints. For any node i, active / reactive residuals are defined and physical losses are formed:

[0064]

[0065] in, Indicates voltage amplitude. Indicates the phase angle difference. Represents the admittance parameter. The active and reactive power injected into the nodes. The difference between the left and right sides of the AC power flow equation. This represents the total residual loss.

[0066] 4) Execute the final representation through the task header to obtain the analysis results of the power system.

[0067] During the training phase, a multi-task joint objective is adopted, simultaneously optimizing the supervision loss of task heads such as state estimation, power flow calculation, and source load prediction along with physical / boundary / consistency constraints. The objective function is:

[0068] in, For the task The supervised loss (MSE / Huber) and its weight in the total loss, Indicates the downstream task set. This represents the penalty coefficient among power flow balance, operational boundary, and observation consistency. Optimization can be achieved using AdamW, learning rate preheating, and cosine annealing; during the inference phase, incremental forward pass is performed along the graph flow, GNN only updates the k-hop neighborhood, LSTM only scrolls within the new window, and the aggregation / attention layer remains online and scalable.

[0069] Example 3 The intelligent analysis system for power systems described in this invention includes: The acquisition module is used to acquire multimodal data of the power grid of the power system, including power grid topology data, power grid static parameters and real-time measurements, and power grid time series data. The preliminary fusion module is used to extract data features from the multimodal data of the power grid to obtain topology representation, equipment measurement representation and time-series dynamic representation, and to perform preliminary fusion of the topology representation and equipment measurement representation to obtain local fusion representation. The fusion module is used to fuse the temporal dynamic representation and the local fusion representation to obtain multi-dimensional features, and apply physical constraints to the multi-dimensional features to obtain the final representation; The execution module is used to execute the final representation to obtain the analysis results of the power system.

[0070] This embodiment also includes: The preprocessing module is used to perform data cleaning, missing value imputation, outlier handling, resampling, slicing, and alignment on the power grid multimodal data.

[0071] In this embodiment, the process of extracting data features from the multimodal data of the power grid to obtain topological representation, equipment measurement representation, and time-series dynamic representation is as follows: The topology representation is obtained by extracting data features from the power grid topology data using a topology encoder. The static parameters and real-time measurements of the power grid are extracted by the equipment measurement encoder to obtain the equipment measurement characterization. The time-series dynamic representation is obtained by extracting data features from the power grid time-series data using a time-series encoder.

[0072] In this embodiment, the process of initially fusing the topological representation and the device measurement representation to obtain a locally fused representation is as follows: The topological representation and the device measurement representation are bilinearly interactively fused to obtain a fusion result. The fusion result is then processed through a gating mechanism to obtain the local fusion representation.

[0073] In this embodiment, the process of fusing the temporal dynamic representation and the local fusion representation to obtain multidimensional features is as follows: The temporal dynamic representation and the local fusion representation are fused through a bidirectional cross-attention mechanism to obtain multidimensional features.

[0074] In this embodiment, the process of executing the final representation to obtain the analysis results of the power system is as follows: The final representation is executed through the task header to obtain the analysis results of the power system.

[0075] The module division in this embodiment is illustrative and represents only one logical functional division. In actual implementation, other division methods may be used. Furthermore, the functional modules in each embodiment of this application can be integrated into a single processor, exist as separate physical entities, or be integrated into a single module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0076] Example 4 A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of an intelligent analysis method for a power system. For example, the method includes: acquiring multimodal power grid data of the power system, including power grid topology data, power grid static parameters and real-time measurements, and power grid time series data; extracting data features from the multimodal power grid data to obtain topology representations, equipment measurement representations, and time-series dynamic representations; performing preliminary fusion of the topology representations and equipment measurement representations to obtain a local fusion representation; fusing the time-series dynamic representations and the local fusion representations to obtain multidimensional features; applying physical constraints to the multidimensional features to obtain a final representation; and executing the final representation to obtain the analysis results of the power system. The memory may include main memory, such as high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device. The processor, network interface, and memory are interconnected via an internal bus, which may be an industry-standard architecture bus, a peripheral component interconnection standard bus, an extended industry-standard architecture bus, etc. The bus may be classified as an address bus, a data bus, a control bus, etc. The memory is used to store programs; specifically, the program may include program code, which includes computer operation instructions. The memory may include main memory and non-volatile memory, and provides instructions and data to the processor.

[0077] Example 5 A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of an intelligent analysis method for a power system. For example, the method includes: acquiring multimodal power grid data of the power system, the multimodal data including power grid topology data, power grid static parameters and real-time measurements, and power grid time series data; extracting data features from the multimodal data to obtain topology representations, equipment measurement representations, and time-series dynamic representations; performing preliminary fusion of the topology representations and equipment measurement representations to obtain local fusion representations; fusing the time-series dynamic representations and local fusion representations to obtain multidimensional features; applying physical constraints to the multidimensional features to obtain a final representation; and executing the final representation to obtain the analysis results of the power system. Specifically, the computer-readable storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. The volatile memory may include random access memory (RAM) and / or cache memory, etc. The non-volatile memory may include read-only memory (ROM), hard disk, flash memory, optical disk, magnetic disk, etc.

[0078] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0079] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0080] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0081] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0082] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and disclosure of the invention. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the following claims.

[0083] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

[0084] The above description is merely a preferred embodiment of the present invention and does not constitute any limitation on the present invention. Any simple modifications, alterations, or equivalent structural changes made to the above embodiments based on the technical essence of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A smart analysis method for power systems, characterized in that, include: Acquire multimodal data of the power grid of the power system, including power grid topology data, power grid static parameters and real-time measurements, and power grid time series data; Data features are extracted from the multimodal data of the power grid to obtain topology representation, equipment measurement representation, and time-series dynamic representation. The topology representation and equipment measurement representation are initially fused to obtain a local fusion representation. The temporal dynamic representation and the local fusion representation are fused to obtain multidimensional features. Physical constraints are then applied to the multidimensional features to obtain the final representation. By executing the final representation, the analysis results of the power system are obtained.

2. The intelligent analysis method for power systems according to claim 1, characterized in that, Before extracting data features from the multimodal data of the power grid, the following steps are also included: The multimodal data of the power grid is subjected to data cleaning, missing value imputation, outlier handling, resampling, slicing and alignment.

3. The intelligent analysis method for power systems according to claim 1, characterized in that, The process of extracting data features from the multimodal data of the power grid to obtain topological representation, equipment measurement representation, and time-series dynamic representation is as follows: The topology representation is obtained by extracting data features from the power grid topology data using a topology encoder. The static parameters and real-time measurements of the power grid are extracted by the equipment measurement encoder to obtain the equipment measurement characterization. The time-series dynamic representation is obtained by extracting data features from the power grid time-series data using a time-series encoder.

4. The intelligent analysis method for power systems according to claim 1, characterized in that, The process of initially fusing the topological representation and the device measurement representation to obtain the local fused representation is as follows: The topological representation and the device measurement representation are bilinearly interactively fused to obtain a fusion result. The fusion result is then processed through a gating mechanism to obtain the local fusion representation.

5. The intelligent analysis method for power systems according to claim 1, characterized in that, The process of fusing the temporal dynamic representation and the local fusion representation to obtain multidimensional features is as follows: The temporal dynamic representation and the local fusion representation are fused through a bidirectional cross-attention mechanism to obtain multidimensional features.

6. The intelligent analysis method for power systems according to claim 1, characterized in that, The process of executing the final representation to obtain the analysis results of the power system is as follows: The final representation is executed through the task header to obtain the analysis results of the power system.

7. An intelligent analysis system for a power system, characterized in that, include: The acquisition module is used to acquire multimodal data of the power grid of the power system, including power grid topology data, power grid static parameters and real-time measurements, and power grid time series data. The preliminary fusion module is used to extract data features from the multimodal data of the power grid to obtain topology representation, equipment measurement representation and time-series dynamic representation, and to perform preliminary fusion of the topology representation and equipment measurement representation to obtain local fusion representation. The fusion module is used to fuse the temporal dynamic representation and the local fusion representation to obtain multi-dimensional features, and apply physical constraints to the multi-dimensional features to obtain the final representation; The execution module is used to execute the final representation to obtain the analysis results of the power system.

8. The intelligent analysis system for power systems according to claim 7, characterized in that, Also includes: The preprocessing module is used to perform data cleaning, missing value imputation, outlier handling, resampling, slicing, and alignment on the power grid multimodal data.

9. The intelligent analysis system for power systems according to claim 7, characterized in that, The process of extracting data features from the multimodal data of the power grid to obtain topological representation, equipment measurement representation, and time-series dynamic representation is as follows: The topology representation is obtained by extracting data features from the power grid topology data using a topology encoder. The static parameters and real-time measurements of the power grid are extracted by the equipment measurement encoder to obtain the equipment measurement characterization. The time-series dynamic representation is obtained by extracting data features from the power grid time-series data using a time-series encoder.

10. The intelligent analysis system for power systems according to claim 7, characterized in that, The process of initially fusing the topological representation and the device measurement representation to obtain the local fused representation is as follows: The topological representation and the device measurement representation are bilinearly interactively fused to obtain a fusion result. The fusion result is then processed through a gating mechanism to obtain the local fusion representation.

11. The intelligent analysis system for power systems according to claim 7, characterized in that, The process of fusing the temporal dynamic representation and the local fusion representation to obtain multidimensional features is as follows: The temporal dynamic representation and the local fusion representation are fused through a bidirectional cross-attention mechanism to obtain multidimensional features.

12. The intelligent analysis system for power systems according to claim 7, characterized in that, The process of executing the final representation to obtain the analysis results of the power system is as follows: The final representation is executed through the task header to obtain the analysis results of the power system.

13. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent analysis method for the power system as described in any one of claims 1-6.

14. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the intelligent analysis method for the power system as described in any one of claims 1-6.

Images