Power system situation awareness method and system based on multi-modal conjoint analysis

Abstract

The invention relates to the technical field of electric power perception, and particularly discloses an electric power system situation perception method and system based on multi-mode conjoint analysis, and the method employs a data processing technology based on deep learning to carry out the statistical analysis of various situation indexes of a power generation side, a power grid side and a demand side of a power distribution network. Semantic embedded feature representations of various situation indexes of a power generation side, a power grid side and a demand side are extracted, and importance evaluation and global feature dynamic aggregation are carried out on data items of the situation indexes to capture key situation features in operation of the power distribution network, so that comprehensive perception of the operation situation of the power distribution network is realized, and the power distribution network performance is improved. Therefore, whether the power grid operates at low carbon is intelligently identified. Through the mode, efficient power distribution network operation situation evaluation can be realized, the intelligent level of power grid operation is effectively improved, and powerful technical support is provided for stable operation and low-carbon development of a power grid.

Description

Technical Field

[0001] The present application relates to the field of power sensing technology, and more specifically, to a power system situation awareness method and system based on multi-modal joint analysis. Background Art

[0002] With the increasing access to renewable energy such as wind and solar energy, the power structure has become more diversified. On the other hand, the application of smart grid technology has promoted the improvement of the informationization and automation level of the power system. However, these have also brought new challenges to the stable operation of the power system. For example, the randomness and intermittency of distributed power sources increase the complexity of system scheduling, and the load characteristics of the user end are also constantly changing, which puts higher requirements on the real-time monitoring and rapid response capabilities of the power grid.

[0003] At present, the operation status monitoring of distribution networks mainly relies on the traditional SCADA (Supervisory Control And Data Acquisition) system. Although the system can provide certain real-time data support, it has obvious deficiencies in data processing capabilities and intelligence. For example, the SCADA system often focuses on single-dimensional data analysis and lacks the ability to comprehensively process multi-source heterogeneous data, making it difficult to accurately grasp the overall operation status of the power grid. In addition, most of the existing situation awareness technologies are based on rule matching or simple statistical models, which have poor adaptability to the complex and changeable power system environment and cannot provide accurate situation assessment.

[0004] Therefore, a power system situation awareness method based on multimodal joint analysis is expected. Summary of the invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a method and system for power system situation awareness based on multimodal joint analysis, which uses data processing technology based on deep learning to perform statistical analysis on various situation indicators on the power generation side, power grid side and demand side of the distribution network, extracts the semantic embedded feature representation of various situation indicators on the power generation side, power grid side and demand side, and through importance evaluation and global feature dynamic aggregation of each situation indicator data item, to capture the key situation characteristics in the operation of the distribution network, to achieve a comprehensive perception of the operation situation of the distribution network, so as to intelligently identify whether the power grid is operating in a low-carbon manner. In this way, efficient distribution network operation situation evaluation can be achieved, the intelligent level of power grid operation can be effectively improved, and strong technical support can be provided for the stable operation and low-carbon development of the power grid.

[0006] Accordingly, according to one aspect of the present application, a method for power system situation awareness based on multimodal joint analysis is provided, which includes: Obtain statistical data sets on the power generation side, statistical data sets on the power grid side, and statistical data sets on the demand side; Embedding and encoding each item of data in the statistical data set of the power generation side, each item of data in the statistical data set of the power grid side, and each item of data in the statistical data set of the demand side to obtain a set of embedded coding vectors for data items on the power generation side, a set of embedded coding vectors for data items on the power grid side, and a set of embedded coding vectors for data items on the demand side; Performing importance modulation on the set of embedded coding vectors of the power grid data items, the set of embedded coding vectors of the power grid data items, and the set of embedded coding vectors of the demand side data items, respectively, to obtain a set of weighted embedded coding vectors of the power grid data items, a set of weighted embedded coding vectors of the demand side data items, and a set of weighted embedded coding vectors of the demand side data items; The set of weighted generation-side data item embedding coding vectors, the set of weighted grid-side data item embedding coding vectors and the set of weighted demand-side data item embedding coding vectors are combined into a set of distribution network data item embedding coding vectors, and the set of distribution network data item embedding coding vectors is dynamically aggregated to obtain a significant aggregate representation vector of all distribution network parameters; Based on the significant aggregation representation vector of all parameters of the distribution network, it is determined whether the power grid is operating in a low-carbon manner.

[0007] In the above-mentioned power system situation awareness method based on multimodal joint analysis, the set of distribution network data item embedded coding vectors is dynamically aggregated, including: determining the initial center vector of the distribution network data item clustering based on the characteristic energy distribution of the set of distribution network data item embedded coding vectors; based on the spatial span of each distribution network data item embedded coding vector in the set of distribution network data item embedded coding vectors relative to the initial center vector of the distribution network data item clustering, performing significance modulation dynamic aggregation on the set of distribution network data item embedded coding vectors to obtain the significant aggregation representation vector of all parameters of the distribution network.

[0008] In the above-mentioned power system situation awareness method based on multimodal joint analysis, the statistical data set on the power generation side includes renewable energy power generation, traditional energy power generation, carbon emissions and power generation efficiency; the statistical data set on the grid side includes transmission and distribution losses, network loss rate, power supply reliability index, voltage quality, network topology and power flow distribution; the statistical data set on the demand side includes power demand forecast value, user power consumption behavior data, distributed energy access data, and energy storage facility status.

[0009] In the above-mentioned power system situation awareness method based on multimodal joint analysis, the set of embedded coding vectors of the power generation side data items, the set of embedded coding vectors of the power grid side data items and the set of embedded coding vectors of the demand side data items are respectively importance modulated to obtain the set of weighted embedded coding vectors of the power generation side data items, the set of weighted embedded coding vectors of the power grid side data items and the set of weighted embedded coding vectors of the demand side data items, including: using fuzzy hierarchical analysis method to perform weight calculation on the set of embedded coding vectors of the power generation side data items, the set of embedded coding vectors of the power grid side data items and the set of embedded coding vectors of the demand side data items to obtain the set of weighted embedded coding vectors of the power generation side data items, the set of weighted embedded coding vectors of the power grid side data items and the set of weighted embedded coding vectors of the demand side data items.

[0010] In the above-mentioned power system situation awareness method based on multimodal joint analysis, the initial center vector of the distribution network data item clustering is determined based on the characteristic energy distribution of the set of distribution network data item embedded coding vectors, including: calculating the static energy factor of each distribution network data item embedded coding vector in the set of distribution network data item embedded coding vectors to obtain a set of distribution network data item static energy factors; selecting the distribution network data item embedded coding vector corresponding to the maximum value in the set of distribution network data item static energy factors as the initial center vector of the distribution network data item clustering.

[0011] In the above-mentioned power system situation awareness method based on multimodal joint analysis, the static energy factor of each distribution network data item embedded coding vector in the set of distribution network data item embedded coding vectors is calculated to obtain a set of distribution network data item static energy factors, including: calculating the kurtosis of the distribution network data item embedded coding vector, and inputting the kurtosis into a sigmoid activation function to obtain the distribution network data item static energy factor.

[0012] In the above-mentioned power system situation awareness method based on multimodal joint analysis, based on the spatial span of each distribution network data item embedded coding vector in the set of distribution network data item embedded coding vectors relative to the initial center vector of the distribution network data item clustering, the set of distribution network data item embedded coding vectors is significantly modulated and dynamically aggregated to obtain the significant aggregation representation vector of all parameters of the distribution network, including: based on the spatial span between each distribution network data item embedded coding vector in the set of distribution network data item embedded coding vectors and the initial center vector of the distribution network data item clustering, and the set of static energy factors of the distribution network data items, the dynamic aggregation energy factor of each distribution network data item embedded coding vector in the set of distribution network data item embedded coding vectors is calculated to obtain a set of dynamic aggregation energy factors of distribution network data items; based on the set of dynamic aggregation energy factors of the distribution network data items, the set of distribution network data item embedded coding vectors is significantly modulated and aggregated to obtain the significant aggregation representation vector of all parameters of the distribution network.

[0013] In the above-mentioned power system situation awareness method based on multimodal joint analysis, the dynamic aggregation energy factor of each distribution network data item embedded coding vector in the set of distribution network data item embedded coding vectors is calculated to obtain a set of dynamic aggregation energy factors of distribution network data items, including: taking the square value of the number of eigenvectors between the distribution network data item embedded coding vector and the initial center vector of the distribution network data item cluster as the spatial span coefficient, and calculating the weighted ratio between the product of the static energy factor of the distribution network data item embedded coding vector and the static energy factor of the initial center vector of the distribution network data item cluster and the spatial span coefficient to obtain the dynamic aggregation energy factor of the distribution network data item.

[0014] In the above-mentioned power system situation awareness method based on multimodal joint analysis, based on the set of dynamic aggregation energy factors of the distribution network data items, the set of embedded coding vectors of the distribution network data items is significantly modulated and aggregated to obtain the significant aggregation representation vector of all parameters of the distribution network, including: inputting the set of dynamic aggregation energy factors of the distribution network data items into a gated mask unit to obtain a set of dynamic aggregation weight factors of the distribution network data items; and calculating the weighted sum of the set of embedded coding vectors of the distribution network data items based on the set of dynamic aggregation weight factors of the distribution network data items to obtain the significant aggregation representation vector of all parameters of the distribution network.

[0015] In the above-mentioned power system situation awareness method based on multimodal joint analysis, whether the power grid is operating in a low-carbon manner is determined based on the significant aggregation representation vector of all parameters of the distribution network, including: inputting the significant aggregation representation vector of all parameters of the distribution network into a classifier-based power grid situation awareness module to obtain a situation awareness result, and the situation awareness result is used to indicate whether the power grid is operating in a low-carbon manner.

[0016] Accordingly, according to another aspect of the present application, a power system situation awareness system based on multimodal joint analysis is provided, comprising: A statistical data set acquisition module is used to acquire statistical data sets on the power generation side, statistical data sets on the power grid side and statistical data sets on the demand side; A statistical data set embedding coding module, used to embed and code each item of data in the statistical data set of the power generation side, each item of data in the statistical data set of the power grid side, and each item of data in the statistical data set of the demand side, so as to obtain a set of embedded coding vectors for data items on the power generation side, a set of embedded coding vectors for data items on the power grid side, and a set of embedded coding vectors for data items on the demand side; A data item importance modulation module, used for respectively performing importance modulation on the set of embedded coding vectors of the power generation side data items, the set of embedded coding vectors of the power grid side data items and the set of embedded coding vectors of the demand side data items to obtain a set of weighted embedded coding vectors of the power generation side data items, a set of weighted embedded coding vectors of the power grid side data items and a set of weighted embedded coding vectors of the demand side data items; A vector set dynamic aggregation module, used to combine the set of weighted generation side data item embedded coding vectors, the set of weighted grid side data item embedded coding vectors and the set of weighted demand side data item embedded coding vectors into a set of distribution network data item embedded coding vectors, and dynamically aggregate the set of distribution network data item embedded coding vectors to obtain a significant aggregation representation vector of all distribution network parameters; A power grid operation state determination module, used to determine whether the power grid is low-carbon operation based on the significant aggregation representation vector of all parameters of the distribution network; Among them, the vector set dynamic aggregation module includes: a distribution network data item clustering initial center determination unit, which is used to determine the distribution network data item clustering initial center vector based on the characteristic energy distribution of the set of distribution network data item embedded coding vectors; a significance modulation dynamic aggregation unit, which is used to perform significance modulation dynamic aggregation on the set of distribution network data item embedded coding vectors based on the spatial span of each distribution network data item embedded coding vector in the set of distribution network data item embedded coding vectors relative to the distribution network data item clustering initial center vector to obtain the distribution network full parameter significant aggregation representation vector.

[0017] Compared with the prior art, the power system situation awareness method and system based on multimodal joint analysis provided by this application use data processing technology based on deep learning to perform statistical analysis on various situation indicators on the power generation side, power grid side and demand side of the distribution network, extract the semantic embedded feature representation of various situation indicators on the power generation side, power grid side and demand side, and through importance evaluation and global feature dynamic aggregation of each situation indicator data item, to capture the key situation characteristics in the operation of the distribution network, to achieve a comprehensive perception of the operation situation of the distribution network, and thus intelligently identify whether the power grid is operating in a low-carbon manner. In this way, efficient distribution network operation situation evaluation can be achieved, the intelligent level of power grid operation can be effectively improved, and strong technical support can be provided for the stable operation and low-carbon development of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0019] Figure 1 The present invention is a flowchart of a method for electric power system situation awareness based on multimodal joint analysis according to an embodiment of the present application.

[0020] Figure 2 Schematic diagram of data flow of a power system situation awareness method based on multimodal joint analysis according to an embodiment of the present application.

[0021] Figure 3 This is a flowchart of step S140 in the power system situation awareness method based on multimodal joint analysis according to an embodiment of the present application.

[0022] Figure 4 This is a flowchart of sub-step S142 in the power system situation awareness method based on multi-modal joint analysis according to an embodiment of the present application.

[0023] Figure 5 It is a block diagram of a power system situation awareness system based on multimodal joint analysis according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] Below, the embodiments of the present application will be described in more detail in conjunction with the accompanying drawings, and the above and other purposes, features and advantages of the present application will become more apparent. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the example embodiments described herein.

[0025] Figure 1 The present invention is a flowchart of a method for electric power system situation awareness based on multimodal joint analysis according to an embodiment of the present application. Figure 2 Schematic diagram of data flow of a power system situation awareness method based on multi-modal joint analysis according to an embodiment of the present application. Figure 1 and Figure 2 As shown, the power system situation awareness method based on multimodal joint analysis according to an embodiment of the present application includes the steps of: S110, obtaining a statistical data set of a power generation side, a statistical data set of a power grid side, and a statistical data set of a demand side; S120, embedding and encoding each item of data in the statistical data set of the power generation side, each item of data in the statistical data set of the power grid side, and each item of data in the statistical data set of the demand side, respectively, to obtain a set of embedded coding vectors for data items on the power generation side, a set of embedded coding vectors for data items on the power grid side, and a set of embedded coding vectors for data items on the demand side; S130, respectively performing importance modulation on the set of embedded coding vectors of the power grid data items, the set of embedded coding vectors of the power grid data items, and the set of embedded coding vectors of the demand side data items to obtain a set of weighted embedded coding vectors of the power grid data items, a set of weighted embedded coding vectors of the demand side data items, and a set of weighted embedded coding vectors of the demand side data items; S140, combining the set of weighted generation-side data item embedded coding vectors, the set of weighted grid-side data item embedded coding vectors and the set of weighted demand-side data item embedded coding vectors into a set of distribution network data item embedded coding vectors, and dynamically aggregating the set of distribution network data item embedded coding vectors to obtain a significant aggregate representation vector of all distribution network parameters; S150, determining whether the power grid is operating in a low-carbon manner based on the significant aggregation representation vector of all parameters of the distribution network.

[0026] In the above-mentioned power system situation awareness method based on multimodal joint analysis, the step S110 obtains a statistical data set on the power generation side, a statistical data set on the power grid side, and a statistical data set on the demand side, wherein the statistical data set on the power generation side includes renewable energy power generation, traditional energy power generation, carbon emissions, and power generation efficiency, the statistical data set on the power grid side includes transmission and distribution losses, network loss rate, power supply reliability index, voltage quality, network topology, and power flow distribution, and the statistical data set on the demand side includes power demand forecast values, user power consumption behavior data, distributed energy access data, and energy storage facility status. It should be understood that power grid dispatching, as a key link in the operation of the power system, needs to understand the operating status of the power system in real time and perform corresponding dispatching operations according to the operating status of the power system. Since the power system covers the entire process from power generation, distribution to power consumption, in order to achieve comprehensive monitoring and evaluation of the operating status of the power system, this application collects various situation indicators on the power generation side, power grid side, and demand side of the power system to build a comprehensive situation awareness evaluation framework, intelligently evaluate whether the power grid is in a low-carbon operating state, so as to provide important decision support for energy conservation and emission reduction of the power grid.

[0027] In the above-mentioned method for power system situation awareness based on multimodal joint analysis, in step S120, each data in the statistical data set of the power generation side, each data in the statistical data set of the power grid side, and each data in the statistical data set of the demand side are embedded and encoded respectively to obtain a set of embedded coding vectors of power generation side data items, a set of embedded coding vectors of power grid side data items, and a set of embedded coding vectors of demand side data items. It should be understood that considering that the sources of various situation data in the power system are diverse, including numerical data (such as power generation, network loss rate), categorical data (such as network topology), etc., it is difficult to directly conduct comprehensive analysis due to the different data types and formats. Therefore, in order to achieve unified data processing, the present application adopts embedded coding technology to embed and encode each data in the statistical data set of the power generation side, the statistical data set of the power grid side, and the statistical data set of the demand side, so as to map each data into a unified semantic embedding feature space, capture the semantic information of each data, and convert different types of data into a unified vector representation, thereby obtaining a set of embedded coding vectors for data items on the power generation side, a set of embedded coding vectors for data items on the power grid side, and a set of embedded coding vectors for data items on the demand side, so as to facilitate subsequent data processing and analysis.

[0028] In the above-mentioned method for electric power system situation awareness based on multimodal joint analysis, the step S130 is to modulate the importance of the set of embedded coding vectors of the power generation side data items, the set of embedded coding vectors of the power grid side data items and the set of embedded coding vectors of the demand side data items to obtain the set of weighted embedded coding vectors of the power generation side data items, the set of weighted embedded coding vectors of the power grid side data items and the set of weighted embedded coding vectors of the demand side data items. In a specific example of the present application, the step S130 includes: using the fuzzy hierarchical analysis method to calculate the weights of the set of embedded coding vectors of the power generation side data items, the set of embedded coding vectors of the power grid side data items and the set of embedded coding vectors of the demand side data items to obtain the set of weighted embedded coding vectors of the power generation side data items, the set of weighted embedded coding vectors of the power grid side data items and the set of weighted embedded coding vectors of the demand side data items. It should be understood that considering that the operating situation of the power system is jointly affected by multiple factors, and there are complex interactions and different importance between the factors. Therefore, in order to more accurately identify the operation status of the power grid, this application further adopts the fuzzy analytic hierarchy process (FAHP) to process the set of the power generation side data item embedded coding vector, the set of the power grid side data item embedded coding vector and the set of the demand side data item embedded coding vector respectively. Specifically, the data in the power system often carries uncertainty. For example, the power generation of renewable energy is greatly affected by weather conditions, and the user's electricity consumption behavior is random. The FAHP method can effectively deal with uncertainty and fuzziness problems by introducing fuzzy set theory, comprehensively consider the relative importance of multiple factors, and assign weights to various data through a hierarchical model to achieve a quantitative evaluation of the importance of different status indicators, thereby more accurately reflecting the actual impact of various factors on the operation status of the power system.

[0029] In the above-mentioned method for electric power system situation awareness based on multimodal joint analysis, the step S140 combines the set of weighted generation side data item embedded coding vectors, the set of weighted grid side data item embedded coding vectors and the set of weighted demand side data item embedded coding vectors into a set of distribution network data item embedded coding vectors, and dynamically aggregates the set of distribution network data item embedded coding vectors to obtain a significant aggregation representation vector of all distribution network parameters. It should be understood that after the above processing, the relative importance of each situation indicator in its corresponding statistical data set is fully considered. However, since the operating situation of the electric power system is jointly affected by the generation side, the grid side and the demand side, it is necessary to further perform global feature aggregation processing on the data on the generation side, the grid side and the demand side to comprehensively evaluate the overall operating situation of the electric power system. Based on this, in the technical solution of the present application, firstly, the set of weighted power generation side data item embedded coding vectors, the set of weighted power grid side data item embedded coding vectors and the set of weighted demand side data item embedded coding vectors are combined into a set of distribution network data item embedded coding vectors, and then, the set of distribution network data item embedded coding vectors is dynamically aggregated by performing feature significance modulation to identify and enhance key situational features in the operation of the distribution network, so as to obtain a more discriminative and representative global feature representation.

[0030] Figure 3 FIG. 1 is a flow chart of step S140 in the power system situation awareness method based on multi-modal joint analysis according to an embodiment of the present application. Figure 3 As shown, the step S140 includes: S141, determining the initial center vector of the distribution network data item clustering based on the characteristic energy distribution of the set of distribution network data item embedded coding vectors; S142, based on the spatial span of each distribution network data item embedded coding vector in the set of distribution network data item embedded coding vectors relative to the initial center vector of the distribution network data item clustering, performing significance modulation dynamic aggregation on the set of distribution network data item embedded coding vectors to obtain the significant aggregation representation vector of all parameters of the distribution network.

[0031] Specifically, the step S141 includes: calculating the static energy factor of each distribution network data item embedded coding vector in the set of distribution network data item embedded coding vectors to obtain a set of distribution network data item static energy factors; selecting the distribution network data item embedded coding vector corresponding to the maximum value in the set of distribution network data item static energy factors as the initial center vector of the distribution network data item clustering. In a specific example of the present application, calculating the static energy factor of the distribution network data item embedded coding vector includes: calculating the kurtosis of the distribution network data item embedded coding vector, and inputting the kurtosis into a sigmoid activation function to obtain the distribution network data item static energy factor.

[0032] That is, firstly, the static energy factor of each distribution network data item embedding coding vector is calculated based on the predefined metric function to evaluate its stability and importance in the feature space, and the distribution network data item embedding coding vector corresponding to the largest static energy factor is selected as the initial center vector of the distribution network data item clustering to determine the cluster core of the distribution network data item feature distribution, thereby providing a stable reference point for subsequent feature aggregation.

[0033] Figure 4 FIG. 1 is a flowchart of sub-step S142 in the method for power system situation awareness based on multi-modal joint analysis according to an embodiment of the present application. Figure 4 As shown, the step S142 includes: S1421, based on the spatial span between each distribution network data item embedded coding vector in the set of distribution network data item embedded coding vectors and the initial center vector of the distribution network data item clustering, and the set of static energy factors of the distribution network data items, calculating the dynamic aggregation energy factor of each distribution network data item embedded coding vector in the set of distribution network data item embedded coding vectors to obtain a set of dynamic aggregation energy factors of distribution network data items; S1422, based on the set of dynamic aggregation energy factors of the distribution network data items, performing significance modulation aggregation on the set of distribution network data item embedded coding vectors to obtain a significant aggregation representation vector of all parameters of the distribution network.

[0034] Specifically, the step S1421 includes: taking the square value of the number of eigenvectors between the distribution network data item embedding coding vector and the initial center vector of the distribution network data item clustering as the spatial span coefficient, and calculating the weighted ratio between the product of the static energy factor of the distribution network data item embedding coding vector and the static energy factor of the initial center vector of the distribution network data item clustering and the spatial span coefficient to obtain the dynamic aggregation energy factor of the distribution network data item.

[0035] Specifically, the step S1422 includes: inputting the set of dynamic aggregation energy factors of the distribution network data items into a gated mask unit to obtain a set of dynamic aggregation weight factors of the distribution network data items; and calculating the weighted sum of the set of embedded coding vectors of the distribution network data items based on the set of dynamic aggregation weight factors of the distribution network data items to obtain the significant aggregation representation vector of all parameters of the distribution network. In a specific example of the present application, the set of dynamic aggregation energy factors of the distribution network data items is input into a gated mask unit to obtain a set of dynamic aggregation weight factors of the distribution network data items, including: inputting the set of dynamic aggregation energy factors of the distribution network data items into a sigmoid function for normalization processing to obtain a set of normalized dynamic aggregation energy factors of the distribution network data items; and based on a preset mask threshold, deactivating the set of normalized dynamic aggregation energy factors of the distribution network data items to obtain the set of dynamic aggregation weight factors of the distribution network data items.

[0036] Here, based on the spatial span of each distribution network data item embedding coding vector relative to the initial center vector of the distribution network data item cluster, and combined with their respective static energy factors, the dynamic aggregation energy factor of each distribution network data item embedding coding vector is further derived. That is, by considering the feature space proximity of each distribution network data item embedding coding vector relative to the core of the distribution network data item feature distribution cluster and the importance difference of the intrinsic attributes, its contribution to the global parameter characteristics of the distribution network is more accurately measured. Then, based on the gated mask mechanism, the obtained dynamic aggregation energy factor is nonlinearly transformed and screened to generate corresponding weights to achieve dynamic weighted aggregation of the set of distribution network data item embedding coding vectors, thereby ensuring that in the process of global feature aggregation, key situation features can be effectively highlighted while suppressing unimportant information to obtain a significant aggregation representation vector of all distribution network parameters. Through this dynamic aggregation method, distribution network information from different data sources can be effectively integrated, key situation features in the operation of the distribution network can be extracted, and more accurate and comprehensive situation awareness capabilities can be provided for power grid dispatching.

[0037] Accordingly, the step S140 includes: processing the set of embedded coding vectors of the distribution network data items using the following feature dynamic aggregation formula to obtain a significant aggregation representation vector of all parameters of the distribution network, wherein the feature dynamic aggregation formula is: , , , , , , , , , in, represents the set of embedding coding vectors of the distribution network data items, , , , and respectively represent the first, second, and third in the set of embedded coding vectors of the distribution network data item , and The distribution network data items are embedded in the coding vector, The value of is the number of embedded coding vectors of the distribution network data item, Indicates the The distribution network data item is embedded in the encoding vector The eigenvalues ​​at the positions, and Respectively represent the The feature mean and square of the feature variance of the embedding vector of the distribution network data item, It means to calculate the expected value of a set. represents the sigmoid activation function, Indicates the The static energy factor of the distribution network data item embedded in the coding vector, Indicates the index corresponding to the maximum value. represents the index of the largest static energy factor of the distribution network data item in the set of static energy factors of the distribution network data item, represents the initial center vector of the distribution network data item clustering, represents the static energy factor of the initial center vector of the distribution network data item clustering, and For different weight parameters, Indicates the The number of feature vectors between the embedded coding vector of the distribution network data item and the initial center vector of the distribution network data item clustering, Indicates the The dynamic aggregation energy factor of the distribution network data item embedded in the encoding vector, Indicates Normalized distribution network data items dynamically aggregate energy factors, Indicates mask processing, is the preset mask threshold, Indicates Dynamic aggregation weight factor of distribution network data items, Represents the significant aggregate representation vector of all parameters of the distribution network.

[0038] In the above-mentioned power system situation awareness method based on multimodal joint analysis, the step S150 determines whether the power grid is running in a low-carbon manner based on the significant aggregation representation vector of all parameters of the distribution network. In a specific example of the present application, the step S150 includes: inputting the significant aggregation representation vector of all parameters of the distribution network into a power grid situation awareness module based on a classifier to obtain a situation awareness result, and the situation awareness result is used to indicate whether the power grid is running in a low-carbon manner. That is, a classification algorithm is used to perform feature analysis and classification judgment on the significant aggregation representation vector of all parameters of the distribution network to identify the current operating state of the power grid. In the technical solution of the present application, the classifier can accurately distinguish the parameter characteristics of the power grid under different operating states, such as low-carbon operating state, high-carbon emission state, etc., through training, so that the feature pattern learning of the significant aggregation representation vector of all parameters of the distribution network can be mapped to the corresponding classification label to intelligently determine whether the power grid is running in a low-carbon manner. In this way, real-time feedback and suggestions can be provided for the operation of the power grid, helping decision makers to adjust the power grid operation strategy in a timely manner to achieve the goal of energy saving and emission reduction.

[0039] Preferably, inputting the distribution network full parameter significant aggregation representation vector into a classifier-based power grid situation awareness module to obtain a situation awareness result comprises: determining the distribution network full parameter significant aggregation probability value obtained by inputting the distribution network full parameter significant aggregation representation vector into the classifier-based power grid situation awareness module, wherein the distribution network full parameter significant aggregation probability value represents the probability value of low-carbon operation of the power grid; calculating the product of a first distribution network full parameter significant aggregation feature value of the distribution network full parameter significant aggregation representation vector and the distribution network full parameter significant aggregation probability value to obtain a first distribution network full parameter significant aggregation feature probability product; calculating the product of a second distribution network full parameter significant aggregation feature value of the distribution network full parameter significant aggregation representation vector and a difference between one and the distribution network full parameter significant aggregation probability value to obtain a second distribution network full parameter significant aggregation feature inverse probability product; calculating the quotient of the first distribution network full parameter significant aggregation feature value divided by the difference between one and the distribution network full parameter significant aggregation probability value to obtain a first distribution network full parameter significant aggregation feature inverse probability quotient; calculating Calculate the quotient of the second distribution network full parameter significant aggregation characteristic value and the distribution network full parameter significant aggregation probability value to obtain the second distribution network full parameter significant aggregation characteristic probability quotient; add the first distribution network full parameter significant aggregation characteristic probability product and the second distribution network full parameter significant aggregation characteristic inverse probability product, and then subtract the difference between the first distribution network full parameter significant aggregation characteristic inverse probability quotient and the second distribution network full parameter significant aggregation characteristic probability quotient to obtain the distribution network full parameter significant aggregation correction characteristic value; for each first distribution network full parameter significant aggregation characteristic value and second distribution network full parameter significant aggregation characteristic value of the distribution network full parameter significant aggregation representation vector, perform matrix multiplication of the distribution network full parameter significant aggregation correction characteristic matrix composed of the distribution network full parameter significant aggregation correction characteristic value and the distribution network full parameter significant aggregation representation vector to obtain the corrected distribution network full parameter significant aggregation representation vector; input the corrected distribution network full parameter significant aggregation representation vector into the classifier-based power grid situation awareness module to obtain the situation awareness result.

[0040] The correction process of the distribution network full parameter significant aggregation representation vector can be expressed as the following correction formula: , , , in, represents the significant aggregation representation vector of all parameters of the distribution network, is the probability value of low-carbon operation of the power grid, and They represent the first and eigenvalues, Indicates the significant aggregate correction eigenvalues ​​of all distribution network parameters, The distribution network full parameter significant aggregation correction characteristic matrix composed of the significant aggregation correction eigenvalues ​​of all distribution network parameters is represented. represents the matrix multiplication operation, A significant aggregation representation vector representing all parameters of the corrected distribution network.

[0041] That is, when using the fuzzy hierarchical analysis method to calculate the weights of the set of embedded coding vectors of the data items on the power generation side, the set of embedded coding vectors of the data items on the power grid side, and the set of embedded coding vectors of the data items on the demand side, the fuzzy hierarchical analysis method is used to perform explicit attention modulation on the set of embedded coding vectors of the data items on the power generation side, the set of embedded coding vectors of the data items on the power grid side, and the set of embedded coding vectors of the data items on the demand side. However, the hierarchical analysis will also cause the embedded coding vectors of each data item to have spatial hierarchical differences, which makes the field dynamic aggregation diversity caused by the spatial layer differences of the significant aggregation representation vectors of all parameters of the distribution network when performing global dynamic aggregation, affecting the iterative consistency of the classification mapping and reducing the accuracy of the classification results.

[0042] Therefore, the present application targets the regional-boundary integral relationship of the high-dimensional feature manifold of the distribution network's full-parameter significant aggregation representation vector, and approximates the simply connected region representation of the high-dimensional manifold of the feature set of the distribution network's full-parameter significant aggregation representation vector composed of eigenvalue pairs by defining the overall probability distribution representation of the feature set of the distribution network's full-parameter significant aggregation representation vector as a manifold constraint boundary, so as to avoid the ambiguity in the mapping of the diversified feature representation of the distribution network's full-parameter significant aggregation representation vector to the local manifold representation in the high-dimensional feature class convergence space, improve the execution iteration consistency of each local feature distribution in the mapping task, improve the feature classification convergence effect, and improve the accuracy of the situation awareness results obtained by the classifier-based power grid situation awareness module.

[0043] In summary, the method for situation awareness of the power system based on multimodal joint analysis according to the embodiment of the present application is explained, which uses data processing technology based on deep learning to perform statistical analysis on various situation indicators on the power generation side, power grid side and demand side of the distribution network, extracts the semantic embedded feature representation of various situation indicators on the power generation side, power grid side and demand side, and through importance evaluation and global feature dynamic aggregation of each situation indicator data item, in order to capture the key situation characteristics in the operation of the distribution network, realize comprehensive perception of the operation situation of the distribution network, and thus intelligently identify whether the power grid is operating in a low-carbon manner. In this way, efficient evaluation of the operation situation of the distribution network can be achieved, the intelligent level of the power grid operation can be effectively improved, and strong technical support can be provided for the stable operation and low-carbon development of the power grid.

[0044] Figure 5 FIG. 1 is a block diagram of a power system situation awareness system based on multi-modal joint analysis according to an embodiment of the present application. Figure 5 As shown, the power system situation awareness system 100 based on multi-modal joint analysis includes: A statistical data set acquisition module 110 is used to acquire a statistical data set of a power generation side, a statistical data set of a power grid side, and a statistical data set of a demand side; A statistical data set embedding coding module 120, used to embed and code each item of data in the statistical data set of the power generation side, each item of data in the statistical data set of the power grid side, and each item of data in the statistical data set of the demand side, respectively, to obtain a set of embedded coding vectors for data items on the power generation side, a set of embedded coding vectors for data items on the power grid side, and a set of embedded coding vectors for data items on the demand side; A data item importance modulation module 130, used to respectively perform importance modulation on the set of embedded coding vectors of the power generation side data items, the set of embedded coding vectors of the power grid side data items, and the set of embedded coding vectors of the demand side data items to obtain a set of weighted embedded coding vectors of the power generation side data items, a set of weighted embedded coding vectors of the power grid side data items, and a set of weighted embedded coding vectors of the demand side data items; A vector set dynamic aggregation module 140 is used to combine the set of weighted generation side data item embedded coding vectors, the set of weighted grid side data item embedded coding vectors and the set of weighted demand side data item embedded coding vectors into a set of distribution network data item embedded coding vectors, and dynamically aggregate the set of distribution network data item embedded coding vectors to obtain a significant aggregation representation vector of all distribution network parameters; The power grid operation state determination module 150 is used to determine whether the power grid is operating in a low-carbon manner based on the significant aggregation representation vector of all parameters of the distribution network.

[0045] In a specific example of the present application, the vector set dynamic aggregation module includes: a distribution network data item clustering initial center determination unit, which is used to determine the distribution network data item clustering initial center vector based on the characteristic energy distribution of the set of distribution network data item embedded coding vectors; a significance modulation dynamic aggregation unit, which is used to perform significance modulation dynamic aggregation on the set of distribution network data item embedded coding vectors based on the spatial span of each distribution network data item embedded coding vector in the set of distribution network data item embedded coding vectors relative to the distribution network data item clustering initial center vector to obtain the distribution network full parameter significance aggregation representation vector.

[0046] Here, those skilled in the art can understand that the specific operations of the various modules and units in the above-mentioned power system situation awareness system based on multi-modal joint analysis have been described in the above reference. Figures 1 to 4 It has been introduced in detail in the description of the power system situation awareness method based on multimodal joint analysis, and therefore, its repeated description will be omitted.

[0047] The basic principle of the present invention is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the present invention to being implemented by adopting the above specific details.

[0048] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed system and method can be implemented in other ways. For example, the system embodiment described above is only schematic. For example, the module division is only a logical function division, and there may be other division methods in actual implementation. The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0049] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference to a figure in a claim should not be considered as limiting the claim to which it relates.

[0050] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units stated in the system claims can also be implemented by one unit through software or hardware.

[0051] Finally, it should be noted that the above description has been given for the purpose of illustration and description. In addition, the above embodiments are only used to illustrate the technical solution of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A power system situation awareness method based on multi-modal joint analysis, characterized in that: include: Obtain statistical data sets on the power generation side, statistical data sets on the power grid side, and statistical data sets on the demand side; Embedding and encoding each item of data in the statistical data set of the power generation side, each item of data in the statistical data set of the power grid side, and each item of data in the statistical data set of the demand side to obtain a set of embedded coding vectors for data items on the power generation side, a set of embedded coding vectors for data items on the power grid side, and a set of embedded coding vectors for data items on the demand side; Performing importance modulation on the set of embedded coding vectors of the power grid data items, the set of embedded coding vectors of the power grid data items, and the set of embedded coding vectors of the demand side data items, respectively, to obtain a set of weighted embedded coding vectors of the power grid data items, a set of weighted embedded coding vectors of the demand side data items, and a set of weighted embedded coding vectors of the demand side data items; The set of weighted generation-side data item embedding coding vectors, the set of weighted grid-side data item embedding coding vectors and the set of weighted demand-side data item embedding coding vectors are combined into a set of distribution network data item embedding coding vectors, and the set of distribution network data item embedding coding vectors is dynamically aggregated to obtain a significant aggregate representation vector of all distribution network parameters; Determining whether the power grid is operating in a low-carbon manner based on the significant aggregation representation vector of all parameters of the distribution network; Among them, the set of distribution network data item embedded coding vectors is dynamically aggregated, including: determining the initial center vector of distribution network data item clustering based on the characteristic energy distribution of the set of distribution network data item embedded coding vectors; based on the spatial span of each distribution network data item embedded coding vector in the set of distribution network data item embedded coding vectors relative to the initial center vector of distribution network data item clustering, the set of distribution network data item embedded coding vectors is dynamically aggregated by significance modulation to obtain the significant aggregation representation vector of all parameters of the distribution network.

2. The method for power system situation awareness based on multimodal joint analysis according to claim 1 is characterized in that: The statistical data set on the power generation side includes renewable energy power generation, traditional energy power generation, carbon emissions and power generation efficiency; the statistical data set on the grid side includes transmission and distribution losses, network loss rate, power supply reliability indicators, voltage quality, network topology and power flow distribution; the statistical data set on the demand side includes electricity demand forecast values, user electricity consumption behavior data, distributed energy access data, and energy storage facility status.

3. The method for power system situation awareness based on multimodal joint analysis according to claim 2 is characterized in that: The set of the generation side data item embedded coding vectors, the set of the grid side data item embedded coding vectors and the set of the demand side data item embedded coding vectors are respectively subjected to importance modulation to obtain a set of weighted generation side data item embedded coding vectors, a set of weighted grid side data item embedded coding vectors and a set of weighted demand side data item embedded coding vectors, including: The fuzzy hierarchical analysis method is used to perform weight calculations on the set of embedded coding vectors of the power generation side data items, the set of embedded coding vectors of the power grid side data items and the set of embedded coding vectors of the demand side data items, respectively, to obtain the set of weighted embedded coding vectors of the power generation side data items, the set of weighted embedded coding vectors of the power grid side data items and the set of weighted embedded coding vectors of the demand side data items.

4. The method for power system situation awareness based on multi-modal joint analysis according to claim 3 is characterized in that: Determining the initial center vector of the distribution network data item clustering based on the characteristic energy distribution of the set of the distribution network data item embedding coding vectors includes: Calculating the static energy factor of each distribution network data item embedding coding vector in the set of distribution network data item embedding coding vectors to obtain a set of distribution network data item static energy factors; The distribution network data item embedding coding vector corresponding to the maximum value in the set of static energy factors of the distribution network data item is selected as the initial center vector of the distribution network data item clustering.

5. The method for electric power system situation awareness based on multi-modal joint analysis according to claim 4 is characterized in that: Calculating the static energy factor of each distribution network data item embedded coding vector in the set of distribution network data item embedded coding vectors to obtain a set of distribution network data item static energy factors, including: The kurtosis of the distribution network data item embedded in the coding vector is calculated, and the kurtosis is input into a sigmoid activation function to obtain a static energy factor of the distribution network data item.

6. The method for power system situation awareness based on multi-modal joint analysis according to claim 5 is characterized in that: Based on the spatial span of each distribution network data item embedding coding vector in the set of distribution network data item embedding coding vectors relative to the initial center vector of the distribution network data item clustering, the set of distribution network data item embedding coding vectors is dynamically aggregated by significance modulation to obtain a significant aggregation representation vector of all parameters of the distribution network, including: Based on the spatial span between each distribution network data item embedding coding vector in the set of distribution network data item embedding coding vectors and the initial center vector of the distribution network data item clustering, and the set of static energy factors of the distribution network data items, the dynamic aggregation energy factor of each distribution network data item embedding coding vector in the set of distribution network data item embedding coding vectors is calculated to obtain a set of dynamic aggregation energy factors of distribution network data items; Based on the set of dynamic aggregation energy factors of the distribution network data items, the set of embedded coding vectors of the distribution network data items is subjected to saliency modulation aggregation to obtain a saliency aggregation representation vector of all parameters of the distribution network.

7. The method for power system situation awareness based on multi-modal joint analysis according to claim 6 is characterized in that: Calculating the dynamic aggregate energy factor of each distribution network data item embedded coding vector in the set of distribution network data item embedded coding vectors to obtain a set of distribution network data item dynamic aggregate energy factors, including: Taking the square value of the number of eigenvectors between the distribution network data item embedding coding vector and the initial center vector of the distribution network data item clustering as the spatial span coefficient, the weighted ratio between the product of the static energy factor of the distribution network data item embedding coding vector and the static energy factor of the initial center vector of the distribution network data item clustering and the spatial span coefficient is calculated to obtain the dynamic aggregation energy factor of the distribution network data item.

8. The method for power system situation awareness based on multi-modal joint analysis according to claim 7 is characterized in that: Based on the set of dynamic aggregation energy factors of the distribution network data items, a set of embedded coding vectors of the distribution network data items is subjected to saliency modulation aggregation to obtain a saliency aggregation representation vector of all parameters of the distribution network, including: Inputting the set of dynamic aggregation energy factors of the distribution network data items into a gated mask unit to obtain a set of dynamic aggregation weight factors of the distribution network data items; The weighted sum of the set of distribution network data item embedding coding vectors is calculated based on the set of dynamic aggregation weight factors of the distribution network data items to obtain the significant aggregation representation vector of all parameters of the distribution network.

9. The method for electric power system situation awareness based on multi-modal joint analysis according to claim 8 is characterized in that: Based on the significant aggregation representation vector of all parameters of the distribution network, determining whether the power grid is in low-carbon operation includes: The distribution network full parameter significant aggregation representation vector is input into a classifier-based power grid situation awareness module to obtain a situation awareness result, and the situation awareness result is used to indicate whether the power grid is operating in a low-carbon manner.

10. The power system situation awareness system based on multi-modal joint analysis is characterized by: include: A statistical data set acquisition module is used to acquire statistical data sets on the power generation side, statistical data sets on the power grid side and statistical data sets on the demand side; A statistical data set embedding coding module, used to embed and code each item of data in the statistical data set of the power generation side, each item of data in the statistical data set of the power grid side, and each item of data in the statistical data set of the demand side, so as to obtain a set of embedded coding vectors for data items on the power generation side, a set of embedded coding vectors for data items on the power grid side, and a set of embedded coding vectors for data items on the demand side; A data item importance modulation module, used for respectively performing importance modulation on the set of embedded coding vectors of the power generation side data items, the set of embedded coding vectors of the power grid side data items and the set of embedded coding vectors of the demand side data items to obtain a set of weighted embedded coding vectors of the power generation side data items, a set of weighted embedded coding vectors of the power grid side data items and a set of weighted embedded coding vectors of the demand side data items; A vector set dynamic aggregation module, used to combine the set of weighted generation side data item embedded coding vectors, the set of weighted grid side data item embedded coding vectors and the set of weighted demand side data item embedded coding vectors into a set of distribution network data item embedded coding vectors, and dynamically aggregate the set of distribution network data item embedded coding vectors to obtain a significant aggregation representation vector of all distribution network parameters; A power grid operation state determination module, used to determine whether the power grid is low-carbon operation based on the significant aggregation representation vector of all parameters of the distribution network; Among them, the vector set dynamic aggregation module includes: a distribution network data item clustering initial center determination unit, which is used to determine the distribution network data item clustering initial center vector based on the characteristic energy distribution of the set of distribution network data item embedded coding vectors; a significance modulation dynamic aggregation unit, which is used to perform significance modulation dynamic aggregation on the set of distribution network data item embedded coding vectors based on the spatial span of each distribution network data item embedded coding vector in the set of distribution network data item embedded coding vectors relative to the distribution network data item clustering initial center vector to obtain the distribution network full parameter significant aggregation representation vector.

Images