Information interaction method for joint simulation of main network and active distribution network

Abstract

The invention relates to the field of intelligent analysis, and provides an information interaction method for joint simulation of a main network and an active distribution network, and the method comprises the steps: extracting a first to-be-evaluated key information item from a key information library, carrying out the word segmentation of a specific target through a deep learning model based on natural language processing, and obtaining a first to-be-evaluated key information item; and performing semantic embedding coding on each specific target word granularity after word segmentation and the first to-be-evaluated key information item, so as to obtain a semantic query result according to comparison between a semantic query matching coefficient between a semantic feature of each specific target word granularity and a semantic feature of the first to-be-evaluated key information item and a preset threshold value. And whether the first to-be-evaluated key information item is included is automatically determined. Therefore, the key information items can be automatically screened, the time and energy required by manual screening are reduced, the information processing speed and efficiency are improved, and meanwhile the response speed of the system is improved.

Description

Technical Field

[0001] This application relates to the field of intelligent analysis, and more specifically, to an information interaction method for the joint simulation of the main grid and the active distribution network. Background Art

[0002] With the wide application of distributed energy sources (such as solar energy, wind energy, etc.), the active distribution network has gradually become an important part of the modern power grid. The active distribution network can not only connect various forms of distributed power sources, but also integrate energy storage devices and other controllable loads, thus improving the flexibility and controllability of the distribution network. However, this also brings new challenges, especially how to achieve effective coordination and management between the main grid and the active distribution network.

[0003] However, the traditional information interaction between the main grid and the active distribution network is usually based on a relatively static and fixed method. Specifically, in traditional information interaction, the information interaction between the main grid and the active distribution network often needs to be achieved through pre-set rules and protocols, which are usually formulated manually, including information collection periods, information types, transmission paths, etc. This will lead to a slow response speed when dealing with emergencies (such as fault detection and response), thus affecting the overall response time and reliability of the system. In addition, when the grid structure or operating conditions change, the original information interaction rules may no longer apply. Modifying these rules requires reconfiguring the system, which is a cumbersome and time-consuming process and is difficult to quickly adapt to new situations.

[0004] Therefore, an optimized information interaction method for the main grid and the active distribution network is desired. Summary of the Invention

[0005] To solve the above technical problems, this application is proposed. Embodiments of this application provide an information interaction method for the joint simulation of the main grid and the active distribution network.

[0006] According to one aspect of this application, there is provided an information interaction method for the joint simulation of the main grid and the active distribution network, which includes:

[0007] Clarify the specific objectives of the joint simulation;

[0008] Based on the specific objectives of the joint simulation, determine the key information that needs to be shared between the main grid and the active distribution network;

[0009] Build a physical model including the main grid and the active distribution network based on the actual grid structure, and create an intelligent component model;

[0010] Integrate the physical model and the intelligent component model into the simulation platform;

[0011] Select a suitable communication protocol to support different types of information exchange;

[0012] Specify the data formats of various types of information to ensure that the sender and the receiver can correctly parse the various types of information;

[0013] Determine the flow paths of the various types of information within the simulation platform;

[0014] Set up the communication network of the simulation platform to simulate data transmission delays and bandwidth limitations;

[0015] Start the co-simulation according to a predetermined plan.

[0016] In the above information interaction method for the co-simulation of the main grid and the active distribution network, based on the specific objectives of the co-simulation, determine the key information that needs to be shared between the main grid and the active distribution network, including: performing word segmentation on the specific objectives to obtain a sequence distribution of specific objective word granularities, and then performing pre-trained semantic embedding to obtain a sequence distribution of specific objective word granularity semantic encoding feature vectors; extracting the first key information item to be evaluated from the key information library; performing pre-trained semantic embedding on the first key information item to be evaluated to obtain a semantic encoding feature vector of the first key information item to be evaluated; using the sequence distribution of the specific objective word granularity semantic encoding feature vectors as a feature vector library and using the semantic encoding feature vector of the first key information item to be evaluated as a query feature vector, performing semantic query encoding on the semantic encoding feature vector of the first key information item to be evaluated and the sequence distribution of the specific objective word granularity semantic encoding feature vectors to obtain a key information semantic matching coefficient; based on the comparison between the key information semantic matching coefficient and a preset threshold, determine whether to include the first key information item to be evaluated.

[0017] In the above information interaction method for the co-simulation of the main grid and the active distribution network, performing word segmentation on the specific objectives to obtain a sequence distribution of specific objective word granularities, and then performing pre-trained semantic embedding to obtain a sequence distribution of specific objective word granularity semantic encoding feature vectors, includes: performing word segmentation on the specific objectives to obtain the sequence distribution of the specific objective word granularities; inputting the sequence distribution of the specific objective word granularities into a pre-trained semantic embedding model to obtain the sequence distribution of the specific objective word granularity semantic encoding feature vectors.

[0018] In the above information interaction method for the co-simulation of the main grid and the active distribution network, inputting the sequence distribution of the specific objective word granularities into a pre-trained semantic embedding model to obtain the sequence distribution of the specific objective word granularity semantic encoding feature vectors, includes: inputting the sequence distribution of the specific objective word granularities into a pre-trained semantic embedding model based on the Bert model to obtain the sequence distribution of the specific objective word granularity semantic encoding feature vectors.

[0019] In the above information interaction method for the joint simulation of the main network and the active distribution network, pre-training semantic embedding of the first key information item to be evaluated to obtain a semantic encoding feature vector of the first key information item to be evaluated includes: inputting the first key information item to be evaluated into the pre-training semantic embedding model based on the Bert model to obtain the semantic encoding feature vector of the first key information item to be evaluated.

[0020] In the above information interaction method for the joint simulation of the main network and the active distribution network, using the sequence distribution of the specific target word granularity semantic encoding feature vectors as a feature vector library and the semantic encoding feature vector of the first key information item to be evaluated as a query feature vector, performing semantic query encoding on the sequence distribution of the semantic encoding feature vector of the first key information item to be evaluated and the specific target word granularity semantic encoding feature vectors to obtain a key information semantic matching coefficient, including: calculating the semantic fine-grained correlation degree between the semantic encoding feature vector of the first key information item to be evaluated and each specific target word granularity semantic encoding feature vector in the sequence distribution of the specific target word granularity semantic encoding feature vectors to obtain a sequence distribution of the semantic fine-grained correlation degrees; arranging the sequence distribution of the semantic fine-grained correlation degrees into a semantic correlation global representation vector; and obtaining the key information semantic matching coefficient based on the semantic correlation global representation vector.

[0021] In the above information interaction method for the joint simulation of the main network and the active distribution network, calculating the semantic fine-grained association degree between each specific target word granularity semantic coding feature vector in the sequence distribution of the semantic coding feature vector of the first key information item to be evaluated and the specific target word granularity semantic coding feature vector to obtain the sequence distribution of the semantic fine-grained association degree includes: calculating the first key information item-target word granularity difference vector by the position difference between the semantic coding feature vector of the first key information item to be evaluated and the specific target word granularity semantic coding feature vector, and calculating the square of the first norm of the first key information item-target word granularity difference vector to obtain the first key information item-target word granularity difference norm value; calculating the difference between the constant one and the square of the first norm of the semantic coding feature vector of the first key information item to obtain the first key information item semantic difference; calculating the difference between the constant one and the square of the first norm of the specific target word granularity semantic coding feature vector to obtain the specific target word granularity semantic difference; calculating the product between the first key information item semantic difference and the specific target word granularity semantic difference to obtain the first key information item-target word granularity product value; after dividing the first key information item-target word granularity difference norm value by the first key information item-target word granularity product value, multiplying the obtained value by the constant two first and then adding the constant one to obtain the first key information item-target word granularity measurement value; inputting the first key information item-target word granularity measurement value into the arccosh function to obtain the semantic fine-grained association degree.

[0022] In the above information interaction method for the joint simulation of the main network and the active distribution network, based on the semantic association global representation vector, obtaining the key information semantic matching coefficient includes: inputting the semantic association global representation vector into the key information matching metric module to obtain the key information semantic matching coefficient.

[0023] In the above information interaction method for the joint simulation of the main network and the active distribution network, the key information matching metric module is a key information matching metric module based on a decoder.

[0024] Compared with the prior art, the information interaction method for the joint simulation of the main network and the active distribution network provided by the present application extracts the first key information item to be evaluated from the key information library, and uses a deep learning model based on natural language processing to perform word segmentation on the specific target. After that, semantic embedding encoding is performed on each specific target word granularity after word segmentation and the first key information item to be evaluated. Based on this, whether to include the first key information item to be evaluated is automatically determined by comparing the semantic query matching coefficient between the semantic features of each specific target word granularity and the semantic features of the first key information item to be evaluated with a preset threshold. In this way, key information items can be automatically screened, reducing the time and effort required for manual screening, improving the speed and efficiency of information processing, and at the same time improving the response speed of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other objects, features, and advantages of the present application will become more obvious. 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 to the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0026] Figure 1 It is a flowchart of the information interaction method for the joint simulation of the main network and the active distribution network according to the embodiments of the present application.

[0027] Figure 2 It is a flowchart of determining the key information that needs to be shared between the main network and the active distribution network based on the specific target of the joint simulation in the information interaction method for the joint simulation of the main network and the active distribution network according to the embodiments of the present application.

[0028] Figure 3 It is a schematic diagram of data flow for determining the key information that needs to be shared between the main network and the active distribution network based on the specific target of the joint simulation in the information interaction method for the joint simulation of the main network and the active distribution network according to the embodiments of the present application.

[0029] Figure 4 It is a flowchart of performing pre-training semantic embedding to obtain a sequence distribution of specific target word granularity semantic encoding feature vectors after performing word segmentation on the specific target to obtain a sequence distribution of specific target word granularity in the information interaction method for the joint simulation of the main network and the active distribution network according to the embodiments of the present application.

[0030] Figure 5In the information interaction method for the joint simulation of the main grid and the active distribution network according to the embodiments of the present application, using the sequence distribution of the semantic encoding feature vectors at the specific target word granularity as the feature vector library and the semantic encoding feature vector of the first key information item to be evaluated as the query feature vector, the semantic query encoding is performed on the semantic encoding feature vector of the first key information item to be evaluated and the sequence distribution of the semantic encoding feature vectors at the specific target word granularity to obtain the key information semantic matching coefficient flowchart. Detailed implementation manners

[0031] Hereinafter, exemplary embodiments of the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0032] In modern power systems, the popularity of distributed energy sources such as solar and wind energy has made active distribution networks a key part of the power grid architecture. Active distribution networks can not only connect diverse distributed power sources but also integrate energy storage devices and other controllable loads, significantly enhancing the flexibility and control capabilities of the power grid. However, this progress has also brought new management challenges, especially in terms of how to achieve coordination and effective management between the main grid and the active distribution network.

[0033] Traditionally, the information interaction between the main grid and the active distribution network has relied on static and fixed interaction patterns. This pattern usually involves manually set rules and protocols, including the frequency of data collection, data types, and transmission routes. This method often reacts slowly when dealing with emergencies such as fault detection and response, reducing the overall response speed and reliability of the system. In addition, any changes in the grid structure or operating conditions may render the existing information exchange rules obsolete. Any modification to these rules requires reconfiguring the system, which is not only cumbersome but also time-consuming and difficult to quickly adapt to the new operating environment.

[0034] It should be understood that joint simulation can simulate the interaction relationship between the main grid and the active distribution network by integrating their models to evaluate and optimize the overall operating performance of the power grid. Based on this, the present application proposes an information interaction method for the joint simulation of the main grid and the active distribution network. Figure 1 Flowchart of the information interaction method for the joint simulation of the main grid and the active distribution network according to the embodiments of the present application. As Figure 1As shown in the figure, the information interaction method for the joint simulation of the main grid and the active distribution network according to the embodiments of the present application includes: S110, clarifying the specific objectives of the joint simulation; S120, determining the key information that needs to be shared between the main grid and the active distribution network based on the specific objectives of the joint simulation; S130, constructing a physical model including the main grid and the active distribution network based on the actual power grid structure, and creating an intelligent component model; S140, integrating the physical model and the intelligent component model into the simulation platform; S150, selecting a suitable communication protocol to support different types of information exchange; S160, specifying the data formats of various types of information to ensure that the sender and the receiver can correctly parse the various types of information; S170, determining the flow paths of the various types of information within the simulation platform; S180, setting up the communication network of the simulation platform to simulate data transmission delays and bandwidth limitations; S190, starting the joint simulation according to a predetermined plan.

[0035] In step S110, the specific objectives of the joint simulation are clarified. Accordingly, considering that clarifying the objectives can avoid waste of resources and deviation of direction, and improve the simulation efficiency, therefore, before starting the joint simulation, it is first necessary to clarify the specific objectives of the simulation, which can ensure that the simulation process has a clear direction.

[0036] In step S120, based on the specific objectives of the joint simulation, the key information that needs to be shared between the main grid and the active distribution network is determined. It should be understood that based on the specific objectives, the key information that needs to be shared between the main grid and the active distribution network is identified, such as the power grid operation status, load prediction, fault information, etc. These information are the basis for realizing effective coordination and management. By identifying the key information, the core elements of the power grid operation can be accurately grasped, and the effectiveness and pertinence of the simulation can be improved. At the same time, unnecessary information transmission is also reduced, and the communication burden is lowered.

[0037] In order to ensure that only the key information highly relevant to the specific objectives is shared, it is necessary to determine the key information shared between the main grid and the active distribution network. However, in traditional methods, the key information items are usually screened out manually from a large amount of data, which requires a large amount of time and manpower investment and is difficult to cope with a large amount of data and real-time changes.

[0038] Accordingly, in determining the key information that needs to be shared between the main grid and the active distribution network based on the specific objectives of the joint simulation, the technical concept of this application is to extract the first key information item to be evaluated from the key information library, and use a deep learning model based on natural language processing to perform word segmentation on the specific objectives. After that, semantic embedding encoding is performed on each specific objective word granularity after word segmentation and the first key information item to be evaluated, so as to automatically determine whether to include the first key information item to be evaluated according to the comparison between the semantic query matching coefficient between each specific objective word granularity semantic feature and the first key information item semantic feature and a preset threshold. In this way, it is possible to automatically screen key information items, reduce the time and effort required for manual screening, improve the speed and efficiency of information processing, and at the same time improve the response speed of the system.

[0039] Figure 2 FIG. is a flowchart of determining the key information that needs to be shared between the main grid and the active distribution network based on the specific objectives of the joint simulation in the information interaction method for the joint simulation of the main grid and the active distribution network according to an embodiment of the present application. Figure 3 FIG. is a schematic diagram of data flow for determining the key information that needs to be shared between the main grid and the active distribution network based on the specific objectives of the joint simulation in the information interaction method for the joint simulation of the main grid and the active distribution network according to an embodiment of the present application. As Figure 2 and Figure 3 shown, determining the key information that needs to be shared between the main grid and the active distribution network based on the specific objectives of the joint simulation includes: S121, performing word segmentation on the specific objectives to obtain a sequence distribution of specific objective word granularities, and then performing pre-training semantic embedding to obtain a sequence distribution of specific objective word granularity semantic encoding feature vectors; S122, extracting the first key information item to be evaluated from the key information library; S123, performing pre-training semantic embedding on the first key information item to be evaluated to obtain a first key information item semantic encoding feature vector; S124, using the sequence distribution of the specific objective word granularity semantic encoding feature vectors as a feature vector library and using the first key information item semantic encoding feature vector as a query feature vector, performing semantic query encoding on the first key information item semantic encoding feature vector and the sequence distribution of the specific objective word granularity semantic encoding feature vectors to obtain a key information semantic matching coefficient; S125, determining whether to include the first key information item to be evaluated based on the comparison between the key information semantic matching coefficient and a preset threshold.

[0040] In step S121, the specific objectives are subjected to word segmentation to obtain a sequence distribution of specific objective word granularities, and then pre-training semantic embedding is performed to obtain a sequence distribution of specific objective word granularity semantic encoding feature vectors. Specifically, Figure 4The flowchart shows the process of tokenizing the specific objective to obtain a sequence distribution of specific objective word granularity and then performing pre-trained semantic embedding to obtain a sequence distribution of specific objective word granularity semantic encoding feature vectors in the information interaction method for the joint simulation of the main grid and the active distribution network according to the embodiments of the present application. As Figure 4 shown, tokenizing the specific objective to obtain a sequence distribution of specific objective word granularity and then performing pre-trained semantic embedding to obtain a sequence distribution of specific objective word granularity semantic encoding feature vectors includes: S1211, tokenizing the specific objective to obtain a sequence distribution of the specific objective word granularity; S1212, inputting the sequence distribution of the specific objective word granularity into a pre-trained semantic embedding model to obtain a sequence distribution of the specific objective word granularity semantic encoding feature vectors. More specifically, in the embodiments of the present application, inputting the sequence distribution of the specific objective word granularity into a pre-trained semantic embedding model to obtain a sequence distribution of the specific objective word granularity semantic encoding feature vectors includes: inputting the sequence distribution of the specific objective word granularity into a pre-trained semantic embedding model based on the Bert model to obtain a sequence distribution of the specific objective word granularity semantic encoding feature vectors.

[0041] It should be understood that the specific objective contains multiple words regarding specific objective information, and these words contain complex context information and semantics. Based on this, in order to more centrally and meticulously analyze the specific objective meaning and semantic information contained in each objective word, in the technical solution of the present application, the specific objective is tokenized to obtain a sequence distribution of specific objective word granularity. Then, considering that each specific objective word granularity has rich semantic information and the association with the context. The Bert model is a deep learning model based on the Transformer architecture, which can provide context-sensitive word embeddings, meaning that the representation of each word depends not only on the word itself but also on its position and context in the sentence. Therefore, the sequence distribution of the specific objective word granularity is input into a pre-trained semantic embedding model based on the Bert model to capture and mine the semantic information in each objective word granularity, obtaining a sequence distribution of specific objective word granularity semantic encoding feature vectors.

[0042] In step S122, the first key information item to be evaluated is extracted from the key information library. To ensure that the included key information items have practical influence in the power grid operation, the key information items to be evaluated can be first extracted from the key information library, and by analyzing the extracted key information items to be evaluated, it can be evaluated whether they should be included. It is worth mentioning that the process of extracting the key information items to be evaluated and analyzing the key information items is not a one-time process but a dynamic process. As the simulation progresses, new data and feedback may prompt a re-evaluation of which information should be included, thus continuously optimizing the information sharing strategy.

[0043] In step S123, pre-training semantic embedding is performed on the first key information item to be evaluated to obtain a semantic encoding feature vector of the first key information item to be evaluated. Specifically, in the embodiment of the present application, performing pre-training semantic embedding on the first key information item to be evaluated to obtain a semantic encoding feature vector of the first key information item to be evaluated includes: inputting the first key information item to be evaluated into the pre-training semantic embedding model based on the Bert model to obtain the semantic encoding feature vector of the first key information item to be evaluated. It should be understood that the first key information item to be evaluated also contains rich deep context semantics and meanings, such as device names, voltage levels, etc. Therefore, in order to capture and extract the deep semantic representations in the first key information item to be evaluated, in the technical solution of the present application, the first key information item to be evaluated is input into the pre-training semantic embedding model based on the Bert model to obtain a semantic encoding feature vector of the first key information item to be evaluated.

[0044] In step S124, using the sequence distribution of the specific target word granularity semantic encoding feature vectors as a feature vector library and using the semantic encoding feature vector of the first key information item to be evaluated as a query feature vector, semantic query encoding is performed on the semantic encoding feature vector of the first key information item to be evaluated and the sequence distribution of the specific target word granularity semantic encoding feature vectors to obtain a key information semantic matching coefficient. Accordingly, considering that the sequence distribution of the specific target word granularity semantic encoding feature vectors reflects the basic meaning of each specific target word and the context semantics in its specific target description. The semantic encoding feature vector of the first key information item to be evaluated represents the core semantic content of the key information item and synthesizes all the semantic information of the key information item to be evaluated. In order to evaluate the semantic similarity between the first key information item to be evaluated and the specific target, and to ensure that the matching is not just based on surface text matching, but goes deep into the semantic level to avoid ignoring important information due to differences in the surface of the text. In the technical solution of the present application, using the sequence distribution of the specific target word granularity semantic encoding feature vectors as a feature vector library and using the semantic encoding feature vector of the first key information item to be evaluated as a query feature vector, semantic query encoding is performed on the semantic encoding feature vector of the first key information item to be evaluated and the sequence distribution of the specific target word granularity semantic encoding feature vectors to obtain a key information semantic matching coefficient.

[0045] Specifically, first, calculate the semantic fine-grained correlation degree between each specific target word granularity semantic encoding feature vector and the semantic encoding feature vector of the first key information item to be evaluated, so as to obtain the sequence distribution of the semantic fine-grained correlation degree. That is to say, by calculating the semantic fine-grained correlation degree, the semantic relationship between each specific target word granularity semantic encoding feature and the semantic encoding feature of the first key information item to be evaluated can be understood more carefully. Since each specific target word granularity may have different weights and contribution degrees in the specific target, and by calculating the fine-grained correlation degree between the two, the contribution size of each word to the overall matching degree can be evaluated, so as to more accurately judge the relevance to the first key information item to be evaluated, which helps to improve the accuracy and detail of the matching. Then, arrange the sequence distribution of the semantic fine-grained correlation degree into a semantic correlation global representation vector to comprehensively evaluate the overall semantic relationship between all specific target word granularities and the first key information item to be evaluated, providing a quantitative index for the overall semantic similarity. Finally, input the semantic correlation global representation vector into the key information matching metric module based on the decoder to effectively process the semantic correlation global representation vector and obtain the key information semantic matching coefficient.

[0046] Specifically, Figure 5 In the information interaction method for main network and active distribution network joint simulation according to the embodiment of the present application, taking the sequence distribution of the specific target word granularity semantic encoding feature vector as the feature vector library and taking the semantic encoding feature vector of the first key information item to be evaluated as the query feature vector, perform semantic query encoding on the semantic encoding feature vector of the first key information item to be evaluated and the sequence distribution of the specific target word granularity semantic encoding feature vector to obtain the flowchart of the key information semantic matching coefficient. As Figure 5 shown, taking the sequence distribution of the specific target word granularity semantic encoding feature vector as the feature vector library and taking the semantic encoding feature vector of the first key information item to be evaluated as the query feature vector, performing semantic query encoding on the semantic encoding feature vector of the first key information item to be evaluated and the sequence distribution of the specific target word granularity semantic encoding feature vector to obtain the key information semantic matching coefficient includes: S1241, calculate the semantic fine-grained correlation degree between the semantic encoding feature vector of the first key information item to be evaluated and each specific target word granularity semantic encoding feature vector in the sequence distribution of the specific target word granularity semantic encoding feature vector to obtain the sequence distribution of the semantic fine-grained correlation degree; S1242, arrange the sequence distribution of the semantic fine-grained correlation degree into a semantic correlation global representation vector; S1243, based on the semantic correlation global representation vector, obtain the key information semantic matching coefficient.

[0047] More specifically, in the embodiments of the present application, calculating the semantic fine-grained association degree between each specific target word granularity semantic encoding feature vector in the sequence distribution of the first key information item to be evaluated semantic encoding feature vector and the specific target word granularity semantic encoding feature vector to obtain the sequence distribution of the semantic fine-grained association degree includes: calculating the position difference between the first key information item to be evaluated semantic encoding feature vector and the specific target word granularity semantic encoding feature vector to obtain the first key information item - target word granularity difference vector, and calculating the square of the first norm of the first key information item - target word granularity difference vector to obtain the first key information item - target word granularity difference norm value; calculating the difference between the constant one and the square of the first norm of the first key information item to be evaluated semantic encoding feature vector to obtain the first key information item semantic difference; calculating the difference between the constant one and the square of the first norm of the specific target word granularity semantic encoding feature vector to obtain the specific target word granularity semantic difference; calculating the product between the first key information item semantic difference and the specific target word granularity semantic difference to obtain the first key information item - target word granularity product value; dividing the first key information item - target word granularity difference norm value by the first key information item - target word granularity product value, then multiplying the obtained value by the constant two and adding the constant one to obtain the first key information item - target word granularity measurement value; inputting the first key information item - target word granularity measurement value into the arccosh function to obtain the semantic fine-grained association degree.

[0048] In the embodiments of the present application, specifically, calculating the semantic fine-grained association degree between each specific target word granularity semantic encoding feature vector in the sequence distribution of the first key information item to be evaluated semantic encoding feature vector and the specific target word granularity semantic encoding feature vector to obtain the sequence distribution of the semantic fine-grained association degree can be expressed by the formula:

[0049]

[0050] where, v1 is the first key information item to be evaluated semantic encoding feature vector, v 2j is the j-th specific target word granularity semantic encoding feature vector in the sequence distribution of the specific target word granularity semantic encoding feature vector, ‖·‖ 2 is the square of the first norm for calculating the vector, arccosh(·) is the inverse hyperbolic cosine function, d P (v1, v 2j ) is for calculating the semantic fine-grained association degree between v1 and v 2j .

[0051] More specifically, in the embodiments of the present application, obtaining the key information semantic matching coefficient based on the semantic association global representation vector includes: inputting the semantic association global representation vector into a key information matching metric module to obtain the key information semantic matching coefficient. Here, the key information matching metric module refers to a key information matching metric module based on a decoder.

[0052] Preferably, in the technical solution of the present application, in the process of calculating the semantic fine-grained association degree between each specific target word granularity semantic encoding feature vector in the sequence distribution of the first key information item to be evaluated semantic encoding feature vector and the specific target word granularity semantic encoding feature vector to obtain the sequence distribution of the semantic fine-grained association degree, due to the relatively significant source domain semantic distribution differences among the specific target word granularity semantic encoding feature vectors in the sequence distribution of the specific target word granularity semantic encoding feature vector, it will cause the obtained semantic association global representation vector to have an unbalanced fine-grained feature distribution, which affects the decoding accuracy of the key information semantic matching coefficient obtained by inputting it into the key information matching metric module based on the decoder.

[0053] Based on this, in a preferred embodiment of the present application, inputting the semantic association global representation vector into a key information matching metric module based on a decoder to obtain the key information semantic matching coefficient includes: optimizing the semantic association global representation vector, including the steps of:

[0054] Subtracting the zero-order cardinality of the semantic association global representation vector from its dimensionality to generate a semantic association global discrete difference component, expressed as:

[0055] n = L - ‖V‖0

[0056] where L represents the dimensionality of the semantic association global representation vector, i.e., the length of the vector, V represents the semantic association global representation vector, ‖V‖0 represents the zero-order cardinality of the semantic association global representation vector, i.e., the zero norm of the vector, and n represents the semantic association global discrete difference component;

[0057] Calculating the quadratic metric of the semantic association global discrete difference component and its binary discrete integral to obtain a semantic association global hierarchical identification quantity α, expressed as:

[0058] α = log2(n 2 + n)

[0059] Constructing an exponential mapping operator with the eigenvalue at each position of the semantic association global representation vector as the base and the semantic association global discrete difference component minus one as the exponent, and performing a Hadamard product with the semantic association global hierarchical identification quantity to generate a semantic association global guiding tensor;

[0060] V1 = V ⊙(n-1) ⊙α

[0061] Wherein, V ⊙(n-1) represents that the eigenvalues at each position for calculating the global semantic association representation vector are used as the base number and the global semantic association discrete difference component minus one is used as the exponent, ⊙ represents the Hadamard product, and V1 represents the global semantic association guiding tensor;

[0062] Perform the Hadamard product on the global semantic association representation vector and the result of subtracting one from the global semantic association discrete difference component, and then through the inverse number field calibration operation, obtain the global semantic association field restriction tensor, which is expressed as:

[0063] V2 = V⊙(n - 1)⊙n -1

[0064] Wherein, V2 represents the global semantic association field restriction tensor;

[0065] Apply the natural base power transformation to the global semantic association field restriction tensor to form an exponential field distribution, and output the global semantic association field compensation tensor, which is expressed as:

[0066] V3 = exp(V2)

[0067] Wherein, V3 represents the global semantic association field compensation tensor;

[0068] Obtain the optimized global semantic association representation vector by element-wise addition of the global semantic association field restriction tensor and the global semantic association field compensation tensor, which is expressed as:

[0069]

[0070] Wherein, represents vector addition, and V′ represents the optimized global semantic association representation vector.

[0071] Finally, input the optimized global semantic association representation vector into the key information matching metric module based on the decoder to obtain the key information semantic matching coefficient.

[0072] For the multi-dimensional attribute topological structure of the global semantic association representation vector, use the statistical indicators of the feature set as the integration benchmark for tensor space representation. Extract the hierarchical identification parameters of the global semantic association representation vector in the discrete airspace superposition amount to achieve index region anchoring, and introduce the compensation factor of the attribute inversion field stability as a feedback mechanism to drive the regression distribution of the global semantic association representation vector to meet the projection trajectory tracing requirements. Thereby, enabling the feature set to perceive the spatial transformation conduction, optimizing the decoding accuracy of the key information semantic matching coefficient obtained by inputting the global semantic association representation vector into the key information matching metric module based on the decoder by enhancing the feature distribution balance.

[0073] In step S125, based on the comparison between the key information semantic matching coefficient and a preset threshold, it is determined whether to include the first key information item to be evaluated. That is, the comparison is made between the key information semantic matching coefficient obtained by performing semantic query encoding matching on the sequence distribution of the semantic encoding feature vectors of the specific target word granularity and the semantic encoding feature vector of the first key information item to be evaluated, so as to automatically determine whether to include the first key information item to be evaluated.

[0074] In step S130, a physical model including a main grid and an active distribution network is constructed based on the actual power grid structure, and an intelligent component model is created. It should be understood that based on the actual power grid structure, a physical model including a main grid and an active distribution network is constructed to reflect the topological structure and equipment parameters of the power grid. At the same time, an intelligent component model, such as an intelligent meter, an energy storage system, a distributed energy controller, etc., is created to simulate the behavior of these devices in the power grid. The combination of the physical model and the intelligent component model enables the simulation to more realistically reflect the actual situation of the power grid. The introduction of intelligent components increases the flexibility and dynamics of the simulation and can better cope with the complex changes of the power grid.

[0075] In step S140, the physical model and the intelligent component model are integrated into the simulation platform. It should be understood that the constructed physical model and intelligent component model are integrated into the simulation platform to form a complete simulation environment, in which various operating states and changes of the power grid can be simulated.

[0076] In step S150, a suitable communication protocol is selected to support different types of information exchange. It should be understood that according to different types of information exchange requirements, a suitable communication protocol (such as MQTT, CoAP, etc.) is selected to support efficient information transmission. That is, selecting a protocol according to requirements can improve the compatibility and efficiency of the system.

[0077] In step S160, the data formats of various types of information are specified to ensure that the sender and the receiver can correctly parse the various types of information. It should be understood that standardized data formats can reduce misunderstandings and errors in information transmission and enhance the reliability of information interaction. In this application, to ensure that the sender and the receiver can correctly parse information, the data formats of various types of information (such as JSON, XML, etc.) are formulated. This standardized information data format can improve the accuracy of information parsing.

[0078] In step S170, determine the flow paths of the various types of information within the simulation platform. It should be understood that within the simulation platform, determining the flow paths of various types of information between the main grid and the active distribution network includes links such as information collection, transmission, processing, and feedback. Clear information flow paths help optimize the information interaction process, reduce information transmission delays and errors. At the same time, it also provides a basis for simulating the communication network in subsequent steps.

[0079] In step S180, set the communication network of the simulation platform to simulate data transmission delays and bandwidth limitations. Correspondingly, considering that in order to more realistically reflect the information interaction situation in power grid operation, communication network parameters can be set in the simulation platform to simulate actual network conditions, such as data transmission delays and bandwidth limitations, which is closer to the actual situation and improves the credibility and practicality of the simulation. At the same time, it can also provide a basis for optimizing information interaction strategies.

[0080] In step S190, start the co-simulation according to a predetermined plan. Start the co-simulation according to a predetermined plan and monitor the information interaction and system response during the simulation process. Through real-time monitoring and adjustment, problems can be discovered and solved in a timely manner. Compared with traditional static simulations, the dynamic adjustment ability is stronger.

[0081] In summary, the information interaction method for co-simulation of the main grid and the active distribution network based on the embodiments of the present application is elucidated. It extracts the first key information item to be evaluated from the key information library, and uses a deep learning model based on natural language processing to perform word segmentation on the specific target. After that, semantic embedding encoding is performed on each specific target word granularity after word segmentation and the first key information item to be evaluated. Based on this, according to the comparison between the semantic query matching coefficient between the semantic features of each specific target word granularity and the semantic features of the first key information item to be evaluated and a preset threshold, it automatically determines whether to include the first key information item to be evaluated. In this way, it can automatically screen key information items, reduce the time and effort required for manual screening, improve the speed and efficiency of information processing, and at the same time improve the system response speed.

[0082] The basic principles of the present application have been described above in conjunction with specific embodiments. However, it should be noted that the advantages, benefits, effects, etc. mentioned in the present application are only examples and not limitations. It cannot be considered that these advantages, benefits, effects, etc. are essential for each embodiment of the present application. In addition, the specific details of the above application are only for the purpose of illustration and easy understanding, and are not limitations. The above details do not limit the present application to necessarily adopt the above specific details to implement.

Claims

1. An information interaction method for joint simulation of a main network and an active distribution network, characterized in that: include: Identify specific goals for joint simulation; Based on the specific objectives of the joint simulation, determine the key information that needs to be shared between the main grid and the active distribution grid; Build a physical model including the main grid and active distribution network based on the actual grid structure, and create an intelligent component model; Integrating the physical model and the smart component model into a simulation platform; Choose appropriate communication protocols to support different types of information exchange; Specify the data format of various types of information to ensure that the sender and receiver can correctly parse the information; Determining the flow paths of the various types of information within the simulation platform; Setting the communication network of the simulation platform to simulate data transmission delay and bandwidth limitation; Start the joint simulation according to the predetermined plan.

2. The information interaction method for joint simulation of the main network and the active distribution network according to claim 1 is characterized in that: Based on the specific objectives of the joint simulation, key information that needs to be shared between the main grid and the active distribution network is determined, including: Performing word segmentation processing on the specific target to obtain a sequence distribution of the specific target word granularity, and then performing pre-training semantic embedding to obtain a sequence distribution of the semantic encoding feature vector of the specific target word granularity; Extracting a first key information item to be evaluated from a key information database; Pre-training semantic embedding of the first key information item to be evaluated to obtain a semantic encoding feature vector of the first key information item to be evaluated; Using the sequence distribution of the semantic encoding feature vector of the specific target word granularity as a feature vector library and the semantic encoding feature vector of the first key information item to be evaluated as a query feature vector, semantic query encoding is performed on the sequence distribution of the semantic encoding feature vector of the first key information item to be evaluated and the semantic encoding feature vector of the specific target word granularity to obtain a key information semantic matching coefficient; Based on the comparison between the key information semantic matching coefficient and a preset threshold, it is determined whether to include the first key information item to be evaluated.

3. The information interaction method for joint simulation of the main network and the active distribution network according to claim 2 is characterized in that: The specific target is segmented to obtain a sequence distribution of the specific target word granularity, and then pre-trained semantic embedding is performed to obtain a sequence distribution of a semantic encoding feature vector of the specific target word granularity, including: Performing word segmentation processing on the specific target to obtain a sequence distribution of the word granularity of the specific target; The sequence distribution of the specific target word granularity is input into a pre-trained semantic embedding model to obtain the sequence distribution of the semantic encoding feature vector of the specific target word granularity.

4. The information interaction method for joint simulation of the main network and the active distribution network according to claim 3 is characterized in that: The sequence distribution of the specific target word granularity is input into a pre-trained semantic embedding model to obtain the sequence distribution of the semantic encoding feature vector of the specific target word granularity, including: inputting the sequence distribution of the specific target word granularity into a pre-trained semantic embedding model based on the Bert model to obtain the sequence distribution of the semantic encoding feature vector of the specific target word granularity.

5. The information interaction method for joint simulation of the main network and the active distribution network according to claim 4 is characterized in that: The first key information item to be evaluated is pre-trained with semantic embedding to obtain a semantic encoding feature vector of the first key information item to be evaluated, including: inputting the first key information item to be evaluated into the pre-trained semantic embedding model based on the Bert model to obtain the semantic encoding feature vector of the first key information item to be evaluated.

6. The information interaction method for joint simulation of the main network and the active distribution network according to claim 5 is characterized in that: Taking the sequence distribution of the semantic encoding feature vector of the specific target word granularity as the feature vector library and taking the semantic encoding feature vector of the first key information item to be evaluated as the query feature vector, performing semantic query encoding on the sequence distribution of the semantic encoding feature vector of the first key information item to be evaluated and the semantic encoding feature vector of the specific target word granularity to obtain the key information semantic matching coefficient, including: Calculating the semantic fine-grained association between the first key information item to be evaluated semantic encoding feature vector and each specific target word granularity semantic encoding feature vector in the sequence distribution of the specific target word granularity semantic encoding feature vector to obtain a sequence distribution of the semantic fine-grained association; Arranging the sequence distribution of the semantic fine-grained association degree into a semantic association global representation vector; Based on the semantic association global representation vector, the key information semantic matching coefficient is obtained.

7. The information interaction method for joint simulation of a main network and an active distribution network according to claim 6, characterized in that: Calculating the semantic fine-grained association between the first key information item to be evaluated semantic encoding feature vector and each specific target word granularity semantic encoding feature vector in the sequence distribution of the specific target word granularity semantic encoding feature vector to obtain a sequence distribution of the semantic fine-grained association, including: Calculating the positional difference between the first key information item to be evaluated semantic coding feature vector and the specific target word granularity semantic coding feature vector to obtain a first key information item to be evaluated-target word granularity differential vector, and calculating the square of a norm of the first key information item to be evaluated-target word granularity differential vector to obtain a first key information item to be evaluated-target word granularity differential norm value; Calculating the difference between a constant one and the square of a norm of the semantic encoding feature vector of the first key information item to be evaluated to obtain a semantic difference value of the first key information item to be evaluated; Calculate the difference between the constant one and the square of the one-norm of the semantic encoding feature vector of the specific target word granularity to obtain the semantic difference of the specific target word granularity; Calculating the product of the semantic difference value of the first key information item to be evaluated and the semantic difference value of the specific target word granularity to obtain a first key information item to be evaluated-target word granularity product value; After dividing the first key information item to be evaluated-target word granularity differential norm value by the first key information item to be evaluated-target word granularity product value, the obtained value is first multiplied by a constant two and then added to a constant one to obtain a first key information item to be evaluated-target word granularity measurement value; The first key information item to be evaluated-target word granularity measurement value is input into an arccosh function to obtain the semantic fine-grained association degree.

8. The information interaction method for joint simulation of the main network and the active distribution network according to claim 7 is characterized in that: Based on the semantic association global representation vector, the key information semantic matching coefficient is obtained, including: inputting the semantic association global representation vector into a key information matching metric module to obtain the key information semantic matching coefficient.

9. The information interaction method for joint simulation of the main network and the active distribution network according to claim 8, characterized in that: The key information matching metric module is a decoder-based key information matching metric module.

Images