Intelligent interaction system and method for power grid production case deduction

Abstract

The invention relates to the field of intelligent analysis, and provides an intelligent interaction system and method for power grid production case deduction, and the system employs a data analysis and processing algorithm based on artificial intelligence to carry out the embedded coding of operation data inputted by a user, carries out the semantic coding of a text description of a first event automatically triggered by the system, and carries out the recognition of a first event. And intelligently obtaining an analysis result according to a multi-scale interactive response between the coding feature and the first event semantic representation feature embedded in the operation data. In this way, potential risk analysis can be carried out on the operation data in real time, and instant feedback is provided. Therefore, the user can immediately know whether the operation of the user is correct or not in the drilling process, and adjustment is carried out according to feedback.

Description

Technical Field

[0001] The present application relates to the field of intelligent analysis, and more specifically, to an intelligent interactive system and method for power grid production case deduction. Background Art

[0002] The power grid is a complex system encompassing multiple links, including power generation, transmission, and distribution, involving numerous devices and facilities. Its operational status is affected by numerous factors. Grid operators require rigorous skills training to ensure they can respond appropriately to various situations in their operations. Therefore, conducting power grid production case studies is an important means of testing the grid's emergency response capabilities. These studies can identify potential issues and enable timely improvements.

[0003] However, traditional production case studies often rely on fixed scripts and pre-set outcomes, lacking flexibility. This prevents drills from fully covering all possible emergency scenarios. Operators may become accustomed to expected outcomes, leaving them unable to cope with unforeseen circumstances in actual emergencies. Furthermore, traditional drills are often limited to a few typical emergency scenarios and fail to cover all possible complexities. This results in a lack of diversity and complexity in the drills.

[0004] Therefore, an intelligent interactive solution for power grid production case deduction is desired. Summary of the Invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiments of the present application provide an intelligent interactive system and method for power grid production case deduction.

[0006] According to one aspect of the present application, an intelligent interactive method for power grid production case deduction is provided, which includes:

[0007] Step 1: In response to receiving an instruction to start a power grid production case simulation, the system automatically triggers a first event;

[0008] Step 2: receiving operation data input by the user;

[0009] Step 3: Based on the operation data, the system updates the grid status in real time;

[0010] Step 4: The system automatically triggers the second event and executes steps 2 and 3 repeatedly until all events are triggered.

[0011] Step 5: The system continuously collects the user's operation data and analyzes the user's operation data based on predefined evaluation indicators to generate a feedback report.

[0012] According to another aspect of the present application, an intelligent interactive system for power grid production case deduction is provided, which includes:

[0013] A simulation instruction event response module is configured to automatically trigger a first event in response to receiving an instruction to start a power grid production case deduction simulation;

[0014] An operation data collection module, used for receiving operation data input by a user;

[0015] A real-time grid status update module, configured to update the grid status in real time based on the operation data;

[0016] An event loop triggering module, configured to automatically trigger a second event in the system, and cyclically execute the operation data acquisition module and the grid status real-time updating module until all events are triggered;

[0017] The feedback report generation module is used for the system to continuously collect user operation data and analyze the user operation data based on predefined evaluation indicators to generate a feedback report.

[0018] Compared with the existing technology, the intelligent interactive system and method for power grid production case deduction provided by this application adopts artificial intelligence-based data analysis and processing algorithms to embed and encode the operation data input by the user, and semantically encode the text description of the first event automatically triggered by the system, so as to intelligently obtain the analysis results based on the multi-scale interactive response between the embedded coding features of the operation data and the semantic representation features of the first event. In this way, the potential risk analysis of the operation data can be performed in real time, and instant feedback can be provided. This means that users can immediately understand whether their operations are correct during the drill and make adjustments based on the feedback. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The above and other purposes, features, and advantages of the present application will become more apparent through a more detailed description of the embodiments of the present application in conjunction with the accompanying drawings. The accompanying drawings are intended 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 drawings, the same reference numerals generally represent the same components or steps.

[0020] Figure 1 This is a flowchart of an intelligent interactive method for power grid production case deduction according to an embodiment of the present application.

[0021] Figure 2 This is a flowchart of step 3 in the intelligent interaction method for power grid production case deduction according to an embodiment of the present application.

[0022] Figure 3This is a flowchart of step 31 in the intelligent interaction method for power grid production case deduction according to an embodiment of the present application.

[0023] Figure 4 This is a flowchart of step 314 in the intelligent interaction method for power grid production case deduction according to an embodiment of the present application.

[0024] Figure 5 This is a system block diagram of an intelligent interactive system for power grid production case deduction according to an embodiment of the present application. DETAILED DESCRIPTION

[0025] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. 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 exemplary embodiments described herein.

[0026] The power grid is a complex system comprised of power generation, transmission, and distribution, and its operational status is affected by numerous factors. To ensure that grid operators can appropriately handle various emergencies, they require rigorous professional training. Simulations of power grid production scenarios are a key method for assessing the grid's emergency response capabilities, helping to uncover potential issues and promote improvements.

[0027] However, traditional production case drills often rely on fixed scripts and preset outcomes, limiting their flexibility. This limitation means that drills may not cover all potential emergency situations, and operators may become reliant on expected outcomes, lacking the ability to respond to unforeseen emergencies. Furthermore, traditional drills typically cover only a few typical emergency scenarios, limiting their diversity and complexity and failing to fully simulate the diverse and complex challenges that the power grid may face.

[0028] Based on this, this application proposes an intelligent interactive method for power grid production case deduction. Figure 1 FIG. 1 is a flow chart of an intelligent interactive method for power grid production case deduction according to an embodiment of the present application. Figure 1 As shown, the intelligent interactive method for power grid production case deduction according to the embodiment of the present application includes: Step 1: in response to receiving an instruction to start a power grid production case deduction simulation, the system automatically triggers a first event; Step 2: receiving operation data input by a user; Step 3: based on the operation data, the system updates the power grid status in real time; Step 4: the system automatically triggers a second event, and cyclically executes Step 2 and Step 3 until all events are triggered; Step 5: the system continuously collects the user's operation data, and analyzes the user's operation data based on predefined evaluation indicators to generate a feedback report.

[0029] In step 1, in response to receiving an instruction to start a power grid production case simulation, the system automatically triggers the first event. It should be understood that when the power grid operator decides to conduct a production case simulation, they will send a startup instruction through the command line tool. After receiving this instruction, the system will immediately respond and trigger the preset "first event". This first event is usually an initial state or emergency event in the simulation scenario, which is used to start the entire simulation process. For example, a "transmission line failure" scenario in a power grid is being simulated. When the system receives the startup instruction, it will automatically trigger the first event - "a short circuit fault occurs in section A of the transmission line". This event will immediately change the current state of the power grid, such as causing a power outage in the relevant area and triggering the action of the protection device.

[0030] In step 2, user-entered operational data is received. It should be understood that during the simulation, grid operators will need to enter operational data based on the current grid status and event prompts. This operational data may include selecting a specific emergency response plan, adjusting equipment parameters, switching to a backup power source, and so on. The system receives this operational data through the user interface and prepares it for subsequent processing and grid status updates.

[0031] In step 3, the system updates the grid status in real time based on the operational data. Based on the operational data entered by the user, the system calculates and updates the current grid status in real time. By updating the grid status in real time, the system can simulate a more realistic and dynamic simulation environment, helping users better understand and respond to various emergency situations.

[0032] Specifically, Figure 2 FIG. 3 is a flow chart of step 3 in the intelligent interaction method for power grid production case deduction according to an embodiment of the present application. Figure 2 As shown, in step 3, it includes: step 31, based on the operation data and the first event, performing a potential risk analysis on the operation data to obtain an analysis result, and the analysis result is used to indicate whether there is a potential danger; step 32, in response to the analysis result indicating that there is a potential danger, the system issues a warning prompt.

[0033] It's understandable that analyzing operational data for potential risks can proactively identify situations that could lead to grid instability or failure, providing users with real-time feedback to help them better understand the impact of their current actions and make better decisions. However, traditional case-based exercises typically conduct evaluation and feedback only after the exercise is complete, failing to provide immediate feedback. This means users can't immediately determine whether their actions were correct, nor can they immediately correct errors.

[0034] Accordingly, in performing a potential risk analysis on the operation data based on the operation data and the first event, the technical concept of the present application is to use an artificial intelligence-based data analysis and processing algorithm to embed the operation data and semantically encode the text description of the first event, so as to intelligently obtain the analysis results based on the multi-scale interactive response between the embedded coding features of the operation data and the semantic representation features of the first event. In this way, the operation data can be analyzed for potential risks in real time and instant feedback can be provided. This means that users can immediately understand whether their operations are correct during the rehearsal and make adjustments based on the feedback.

[0035] In step 31, based on the operation data and the first event, a potential risk analysis is performed on the operation data to obtain an analysis result, wherein the analysis result is used to indicate whether there is a potential danger. Specifically, Figure 3 FIG3 is a flow chart of step 31 in the intelligent interaction method for power grid production case deduction according to an embodiment of the present application. Figure 3 As shown, in step 31, it includes: step 311, using the operation data embedding coding matrix to embed the operation data to obtain a high-dimensional embedding coding vector of the operation data; step 312, extracting the text description of the first event; step 313, using a semantic encoder including a Bert model to semantically encode the text description of the first event to obtain a first event semantic information representation vector; step 314, performing feature decoupled local-global interaction response analysis on the operation data high-dimensional embedding coding vector and the first event semantic information representation vector to obtain an operation data-first event multi-scale interaction response matrix; step 315, obtaining the analysis result based on the operation data-first event multi-scale interaction response matrix.

[0036] In step 311, the operation data is embedded and encoded using the operation data embedding encoding matrix to obtain a high-dimensional embedded encoding vector for the operation data. Accordingly, considering that the operation data may be multidimensional and may have highly nonlinear semantic associations, based on this, in order to map this high-dimensional data into a lower-dimensional continuous space while preserving the relative distances and structural information between the data, so as to better capture the complex relationships and patterns in the data, the technical solution of this application uses the operation data embedding encoding matrix to embed the operation data to obtain a high-dimensional embedded encoding vector for the operation data.

[0037] In step 312, a text description of the first event is extracted. It should be understood that the text description of the first event includes information such as the event type and detailed event information. By extracting the text description of the first event and analyzing it in detail in subsequent steps, the nature and potential risk of the event can be more accurately assessed, thereby improving the efficiency and safety of power grid emergency response.

[0038] In step 313, a semantic encoder including a Bert model is used to semantically encode the text description of the first event to obtain a first event semantic information representation vector. It should be understood that the text description of the first event contains semantic information between each word and semantic associations between contexts. Considering that the Bert model is a deep learning model based on the Transformer architecture, it can perform bidirectional encoding on input data while considering contextual information. Based on this, a semantic encoder including a Bert model is used to semantically encode the text description of the first event to capture and extract the true semantic meaning contained in the text description, thereby obtaining a first event semantic information representation vector.

[0039] In step 314, a feature-decoupled local-global interactive response analysis is performed on the high-dimensional embedded coding vector of the operation data and the first event semantic information representation vector to obtain an operation data-first event multi-scale interactive response matrix. Accordingly, considering that the high-dimensional embedded coding vector of the operation data is a semantic feature representation of the user operation data, it reflects the important features and mutual relationships in the operation data. The first event semantic information representation vector captures the semantic information and contextual relationship in the event description. There is a mutual interactive response relationship between the two. Therefore, in order to obtain the analysis result more accurately, in the technical solution of the present application, a feature-decoupled local-global interactive response analysis is performed on the high-dimensional embedded coding vector of the operation data and the first event semantic information representation vector to obtain an operation data-first event multi-scale interactive response matrix.

[0040] Specifically, the first event semantic information representation vector is feature decoupled to segment it into a sequence of first-event local semantic information representation vectors containing multiple local semantic information, providing a foundation for subsequent fine-grained analysis and association. Next, the high-dimensional embedding code vector of the operation data is nonlinearly transformed to capture and extract more complex information from the operation data, resulting in a nonlinearly transformed high-dimensional embedding code vector for the operation data. Next, to more fine-grainedly analyze the association and interaction between each first-time local semantic feature and the embedded feature of the operation data, an association matrix is calculated between each first-event local semantic information representation vector and the nonlinearly transformed high-dimensional embedding code vector of the operation data to obtain a sequence of operation data-first-event semantic association matrices. The feature interaction energy factor of each association matrix is then calculated to obtain a sequence of operation data-first-event semantic feature interaction energy factors. In other words, the feature interaction energy factor helps identify and strengthen the semantic feature interactions between the operation data and the first event that are most critical for model prediction. Each feature interaction energy factor is then normalized to fall within the (0, 1) interval and converted into an attention weight. In particular, normalization ensures weight balance, allowing the model to more appropriately allocate attention to interactions between different features. Finally, the normalized factor weights are used to weight each action data-first event semantic association matrix and sum them to obtain the action data-first event multi-scale interaction response matrix, which integrates local features and multi-scale interaction information.

[0041] Specifically, Figure 4 FIG3 is a flow chart of step 314 in the intelligent interaction method for power grid production case deduction according to an embodiment of the present application. Figure 4As shown, in step 314, it includes: step 3141, performing feature decoupling on the first event semantic information representation vector to obtain a sequence of first event local semantic information representation vectors; step 3142, performing nonlinear transformation on the high-dimensional embedding coding vector of the operation data to obtain a high-dimensional embedding coding vector of the operation data after nonlinear transformation; step 3143, calculating the association matrix between the high-dimensional embedding coding vector of the operation data after nonlinear transformation and each first event local semantic information representation vector in the sequence of the first event local semantic information representation vector to obtain a sequence of operation data-first event semantic association matrices; step 3144, calculating the operation data-first The characteristic interaction energy factors of each operation data-first event semantic association matrix in the sequence of event semantic association matrices are used to obtain a sequence of operation data-first event semantic feature interaction energy factors; step 3145, normalizing the sequence of operation data-first event semantic feature interaction energy factors to obtain a sequence of normalized operation data-first event semantic feature interaction energy factors; step 3146, using the sequence of normalized operation data-first event semantic feature interaction energy factors as a sequence of weights, calculating the position-weighted sum of the sequence of operation data-first event semantic association matrices to obtain the operation data-first event multi-scale interaction response matrix.

[0042] More specifically, in an embodiment of the present application, a nonlinear transformation is performed on the high-dimensional embedded coding vector of the operation data to obtain the high-dimensional embedded coding vector of the operation data after the nonlinear transformation, including: multiplying the high-dimensional embedded coding vector of the operation data with the operation data transformation matrix, and then adding the obtained operation data transformation vector and the operation data bias vector by position to obtain the high-dimensional embedded coding vector of the operation data after the nonlinear transformation.

[0043] More specifically, in an embodiment of the present application, the association matrix between the high-dimensional embedded coding vector of the operation data after the nonlinear transformation and each first event local semantic information representation vector in the sequence of the first event local semantic information representation vector is calculated to obtain a sequence of operation data-first event semantic association matrices, including: vector multiplying the transpose vector of the high-dimensional embedded coding vector of the operation data after the nonlinear transformation and each first event local semantic information representation vector in the sequence of the first event local semantic information representation vector to obtain the sequence of the operation data-first event semantic association matrices.

[0044] More specifically, in an embodiment of the present application, the characteristic interaction energy factors of each operation data-first event semantic association matrix in the sequence of the operation data-first event semantic association matrix are calculated to obtain a sequence of operation data-first event semantic characteristic interaction energy factors, including: respectively calculating the mean and variance of the operation data-first event semantic association matrix to obtain the operation data-first event semantic association mean and the operation data-first event semantic association variance; calculating the positional difference between the operation data-first event semantic association matrix and the operation data-first event semantic association mean to obtain the operation data-first event semantic association difference matrix; calculating the fourth power of each eigenvalue in the operation data-first event semantic association difference matrix to obtain the operation data-first event semantic association modulation difference matrix; calculating the expected value of the operation data-first event semantic association modulation difference matrix to obtain the operation data-first event semantic association expected value; and adding the operation data-first event semantic association expected value to the matrix. Divide it by the square of the operation data-first event semantic association variance to obtain the operation data-first event semantic association kurtosis value; calculate the sum of the operation data-first event semantic association variance and the hyperparameter to obtain the operation data-first event semantic association first energy factor; calculate the square of the difference between the operation data-first event semantic association kurtosis value and the operation data-first event semantic association mean to obtain the operation data-first event semantic association difference value; add the value obtained by multiplying the operation data-first event semantic association variance by a constant two and the value obtained by multiplying the hyperparameter by a constant two to the operation data-first event semantic association difference value to obtain the operation data-first event semantic association second energy factor; divide the operation data-first event semantic association first energy factor by the operation data-first event semantic association second energy factor to obtain the operation data-first event semantic feature interaction energy factor corresponding to the operation data-first event semantic association matrix.

[0045] More specifically, in an embodiment of the present application, the sequence of the operation data-first event semantic feature interaction energy factors is normalized to obtain a sequence of normalized operation data-first event semantic feature interaction energy factors, including: using a Sigmoid function to process the sequence of the operation data-first event semantic feature interaction energy factors to obtain the sequence of normalized operation data-first event semantic feature interaction energy factors.

[0046] In the embodiment of the present application, specifically, a feature-decoupled local-global interaction response analysis is performed on the high-dimensional embedded coding vector of the operation data and the first event semantic information representation vector to obtain an operation data-first event multi-scale interaction response matrix, which can be expressed as:

[0047] Decouple(f2)={f 21 ,f 22 ,...,f 2i ,...,f 2n}

[0048] f 1t =Wf1+b

[0049]

[0050] E norm,12i =sigmoid(E 12i )

[0051] M=∑ i M 12i ⊙E norm,12i

[0052] Where f2 is the semantic information representation vector of the first event, Decouple(·) is the feature decoupling operation, and f 21 ,f 22 ,...,f 2i ,...,f 2n are the 1st, 2nd, ..., ith, ..., nth first event local semantic information representation vectors in the sequence of the first event local semantic information representation vectors, n is the number of feature vectors in the sequence of the first event local semantic information representation vectors, f1 is the high-dimensional embedding coding vector of the operation data, W and b are the operation data transformation matrix and the operation data bias vector, respectively, f 1t is the high-dimensional embedding coding vector of the operation data after the nonlinear transformation, f 1t T is f 1t The transposed vector of Represents vector multiplication, M 12i is the operation data-first event semantic association matrix between the local semantic information representation vector of the i-th first event and the high-dimensional embedding encoding vector of the operation data after the nonlinear transformation, M 12i (t) is M 12i The eigenvalues at each position in 12i It's M 12i The mean of the matrix, E(·) is the expected value of the matrix, σ 12i 2 It's M 12i The variance of k 12i M 12i The kurtosis value, λ is a hyperparameter, E 12i It's M 12i The corresponding operation data-first event semantic feature interaction energy factor, sigmoid(·) is the sigmoid function, Enorm,12i It's M 12i The corresponding normalized operation data-first event semantic feature interaction energy factor, ⊙ is the position point multiplication, and M is the operation data-first event multi-scale interaction response matrix.

[0053] In step 315, the analysis result is obtained based on the operation data-first event multi-scale interaction response matrix. Specifically, in an embodiment of the present application, the analysis result is obtained based on the operation data-first event multi-scale interaction response matrix, including: inputting the operation data-first event multi-scale interaction response matrix into a potential risk analyzer based on a classifier to obtain the analysis result. That is, the operation data-first event multi-scale interaction response matrix obtained by interactive response using the high-dimensional embedded coding vector of the operation data and the first event semantic information representation vector is classified and processed, so as to intelligently obtain the analysis result. In this way, the potential risk analysis of the operation data can be performed in real time, and immediate feedback can be provided. This means that users can immediately understand whether their operations are correct during the rehearsal process and make adjustments based on the feedback.

[0054] In the technical solution of the present application, the high-dimensional embedding coding vector of the operation data and the first event semantic information representation vector represent the operation time semantic embedding coding features and the first time semantic information coding features. When the feature local-global interaction response based on feature decoupling is input thereto, the significant differences in feature dimensions and feature modalities will cause the local-global interaction response to have a local-global responsiveness imbalance, affecting the classification judgment accuracy of the analysis results obtained by the potential risk analyzer based on the classifier.

[0055] Based on this, in a preferred example of the present application, in the process of inputting the operation data-first event multi-scale interaction response matrix into a classifier-based potential risk analyzer to obtain the analysis result, the operation data-first event multi-scale interaction response matrix is first subjected to local-global equilibrium modulation to obtain an optimized operation data-first event multi-scale interaction response vector. The specific process includes the following steps:

[0056] Expanding the operation data-first event multi-scale interaction response matrix into an operation data-first event multi-scale interaction response vector;

[0057] The intrinsic stability and change tendency of each set of eigenvalue pairs in the multi-scale interactive response vector of the operation data-first event are quantified to obtain the interactive response inherent structure constraint characterization matrix and the interactive response space dimension diffusion trade-off matrix, which are expressed as:

[0058]

[0059] v i,v j ∈V

[0060] Wherein, V represents the multi-scale interactive response vector of the operation data-first event, v i and v j They represent the eigenvalues of the i-th and j-th positions of the multi-scale interaction response vector of the operation data-first event, w1, w2, w3, and w4 represent different weight hyperparameters. represents the value at position (i, j) in the interaction response inherent structural constraint characterization matrix, Represents the value at position (i, j) in the diffusion trade-off matrix of the interaction response space dimension.

[0061] Based on the interaction response inherent structure constraint characterization matrix, the core trend capture of the operation data-first event multi-scale interaction response vector is performed to obtain the interaction response orthogonal projection base vector, which is expressed as:

[0062]

[0063] Among them, Sigmoid represents the activation function, M1 and b1 represent the first weight matrix and the first bias vector, represents matrix multiplication, and V1 represents the interaction response orthogonal projection basis vector.

[0064] Based on the interactive response space dimensional diffusion trade-off matrix, deviation and compensation information mining is performed on the operation data-first event multi-scale interactive response vector to obtain an interactive response projection compensation residual vector, which is expressed as:

[0065]

[0066] Wherein, M2 and b2 represent the second weight matrix and the second bias vector, and V2 represents the interaction response projection compensation residual vector.

[0067] A synergistic variation pattern explicit model between the interactive response orthogonal projection base vector and the interactive response projection compensation residual vector is constructed to obtain a heterogeneous co-action map of the interactive response projection components, which is expressed as:

[0068]

[0069] v 1i ∈V1

[0070] v 2j ∈V2

[0071] Among them, v 1i represents the eigenvalue of the i-th position of the orthogonal projection basis vector of the interaction response, v 2jrepresents the eigenvalue of the j-th position of the interaction response projection compensation residual vector, d(v 1i ,v 2j ) represents the calculation of v 1i and v 2j The absolute value of the difference between the two, softmax represents the normalized exponential function, D i,j Represents the value of the (i, j) position in the heterogeneous co-action map of the interactive response projection component.

[0072] After feature fusion of the interaction response orthogonal projection base vector and the interaction response projection compensation residual vector, they are mapped to the atlas space of the interaction response projection component heterogeneous interaction map to obtain the optimized operation data-first event multi-scale interaction response vector, which is expressed as:

[0073]

[0074] Among them, W r and b represent the third weight matrix and the third bias vector, [V1, V2] represents the cascading of V1 and V2, D represents the heterogeneous co-action map of the interaction response projection component, and V′ represents the optimized operation data-first event multi-scale interaction response vector.

[0075] Specifically, based on the distribution potential field description of each eigenvalue of the multi-scale interactive response vector of the operational data-first event, which is analyzed through the inherent structural constraints determined based on linear and nonlinear statistical analysis, a representative sampling point sampling framework is used to correlate and analyze the orthogonal heterogeneity between local structural features and the characteristic dimensions of the eigenvector. Thus, the expected category offset threshold is calibrated with the help of autocorrelation to establish the basis for how the structural element correlation corresponds to the specified steady-state classification criteria, thereby enhancing the ability to determine the steady-state classification through the configuration of source elements and achieving concurrent multi-axis calibration from local structural features to the overall system behavior. In this way, the balanced characteristics of the local-global responsiveness of the multi-scale interactive response features of the operational data-first event are improved, thereby improving the classification judgment accuracy of the analysis results obtained by the classifier-based potential risk analyzer.

[0076] In step S32, in response to the analysis result indicating the presence of a potential hazard, the system issues a warning. It should be understood that warnings can instantly convey information about potential hazards discovered during the current simulation to power grid operators. This immediate feedback mechanism helps operators quickly identify problems and avoid delayed responses due to information lags. The issuance of warnings is also recorded by the system, serving as an important basis for subsequent analysis of the simulation process and evaluation of operator performance. By analyzing this data, the simulation system's algorithms and models can be further improved, enhancing its ability to identify potential hazards and the accuracy of early warnings.

[0077] In step 4, the system automatically triggers the second event and loops through steps 2 and 3 until all events have been triggered. It should be understood that as the simulation progresses, the system automatically triggers subsequent events based on the current grid state and pre-set simulation logic. These subsequent events may be chain reactions triggered by user actions or automatically generated by the system based on pre-set conditions. The system loops through steps 2 and 3 until all pre-set events have been triggered and completed. For example, in the scenario of a "short circuit fault in transmission line section A," the system might automatically trigger the second event—"backup power supply overload protection activated." This is because the backup power supply may reach its capacity limit when carrying additional load, triggering overload protection. At this point, the system again requests user input to address this new situation (such as further adjusting load distribution, activating additional backup power sources, etc.) and updates the grid state in real time to reflect these changes. This process continues until all pre-set events have been successfully triggered and completed.

[0078] In step 5, the system continuously collects user operational data and analyzes it based on predefined evaluation metrics to generate a feedback report. It should be understood that throughout the simulation process, the system continuously collects user operational data and analyzes it based on predefined evaluation metrics. These metrics may include response time, operational accuracy, resource utilization efficiency, etc. By analyzing this data, the system can assess the user's performance in the emergency situation and generate a detailed feedback report. This report will help users understand their strengths and weaknesses, allowing for subsequent improvements and enhancements.

[0079] The intelligent interactive method for power grid production case simulations provided in this application can simulate more diverse and complex emergency scenarios by dynamically adjusting event triggers and updating the grid status in real time. Furthermore, by dynamically adjusting event triggers based on user-entered operational data, a wider range of emergency scenarios can be simulated, including unconventional and complex situations. This makes the drills more comprehensive and covers a wider range of possible emergency scenarios.

[0080] In summary, an intelligent interactive method for power grid production case deduction based on an embodiment of the present application is illustrated, which uses an artificial intelligence-based data analysis and processing algorithm to embed and encode the operation data input by the user, and semantically encode the text description of the first event automatically triggered by the system, so as to intelligently obtain the analysis results based on the multi-scale interactive response between the embedded coding features of the operation data and the semantic representation features of the first event. In this way, the potential risk analysis of the operation data can be performed in real time, and instant feedback can be provided. This means that users can immediately understand whether their operations are correct during the rehearsal and make adjustments based on the feedback.

[0081] Figure 5 FIG is a system block diagram of an intelligent interactive system for power grid production case deduction according to an embodiment of the present application. Figure 5 As shown, according to an embodiment of the present application, the intelligent interactive system 100 for power grid production case deduction includes: a simulation instruction event response module 110, which is used to automatically trigger a first event in response to receiving an instruction to start a simulation of a power grid production case deduction; an operation data acquisition module 120, which is used to receive operation data input by a user; a power grid status real-time update module 130, which is used to update the power grid status in real time based on the operation data; an event loop triggering module 140, which is used to automatically trigger a second event in the system, and cyclically execute the operation data acquisition module and the power grid status real-time update module until all events are triggered; a feedback report generation module 150, which is used for the system to continuously collect the user's operation data, and analyze the user's operation data based on predefined evaluation indicators to generate a feedback report.

[0082] Here, those skilled in the art will appreciate that the specific functions and operations of the various units and modules in the above-mentioned intelligent interactive system 100 for power grid production case deduction have been described in the above reference. Figures 1 to 4 The description of the intelligent interaction method for power grid production case deduction is introduced in detail, and therefore, its repeated description will be omitted.

[0083] In summary, the intelligent interactive system 100 for power grid production case deduction based on the embodiment of the present application is explained, which adopts an artificial intelligence-based data analysis and processing algorithm to embed the operation data input by the user and semantically encode the text description of the first event automatically triggered by the system, so as to intelligently obtain the analysis results based on the multi-scale interactive response between the embedded coding characteristics of the operation data and the semantic representation characteristics of the first event. In this way, the potential risk analysis of the operation data can be performed in real time, and instant feedback can be provided. This means that users can immediately understand whether their operations are correct during the rehearsal and make adjustments based on the feedback.

[0084] As described above, the intelligent interactive system 100 for power grid production case deduction according to the embodiment of the present application can be implemented in various wireless terminals, such as a server for intelligent interaction for power grid production case deduction. In one example, the intelligent interactive system 100 for power grid production case deduction according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or hardware module. For example, the intelligent interactive system 100 for power grid production case deduction can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the intelligent interactive system 100 for power grid production case deduction can also be one of the many hardware modules of the wireless terminal.

[0085] Alternatively, in another example, the intelligent interactive system 100 for power grid production case deduction and the wireless terminal may also be separate devices, and the intelligent interactive system 100 for power grid production case deduction may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

Claims

1. An intelligent interactive method for power grid production case deduction, characterized by: include: Step 1: In response to receiving an instruction to start a power grid production case simulation, the system automatically triggers a first event; Step 2: receiving operation data input by the user; Step 3: Based on the operation data, the system updates the grid status in real time; Step 4: The system automatically triggers the second event and executes steps 2 and 3 repeatedly until all events are triggered. Step 5: The system continuously collects the user's operation data and analyzes the user's operation data based on predefined evaluation indicators to generate a feedback report.

2. The intelligent interactive method for power grid production case deduction according to claim 1 is characterized in that: The step 3 further includes: Based on the operation data and the first event, performing a potential risk analysis on the operation data to obtain an analysis result, wherein the analysis result is used to indicate whether there is a potential danger; In response to the analysis result indicating that there is a potential danger, the system issues a warning prompt.

3. The intelligent interactive method for power grid production case deduction according to claim 2 is characterized in that: Based on the operation data and the first event, performing a potential risk analysis on the operation data to obtain an analysis result, including: Embedding the operation data using an operation data embedding coding matrix to obtain a high-dimensional embedding coding vector of the operation data; extracting a text description of the first event; Using a semantic encoder including a Bert model, semantically encode the text description of the first event to obtain a first event semantic information representation vector; Performing feature-decoupled local-global interactive response analysis on the high-dimensional embedding coding vector of the operation data and the first event semantic information representation vector to obtain an operation data-first event multi-scale interactive response matrix; The analysis result is obtained based on the operation data-first event multi-scale interaction response matrix.

4. The intelligent interactive method for power grid production case deduction according to claim 3 is characterized in that: Performing feature-decoupled local-global interactive response analysis on the high-dimensional embedding coding vector of the operation data and the first event semantic information representation vector to obtain an operation data-first event multi-scale interactive response matrix, including: Performing feature decoupling on the first event semantic information representation vector to obtain a sequence of first event local semantic information representation vectors; Performing a nonlinear transformation on the high-dimensional embedding coding vector of the operation data to obtain a high-dimensional embedding coding vector of the operation data after nonlinear transformation; Calculating an association matrix between the high-dimensional embedding coding vector of the operation data after the nonlinear transformation and each first event local semantic information representation vector in the sequence of the first event local semantic information representation vectors to obtain a sequence of operation data-first event semantic association matrices; Calculating the feature interaction energy factor of each operation data-first event semantic association matrix in the sequence of the operation data-first event semantic association matrix to obtain a sequence of operation data-first event semantic feature interaction energy factors; Normalizing the sequence of the operation data-first event semantic feature interaction energy factors to obtain a sequence of normalized operation data-first event semantic feature interaction energy factors; Using the sequence of normalized operation data-first event semantic feature interaction energy factors as a sequence of weights, the position-weighted sum of the sequence of operation data-first event semantic association matrices is calculated to obtain the operation data-first event multi-scale interaction response matrix.

5. The intelligent interactive method for power grid production case deduction according to claim 4 is characterized in that: The nonlinear transformation is performed on the high-dimensional embedding coding vector of the operation data to obtain the high-dimensional embedding coding vector of the operation data after the nonlinear transformation, including: multiplying the high-dimensional embedding coding vector of the operation data with the operation data transformation matrix, and adding the obtained operation data transformation vector and the operation data bias vector according to position to obtain the high-dimensional embedding coding vector of the operation data after the nonlinear transformation.

6. The intelligent interactive method for power grid production case deduction according to claim 5 is characterized in that: Calculating the association matrix between the high-dimensional embedding coding vector of the operation data after the nonlinear transformation and each first event local semantic information representation vector in the sequence of the first event local semantic information representation vector to obtain a sequence of operation data-first event semantic association matrices, including: vector multiplying the transpose vector of the high-dimensional embedding coding vector of the operation data after the nonlinear transformation and each first event local semantic information representation vector in the sequence of the first event local semantic information representation vector to obtain the sequence of operation data-first event semantic association matrices.

7. The intelligent interactive method for power grid production case deduction according to claim 6 is characterized in that: Calculating the feature interaction energy factor of each operation data-first event semantic association matrix in the sequence of the operation data-first event semantic association matrix to obtain a sequence of operation data-first event semantic feature interaction energy factors includes: respectively calculating the mean and variance of the operation data-first event semantic association matrix to obtain the operation data-first event semantic association mean and the operation data-first event semantic association variance; Calculating the positional difference between the operation data-first event semantic association matrix and the operation data-first event semantic association mean to obtain an operation data-first event semantic association difference matrix; Calculating the fourth power of each eigenvalue in the operation data-first event semantic association difference matrix to obtain an operation data-first event semantic association modulation difference matrix; Calculating the expected value of the operation data-first event semantic association modulation difference matrix to obtain the operation data-first event semantic association expected value; Dividing the expected value of the operation data-first event semantic association by the square of the operation data-first event semantic association variance to obtain a kurtosis value of the operation data-first event semantic association; Calculating the sum of the variance of the operation data-first event semantic association and a hyperparameter to obtain a first energy factor of the operation data-first event semantic association; Calculating the square of the difference between the kurtosis value of the operation data-first event semantic association and the mean value of the operation data-first event semantic association to obtain a difference value of the operation data-first event semantic association; A second energy factor of the operation data-first event semantic association is obtained by adding a value obtained by multiplying the operation data-first event semantic association variance by a constant of two and a value obtained by multiplying the hyperparameter by a constant of two to the operation data-first event semantic association difference value; The first energy factor of the operation data-first event semantic association is divided by the second energy factor of the operation data-first event semantic association to obtain the operation data-first event semantic feature interaction energy factor corresponding to the operation data-first event semantic association matrix.

8. The intelligent interactive method for power grid production case deduction according to claim 7 is characterized in that: The sequence of the operation data-first event semantic feature interaction energy factors is normalized to obtain a sequence of normalized operation data-first event semantic feature interaction energy factors, including: using a Sigmoid function to process the sequence of the operation data-first event semantic feature interaction energy factors to obtain the sequence of normalized operation data-first event semantic feature interaction energy factors.

9. The intelligent interactive method for power grid production case deduction according to claim 8, characterized in that: Obtaining the analysis result based on the operation data-first event multi-scale interaction response matrix includes: inputting the operation data-first event multi-scale interaction response matrix into a classifier-based potential risk analyzer to obtain the analysis result.

10. An intelligent interactive system for power grid production case deduction, characterized by: include: A simulation instruction event response module is configured to automatically trigger a first event in response to receiving an instruction to start a power grid production case deduction simulation; An operation data collection module, used for receiving operation data input by a user; A real-time grid status update module, configured to update the grid status in real time based on the operation data; An event loop triggering module, configured to automatically trigger a second event in the system, and cyclically execute the operation data acquisition module and the grid status real-time updating module until all events are triggered; The feedback report generation module is used for the system to continuously collect user operation data and analyze the user operation data based on predefined evaluation indicators to generate a feedback report.

Images