A power grid planning scenario generation method, device and equipment based on a rule-based thinking chain, medium and product

Abstract

The application discloses a power grid planning scene generation method and device based on a rule thinking chain, equipment, a medium and a product, relates to the technical field of power systems, and comprises the following steps: acquiring a text data set and a power grid working condition data set; performing structural analysis on the text data set to obtain a planning boundary element set; performing constraint construction based on the planning boundary element set to obtain a rule boundary thinking chain; performing variable mapping on the rule boundary thinking chain to obtain a rule boundary thinking chain with completed scene variable mapping assignment and corresponding scene disturbance operators; performing scene generation based on the power grid working condition data set, the rule boundary thinking chain with completed scene variable mapping assignment and the scene disturbance operators to obtain a candidate scene; and performing simulation verification and multi-dimensional index screening on the candidate scene to obtain a power grid planning target scene library. The method can generate a power grid planning scene that meets planning boundary requirements.

Description

Technical Field

[0001] This application relates to the field of power system technology, and in particular to a method, device, equipment, medium and product for generating power grid planning scenarios based on procedural thinking chains. Background Technology

[0002] Against the backdrop of new power system construction, power grid planning faces multiple challenges, including the integration of high-proportion renewable energy sources, flexible participation of energy storage and demand response resources, and the impact of extreme weather. When conducting grid structure optimization, renewable energy transmission scheme design, load development adaptability analysis, and safety and stability verification, power grid planners and designers need to construct a database of typical scenarios that represent the annual operating boundaries, resource structure, and risk characteristics of the planned scenario. The coverage and quality of this database directly affect the safety, economy, and adaptability of the planning scheme.

[0003] Currently, the construction of power grid planning scenarios mainly adopts data-driven methods, including scenario clustering and typical day selection methods based on historical operating data, scenario simulation methods based on load and renewable energy output forecasts, and scenario construction methods that consider source-load uncertainties. For example, related technologies use principal component analysis to reduce the dimensionality of generator nodes and load nodes, and use clustering algorithms to select real operating scenarios as typical scenarios; other solutions simulate the main behavior of the power system, extract basic operating modes from historical operating data, combine them with predictive models to generate future operating data, and then screen scenarios covering different renewable energy penetration rates, load levels, and the impact of extreme weather to construct a typical scenario library for power grid planning.

[0004] However, the scenarios generated by the aforementioned existing technologies mainly rely on historical operational data features and predicted data features, which cannot guarantee the correspondence between the constructed scenario library and the boundary requirements in the planning documents and technical guidelines. As a result, the generated scenario library cannot directly meet the requirements of the planning scheme demonstration for scenario boundary conditions. Summary of the Invention

[0005] This application provides a method, apparatus, equipment, medium, and product for generating power grid planning scenarios based on procedural thinking chains, which can generate power grid planning scenarios that meet the planning boundary requirements.

[0006] To achieve the above objectives, this application adopts the following technical solution: Firstly, this application provides a method for generating power grid planning scenarios based on procedural thinking chains, including: Obtain text-based datasets and power grid operating condition datasets; The text dataset is structured and parsed to obtain a set of planning boundary elements; constraints are then constructed based on the set of planning boundary elements to obtain the procedure boundary thought chain. By performing variable mapping on the procedure boundary thinking chain, we obtain the procedure boundary thinking chain and the corresponding scene perturbation operator after completing the scene variable mapping assignment. Based on the power grid operating condition dataset, the procedural boundary thinking chain that completes the scene variable mapping and assignment, and the scene disturbance operator, scene generation is performed to obtain candidate scenes. The candidate scenarios are simulated and verified to obtain valid candidate scenarios; the valid candidate scenarios are then screened using multi-dimensional indicators to obtain a target scenario library for power grid planning.

[0007] In some possible implementations, variable mapping is performed on the procedure boundary thinking chain to obtain the procedure boundary thinking chain with completed scene variable mapping assignment and the corresponding scene perturbation operator, including: Using the constraint variables in the procedure boundary thinking chain as search conditions, the target mapping variable is determined by searching in the preset planning variable dictionary; the target mapping variable is written into the scenario variable mapping bit in the procedure boundary thinking chain to obtain the procedure boundary thinking chain with completed scenario variable mapping assignment; the disturbance action in the procedure boundary thinking chain is bound to the written target mapping variable to obtain the scenario disturbance operator.

[0008] In some possible implementations, scenario generation is performed based on the power grid operating condition dataset, the procedural boundary thinking chain that completes the scenario variable mapping assignment, and the scenario perturbation operator, resulting in candidate scenarios, including: Based on the power grid operating condition dataset, scenario clustering is performed to obtain basic scenarios. The scenario parameters of each basic scenario are substituted into the Boolean trigger conditions in the procedural boundary thinking chain that completes the scenario variable mapping assignment for logical evaluation to obtain the evaluation results. The thinking chain in the enabled state is determined based on the evaluation results. According to the scenario variable mapping bits and scenario perturbation operators corresponding to the thinking chain in the enabled state, the corresponding basic scenarios are subjected to variable perturbation processing to obtain perturbed intermediate scenarios. The perturbed intermediate scenarios are subjected to scenario derivation processing to obtain the corresponding derived scenarios. The basic scenarios, perturbed intermediate scenarios, and derived scenarios are used as candidate scenarios.

[0009] In some possible implementations, the text dataset is structured and parsed to obtain a set of planning boundary features, including: The text dataset is sliced ​​into document slices to obtain knowledge slices; each knowledge slice is classified into sentences, and the sentences to be extracted are obtained based on the classification results; boundary elements are extracted from the sentences to be extracted to obtain original boundary elements; the original boundary elements are validated for field validity to obtain validated original boundary elements; the validated original boundary elements are aggregated to obtain the planning boundary element set.

[0010] In some possible implementations, candidate scenarios are simulated and verified to obtain valid candidate scenarios, including: Each candidate scenario is written into the power grid simulation model and verified by power flow calculation to obtain candidate scenarios that pass the power flow calculation verification; the candidate scenarios that pass the power flow calculation verification are then verified by safety and stability to obtain candidate scenarios that pass the safety and stability verification; the candidate scenarios that pass the safety and stability verification are determined as valid candidate scenarios.

[0011] In some possible implementations, multi-dimensional indicators are used to screen effective candidate scenarios to obtain a database of target scenarios for power grid planning, including: Calculate the multidimensional evaluation index of each valid candidate scenario to obtain the comprehensive score corresponding to each valid candidate scenario; based on the mandatory coverage constraint and the comprehensive score, screen each valid candidate scenario to obtain the target scenario library for power grid planning.

[0012] Secondly, this application provides a power grid planning scenario generation device based on procedural thinking chain, comprising: The acquisition module is used to acquire text-based datasets and power grid operating condition datasets. The data processing module is used to perform structured parsing on text datasets to obtain a set of planning boundary elements; to construct constraints based on the set of planning boundary elements to obtain a procedure boundary thinking chain; and to perform variable mapping on the procedure boundary thinking chain to obtain a procedure boundary thinking chain with completed scene variable mapping assignment and the corresponding scene perturbation operator. The scenario generation module is used to generate scenarios based on the power grid operating condition dataset, the procedural boundary thinking chain that completes the scenario variable mapping assignment, and the scenario disturbance operator to obtain candidate scenarios; to perform simulation verification on the candidate scenarios to obtain valid candidate scenarios; and to screen the valid candidate scenarios using multi-dimensional indicators to obtain the power grid planning target scenario library.

[0013] Thirdly, this application provides a computing device, including a memory and a processor; The memory stores one or more computer programs, the one or more computer programs including instructions; when the instructions are executed by the processor, the computing device performs the method as described in any one of the first aspects.

[0014] Fourthly, this application provides a computer-readable storage medium for storing a computer program for performing the method as described in any one of the first aspects.

[0015] Fifthly, this application provides a computer program product comprising one or more computer instructions, wherein when the computer instructions are executed by a computer, the computer performs the method as described in any one of the first aspects.

[0016] As can be seen from the above technical solution, this application has at least the following beneficial effects: In this application, a text-based dataset and a power grid operating condition dataset are acquired. The text-based dataset is a static, unstructured knowledge dataset used for power grid planning, while the power grid operating condition dataset is a structured operating condition dataset used for power grid planning analysis. Then, the text-based dataset is structured and parsed to obtain a set of planning boundary elements. Based on this set, constraints are constructed to obtain a procedure boundary thought chain. Each procedure boundary thought chain includes source evidence, reasoning basis, Boolean trigger conditions, constraint variables, boundary thresholds, scenario variable mapping bits, and disturbance actions. Next, the procedure boundary thought chains are variable-mapped to obtain a procedure boundary thought chain with completed scenario variable mapping assignment and a corresponding scenario disturbance operator. Then, based on the power grid operating condition dataset, the procedure boundary thought chains with completed scenario variable mapping assignment, and the scenario disturbance operator, scenarios are generated to obtain candidate scenarios. Finally, the candidate scenarios are simulated and verified to obtain valid candidate scenarios. These valid candidate scenarios are then filtered using multi-dimensional indicators to obtain a power grid planning target scenario library. This scheme establishes the correspondence between constraint variables and scenario parameter variables in the planning variable dictionary through variable mapping, and binds disturbance actions with target mapping variables to generate scenario disturbance operators, thus establishing a mapping channel between the planning boundary and the simulation parameter domain. In the scenario generation stage, Boolean trigger conditions are logically evaluated to determine the activation state chain. Then, variable disturbance processing and scenario derivation processing are performed based on the scenario variable mapping bits and scenario disturbance operators to obtain candidate scenarios. Through simulation verification and multi-dimensional index screening, a power grid planning target scenario library is obtained from the candidate scenarios, ultimately generating power grid planning scenarios that meet the planning boundary requirements.

[0017] It should be understood that the descriptions of technical features, technical solutions, beneficial effects, or similar language in this application do not imply that all features and advantages can be achieved in any single embodiment. Rather, it is understood that the description of a feature or beneficial effect means that a specific technical feature, technical solution, or beneficial effect is included in at least one embodiment. Therefore, the descriptions of technical features, technical solutions, or beneficial effects in this specification do not necessarily refer to the same embodiment. Furthermore, the technical features, technical solutions, and beneficial effects described in this embodiment can be combined in any suitable manner. Those skilled in the art will understand that embodiments can be implemented without one or more specific technical features, technical solutions, or beneficial effects of a particular embodiment. In other embodiments, additional technical features and beneficial effects may be identified in specific embodiments that do not embody all embodiments. Attached Figure Description

[0018] Figure 1 A flowchart illustrating a method for generating power grid planning scenarios based on procedural thinking chains, provided for an embodiment of this application; Figure 2 A structural diagram of a power grid planning scenario generation device based on procedural thinking chain provided in this application embodiment; Figure 3 This is a schematic diagram of a computing device provided in an embodiment of this application. Detailed Implementation

[0019] The terms "first," "second," and "third," etc., used in this application specification and accompanying drawings are used to distinguish different objects, not to limit a specific order.

[0020] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0021] To make the technical solution of this application clearer and easier to understand, the following describes a method for generating power grid planning scenarios based on procedural thinking chains, provided by an embodiment of this application. Figure 1 As shown, this figure is a flowchart illustrating a method for generating power grid planning scenarios based on procedural thinking chains, provided in an embodiment of this application. This method for generating power grid planning scenarios based on procedural thinking chains includes: S201. Obtain text-based datasets and power grid operating condition datasets.

[0022] Text-based datasets are static, unstructured knowledge datasets used for power grid planning, each with a unique identifier. Power grid operating condition datasets are structured operating condition datasets used for power grid planning and analysis, comprising multiple operating condition data points, each with a unique identifier. The data sources for power grid operating condition datasets include basic planning data, historical operating data, and future forecast data.

[0023] The processing equipment acquires text-based data files (such as power grid planning regulations, technical guidelines, industry standards, planning documents, and design specifications). These text-based data files include one or more types of editable documents and scanned documents. For editable documents, the processing equipment directly parses them, extracting paragraph and table content. For scanned documents, the processing equipment uses an OCR (Optical Character Recognition) model to recognize the text and combines this with layout analysis to determine the text content and its reading order. The processing equipment then summarizes the extracted and / or recognized text content to obtain a text-based dataset and assigns a unique number to each dataset.

[0024] The processing equipment retrieves one or more of the following from the power grid business database through a database query interface: basic planning data, historical operating data, and future forecast data. The processing equipment then performs structured integration on the retrieved data, such as data cleaning and format standardization according to preset fields. The processed data is organized into multiple operating condition data sets, resulting in a power grid operating condition dataset. Each operating condition data set includes scenario parameters such as load parameters, grid parameters, renewable energy output parameters, and meteorological parameters. The processing equipment assigns a unique number to each power grid operating condition dataset.

[0025] S202. Perform structured parsing on the text dataset to obtain the set of planning boundary elements.

[0026] A planning boundary element set is a data set containing at least one planning boundary element. Each planning boundary element includes at least the following fields: source identifier, applicable objects, constraint variables, boundary threshold, time range, area range, constraint type, priority, original text summary, and statement category. The source identifier records the source document identifier, chapter identifier, clause identifier, or table number identifier of the planning boundary element. The applicable objects record the object category to which the planning boundary element applies. The constraint variables record the planning variables constrained by the planning boundary element. The boundary threshold records the numerical threshold, target value, or verification condition of the planning boundary element. The time range records the... The applicable year, season, or period (e.g., the peak summer season is from July 1st to August 31st each year); the area scope is used to record the applicable area or power grid level of the planning boundary element; the constraint type is used to record the constraint category of the planning boundary element, including target constraints, upper limit constraints, lower limit constraints, or verification constraints; the priority is used to record the mandatory level, recommended level, or reference level of the planning boundary element; the original text summary is a concise text obtained by semantically compressing the extracted statements, removing decorative and preparatory text within the statements, and not introducing semantics outside the original text; the statement category is used to record the statement category from which the planning boundary element originates, including target statements, procedural constraint statements, or other statements.

[0027] One possible approach involves slicing a text-based dataset into document slices to obtain knowledge slices; classifying each knowledge slice into sentences and determining the sentences to be extracted based on the classification results; extracting boundary elements from the sentences to be extracted to obtain raw boundary elements; validating the fields of the raw boundary elements to obtain valid raw boundary elements; and aggregating the valid raw boundary elements to obtain a planning boundary element set. Each planning boundary element in the planning boundary element set carries the knowledge slice number, the text-based dataset number, and the power grid operating condition dataset number that generated the planning boundary element, thus enabling data traceability of the knowledge slices.

[0028] The processing device first performs document slicing on the acquired complete text dataset. Specifically, the character threshold for a single knowledge slice is preset to any value between 800 and 1200 characters. The processing device reads the text dataset character by character. When the cumulative number of characters read reaches the character threshold, it scans backward to find the position of the first newline character, the first chapter title mark, the first clause number mark, or the first period, and uses the first position found as the slice truncation boundary. The maximum offset of the processing device scanning backward can be preset to 200 characters to limit the maximum length of the slice. If no truncation position is found within the maximum offset range of 200 characters, forced truncation is performed directly at the position corresponding to the character threshold. After the processing device completes the slicing, it assigns a unique slice number to each knowledge slice and synchronously records the start and end character positions of the knowledge slice in the text dataset to achieve data traceability of the knowledge slice.

[0029] The processing equipment performs attribute determination on each individual statement within each knowledge slice, classifying the statement into categories based on its semantic attributes. These categories cover target statements, procedural constraint statements, and other statements. After determination, only target statements and procedural constraint statements are selected and retained, and these selected statements are grouped into statements to be extracted; other statements are discarded. Procedural constraint statements are those containing constraint keywords such as "should not," "must not," "should," "appropriate," "must," "strictly prohibited," "limit," "not greater than," or "not less than," and have a clearly defined applicable object, time range, or regional scope. Target statements contain goal-oriented keywords such as "achieve," "control within," "satisfy," or "should be achieved." Statements that are neither procedural constraint statements nor target statements are classified as "other statements." When a single statement matches keywords corresponding to multiple categories, the determination is performed according to a preset statement priority, with procedural constraint statements having higher priority than target statements.

[0030] The processing device performs boundary feature extraction for all statements to be extracted. It can call the locally deployed large language model and preset feature extraction rules to extract the source identifier, applicable object, constraint variables, boundary threshold, time range, regional range, constraint type, priority, original text summary, and statement category from each statement to be extracted (field extraction is first completed by the preset feature extraction rules, and semantic completion and error correction are only performed by the large language model in cases where there is ambiguity or missing fields in the feature extraction rules). The extracted fields are combined to generate a single original boundary feature. After traversing all statements to be extracted, a batch of original boundary features is obtained. The large language model is deployed offline locally and does not require calling external cloud interfaces. The inference calculation is performed only on the processing device.

[0031] The preset element extraction rules include field-level regular expression matching rules and keyword mapping rules. The source document identifier, chapter identifier, clause identifier, or table number identifier is extracted using regular expression matching rules (e.g., if the statement to be extracted is “(Q / GDW 1799.1—2013) Chapter 5, Article 5.2.1, Table 3-1 stipulates: The short-circuit current level of the 220kV power grid in the region should not exceed 50kA,” the regular expression matches the source document identifier “Q / GDW 1799.1—2013”, the chapter identifier “Chapter 5”, the clause identifier “Article 5.2.1”, and the table number identifier “Table 3-1”).

[0032] The applicable objects are determined through keyword mapping rules at the scenario, equipment, and power grid levels (e.g., if the statement to be extracted is "the short-circuit current of the 220kV substation busbar should not exceed 50kA", and the equipment keyword "220kV substation busbar" is matched, then the applicable object is "220kV substation busbar"); constraint variables are determined through constraint-type word mapping (e.g., if the statement to be extracted is "the short-circuit current level of the 220kV power grid within the region should not exceed 50kA", and the constraint object word "short-circuit current level" is matched, then the constraint variable is "short-circuit current level"); Boundary thresholds are extracted by matching numerical values, units, and comparison operators (e.g., if the statement to be extracted is "short-circuit current should not exceed 50kA", and the comparison operator "should not exceed", the numerical value "50", and the unit "kA" are matched, then the boundary threshold is "should not exceed 50kA"); the time range is determined by mapping year, time period, and season keywords (e.g., if the statement to be extracted is "the main transformer load rate should not exceed 80% during peak summer season", and the seasonal time period keyword "during peak summer season" is matched, then the time range is "during peak summer season"); the time range is determined by region, voltage level, and electricity... Network-level keyword mapping determines the regional scope (e.g., if the statement to be extracted is "the pass rate of 110kV line N-1 in the first-level power grid should reach 100%", and the network-level keyword "first-level power grid" and the voltage level keyword "110kV" are matched, then the regional scope is "first-level power grid, 110kV"); threshold feature mapping is used to classify constraint types (e.g., if the statement to be extracted contains upper limit expressions such as "should not be greater than" or "must not exceed", then it is determined to be an upper limit constraint; if the statement to be extracted contains lower limit expressions such as "should not be less than" or "should not be lower than", then...) If the statement to be extracted contains target expressions such as "should be achieved" or "should be realized," it is determined to be a target constraint; if the statement to be extracted contains verification expressions such as "should be satisfied" or "must be verified," it is determined to be a verification constraint. For priority fields, the corresponding priority level is mapped through constraint strength keywords. The mapping relationship is fixed as follows: keywords such as "must," "strictly prohibited," "must not," and "should not" correspond to the mandatory level; keywords such as "should" and "should" correspond to the recommendation level; statements without mandatory constraint keywords that only describe planning goals and development trends correspond to the reference level. When the field results output by the preset element extraction rules are ambiguous or missing, the statement to be extracted and the extraction results based on the element extraction rules are jointly fed into the large language model. The large language model combines the context semantics to complete the field and eliminate ambiguity. The extraction results based on the element extraction rules and the results of the large language model are fused to obtain the final content of each field, i.e., the original boundary elements.

[0033] The processing equipment performs field validity checks on each item of the batch of original boundary elements, sequentially verifying whether there are any issues such as missing content, format errors, out-of-bounds values, or logical conflicts in each field of each original boundary element. Specific verification includes: checking for missing content in fields such as source identifier, constraint variables, boundary thresholds, and constraint types; for example, if the boundary threshold field of an original boundary element is empty, it is considered a content missing anomaly. It also checks whether the numerical format, unit of measurement, and comparison logic corresponding to the boundary threshold are compliant and valid; for example, if the boundary threshold is filled in as "should not be greater than AkA" and the numerical part is a non-numeric character, it is considered a format error anomaly. Furthermore, it checks whether the constraint type matches the boundary threshold, excluding cases where the upper or lower limit constraints contradict the boundary threshold logic; for example, if the boundary threshold is "should not be less than 80%" but the constraint type is an upper limit constraint, it is considered a logical conflict anomaly. Finally, it checks whether the priority field is consistent with the constraint strength of the original statement to be extracted, excluding conflicts where the keyword and priority level do not match; for example, if the original statement to be extracted contains the keyword "must" and the priority field is marked as a recommendation level, it is considered a logical conflict anomaly. The processing device will remove any original boundary elements with any abnormal field, and retain original boundary elements with complete fields, compliant format, and correct logic, thus obtaining the original boundary elements that pass the verification.

[0034] The processing device adds a data reference tag to each verified original boundary feature. This data reference tag is bound to the knowledge slice number that generated the original boundary feature, the number of the text dataset to which the knowledge slice belongs, and the number of the corresponding power grid operating condition dataset, thus enabling data traceability. Finally, the processing device collects and integrates all verified original boundary features carrying data reference tags to form a complete set of planning boundary features.

[0035] S203. Based on the set of planning boundary elements, a constraint-based construction is carried out to obtain the procedure boundary thinking chain.

[0036] Each procedural boundary thinking chain includes source evidence, reasoning basis, Boolean triggering conditions, constraint variables, boundary thresholds, scenario variable mapping bits, and disturbance actions.

[0037] The specific construction process is as follows: The processing equipment extracts the source identifier and original text summary corresponding to a single planning boundary element, concatenates and combines these two types of field information to generate source evidence for the corresponding procedural boundary thinking chain; it extracts the source identifier, constraint type, and boundary threshold of the planning boundary element, and combines them to generate the reasoning basis for the procedural boundary thinking chain; based on the constraint type, boundary threshold, time range, and regional range of the planning boundary element, it generates the corresponding Boolean trigger condition. For example, if the constraint variable of a planning boundary element is "short-circuit current level," the boundary threshold is "should not exceed 50kA," the regional range is "first-level power grid, 220kV," the time range is "during peak summer season," and the constraint type is "upper limit constraint," then the generated Boolean trigger condition is: the measured short-circuit current value under the current power grid operating scenario during peak summer season and at the first-level power grid 220kV voltage level is greater than 50kA. If this Boolean trigger condition is true, the thinking chain is activated; if it is false, it is not activated.

[0038] When constructing the procedural boundary thinking chain, the processing device also reads the statement category field of the planning boundary element. If the statement category is a procedural constraint statement, the generated procedural boundary thinking chain is marked as a mandatory constraint thinking chain; if the statement category is a goal statement, the generated procedural boundary thinking chain is marked as a goal-oriented thinking chain.

[0039] The processing device directly inputs the constraint variables and boundary thresholds of a single planning boundary element into the corresponding fields of the currently assembled procedure boundary thinking chain; it reserves scenario variable mapping positions in the procedure boundary thinking chain; and it configures the corresponding disturbance actions according to the constraint type and priority of the planning boundary element. The processing device traverses all verified planning boundary elements in the planning boundary element set, and completes the structured assembly of the procedure boundary thinking chain one by one, resulting in a set of procedure boundary thinking chains.

[0040] S204. Perform variable mapping on the procedure boundary thinking chain to obtain the procedure boundary thinking chain and the corresponding scene disturbance operator after completing the scene variable mapping assignment.

[0041] One possible approach is to use the constraint variables in the procedure boundary thinking chain as search conditions to search in a preset planning variable dictionary to determine the target mapping variable; write the target mapping variable into the scenario variable mapping bit in the procedure boundary thinking chain to obtain the procedure boundary thinking chain with completed scenario variable mapping assignment; and bind the disturbance action in the procedure boundary thinking chain with the written target mapping variable to obtain the scenario disturbance operator.

[0042] The planning variable dictionary is a pre-defined standardized variable mapping database that stores a one-to-one correspondence between all constraint variables and power grid operating condition variables. The target mapping variables are standardized power grid operating condition variables that match the constraint variables within the procedure boundary thought chain. The scenario perturbation operator is used to perform quantitative variable perturbation processing on the basic scenario.

[0043] The processing equipment retrieves a pre-set planning variable dictionary and reads each constraint variable recorded in the procedure boundary thinking chain. Using this constraint variable as a search condition, it searches the planning variable dictionary to find all power grid operating condition variables that match the constraint variable, and these are selected as candidate operating condition variables. Then, it compares the applicable objects and regional scope recorded in the procedure boundary thinking chain with the applicable objects and regional scope of each candidate operating condition variable. If the applicable objects or regional scope of a candidate operating condition variable do not match the content recorded in the procedure boundary thinking chain, the candidate operating condition variable is discarded; only candidate operating condition variables whose applicable objects and regional scopes match are retained. If only one candidate operating condition variable is retained, then that candidate operating condition variable is determined as the target mapping variable; if multiple candidate operating condition variables are retained, then the candidate operating condition variable with the smallest mapping sequence number is selected as the target mapping variable according to the mapping sequence number pre-configured in the planning variable dictionary; the mapping sequence number is pre-written when the planning variable dictionary is constructed, and the mapping sequence numbers of each candidate operating condition variable under the same constraint variable are different; if no candidate operating condition variable is retained, then the procedure boundary thinking chain is marked as mapping failure or pending manual confirmation.

[0044] For example, if the constraint variable within the boundary thinking chain of the procedure is "short-circuit current level", the applicable object is "220kV substation bus", and the regional scope is "first-level power grid", the processing equipment uses the constraint variable "short-circuit current level" as the search condition to find all power grid operating condition variables whose meaning is short-circuit current level in the planning variable dictionary. Then, it checks the applicable object and regional scope of these power grid operating condition variables one by one, discarding power grid operating condition variables whose applicable object is 110kV substation bus and whose regional scope is the second-level power grid. If there is one remaining power grid operating condition variable, it is directly determined as the target mapping variable "measured value of 220kV bus short-circuit current". If there are multiple remaining variables, the one with the smallest mapping sequence number is taken as the target mapping variable "measured value of 220kV bus short-circuit current".

[0045] The processing device writes the target mapping variable into the reserved scene variable mapping bit in the procedure boundary thinking chain, generating a procedure boundary thinking chain with completed scene variable mapping assignment. The processing device extracts the perturbation action configured in the procedure boundary thinking chain and binds this perturbation action to the target mapping variable already written into the scene variable mapping bit (i.e., specifying the operation object of the perturbation action as the target mapping variable written in the scene variable mapping bit, so that the perturbation action explicitly acts on that target mapping variable, rather than other target mapping variables), generating a scene perturbation operator. The processing device traverses all procedure boundary thinking chains, repeatedly executing the retrieval, assignment, and binding operations to complete the variable mapping processing for all thinking chains.

[0046] S205. Based on the power grid operating condition dataset, the procedural boundary thinking chain and scenario disturbance operator that complete the scenario variable mapping assignment are used to generate scenarios and obtain candidate scenarios.

[0047] One possible approach involves clustering scenarios based on a power grid operating condition dataset to obtain basic scenarios; substituting the scenario parameters of each basic scenario into the Boolean trigger conditions in the procedural boundary thinking chain that has completed scenario variable mapping assignment for logical evaluation to obtain the evaluation results; determining the thinking chain in the enabled state based on the evaluation results; performing variable perturbation processing on the corresponding basic scenarios based on the scenario variable mapping bits and scenario perturbation operators corresponding to the thinking chain in the enabled state to obtain perturbed intermediate scenarios; performing scenario derivation processing on the perturbed intermediate scenarios to obtain the corresponding derived scenarios; and using the basic scenarios, perturbed intermediate scenarios, and derived scenarios as candidate scenarios.

[0048] The basic scenario is the initial power grid planning operating condition data obtained from the power grid operating condition dataset through scenario clustering. Scenario parameters are parameter data used to characterize the power grid operating condition corresponding to the basic scenario.

[0049] The post-disturbance intermediate scenario refers to the operating condition data obtained after adjusting the target mapping variables indicated by the scenario variable mapping bits in the basic scenario according to the scenario disturbance operator corresponding to the activated procedure boundary thinking chain, while keeping the other scenario parameters unchanged. The derived scenario refers to multiple sets of differentiated operating condition data generated based on the post-disturbance intermediate scenario, while keeping the grid parameters and meteorological parameters unchanged, and shifting one or more of the load parameters and renewable energy output parameters upwards and downwards according to preset shift ratios. The candidate scenario is a set of power grid planning operating condition data to be simulated and verified, including the basic scenario, the post-disturbance intermediate scenario, and the derived scenario.

[0050] The processing equipment extracts operating condition data from the power grid operating condition dataset. It calculates the Euclidean distance between any two data points and normalizes the distance, converting it into a similarity value between 0 and 1. A value closer to 1 indicates a closer similarity between the two data points. The processing equipment then divides all operating condition data into different groups based on a similarity threshold, grouping samples with similarity values ​​greater than the threshold into the same group, resulting in multiple groups. For each group, it calculates the cluster center and selects the data point closest to the cluster center as the representative sample for that group. These representative samples are then used to define multiple basic scenarios. For example, clustering the operating condition data in the power grid dataset according to a similarity threshold yields four basic scenarios: summer high-load operating condition, winter high-load operating condition, renewable energy high-generation operating condition, and normal stable operating condition. The preset similarity threshold can be manually configured by operators based on the data dispersion of the power grid operating condition dataset. If no manual configuration is performed, the system's built-in default value will be used. For example, the configurable range of the preset similarity threshold is 0.6 to 0.9, and the system default value is 0.75.

[0051] For each completed scenario variable mapping and assignment procedure boundary thought chain, the processing device reads its Boolean trigger condition. Based on the target mapping variable corresponding to the Boolean trigger condition, the processing device obtains the corresponding parameter value from the current basic scenario's scenario parameters and compares this parameter value with the boundary threshold. For the actual time range and / or area range included in the Boolean trigger condition, it also determines whether the basic scenario's time information falls within that time range and whether the area information matches that area range. If the parameter value satisfies the logical relationship defined by the boundary threshold, and the actual time range and area range included in the trigger condition also match, the logical evaluation result of the Boolean trigger condition is assigned as true; if the parameter value does not satisfy the logical relationship defined by the boundary threshold, or the actual time range included in the trigger condition does not match, or the actual area range included in the trigger condition does not match, the logical evaluation result is assigned as false.

[0052] For example, the Boolean trigger condition for a certain procedure boundary thought chain is "the measured value of the 220kV short-circuit current in the first-level power grid is greater than 50kA". The processing device obtains the "measured value of the 220kV bus short-circuit current" from the scenario parameters of the current basic scenario. If the value is 53kA, which is greater than 50kA, the result is true; if the value is 48kA, which is not greater than 50kA, the result is false. If the trigger condition also includes the time range "during peak summer season" and the regional range "first-level power grid, 220kV", then the time information of the current basic scenario must fall within "during peak summer season" and the regional information must match "first-level power grid, 220kV" before the logical evaluation result is assigned a true value.

[0053] The processing device compares the evaluation results corresponding to each procedure boundary thinking chain, marks the procedure boundary thinking chain with a true evaluation result as an enabled thinking chain, and discards the procedure boundary thinking chain with a false evaluation result.

[0054] For each enabled procedure boundary thought chain, the processing device reads the target mapping variable stored in the scenario variable mapping bit of that procedure boundary thought chain. Using the name of this target mapping variable as a search keyword, the processing device searches the scenario parameter list of the basic scenario for a matching scenario parameter and reads its current value. If no matching scenario parameter is found, the basic scenario or the procedure boundary thought chain is marked as a mapping anomaly, and the current disturbance processing ends. For example, if the target mapping variable written in the scenario variable mapping bit is "measured value of 220kV bus short-circuit current", the processing device reads the scenario parameter value named "measured value of 220kV bus short-circuit current" from the scenario parameter list of the basic scenario.

[0055] The processing device reads the scene perturbation operator corresponding to the procedural boundary thought chain of the enabled state, and obtains the perturbation direction and preset perturbation amplitude from the scene perturbation operator. The perturbation direction is divided into two categories: upward adjustment and downward adjustment. The preset perturbation amplitude is a fixed percentage value. The upward adjustment calculation method is: perturbation parameter value = scene parameter value × (1 + preset perturbation amplitude); the downward adjustment calculation method is: perturbation parameter value = scene parameter value × (1 + preset perturbation amplitude). (Preset disturbance amplitude); For example, if the scenario parameter value read as "Measured value of 220kV bus short-circuit current" is 50kA, the scenario disturbance operator configuration is increased, the preset disturbance amplitude is set to 0.1, and the parameter value after disturbance is equal to 50kA×(1+0.1)=55kA. The processing device replaces the scenario parameter value corresponding to the target mapping variable in the basic scenario with the disturbed parameter value, while keeping the other scenario parameters unchanged, generating the disturbed intermediate scenario. The preset disturbance amplitude can be manually configured by the operator in conjunction with the procedural boundary margin of the corresponding constraint variables; if not manually configured, the system default value is used. The configurable range of the preset disturbance amplitude is 0.05-0.2, and the system default value is 0.1.

[0056] When performing scenario derivation processing, the processing equipment retains the network structure parameters and meteorological parameters of the intermediate scenario after the disturbance unchanged, and selects the load parameters and renewable energy output parameters in the intermediate scenario after the disturbance as the parameters to be shifted. The parameters to be shifted can be shifted upwards and downwards according to a preset shift ratio, generating multiple sets of new operating condition data with different parameters but the same operating conditions, thus obtaining derivation scenarios. Each derivation scenario differs from the corresponding intermediate scenario after the disturbance only in the numerical values ​​of the load parameters and / or renewable energy output parameters; the network structure parameters and meteorological parameters remain unchanged. For example, based on the load parameters of the intermediate scenario after the disturbance, shifting upwards by 3% and downwards by 3% respectively, while simultaneously making small numerical adjustments to the renewable energy output parameters, generates multiple sets of derivation scenarios with the same source but different characteristics. The preset shift ratio can be configured to range from 0.02 to 0.08, with the system default value being 0.03.

[0057] The processing equipment collects all basic scenarios, all intermediate scenarios after disturbance, and all derived scenarios, and merges them to form candidate scenarios.

[0058] S206. Simulate and verify the candidate scenarios to obtain valid candidate scenarios; perform multi-dimensional index screening on the valid candidate scenarios to obtain the power grid planning target scenario library.

[0059] The simulation verification process can be as follows: Each candidate scenario is written into the power grid simulation model and verified by power flow calculation to obtain candidate scenarios that pass the power flow calculation verification; the candidate scenarios that pass the power flow calculation verification are then verified by safety and stability to obtain candidate scenarios that pass the safety and stability verification; the candidate scenarios that pass the safety and stability verification are determined as valid candidate scenarios.

[0060] The process of selecting multidimensional indicators can be as follows: Calculate the multidimensional evaluation index of each valid candidate scenario to obtain the comprehensive score corresponding to each valid candidate scenario; based on the mandatory coverage constraint and the comprehensive score, screen each valid candidate scenario to obtain the target scenario library for power grid planning.

[0061] The power grid simulation model is a simulation model that includes the power grid topology and equipment parameters, used for power flow calculations and safety and stability verification. Multidimensional evaluation metrics are used to evaluate valid candidate scenarios. Forced coverage constraints are preset screening rules used to limit the admission criteria for the power grid planning target scenario library.

[0062] The processing equipment imports the scenario parameters of each candidate scenario into a preset power grid simulation model. The power grid simulation model performs power flow calculations based on the imported scenario parameters (using the Newton-Raphson iterative algorithm to solve the power flow problem), and outputs the voltage of each node (such as voltage magnitude and voltage phase angle) and the power flow of each branch (such as active power flow and reactive power flow).

[0063] After the power flow calculation is completed, the processing device calls the preset grid operation thresholds to determine the threshold values ​​for each node voltage and each branch power flow one by one. The preset grid operation thresholds are fixed limits in the regulations, including the upper limit of node voltage, the lower limit of node voltage, the upper limit of branch active power, and the upper limit of branch reactive power. Specifically, the processing device determines whether all node voltages in the current candidate scenario are within the voltage threshold range formed by the upper and lower limits of node voltage, whether the active power flow of all branches in the current candidate scenario does not exceed the corresponding upper limit of branch active power, whether the reactive power flow of all branches in the current candidate scenario does not exceed the corresponding upper limit of branch reactive power, and whether the power flow iteration process of the current candidate scenario converges normally and without numerical distortion or solution anomalies.

[0064] If all of the above conditions are met, the current candidate scenario is determined to have passed the power flow calculation verification. If the current candidate scenario has any of the following conditions: voltage exceeding the limit at any node, active or reactive power flow exceeding the limit in any branch, power flow solution not converging, numerical distortion, or solution anomaly, the current candidate scenario is determined to have failed the power flow calculation verification and is removed from the candidate scenario.

[0065] For candidate scenarios verified through power flow calculations, further safety and stability verification is conducted. Specifically, the processing equipment imports the current candidate scenario into the power grid simulation model and calls the standard fault condition library built into the power grid simulation model to sequentially apply preset fault conditions to the current candidate scenario. Preset fault conditions may include one or more of the following: instantaneous single-phase ground fault, three-phase short-circuit fault, and transformer short-time fault. After a fault occurs, it is cleared according to a preset fault clearing time, and simulation continues for a preset duration after fault clearing, collecting raw data on the dynamic operation of the power grid. The raw data on the dynamic operation of the power grid includes at least one or more of the following: voltage time-series data of each node, power angle time-series data of each generator unit, and system frequency time-series data. Specifically, the preset fault clearing time for an instantaneous single-phase ground fault can be 0.1 seconds, the preset fault clearing time for a three-phase short-circuit fault can be 0.15 seconds, and the preset fault clearing time for a transformer short-time fault can be 0.2 seconds; the preset duration of simulation continuing after fault clearing can be pre-set according to the power grid scale and stability analysis requirements, such as 10 seconds.

[0066] The processing equipment performs safety and stability assessments based on the collected raw data of the power grid's dynamic operation: when the voltage drop amplitude and post-fault voltage recovery time of all network nodes meet the preset voltage stability thresholds, the maximum power angle difference between generator units does not exceed the preset power angle stability threshold, the operating frequency deviation remains within the preset frequency stability range, and no safety issues such as grid disconnection, continuous parameter exceeding limits, or power oscillation instability occur during the simulation process, the current candidate scenario is deemed to have passed the safety and stability verification. If any dynamic operation indicator does not meet the corresponding stability threshold requirements, the candidate scenario is directly eliminated. The preset voltage stability threshold, preset power angle stability threshold, and preset frequency stability range are all pre-set according to fixed limits in the power grid planning regulations or pre-set according to the safety and stability standards built into the power grid simulation model. The processing equipment will determine the candidate scenarios that pass both the power flow calculation verification and the safety and stability verification as valid candidate scenarios.

[0067] The voltage sag amplitude of all nodes in the network can be determined as follows: For each node, calculate the voltage sag amplitude of the node based on the steady-state voltage value before the fault and the lowest voltage value after the fault; then take the maximum value among all node voltage sag amplitudes as the voltage sag amplitude of all nodes in the network.

[0068] The voltage recovery time after a fault can be determined as follows: Starting from the fault clearing moment, for each node, continuously check whether its voltage has recovered to a preset voltage recovery range. When the node's voltage first enters this preset range, the time between this moment and the fault clearing moment is taken as the node's voltage recovery time. If a preset duration is required after entering this range, the moment it first enters and maintains this preset duration is taken as the recovery moment. If the node's voltage never enters the preset voltage recovery range before the simulation ends, the node is considered not to have recovered. Finally, the longest voltage recovery time among all nodes in the entire network is taken as the post-fault voltage recovery time.

[0069] The maximum power angle difference between generator sets can be determined as follows: at each moment during the simulation period after the fault, calculate the difference between the maximum and minimum power angles of all generator sets to obtain the power angle difference between the generator sets at that moment; then take the maximum value of the power angle difference between the generator sets at all moments during the entire simulation period as the maximum power angle difference between each generator set.

[0070] The operating frequency deviation can be determined based on the difference between the actual system frequency and the rated frequency.

[0071] For each valid candidate scenario, the power grid reliability index, grid economic index, renewable energy absorption capacity index, and operational safety margin index are calculated one by one. Among them, the power grid reliability index is used to characterize the continuous power supply capacity of load nodes under the valid candidate scenario; the grid economic index is used to characterize the grid loss level under the valid candidate scenario; the renewable energy absorption capacity index is used to characterize the utilization degree of renewable energy generation under the valid candidate scenario; and the operational safety margin index is used to characterize the safety margin of branch power relative to the power limit under the valid candidate scenario.

[0072] The power grid reliability index is calculated using the proportion of effective power supply time. Specifically, within the simulation period of the effective candidate scenario, for each load node, the normal power supply time of that load node is statistically analyzed, and the ratio of this normal power supply time to the total simulation time is determined as the power supply reliability rate of that load node. Then, the average power supply reliability rate of all load nodes is calculated to obtain the power grid reliability index for the current effective candidate scenario. A higher ratio indicates higher power supply reliability. The energy efficiency index of the power grid is calculated using the unit load power grid loss rate. Specifically, it is calculated as the ratio of the total active power loss to the total active power load of the entire network during the simulation period for the current effective candidate scenario; or it is calculated as the ratio of the total active power loss to the total active power load of the entire network at the same time section. The lower the loss rate, the better the energy efficiency of the power grid operation.

[0073] The renewable energy absorption capacity index is calculated using the renewable energy utilization rate method. Specifically, it is the ratio of the actual grid-connected power generation of renewable energy in the current valid candidate scenarios to the theoretical maximum power generation of renewable energy. The theoretical maximum power generation of renewable energy is determined based on the sum of the actual grid-connected power generation of renewable energy and the amount of wind and solar power curtailed, or based on renewable energy resource forecast data. A higher ratio indicates a stronger renewable energy absorption capacity.

[0074] The operational safety margin index is calculated using the average power margin of all branches in the network. Specifically, the power margin ratio of each branch is calculated as the ratio of the difference between the branch power limit and the current power flow to the power limit. The average power margin ratio of all branches in the network is then taken as the operational safety margin index. Alternatively, the power margin ratio of each branch is calculated as the ratio of the difference between the branch power limit and the current power flow to the power limit, and the minimum value among the power margin ratios of all branches in the network is taken as the operational safety margin index.

[0075] The system presets fixed weight coefficients for four indicators. The sum of the weight coefficients for the four indicators equals 1. The weight coefficients are fixed parameters preset by the system and can be manually adjusted by operators.

[0076] For all valid candidate scenarios, extreme value normalization is performed for each multi-dimensional evaluation index: For the three positive indices—grid power supply reliability, renewable energy absorption capacity, and operational safety margin—where larger values ​​are better, the normalization calculation method is: Normalized index value = (Original value of the index in the current scenario - Minimum value of the index in all valid candidate scenarios) ÷ (Maximum value of the index in all valid candidate scenarios - Minimum value of the index in all valid candidate scenarios); For the negative index—grid structure economics—where smaller values ​​are better, the normalization calculation method is: Normalized index value = (Maximum value of the index in all valid candidate scenarios - Original value of the index in the current scenario) ÷ (Maximum value of the index in all valid candidate scenarios - Minimum value of the index in all valid candidate scenarios). When the maximum and minimum values ​​of an index are equal across all valid candidate scenarios, the normalized index value for that index across all scenarios is fixed at 1. After the above processing, normalized index values ​​with a range of 0-1 are obtained. Each normalized index value is multiplied by its corresponding preset weight coefficient and then summed to directly obtain the comprehensive score of the current effective candidate scenario. This comprehensive score is a dimensionless evaluation value used to characterize the comprehensive performance of the effective candidate scenario across multiple evaluation dimensions. The higher the comprehensive score, the better the performance of the effective candidate scenario in adapting to power grid planning.

[0077] The processing equipment filters based on the label type of the procedural boundary thinking chain. Specifically, the constraint variables, boundary thresholds, and constraint types corresponding to the mandatory constraint type thinking chain are used as mandatory coverage constraints. Each valid candidate scenario's corresponding scenario parameters are checked to see if they satisfy the constraint relationship defined by the mandatory coverage constraint. If any mandatory coverage constraint is not satisfied in a valid candidate scenario, that valid candidate scenario is removed. Goal-oriented thinking chains do not participate in the mandatory coverage constraint screening, nor in the comprehensive scoring calculation. The processing equipment uses the constraint variables and boundary thresholds corresponding to the goal-oriented thinking chain as target reference items, outputting them along with the power grid planning target scenario library for operators to review or refer to.

[0078] Based on the above, the power grid planning scenario generation method based on the procedural thinking chain obtains a text-based dataset and a power grid operating condition dataset. The text-based dataset is a static, unstructured knowledge dataset used for power grid planning, while the power grid operating condition dataset is a structured operating condition dataset used for power grid planning analysis. Then, the text-based dataset is structured and parsed to obtain a set of planning boundary elements. Constraints are then constructed based on this set to obtain a procedural boundary thinking chain. Each procedural boundary thinking chain includes source evidence, reasoning basis, Boolean trigger conditions, constraint variables, boundary thresholds, scenario variable mapping bits, and disturbance actions. Next, variable mapping is performed on the procedural boundary thinking chains to obtain procedural boundary thinking chains with completed scenario variable mapping assignments and corresponding scenario disturbance operators. Then, based on the power grid operating condition dataset, the procedural boundary thinking chains with completed scenario variable mapping assignments, and the scenario disturbance operators, scenarios are generated to obtain candidate scenarios. Finally, simulation verification is performed on the candidate scenarios to obtain valid candidate scenarios. Multi-dimensional index screening is then applied to the valid candidate scenarios to obtain a power grid planning target scenario library. This scheme establishes the correspondence between constraint variables and scenario parameter variables in the planning variable dictionary through variable mapping, and binds disturbance actions with target mapping variables to generate scenario disturbance operators, thus establishing a mapping channel between the planning boundary and the simulation parameter domain. In the scenario generation stage, Boolean trigger conditions are logically evaluated to determine the activation state chain. Then, variable disturbance processing and scenario derivation processing are performed based on the scenario variable mapping bits and scenario disturbance operators to obtain candidate scenarios. Through simulation verification and multi-dimensional index screening, a power grid planning target scenario library is obtained from the candidate scenarios, ultimately generating power grid planning scenarios that meet the planning boundary requirements.

[0079] The above text combined Figure 1 The method for generating power grid planning scenarios based on procedural thinking chains provided in this application embodiment has been described in detail. The apparatus and equipment provided in this application embodiment will be described below with reference to the accompanying drawings.

[0080] This application also provides a power grid planning scenario generation device based on procedural thinking chain, such as... Figure 2 As shown, this figure is a structural diagram of a power grid planning scenario generation device based on procedural thinking chain provided in an embodiment of this application. The device includes: Module 301 is used to acquire text-based datasets and power grid operating condition datasets. Data processing module 302 is used to perform structured parsing on text datasets to obtain a set of planning boundary elements; to construct constraints based on the set of planning boundary elements to obtain a procedure boundary thinking chain; and to perform variable mapping on the procedure boundary thinking chain to obtain a procedure boundary thinking chain with completed scene variable mapping assignment and the corresponding scene perturbation operator. The scenario generation module 303 is used to generate scenarios based on the power grid operating condition dataset, the procedure boundary thinking chain that completes the scenario variable mapping assignment, and the scenario disturbance operator to obtain candidate scenarios; to perform simulation verification on the candidate scenarios to obtain valid candidate scenarios; and to perform multi-dimensional index screening on the valid candidate scenarios to obtain the power grid planning target scenario library.

[0081] In some possible implementations, the data processing module 302 is specifically used for: Using the constraint variables in the procedure boundary thinking chain as search conditions, the target mapping variable is determined by searching in the preset planning variable dictionary; the target mapping variable is written into the scenario variable mapping bit in the procedure boundary thinking chain to obtain the procedure boundary thinking chain with completed scenario variable mapping assignment; the disturbance action in the procedure boundary thinking chain is bound to the written target mapping variable to obtain the scenario disturbance operator.

[0082] In some possible implementations, the scene generation module 303 is specifically used for: Based on the power grid operating condition dataset, scenario clustering is performed to obtain basic scenarios. The scenario parameters of each basic scenario are substituted into the Boolean trigger conditions in the procedural boundary thinking chain that completes the scenario variable mapping assignment for logical evaluation to obtain the evaluation results. The thinking chain in the enabled state is determined based on the evaluation results. According to the scenario variable mapping bits and scenario perturbation operators corresponding to the thinking chain in the enabled state, the corresponding basic scenarios are subjected to variable perturbation processing to obtain perturbed intermediate scenarios. The perturbed intermediate scenarios are subjected to scenario derivation processing to obtain the corresponding derived scenarios. The basic scenarios, perturbed intermediate scenarios, and derived scenarios are used as candidate scenarios.

[0083] In some possible implementations, the data processing module 302 is specifically used for: The text dataset is sliced ​​into document slices to obtain knowledge slices; each knowledge slice is classified into sentences, and the sentences to be extracted are obtained based on the classification results; boundary elements are extracted from the sentences to be extracted to obtain original boundary elements; the original boundary elements are validated for field validity to obtain validated original boundary elements; the validated original boundary elements are aggregated to obtain the planning boundary element set.

[0084] In some possible implementations, the scene generation module 303 is specifically used for: Each candidate scenario is written into the power grid simulation model and verified by power flow calculation to obtain candidate scenarios that pass the power flow calculation verification; the candidate scenarios that pass the power flow calculation verification are then verified by safety and stability to obtain candidate scenarios that pass the safety and stability verification; the candidate scenarios that pass the safety and stability verification are determined as valid candidate scenarios.

[0085] In some possible implementations, the scene generation module 303 is specifically used for: Calculate the multidimensional evaluation index of each valid candidate scenario to obtain the comprehensive score corresponding to each valid candidate scenario; based on the mandatory coverage constraint and the comprehensive score, screen each valid candidate scenario to obtain the target scenario library for power grid planning.

[0086] The power grid planning scenario generation device based on procedural thinking chain according to the embodiments of this application can correspond to the execution of the method described in the embodiments of this application, and the other operations and / or functions of each module / unit of the power grid planning scenario generation device based on procedural thinking chain are respectively for implementing Figure 1 For the sake of brevity, the corresponding processes of each method in the illustrated embodiments will not be described in detail here.

[0087] This application also provides a computing device. For example... Figure 3 As shown in the figure, this is a schematic diagram of a computing device provided in an embodiment of this application. The computing device 400 includes a bus 401, a processor 402, a communication interface 403, and a memory 404. The processor 402, the memory 404, and the communication interface 403 communicate with each other via the bus 401.

[0088] Bus 401 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, Figure 3 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0089] Processor 402 can be any one or more of the following processors: central processing unit (CPU), graphics processing unit (GPU), microprocessor (MP), or digital signal processor (DSP).

[0090] Communication interface 403 is used for communication with external devices.

[0091] Memory 404 may include volatile memory, such as random access memory (RAM). Memory 404 may also include non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid state drive (SSD).

[0092] The memory 404 stores executable code, and the processor 402 executes the executable code to perform the aforementioned power grid planning scenario generation method based on procedural thinking chain.

[0093] Specifically, in achieving Figure 2 In the case of the illustrated embodiment, and Figure 2 When the modules or units of the power grid planning scenario generation device based on procedural thinking chain described in the embodiment are implemented through software, the execution... Figure 2 The software or program code required for the functions of each module / unit can be partially or entirely stored in memory 404. Processor 402 executes the program code corresponding to each unit stored in memory 404, and executes the aforementioned power grid planning scenario generation method based on procedural thinking chain.

[0094] This application also provides a computer-readable storage medium. The computer-readable storage medium can be any available medium that a computing device can store, or a data storage device such as a data center containing one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive). The computer-readable storage medium includes instructions that instruct the computing device to execute the above-described method for generating power grid planning scenarios based on procedural thinking chains.

[0095] This application also provides a computer program product comprising one or more computer instructions. When the computer instructions are loaded and executed on a computing device, all or part of the processes or functions described in this application are generated.

[0096] The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, or data center to another website, computer, or data center via wired (e.g., coaxial cable, fiber optic) or wireless (e.g., infrared, wireless, microwave, etc.) means.

[0097] When the computer program product is executed by a computer, the computer executes any of the aforementioned methods of the power grid planning scenario generation method based on procedural thinking chains. The computer program product can be a software installation package; when any of the aforementioned methods of the power grid planning scenario generation method based on procedural thinking chains is required, the computer program product can be downloaded and executed on the computer.

[0098] The descriptions of the processes or structures corresponding to the above figures each have their own emphasis. For parts of a process or structure that are not described in detail, please refer to the relevant descriptions of other processes or structures.

[0099] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be covered within the scope of protection of this application.

Claims

1. A method for generating power grid planning scenarios based on procedural thinking chains, characterized in that, The method includes: Obtain text-based datasets and power grid operating condition datasets; The text dataset is structured and parsed to obtain a set of planning boundary elements; constraints are then constructed based on the set of planning boundary elements to obtain the procedure boundary thought chain. By performing variable mapping on the procedure boundary thinking chain, we obtain the procedure boundary thinking chain and the corresponding scene perturbation operator after completing the scene variable mapping assignment. Based on the power grid operating condition dataset, the procedural boundary thinking chain that completes the scene variable mapping and assignment, and the scene disturbance operator, scene generation is performed to obtain candidate scenes. The candidate scenarios are simulated and verified to obtain valid candidate scenarios; the valid candidate scenarios are then screened using multi-dimensional indicators to obtain a target scenario library for power grid planning.

2. The method according to claim 1, characterized in that, The process of mapping variables to the procedure boundary thinking chain to obtain the procedure boundary thinking chain with completed scene variable mapping assignment and the corresponding scene perturbation operator includes: Using the constraint variables in the boundary thinking chain of the procedure as search conditions, the target mapping variable is determined by searching in the preset planning variable dictionary; Write the target mapping variable into the scenario variable mapping bit in the procedure boundary thinking chain to obtain the procedure boundary thinking chain that has completed the scenario variable mapping assignment. By binding the perturbation actions in the procedure boundary thinking chain with the written target mapping variables, the scene perturbation operator is obtained.

3. The method according to claim 1, characterized in that, The process of generating candidate scenarios based on the power grid operating condition dataset, the procedural boundary thinking chain that completes the scenario variable mapping assignment, and the scenario perturbation operator yields the following: Basic scenarios are obtained by clustering scenarios based on power grid operating condition datasets; Substitute the scenario parameters of each basic scenario into the Boolean trigger conditions in the procedural boundary thinking chain of each completed scenario variable mapping assignment, perform logical evaluation, and obtain the evaluation result; determine the thinking chain of the enabled state based on the evaluation result; Based on the scene variable mapping bits and scene perturbation operators corresponding to the enabled mind chain, the corresponding basic scene is subjected to variable perturbation processing to obtain the perturbated intermediate scene. The intermediate scene after the disturbance is processed to obtain the corresponding derived scene; The basic scenario, the intermediate scenario after disturbance, and the derived scenario are selected as candidate scenarios.

4. The method according to claim 1, characterized in that, The structured parsing of the text dataset yields a set of planning boundary features, including: Document slicing is performed on text datasets to obtain knowledge slices; Each knowledge slice is classified into different sentence types, and the sentences to be extracted are obtained based on the results of the sentence type classification. Boundary features are extracted from the statements to be extracted, resulting in the original boundary features. Perform field validity validation on the original boundary features to obtain the original boundary features that pass the validation; The original boundary elements that have passed the verification are aggregated to obtain the planning boundary element set.

5. The method according to claim 1, characterized in that, The process of simulating and validating candidate scenes to obtain effective candidate scenes includes: Each candidate scenario is written into the power grid simulation model and power flow calculation is performed to verify the candidate scenarios that pass the power flow calculation verification. The candidate scenarios verified by power flow calculation will be subjected to security and stability verification to obtain the candidate scenarios that pass the security and stability verification. Candidate scenarios that pass the security and stability verification will be identified as valid candidate scenarios.

6. The method according to claim 1, characterized in that, The process of screening effective candidate scenarios using multi-dimensional indicators to obtain a target scenario library for power grid planning includes: Calculate the multidimensional evaluation index for each valid candidate scenario to obtain the comprehensive score corresponding to each valid candidate scenario; Based on the mandatory coverage constraints and comprehensive scores, each valid candidate scenario is screened to obtain the target scenario library for power grid planning.

7. A power grid planning scenario generation device based on procedural thinking chain, characterized in that, The device includes: The acquisition module is used to acquire text-based datasets and power grid operating condition datasets. The data processing module is used to perform structured parsing on text datasets to obtain a set of planning boundary elements; to construct constraints based on the set of planning boundary elements to obtain a procedure boundary thinking chain; and to perform variable mapping on the procedure boundary thinking chain to obtain a procedure boundary thinking chain with completed scene variable mapping assignment and the corresponding scene perturbation operator. The scenario generation module is used to generate scenarios based on the power grid operating condition dataset, the procedural boundary thinking chain that completes the scenario variable mapping assignment, and the scenario disturbance operator to obtain candidate scenarios; to perform simulation verification on the candidate scenarios to obtain valid candidate scenarios; and to screen the valid candidate scenarios using multi-dimensional indicators to obtain the power grid planning target scenario library.

8. A computing device, characterized in that, Including memory and processor; The memory stores one or more computer programs, the one or more computer programs including instructions; when the instructions are executed by the processor, the computing device performs the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store a computer program for performing the method as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, The computer program product includes one or more computer instructions that, when executed by a computer, perform the method as described in any one of claims 1 to 6.

Images