A reservoir supervision method and device based on digital twinning and a medium

Abstract

The application discloses a reservoir supervision method and device based on digital twinning, and a medium, relates to the technical field of digital twinning supervision, and comprises the following steps: based on the current digital twinning supervision context, receiving supervision intention input information, carrying out semantic disintegration, object recognition, time period normalization and boundary completion, and generating a supervision action token sequence; extracting a supervision constraint condition set from the current digital twinning supervision context, and carrying out condition expansion, conflict resolution and path connection on the supervision constraint condition set to generate a supervision script graph. Through condition expansion, conflict resolution and path connection on the supervision constraint condition set, the supervision script graph is generated, the supervision state, the control action and the boundary condition can be organized into a continuous-executable data link, the candidate supervision path expansion, the script switching and the feedback correction are supported, and the coherence, the calculability and the closed-loop processing capability of the supervision process are improved.

Description

Technical Field

[0001] This invention relates to the field of digital twin supervision technology, and in particular to a reservoir supervision method, equipment and medium based on digital twins. Background Technology

[0002] With the advancement of water conservancy informatization and smart water conservancy construction, reservoir supervision methods are gradually evolving from decentralized monitoring to digital twin collaborative monitoring. The supervision methods are now able to aggregate topographic, structural, operational, and forecast data, and support scheduling analysis, risk identification, and operational control through state mapping, scenario reconstruction, and process simulation.

[0003] Existing methods mainly focus on status display, judgment, and single-time simulation. Although they can form monitoring conclusions and scheduling suggestions, they lack a processing method to unify regulatory intent, regulatory constraints, and operational status into a continuously executable path. This makes it difficult to complete the generation of actions, path switching, and closed-loop correction in the regulatory process within the same data chain, making it difficult to support continuous supervision in high-timeliness and high-constraint scenarios. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a reservoir supervision method based on digital twins to solve the problem that the supervision intent and constraints in digital twin reservoir supervision are difficult to transform into a continuously executable supervision path.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a reservoir supervision method based on digital twins, comprising: collecting multi-source reservoir supervision data, performing time alignment, spatial mapping, and association with supervised objects to construct a current digital twin supervision context; based on the current digital twin supervision context, receiving supervision intent input information, performing semantic decomposition, object identification, time period normalization, and boundary completion to generate a supervision action token sequence; extracting a set of supervision constraints from the current digital twin supervision context, and performing condition expansion, conflict resolution, and path connection on the set of supervision constraints to generate a supervision script graph; mapping the current digital twin supervision context to supervision state nodes in the supervision script graph, expanding candidate supervision paths along action edges according to the supervision action token sequence, and performing digital twin deduction and constraint verification to determine a target supervision script group; executing gate control and observation based on the main supervision script in the target supervision script group, collecting execution feedback data, switching scripts based on the execution feedback data, and correcting the supervision action token sequence and supervision script graph based on the execution feedback data to obtain the current digital twin supervision context for the next supervision cycle.

[0008] As a preferred embodiment of the reservoir supervision method based on digital twins described in this invention, the specific steps for constructing the current digital twin supervision context are as follows:

[0009] Unify the temporal, spatial location, and object identification of multi-source data for reservoir supervision, clean and convert missing, duplicate, and conflicting data to obtain a spatiotemporally correlated data set.

[0010] Based on the objects, states, and related information in the spatiotemporal correlated data set, the data is extracted, merged, and compiled to generate the current digital twin regulatory context.

[0011] As a preferred embodiment of the reservoir supervision method based on digital twins described in this invention, the supervision intent input information is interactive input data composed of natural language text information, timeline information, spatial selection information, control object selection information, and risk boundary information.

[0012] As a preferred embodiment of the reservoir supervision method based on digital twins described in this invention, the specific steps for generating the supervision action token sequence are as follows:

[0013] Given the current digital twin regulatory context, obtain the regulatory interaction data from the regulatory intent input information, perform format unification, noise removal and time alignment to obtain parsable intent data;

[0014] Based on the parsable intent data, intent element identifiers are extracted, and the intent element identifiers are associated, reorganized, and missing information is filled in to obtain structured intent data.

[0015] Based on structured intent data, field mapping is used to sequentially encode and tokenize the regulatory objects, time ranges, action sequences, and boundary restriction identifiers to generate a regulatory action token sequence.

[0016] As a preferred embodiment of the reservoir supervision method based on digital twins described in this invention, the specific steps for generating the supervision scenario diagram are as follows:

[0017] Based on the current regulatory context of digital twins, regulatory constraint element data is extracted, and then filtered, classified, and semantically merged to form a set of regulatory constraint conditions.

[0018] Based on the set of regulatory constraints, restrictions targeting the same regulatory object are split, conflicted, and their sequential relationships are sorted to obtain a set of connectable constraint fragments.

[0019] By using a set of connectable constraint fragments, links are spliced ​​and nodes are reorganized according to the regulated object, trigger time period, and action sequence to generate a regulatory script diagram.

[0020] As a preferred embodiment of the reservoir supervision method based on digital twins described in this invention, the target supervision script group refers to a set of scripts consisting of a main supervision script, a backup supervision script, and a rollback supervision script determined by digital twin deduction and constraint verification of candidate supervision paths.

[0021] As a preferred embodiment of the reservoir supervision method based on digital twins described in this invention, the specific steps for determining the target supervision script group are as follows:

[0022] Based on the current digital twin regulatory context and regulatory script diagram, regulatory status element data is extracted, and node matching and status reorganization are performed on the regulatory status element data to obtain regulatory status node mapping information.

[0023] Based on the regulatory status node mapping information and the regulatory action token sequence, obtain the path expansion element data, and perform sequential expansion, reachability filtering and path splicing to generate a candidate regulatory path set;

[0024] Extract the inference input data and constraint verification input data of the candidate regulatory path set, and perform time-series evolution, consistency comparison and boundary screening to obtain the target regulatory script group.

[0025] As a preferred embodiment of the reservoir supervision method based on digital twins described in this invention, the specific steps for obtaining the current digital twin supervision context for the next supervision cycle are as follows:

[0026] Based on the main regulatory script in the target regulatory script group, extract the gate control data and observation task data corresponding to the main regulatory script, and perform time-series processing, object matching and format conversion to obtain the execution task data;

[0027] Based on the task execution data, obtain the execution feedback element data, and perform time alignment, anomaly removal, and correlation reorganization on the execution feedback element data to form execution feedback data;

[0028] Based on execution feedback data, script switching information is generated by matching with the target regulatory script group. The script switching information is then used to rearrange and correct the regulatory action token sequence and regulatory script diagram to obtain the current digital twin regulatory context for the next regulatory cycle.

[0029] In a second aspect, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the reservoir supervision method based on digital twins as described in the first aspect of the present invention.

[0030] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the reservoir supervision method based on digital twins as described in the first aspect of the present invention.

[0031] The beneficial effects of this invention are as follows: by expanding, resolving conflicts, and connecting paths in the set of regulatory constraints, a regulatory script graph is generated, which can organize the regulatory state, control actions, and boundary conditions into a continuously executable data link, supporting the expansion of candidate regulatory paths, script switching, and feedback correction, thereby improving the coherence, computability, and closed-loop processing capabilities of the regulatory process. Attached Figure Description

[0032] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a flowchart of a reservoir supervision method based on digital twins.

[0034] Figure 2 A flowchart for building a regulatory context for digital twins.

[0035] Figure 3 A flowchart for generating regulatory action token sequences and script diagrams.

[0036] Figure 4 A flowchart for digital twin simulation and script switching. Detailed Implementation

[0037] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0038] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0039] Secondly, the term "one embodiment" or "example" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the invention. The appearance of an embodiment in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that mutually excludes other embodiments.

[0040] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides a reservoir supervision method based on digital twins, comprising the following steps:

[0041] S1: Collect multi-source data for reservoir monitoring, perform time alignment, spatial mapping, and association with monitored objects to construct the current digital twin monitoring context.

[0042] S1.1: Unify the time, spatial location and object identification of multi-source data for reservoir supervision, clean and convert missing, duplicate and conflicting data to obtain a spatiotemporally correlated data set.

[0043] Furthermore, the multi-source data for reservoir supervision is time-aligned, spatially mapped, and associated with the monitored objects to establish a unified correspondence between time information, spatial location, and object identifiers. The multi-source data for reservoir supervision is also cleaned, including filling in missing data, merging duplicate data, and filtering out conflicting data. After cleaning, the remaining data is format-converted and uniformly expressed to ensure consistency between time information, spatial location, and object identifiers, thus obtaining a spatiotemporally correlated data set.

[0044] It should be noted that the multi-source data for reservoir monitoring includes water level, flow rate, rainfall, dam deformation, seepage, water quality parameters, meteorological parameters, images and videos, operational status, and inspection records.

[0045] Unified correspondence refers to normalizing reservoir monitoring data from different sources, in different formats, and collected in different ways, according to the same time benchmark, the same spatial reference, and the same monitoring object identification rules. This ensures that any piece of data can be associated with a time, spatial location, and related monitoring object, and that data from different sources are consistently matched, comparable, and correlated across time, space, and object dimensions.

[0046] S1.2: Extract, merge, and organize the objects, states, and related information in the spatiotemporal associated data set to generate the current digital twin regulatory context.

[0047] Furthermore, based on the objects, states, and associated information in the spatiotemporal associated data set, the objects in the spatiotemporal associated data set are extracted according to the object identifier. The state information and associated information corresponding to the same object identifier are merged, and the merged object information, state information, and associated information are compiled according to the time sequence and spatial location to form a corresponding relationship between the object information, state information, and associated information, thereby generating the current digital twin regulatory context.

[0048] S2: Based on the current digital twin regulatory context, receive regulatory intent input information, perform semantic decomposition, object identification, time period normalization and boundary completion, and generate a regulatory action token sequence.

[0049] S2.1: Regulatory intent input information is interactive input data consisting of natural language text information, timeline information, spatial selection information, control object selection information, and risk boundary information.

[0050] Given the current digital twin regulatory context, regulatory interaction data from the regulatory intent input information is obtained, and then formatted, noise-removed, and time-aligned to obtain parsable intent data.

[0051] Furthermore, within the current digital twin regulatory context, the system acquires natural language text information, timeline information, spatial selection information, controlled object selection information, and risk boundary information from the regulatory intent input information. This information is then converted into a unified format. Duplicate, invalid, and blank content in the natural language text information is removed; overlapping and fragmented content in the timeline information is organized; and offset and discrete content in the spatial selection information is cleaned up. Finally, the natural language text information, timeline information, spatial selection information, controlled object selection information, and risk boundary information are aligned with the time information in the current digital twin regulatory context to obtain parseable intent data.

[0052] S2.2: Based on the parsable intent data, extract intent element identifiers, perform association, recombination, and missing completion on the intent element identifiers to obtain structured intent data.

[0053] Furthermore, based on the parseable intent data, text fragments, time fragments, spatial selection fragments, and selection fragments corresponding to the regulatory objects, time ranges, action sequences, and boundary restrictions are extracted from the parseable intent data to form intent element identifiers. Then, according to the relationship that the same regulatory object corresponds to the same time range and the same action sequence corresponds to the same boundary restriction, the intent element identifiers are associated and recombined, merging the scattered expressions into the same data item. For regulatory objects, time ranges, action sequences, and boundary restrictions that are missing after association and recombination, the missing information is filled in by combining the objects, states, and related information in the current digital twin regulatory context to obtain structured intent data.

[0054] The complete numerical calculation formula for intent data is as follows:

[0055] ;

[0056] in, Indicates the first The completeness value of the input information for the group's regulatory intent. The input group number indicates the regulatory intent. Indicates the type number of the input information. Indicates the total number of input information types. Indicates the first The number of valid data items in the input information. Indicates the first Group 1 Alignment ratio of input information.

[0057] It should be noted that the completeness value represents the degree of completeness of the regulatory object, time range, action sequence and boundary restriction identifier in each group of structured intent data. After obtaining each group of structured intent data, the structured intent data in each group is sorted according to the completeness value, and the structured intent data with the highest ranking is processed by field mapping, sequential encoding and tokenization conversion.

[0058] The alignment ratio refers to the proportion of the data. Group 1 Among the input information for regulatory intent, the number of data items that, after format standardization, noise removal, and time alignment, can establish a correspondence with the time information in the current digital twin regulatory context and enter the intent element extraction, accounts for the [percentage missing]. The proportion of valid data items in the input information.

[0059] S2.3: Based on the structured intent data, use field mapping to sequentially encode and tokenize the regulatory object, time range, action sequence, and boundary restriction identifier to generate a regulatory action token sequence.

[0060] Furthermore, based on the regulatory objects, time ranges, action sequences, and boundary restriction identifiers in the structured intent data obtained after sorting by completeness values, field mapping is used to map the regulatory objects, time ranges, action sequences, and boundary restriction identifiers to fixed field positions. The regulatory objects, time ranges, action sequences, and boundary restriction identifiers in the fixed field positions are sequentially encoded. The sequentially encoded regulatory objects, time ranges, action sequences, and boundary restriction identifiers are then tokenized according to their field positions. Finally, the tokenized regulatory objects, time ranges, action sequences, and boundary restriction identifiers are sequentially concatenated according to their action sequences and time ranges to generate a regulatory action token sequence.

[0061] Specifically, field mapping refers to matching and uniformly converting the field names, meanings, formats, and values ​​of data fields from different sources in reservoir monitoring multi-source data. This maps similar information from different data sources to the same target field, ensuring that time information, spatial location, object identification, and monitoring data have a consistent data expression method during the fusion processing.

[0062] Sequential coding is a process of arranging and marking the positions of various intent elements in structured intent data according to the regulatory object, time range, sequence of actions, and boundary restrictions.

[0063] Tokenization is the process of converting sequentially encoded regulatory objects, time ranges, action sequences, and boundary restriction identifiers into discrete tokens arranged in a uniform field format.

[0064] The regulatory action token sequence is a data sequence that represents the content and order of regulatory actions. It is formed by semantically decomposing, identifying objects, normalizing time periods, and completing boundaries of the regulatory intent input information, and then sequentially encoding and tokenizing it according to the regulatory object, time range, action sequence, and boundary restriction identifier.

[0065] S3: Extract the set of regulatory constraints from the current digital twin regulatory context, and expand, resolve conflicts and connect paths to generate a regulatory script graph.

[0066] S3.1: Based on the current digital twin regulatory context, extract regulatory constraint element data, and perform filtering, classification and semantic merging to form a set of regulatory constraint conditions.

[0067] Furthermore, object information, status information, and associated information in the current digital twin regulatory context are extracted. The content corresponding to the regulatory object, triggering time period, action sequence, and boundary restrictions in the object information, status information, and associated information is used as regulatory constraint element data. The regulatory constraint element data is filtered according to the regulatory object, and the filtered regulatory constraint element data is classified according to the triggering time period and action sequence. Then, semantically consistent and semantically repetitive content in the classified regulatory constraint element data is semantically merged to form a set of regulatory constraint conditions.

[0068] Specifically, the triggering period refers to the time range corresponding to when data, events, and processing actions are identified, recorded, and initiated, representing the specific time interval in which content occurs and takes effect.

[0069] Action sequence refers to the order in which multiple data, events, and processing actions are arranged according to their occurrence, execution, and processing sequence, representing the sequential relationship between various elements.

[0070] By clearly defining the triggering time period and the sequence of actions, data from different sources can be presented in a distinguishable and correlated manner over time, facilitating link splicing, node reorganization, and process analysis.

[0071] S3.2: Based on the set of regulatory constraints, perform condition splitting, conflict comparison, and sequence organization on the restrictions that point to the same regulatory object to obtain a set of connectable constraint fragments.

[0072] Furthermore, based on the set of regulatory constraints, the restrictions are categorized according to the regulatory objects. Restrictions pointing to the same regulatory object are split into conditions based on trigger time, action sequence, and boundary restriction identifiers, forming restrictions corresponding to different trigger time, different action sequences, and different boundary restriction identifiers. Conflict comparisons are then performed on the restrictions after condition splitting to identify restrictions with inconsistent action sequences, inconsistent boundary restrictions, and inconsistent restriction directions within the same trigger time. The sequential relationships of the restrictions after conflict comparison are then organized, and restrictions that can be continuously connected according to action sequence are linked to obtain a set of connectable constraint fragments.

[0073] The formula for calculating the success rate of constrained segment connection is:

[0074] ;

[0075] in, Indicates the first The degree of connectivity of constraint segments corresponding to each regulatory status node represents the formation of connectable constraint segments after condition splitting, conflict comparison, and sequence relationship sorting. Indicates the regulatory status node number, Indicates the constraint feature type number. Indicates the total number of constraint feature types. Indicates the first The number of condition split items for a class constraint element. Indicates the first The node The proportion of conflict-free elements in a class of constraint elements.

[0076] It should be noted that the degree of constraint segment connectivity is used to characterize the formation of connectable constraint segments corresponding to each regulatory status node. After obtaining the set of connectable constraint segments, the set of connectable constraint segments corresponding to the same regulatory object is sorted according to the degree of constraint segment connectivity, and the sorted set of connectable constraint segments is then processed by link splicing and node reorganization.

[0077] The conflict-free ratio refers to the first The first regulatory status node corresponding to the Among the constraint elements, the number of constraints that remain after conflict comparison after condition decomposition accounts for the largest proportion. The proportion of the number of condition split items of class constraint elements, the conflict comparison includes identifying constraint items with inconsistent action order, inconsistent boundary restrictions, and inconsistent restriction directions.

[0078] S3.3: Using a set of connectable constraint fragments, link splicing and node reorganization are performed according to the regulatory object, trigger time period and action sequence to generate a regulatory script diagram.

[0079] Furthermore, using the set of connectable constraint fragments sorted by the degree of connection of constraint fragments, the set of connectable constraint fragments is classified according to the regulatory object, and the set of connectable constraint fragments corresponding to the same regulatory object is sorted according to the trigger time period. The sorted set of connectable constraint fragments is spliced ​​together according to the action sequence. The set of connectable constraint fragments with continuous trigger time periods and connected action sequences is combined into regulatory state nodes. The connection relationship formed by the link splicing is organized into action edges, and the classified regulatory state nodes are reorganized to generate a regulatory script graph.

[0080] It should be noted that link splicing is based on time information, spatial location and object identification that have been uniformly processed, to connect related content scattered in different data, and to connect data segments that belong to the same regulatory object and can be matched to form a continuous data link.

[0081] Node reorganization is the process of reorganizing and repackaging the data nodes in the link after the link is spliced, according to the unified time information, spatial location, object identifier and field expression method, so as to make the correspondence between the nodes clearer and more consistent.

[0082] For example, during reservoir monitoring, water level monitoring data, gate control data, and inspection record data are obtained at the same time. Water level monitoring data shows that the water level in front of the dam is 126.5m at 08:00, gate control data shows that gate No. 2 is 20% open at 08:05, and inspection record data shows that gate No. 2 is operating normally at 08:10.

[0083] Aligning the temporal information of multi-source data, uniformly mapping spatial locations, and uniformly associating object identifiers enable the three types of data to be mapped to the same reservoir area and the same monitored object. Connecting corresponding data content from different periods forms a continuous data link reflecting the state changes of the monitored object; this process is called link splicing. After the link splicing is completed, the names, field formats, and organization of the node data in the link are uniformly organized. For example, different expressions such as "Gate 2," "Gate No. 2," and "Gate Two" are merged into a consistent object identifier, and the content of each node is reorganized according to unified rules; this process is called node reorganization.

[0084] S4: Map the current digital twin regulatory context to the regulatory state node in the regulatory script graph, expand the candidate regulatory path along the action edge according to the regulatory action token sequence, and perform digital twin inference and constraint verification to determine the target regulatory script group.

[0085] S4.1: The target regulatory script group refers to the set of scripts consisting of the main regulatory script, the backup regulatory script, and the rollback regulatory script, which are determined by digital twin simulation and constraint verification of candidate regulatory paths.

[0086] Based on the current digital twin regulatory context and regulatory script diagram, regulatory status element data is extracted, and node matching and status reorganization are performed on the regulatory status element data to obtain regulatory status node mapping information.

[0087] Furthermore, based on the current digital twin regulatory context and regulatory script diagram, the objects, states, and associated information corresponding to the regulatory state nodes are extracted from the current digital twin regulatory context to form regulatory state element data. Then, according to the regulatory objects, triggering time periods, and action sequences corresponding to the regulatory state nodes in the regulatory script diagram, node matching is performed on the regulatory state element data. The objects, states, and associated information corresponding to the same regulatory state node are merged. The regulatory state element data that have completed node matching are reorganized into states. The objects, states, and associated information corresponding to the same regulatory state node are organized according to the connection relationships in the regulatory script diagram to obtain regulatory state node mapping information.

[0088] S4.2: Based on the regulatory status node mapping information and the regulatory action token sequence, obtain the path expansion element data, and perform sequential expansion, reachability filtering and path splicing to generate a candidate regulatory path set.

[0089] Furthermore, based on the regulatory status node mapping information, the regulatory status nodes that have been mapped in the regulatory script graph are determined, and the action sequence and boundary restriction identifiers corresponding to the regulatory status nodes are extracted based on the regulatory action token sequence. The action edge information connected to the regulatory status nodes is obtained from the regulatory script graph, and the action edge information, action sequence, and boundary restriction identifiers are integrated into path expansion element data.

[0090] When sequentially expanding the path expansion element data, the regulatory status nodes and action edges are connected in the order of actions in the regulatory action token sequence. When performing reachability filtering on the path expansion element data, action edges and interrupted links that cannot be reached from the current digital twin regulatory context are filtered out based on the regulatory status node mapping information. When performing path splicing on the path expansion element data, the remaining action edges are spliced ​​in the order of actions in the regulatory action token sequence according to the action edge connection relationship in the regulatory script diagram, generating a candidate regulatory path set.

[0091] S4.3: Extract the inference input data and constraint verification input data of the candidate regulatory path set, and perform time-series evolution, consistency comparison and boundary screening to obtain the target regulatory script group.

[0092] Furthermore, the regulatory status nodes, action edges, triggering time periods, and action sequences contained in each candidate regulatory path are extracted from the candidate regulatory path set. Combined with the object status information in the current digital twin regulatory context, the inference input data is formed. The regulatory constraint condition set with each candidate regulatory path is extracted from the regulatory script graph to form the constraint verification input data.

[0093] The simulation input data is subjected to temporal evolution according to the trigger time period and action sequence to obtain the state evolution data of each candidate regulatory path. The state evolution data is compared with the constraint verification input data to identify conflicts and inconsistencies between regulatory state nodes, action edges and trigger time periods. Boundary filtering is performed on the state evolution data to retain candidate regulatory paths that meet the regulatory constraint condition set. The candidate regulatory paths that meet the regulatory constraint condition set are merged according to the execution priority relationship and switching relationship to obtain the target regulatory script group.

[0094] It should be noted that the simulation input data is formed by extracting the regulatory status nodes, action edges, triggering time periods and action sequences contained in each candidate regulatory path from the candidate regulatory path set, and combining them with the object status information in the current digital twin regulatory context.

[0095] The constraint verification input data is data extracted from the regulatory script diagram and formed by the set of regulatory constraints corresponding to each candidate regulatory path; the constraint verification input data includes the content of the regulatory object, the content of the triggering time period, the content of the action sequence, and the content of the boundary restriction.

[0096] For example, in reservoir monitoring, in response to the situation of increased upstream water flow and continuous rise in water level in front of the dam, three candidate monitoring paths were generated. The first path is to issue an early warning at 09:00 → open gate No. 2 by 20% at 09:10 → arrange inspection at 09:20. The second path is to issue an early warning at 09:00 → open gates No. 1 and No. 2 by 30% at 09:05. The third path is to issue an early warning at 09:00 → wait 30 minutes before opening the gates.

[0097] The regulatory scenario diagram sets constraints, such as the gate opening range not exceeding the downstream flow rate in the same period, the downstream capacity limit, and the need to implement control measures within 20 minutes after the warning is issued. After analyzing the three paths, it was found that although the second path is fast, it would cause the downstream flow rate to exceed the limit. The third path is too delayed and does not meet the time limit for handling after the warning. Only the first path meets both the restrictions and the downstream flow constraints and response time requirements. Therefore, the second and third paths are eliminated, and only the first path is kept. The candidate regulatory paths that meet the set of regulatory constraints are retained. In other words, schemes that do not conform to actual regulation are screened out, leaving only executable, compliant, and safe paths.

[0098] S5: Based on the main regulatory script execution gate control and observation in the target regulatory script group, collect execution feedback data, switch scripts according to the execution feedback data, and correct the regulatory action token sequence and regulatory script diagram according to the execution feedback data to obtain the current digital twin regulatory context for the next regulatory cycle.

[0099] S5.1: Based on the main regulatory script in the target regulatory script group, extract the gate control data and observation task data corresponding to the main regulatory script, and perform time-series processing, object matching and format conversion to obtain the execution task data.

[0100] Furthermore, based on the main regulatory script in the target regulatory script group, the gate control data and observation task data corresponding to the main regulatory script are extracted according to the action sequence, trigger time period, and regulatory object in the main regulatory script. The control time period, control object, and control content in the gate control data are correlated with the observation time period, observation object, and observation content in the observation task data. The correlated gate control data and observation task data are then time-series organized. The gate control data and observation task data corresponding to the same regulatory object are matched. Finally, the field content, field order, and data representation of the gate control data and observation task data are format-converted to obtain the execution task data.

[0101] S5.2: Based on the task execution data, obtain the execution feedback element data, and perform time alignment, anomaly removal, and correlation reorganization on the execution feedback element data to form execution feedback data.

[0102] Furthermore, based on the time sequence organization, object pairing, and format conversion information in the execution task data, execution feedback element data corresponding to the gate control data and observation task data are obtained. The execution feedback element data is time-aligned according to the time sequence organization information. The time-aligned execution feedback element data is then identified and anomalies are removed according to the object pairing information. Finally, the anomaly-removed execution feedback element data is associated and recombined according to the gate control data, observation task data, time alignment information, and object pairing information to form execution feedback data.

[0103] It should be noted that the execution feedback element data refers to the set of data items corresponding to the gate control data and observation task data, obtained based on the time sequence organization information, object pairing information and format conversion information in the execution task data.

[0104] Object pairing information refers to the correspondence between gate control data and observation task data in the execution task data for the same regulatory object, and the corresponding identification and association reorganization of execution feedback element data.

[0105] S5.3: Based on execution feedback data, generate script switching information by matching with the target regulatory script group, and use the script switching information to rearrange and correct the regulatory action token sequence and regulatory script diagram to obtain the current digital twin regulatory context for the next regulatory cycle.

[0106] Furthermore, based on the execution feedback data, the execution feedback data is matched with the main regulatory script, backup regulatory script, and rollback regulatory script in the target regulatory script group. After the execution feedback data has been time-aligned, anomaly removed, and association reorganized, the object information, status information, and association information in the execution feedback data are matched item by item with the regulatory status nodes, action edges, triggering time periods, and action sequences corresponding to the main regulatory script, backup regulatory script, and rollback regulatory script in the target regulatory script group.

[0107] When the execution feedback data corresponds to continuous regulatory status nodes and action edges on the candidate regulatory path, the location of the regulatory path corresponding to the execution feedback data is determined. When the execution feedback data corresponds to the sequential connection relationship before and after the action edge, the action sequence position corresponding to the execution feedback data is determined. When the continuous matching relationship between the execution feedback data and the regulatory status nodes, action edges, triggering time periods, and action sequences corresponding to the current main regulatory script is interrupted, and a continuous matching relationship is formed between the execution feedback data and the regulatory status nodes, action edges, triggering time periods, and action sequences corresponding to the backup regulatory script and the rollback regulatory script, the script switching position corresponding to the execution feedback data is determined, and script switching information is generated accordingly.

[0108] By utilizing script switching information, the order of actions, retained content, and replaced content in the regulatory action token sequence are rearranged and corrected. The connection relationships of regulatory status nodes, action edges, and switching edges in the regulatory script graph are also rearranged and corrected. The rearranged and corrected regulatory action token sequence and the rearranged and corrected regulatory script graph are mapped to the object information, status information, and association information in the execution feedback data to obtain the current digital twin regulatory context for the next regulatory cycle.

[0109] It should be noted that the criteria for determining whether something corresponds or no longer corresponds are that, after time alignment, anomaly removal, and correlation reorganization, the object information, status information, and correlation information in the execution feedback data are matched item by item with the regulatory status nodes, action edges, triggering time periods, and action sequences corresponding to the current main regulatory script, backup regulatory script, and rollback regulatory script.

[0110] When the execution feedback data can be continuously mapped to adjacent regulatory state nodes and action edges according to the path connection relationship in the current main regulatory script, it is determined that the execution feedback data corresponds to the current main regulatory script; when the path connection relationship of the execution feedback data in the current main regulatory script is not continuous, but can be continuously mapped to adjacent regulatory state nodes and action edges according to the path connection relationship in the backup regulatory script and the rollback regulatory script, it is determined that the execution feedback data no longer corresponds to the current main regulatory script but corresponds to the backup regulatory script and the rollback regulatory script.

[0111] This embodiment also provides a computer device applicable to the reservoir supervision method based on digital twins, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the reservoir supervision method based on digital twins as proposed in the above embodiment.

[0112] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC, or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0113] This embodiment also provides a storage medium on which a computer program is stored. When executed by a processor, the program implements the reservoir supervision method based on digital twins as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0114] In summary, this invention generates a regulatory script graph by expanding, resolving conflicts, and connecting paths in the set of regulatory constraints. This graph organizes regulatory states, control actions, and boundary conditions into continuously executable data links, supporting candidate regulatory path expansion, script switching, and feedback correction, thereby improving the coherence, computability, and closed-loop processing capabilities of the regulatory process.

[0115] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A reservoir monitoring method based on digital twins, characterized in that, include: Collect multi-source data on reservoir monitoring, perform time alignment, spatial mapping, and association with monitored objects to construct the current digital twin monitoring context; Based on the current digital twin regulatory context, it receives regulatory intent input information, performs semantic decomposition, object identification, time period normalization and boundary completion, and generates a regulatory action token sequence; Extract the set of regulatory constraints from the current digital twin regulatory context, and expand, resolve conflicts and connect paths to generate a regulatory script graph. Map the current digital twin regulatory context to the regulatory state node in the regulatory script graph, expand candidate regulatory paths along the action edges according to the regulatory action token sequence, and perform digital twin simulation and constraint verification to determine the target regulatory script group. Based on the gate control and observation of the main regulatory script in the target regulatory script group, execution feedback data is collected, script switching is performed according to the execution feedback data, and the regulatory action token sequence and regulatory script diagram are corrected according to the execution feedback data to obtain the current digital twin regulatory context for the next regulatory cycle.

2. The reservoir supervision method based on digital twin as described in claim 1, characterized in that, The specific steps for constructing the current digital twin regulatory context are as follows: Unify the temporal, spatial location, and object identification of multi-source data for reservoir supervision, clean and convert missing, duplicate, and conflicting data to obtain a spatiotemporally correlated data set. Based on the objects, states, and related information in the spatiotemporal correlated data set, the data is extracted, merged, and compiled to generate the current digital twin regulatory context.

3. The reservoir supervision method based on digital twins as described in claim 1, characterized in that, The regulatory intent input information is interactive input data composed of natural language text information, timeline information, spatial selection information, control object selection information, and risk boundary information.

4. The reservoir supervision method based on digital twin as described in claim 3, characterized in that, The specific steps for generating the regulatory action token sequence are as follows: Given the current digital twin regulatory context, obtain the regulatory interaction data from the regulatory intent input information, perform format unification, noise removal and time alignment to obtain parsable intent data; Based on the parsable intent data, intent element identifiers are extracted, and the intent element identifiers are associated, reorganized, and missing information is filled in to obtain structured intent data. Based on structured intent data, field mapping is used to sequentially encode and tokenize the regulatory objects, time ranges, action sequences, and boundary restriction identifiers to generate a regulatory action token sequence.

5. The reservoir supervision method based on digital twin as described in claim 1, characterized in that, The specific steps for generating the regulatory script diagram are as follows: Based on the current regulatory context of digital twins, regulatory constraint element data is extracted, and then filtered, classified, and semantically merged to form a set of regulatory constraint conditions. Based on the set of regulatory constraints, restrictions targeting the same regulatory object are split, conflicted, and their sequential relationships are sorted to obtain a set of connectable constraint fragments. By using a set of connectable constraint fragments, links are spliced ​​and nodes are reorganized according to the regulated object, trigger time period, and action sequence to generate a regulatory script diagram.

6. The reservoir supervision method based on digital twin as described in claim 1, characterized in that, The target regulatory script group refers to a set of scripts consisting of the main regulatory script, backup regulatory script, and rollback regulatory script, which are determined by digital twin simulation and constraint verification of candidate regulatory paths.

7. The reservoir supervision method based on digital twin as described in claim 6, characterized in that, The specific steps for identifying the target monitoring script group are as follows: Based on the current digital twin regulatory context and regulatory script diagram, regulatory status element data is extracted, and node matching and status reorganization are performed on the regulatory status element data to obtain regulatory status node mapping information. Based on the regulatory status node mapping information and the regulatory action token sequence, obtain the path expansion element data, and perform sequential expansion, reachability filtering and path splicing to generate a candidate regulatory path set; Extract the inference input data and constraint verification input data of the candidate regulatory path set, and perform time-series evolution, consistency comparison and boundary screening to obtain the target regulatory script group.

8. The reservoir supervision method based on digital twin as described in claim 1, characterized in that, The specific steps for obtaining the current digital twin regulatory context for the next regulatory cycle are as follows: Based on the main regulatory script in the target regulatory script group, extract the gate control data and observation task data corresponding to the main regulatory script, and perform time-series processing, object matching and format conversion to obtain the execution task data; Based on the task execution data, obtain the execution feedback element data, and perform time alignment, anomaly removal, and correlation reorganization on the execution feedback element data to form execution feedback data; Based on execution feedback data, script switching information is generated by matching with the target regulatory script group. The script switching information is then used to rearrange and correct the regulatory action token sequence and regulatory script diagram to obtain the current digital twin regulatory context for the next regulatory cycle.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the reservoir supervision method based on digital twins as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the reservoir supervision method based on digital twins as described in any one of claims 1 to 8.

Images