Power grid operation maintenance risk pre-judgment method, device and equipment and storage medium

Abstract

The invention discloses a power grid operation maintenance risk pre-judgment method and device, equipment and a storage medium, and the method comprises the steps: obtaining a dispatching instruction generated by a power system, carrying out the feature analysis of the dispatching instruction, and extracting dispatching operation information; based on the scheduling operation information, matching a scheduling deduction rule matched with the scheduling instruction from a preset switching operation deduction rule base; based on the scheduling operation information, performing switching operation deduction on the scheduling instruction by using a scheduling deduction rule to generate a switching operation sequence; and based on a preset power grid model and an anti-error rule, performing anti-error judgment on the switching operation sequence in combination with the scheduling operation information, and outputting a risk pre-judgment result based on an anti-error judgment result. The method can significantly improve the scheduling processing efficiency, reduces human errors, adapts to the dynamic change and complex working conditions of the power grid, accurately carries out graded early warning, guarantees the data safety, and improves the operation and maintenance safety and reliability of the power grid in an all-round manner.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power grid maintenance, and particularly relates to a power grid operation and maintenance risk prediction method, device, equipment and storage medium. BACKGROUND

[0002] With the continuous expansion and complication of the power grid, faults caused by external environmental factors during power grid operation occur frequently, especially under extreme weather conditions, and the power grid failure rate significantly increases. How to effectively improve the power grid dispatching decision-making ability under extreme weather conditions and reduce the power grid failure risk has become an important topic in current power grid safety research.

[0003] The current mainstream solution is to analyze historical operation data, summarize the rules and build an evaluation model. Although this method can reduce the failure rate to a certain extent, it has obvious limitations. First, as disclosed in document 1 (publication number CN120634087A), a power grid situation deduction risk assessment method based on reinforcement learning is disclosed, a power grid situation model is constructed based on device health state assessment results and historical operation data, the power grid operation state under different maintenance plans is predicted, and the non-planned power outage risk probability is calculated. The model construction in this method depends on a large amount of historical data, but the actual available data often cannot cover all possible abnormal situations, resulting in insufficient model accuracy and generalization ability. Secondly, with the advancement of the modernization process of the power grid, the structure and operation mode of the power grid are becoming increasingly complex, and the traditional evaluation scheme is difficult to meet the high standard requirements of modern power grids for safety and reliability. In particular, in key links such as switching operation, there is a lack of intelligent risk prediction mechanism, and operation personnel often need to rely on experience to judge, which has a large risk of human error.

[0004] In addition, the existing dispatching instruction processing method is mostly artificial analysis and judgment, which is low in efficiency and prone to errors. In terms of error prevention judgment, the conventional method such as the dispatching operation knowledge base in document 2 (publication number CN120636390A) is constructed based on historical data and expert experience, which belongs to a static rule base and cannot update the power grid topology changes (such as network reconfiguration caused by fault tripping) in real time. Similar to the prior art of document 1, the existing technology usually adopts static rule checking, which is difficult to fully consider the potential risks brought by the dynamic changes of the power grid topology. These problems seriously restrict the safety of power grid operation and the efficiency of emergency disposal. SUMMARY

[0005] To solve the problems in the prior art, the present application provides a power grid operation and inspection risk prediction method, device, equipment and storage medium, obtains a dispatching instruction generated by a power system, analyzes the features of the dispatching instruction, and extracts dispatching operation information; based on the dispatching operation information, a dispatching deduction rule matched with the dispatching instruction is matched from a preset breaker switching operation deduction rule library; based on the dispatching operation information, the dispatching deduction rule is used to deduce the breaker switching operation of the dispatching instruction, and a breaker switching operation sequence is generated; based on a preset power grid model and a mistake prevention rule, the breaker switching operation sequence is judged based on the dispatching operation information, and a risk prediction result is output based on the result of the mistake prevention judgment. The present application can significantly improve the dispatching processing efficiency, reduce human errors, adapt to power grid dynamic changes and complex working conditions, accurately grade early warning and ensure data security, and comprehensively enhance the safety and reliability of power grid operation and inspection.

[0006] The present application adopts the following technical solutions.

[0007] The present application provides a power grid operation and inspection risk prediction method, which comprises the following steps: obtaining a dispatching instruction, analyzing the features of the dispatching instruction, and extracting dispatching operation information; wherein the dispatching operation information comprises an operation object, an operation type, an operation initial state and a target operation state; based on the dispatching operation information, a dispatching deduction rule matched with the dispatching instruction is matched from a preset breaker switching operation deduction rule library; the dispatching deduction rule is a constraint condition set for adjusting the execution order of the operation object; based on the dispatching operation information, the dispatching deduction rule is used to deduce the breaker switching operation of the dispatching instruction, and a breaker switching operation sequence is generated; based on a preset power grid model and a mistake prevention rule, the breaker switching operation sequence is judged based on the dispatching operation information, and a risk prediction result is output based on the result of the mistake prevention judgment.

[0008] Further preferably, the dispatching instruction is converted into a unified structured data format, and the dispatching instruction after format conversion is analyzed using NLP technology and named entity recognition technology to extract the dispatching operation information.

[0009] Further preferably, based on the operation object and the operation type, a plurality of candidate rules are retrieved from a preset breaker switching operation deduction rule library; a target deduction rule with a state element being the operation initial state and the operation target state is selected from the plurality of candidate rules, and the target deduction rule is used as the dispatching deduction rule corresponding to the dispatching instruction; If no rule completely matches the dispatching instruction, a similarity algorithm is used to select the closest rule from the candidate rules as the target deduction rule, and the difference part in the target deduction rule is adjusted to obtain a dispatching deduction rule matching the dispatching instruction.

[0010] Further preferably, the candidate rules refer to a rule set preliminarily retrieved from the switching operation deduction rule library and associated with the operation object and the operation type; The difference part refers to the rule content in the target deduction rule that is inconsistent with the dispatching operation information in the state element or operation logic; Adjusting according to the marked difference part includes modifying the device state threshold or adjusting the operation sequence to finally generate a deduction rule adapted to the current dispatching instruction; The similarity scores of each candidate rule and the dispatching instruction in the operation object, operation type, operation initial state, and operation target state are calculated; The operation object, operation type, operation initial state, and operation target state are assigned weights, and the total similarity score is obtained by weighted sum of each similarity score; The candidate rule with the highest total similarity score is selected as the target deduction rule.

[0011] Further preferably, the object sequence is obtained by sorting all operation objects in the dispatching operation information according to the dispatching deduction rule; the object sequence refers to a linear execution order generated according to the correlation between operation objects; Based on the operation initial state of each operation object in the dispatching operation information, the starting state of each operation object in the object sequence is set; The object sequence after setting the starting state is gradually deduced until the final state of each operation object is the corresponding operation target state, and a complete switching operation sequence is obtained; the state change logic refers to the conversion condition followed by the operation object from the operation initial state to the operation target state.

[0012] Further preferably, before the anti-misoperation judgment, the switching operation sequence is preliminarily checked, including: Load the preset verification rule library, which contains basic logical constraints and typical operation modes of power grid operation; Step-by-step analysis of the switching operation sequence is performed in combination with the verification rule library, and simulation deduction is performed for the execution condition of each operation step; When an abnormal situation occurs that does not conform to the verification rule library, a corresponding verification report is generated and a sequence reconstruction mechanism is triggered, including inserting a missing step or adjusting the operation sequence to perfect the switching operation sequence.

[0013] Further preferably, the preset power grid model and the anti-misoperation rule are combined with the dispatching operation information to perform anti-misoperation judgment on the switching operation sequence, and based on the result of the anti-misoperation judgment, a comprehensive risk value is calculated by using a multi-index weighted scoring model based on a scenario coefficient, and a risk prediction result is output according to the comprehensive risk value, specifically including: The power grid model is constructed based on graph theory to connect power equipment, and uses an adjacency matrix or an adjacency list data structure to implement; The switching operation sequence is mapped into the power grid model to generate a topological change graph of all operation objects in the switching operation sequence; The topological change graph includes device connection state, electrical parameters and operation steps.

[0014] Further preferably, based on the topological change graph and the anti-misoperation rule, a depth-first search method and a breadth-first search method are used to perform topological search on the switching operation sequence, and risk prediction is performed based on the search result, specifically including: When initializing the search queue, the depth limit and the breadth limit of the search are set; Based on the depth limit and the breadth limit, the path and node information of each operation object in the topological change graph are searched; The recorded path and node information are judged whether they violate the anti-misoperation rule, and a risk prediction result is output based on the result of the anti-misoperation judgment.

[0015] Further preferably, the path and the node information refer to the state change trajectory of the operation object in the topological change graph, and are stored by using an adjacency list or a matrix data structure; The definition of the anti-misoperation rule specifically includes: Actual cases and feedback information in the process of power grid operation are collected, and the anti-misoperation rule is recorded and archived; and the rule is version managed, and the modification history of the rule is recorded.

[0016] Further preferably, the risk prediction based on the result of the judgment includes: If there is an operation object that violates the anti-misoperation rule in the result of the judgment, the risk level of the operation object that violates the anti-misoperation rule is determined based on the severity level of the potential risk; and corresponding warning prompt information is generated based on the risk level; The process of obtaining the severity level of the potential risk is as follows: for the potential risk, a multi-index weighted scoring model based on a scenario coefficient is used to calculate a comprehensive risk value, the comprehensive risk value is compared with a set level threshold interval, and the risk level is obtained; The pre-warning prompt information refers to the warning content automatically matched according to different risk levels, including risk type, level, influence range and suggested measures.

[0017] Further preferably, the comprehensive risk value is calculated based on a multi-index weighted scoring model of a scene coefficient, and the specific steps are as follows: For different scene types, corresponding scene factors are set according to experience and actual situation; the specific scene of the operation object violating the anti-misoperation rule is determined, the specific scene is matched with the different scene types, the scene factors corresponding to the matched several scene types are summed, and the sum is added to 1 to obtain the scene coefficient; The multi-indexes include a number of power-off users, equipment maintenance cost and rush repair response time; wherein, the ratio of the number of power-off users to the total number of users is calculated, multiplied by 100 times, as the number of power-off users; the number of equipment needing maintenance is multiplied by the corresponding maintenance cost and summed, and the sum is divided by the average total cost threshold of the equipment, and the average total cost threshold is multiplied by 100 times to obtain the equipment maintenance cost index; the estimated rush repair time is divided by the standard rush repair time, and the result is multiplied by 100 times to obtain the rush repair response time index; The weight coefficients of the number of power-off users, the equipment maintenance cost and the rush repair response time are set, and based on the weight coefficients, each index is weighted and summed, and the sum is multiplied by the scene coefficient to obtain the comprehensive risk value; the weight coefficients are set based on actual situation.

[0018] The application also provides a power grid operation and inspection risk prediction system, which comprises an operation information acquisition module, a dispatching deduction rule matching module, a switching operation sequence generation module and a risk prediction module: The operation information acquisition module acquires the dispatching instruction generated by the power system, analyzes the characteristics of the dispatching instruction, and extracts the dispatching operation information; wherein, the dispatching operation information includes operation object, operation type, operation initial state and target operation state; The dispatching deduction rule matching module matches the dispatching deduction rule matched with the dispatching instruction from the preset switching operation deduction rule library based on the dispatching operation information; the dispatching deduction rule refers to the constraint condition set of adjusting the operation object execution sequence; The risk prediction module, the switching operation sequence generation module, based on the dispatching operation information, deduces the switching operation of the dispatching instruction by using the dispatching deduction rule, and generates the switching operation sequence; Based on the preset power grid model and the anti-misoperation rule, the switching operation sequence is judged based on the dispatching operation information, and the risk prediction result is output based on the result of the anti-misoperation judgment.

[0019] The present invention also proposes a device comprising a processor and a storage medium: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps according to the method described above.

[0020] The present invention also proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0021] 1. This invention acquires dispatch instructions generated by the power system (OMS system), performs feature parsing on the dispatch instructions, and extracts dispatch operation information. Based on the dispatch operation information, it matches dispatch deduction rules that match the dispatch instructions from a preset switching operation deduction rule base. Based on the dispatch operation information, it uses the dispatch deduction rules to perform switching operation deduction on the dispatch instructions, generating a switching operation sequence. Based on a preset power grid model and error prevention rules, it combines the dispatch operation information to perform error prevention judgment on the switching operation sequence, and outputs a risk prediction result based on the error prevention judgment. By automatically parsing dispatch instructions, intelligently matching deduction rules to generate operation sequences, and combining a dynamic error prevention judgment mechanism, this invention solves the problems of relying on human experience and insufficient static rule verification in existing technologies, thereby improving operational accuracy, reducing the risk of human error, and enhancing the safety of power grid operation.

[0022] 2. This invention constructs a structured switching operation simulation rule base and achieves rapid rule location through a two-layer mechanism: initial retrieval based on "operation object + operation type" and precise matching based on "initial state + target state." For scenarios without perfectly matching rules, a similarity algorithm is introduced to weightedly calculate the suitability of candidate rules, and manual adjustment of differences is supported, forming a collaborative mode of "automatic matching + manual verification," thus solving the problem that static rule bases cannot cover complex scheduling scenarios. Simultaneously, the rule base has a regular review and dynamic update mechanism, allowing for continuous optimization as the power grid structure changes, new equipment is introduced, and fault cases accumulate, ensuring that the rule system iterates synchronously with the actual needs of power grid operation.

[0023] 3. This invention, based on the power grid topology and equipment relationships, generates a sequence of operational objects through topological sorting of a directed acyclic graph. Combined with a finite state machine model, it achieves step-by-step state deduction, ensuring that the operational steps comply with equipment operating specifications and electrical logic. A preliminary verification step for the operational sequence is added. By loading a verification rule base, the logicality and completeness of the steps are simulated and deduced, automatically identifying and correcting issues such as missing steps and incorrect sequences. This avoids the risk of equipment malfunction due to operational sequence loopholes, significantly improving the standardization and safety of switching operations.

[0024] 4. This invention abandons the traditional approach of relying on historical data to build models. Instead, it achieves risk prediction through a combination of rule base and topology simulation. It effectively addresses new fault modes in complex scenarios such as typhoons and snowstorms without requiring the accumulation of large amounts of scarce data on extreme weather and abnormal operating conditions. The power grid model uses an adjacency matrix or adjacency list data structure, supporting rapid adaptation to power grid structure adjustments (such as adding substations or upgrading lines). The topology change graph is updated in real time through a dynamic graph database, ensuring accurate identification of hidden risks even in dynamic scenarios such as power grid reconfiguration, thus improving the efficiency of power grid emergency response and operational reliability. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the power grid operation and maintenance risk prediction method in Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the power grid operation and maintenance risk prediction method in Embodiment 2 of the present invention; Figure 3 This is a flowchart illustrating one method for preventing false judgments in an embodiment of the present invention; Figure 4 This is a schematic diagram of one embodiment of the power grid operation and maintenance risk prediction device in this invention. Figure 5 This is a schematic diagram of one embodiment of the electronic device in this invention. Detailed Implementation

[0026] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar elements and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms “comprising” or “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0027] In existing technologies, power grid risk assessment mainly relies on historical data to build models. However, due to insufficient data on extreme weather and other abnormal operating conditions, the models have limited coverage and their prediction accuracy is insufficient to meet the reliability requirements of modern power grids. When encountering typhoons or snowstorms, traditional models cannot effectively identify new failure modes, and operators must rely on experience to formulate temporary dispatch plans, which carries the risks of response delays and human error.

[0028] To address the aforementioned issues, given the inherent limitations of relying solely on data-driven models, this application shifts its focus to transforming human experience into executable computer rules. By analyzing typical switching operation cases, state transition logic and equipment association rules are extracted, and a structured deduction knowledge base is constructed. For the instruction parsing stage, natural language processing technology is employed to automate the recognition of unstructured text, eliminating errors from manual transcription. In the deduction and verification stage, a power grid topology model is introduced to achieve virtual simulation of operation steps and risk prediction.

[0029] Therefore, this application proposes a method for predicting power grid operation and maintenance risks, including: obtaining dispatch instructions generated by the power system and extracting operation information through feature analysis; matching switching operation deduction rules based on the operation information; generating switching operation sequences using the rules; and making risk predictions by combining the power grid model and anti-misoperation rules.

[0030] Example 1 For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 The first embodiment of the power grid operation and maintenance risk prediction method in this application includes: 101. Obtain dispatch instructions generated by the power system, and perform feature parsing on the dispatch instructions to extract dispatch operation information; In this embodiment, when obtaining scheduling instructions, scheduling instructions can be collected through an interface or operation interface. Specifically, scheduling data can be obtained through various methods such as manual input, text import, and system interface.

[0031] When parsing and extracting scheduling operation information, natural language processing technology is used to identify the operation object, operation type, and status parameters in the text. The status parameters include the initial operation state and the target operation state. Specifically, named entity recognition algorithm can be used to extract key fields to achieve standardized conversion of unstructured scheduling instructions, thereby extracting the required instruction features, such as a set of operation information including the operation object (which can be replaced by the device number), the initial operation state, and the target operation state.

[0032] 102. Based on the scheduling operation information, the scheduling deduction rule that matches the scheduling instruction is matched from the preset switching operation deduction rule base; It should be noted that this switching operation deduction rule base refers to a database that stores the state transition logic of equipment. Specifically, rule entries are formed by summarizing expert experience to achieve automated deduction of operation steps. For example, a detailed analysis of the power grid model and equipment characteristics is conducted to determine the operation rules for different equipment under different operation types. The rules are then categorized and stored according to operation type and equipment type, and the resulting rule base is updated and maintained regularly.

[0033] In this embodiment, the rule base includes the following rule examples: Rule 1: If the object of operation is a bus-side disconnector, the bus must be verified to be de-energized first.

[0034] Rule 2: If the operation type is closing, the grounding switch status must be checked to be open.

[0035] Rule 3: If the target status is running, it is necessary to verify that the protection device is engaged.

[0036] During matching, matching rules are filtered from the rule base based on the operation object (such as disconnecting switch) and operation type (closing) in the scheduling operation information. In order to improve the matching success rate, existing technologies such as fuzzy matching algorithms can also be used to process semantic differences in instructions (such as "close" and "closing") for matching, thereby ensuring the success rate of rule matching.

[0037] Furthermore, the regular update and maintenance mechanism for the rule base includes: 1. Rule base update mechanism design Periodic review plan: Review cycle setting: Based on the actual needs of power grid operation and the frequency of historical data updates, the review cycle of the rule base is set to a comprehensive review once a quarter and a small-scale verification once a month.

[0038] Review content planning: Before each review, develop a detailed review content list, including but not limited to the validity, accuracy, coverage, and compatibility with new equipment and technologies of the rules.

[0039] Review team formation: A review team composed of personnel with multidisciplinary backgrounds, including power grid dispatching experts, equipment maintenance engineers, and software engineers, will be formed to ensure the comprehensiveness and professionalism of the review.

[0040] Update trigger conditions: Changes in power grid structure: When there are major adjustments to the power grid structure, such as the addition of new substations or line upgrades, the rule base update process is immediately triggered.

[0041] New equipment deployment: When new equipment is put into operation, relevant rules should be updated in a timely manner based on its characteristics and operational requirements.

[0042] Accident and Failure Analysis: After a major power grid accident or frequent occurrence of similar failures, conduct in-depth analysis of the causes and extract new rules and regulations to prevent similar problems from recurring.

[0043] 2. Data Collection and Analysis Runtime data collection: Data source integration: Integrate data from multiple sources such as OMS, SCADA, and EMS systems to ensure the comprehensiveness and accuracy of the data.

[0044] Real-time data acquisition: Real-time data acquisition technology is used to ensure the timeliness and continuity of data, providing data support for the dynamic adjustment of the rule base.

[0045] Data cleaning and preprocessing: Cleaning and preprocessing the collected raw data to remove noisy data, fill in missing values, and improve data quality.

[0046] Case Analysis: Case library construction: Establish a power grid operation case library, classify and store normal operation cases, fault cases, accident cases, etc., to provide rich materials for rule extraction.

[0047] Case study methodology: Employing a combination of qualitative and quantitative analysis methods, we delve into the underlying causes and patterns of cases to extract universally applicable and instructive rules.

[0048] Rule extraction process: Establish a standard process for rule extraction, including case selection, cause analysis, rule extraction, and rule verification, to ensure that the extracted rules are scientific and reasonable.

[0049] 3. Expert review and summary Expert team formation: Expert selection criteria: Clearly define the standards and conditions for expert selection, including professional background, work experience, and technical level, to ensure the professionalism and authority of the expert team.

[0050] Expert team composition: An expert team will be formed, consisting of power grid dispatching experts, equipment manufacturer representatives, university researchers, and other stakeholders, to achieve knowledge complementarity and experience sharing.

[0051] Expert training and exchange: Regularly organize expert training and exchange activities to improve experts' professional competence and review capabilities, and promote knowledge sharing and experience transfer among experts.

[0052] Rule review: Review process design: Develop detailed rules and review processes, including preliminary review, secondary review, and final review, to ensure the rigor and fairness of the review.

[0053] Evaluation criteria development: Clarify the evaluation criteria and basis, including the scientific nature, rationality, and operability of the rules, to provide clear guidance for expert evaluation.

[0054] Review feedback: Establish a review feedback mechanism to promptly provide expert review opinions to the rule-making personnel, thereby promoting the continuous optimization and improvement of the rules.

[0055] Rule summary: Rule representation: The rule representation adopts a structured form, such as production rules or frame representation, which facilitates the storage, querying and application of rules.

[0056] Rule classification and coding: The extracted rules are classified and coded for management, which facilitates subsequent retrieval and maintenance.

[0057] Rule document writing: Write detailed rule documents, including the description of the rule, applicable conditions, and operation steps, to provide clear guidance for the application of the rule.

[0058] 4. Rule base update implementation Version control: Version control strategy: A version control strategy is adopted to manage the rule base, recording information such as the time, content, and updater of each update, which facilitates traceability and rollback.

[0059] Version release process: Establish version release processes and standards to ensure that the new version of the rule base can be deployed and applied smoothly.

[0060] Version compatibility testing: Conduct thorough compatibility testing before releasing a new version to ensure seamless integration of the new rules with the existing system.

[0061] Incremental update: Incremental update strategy: The rule base is updated using an incremental update method, which only updates rule entries that have changed or been added, reducing the impact of updates on the system and downtime.

[0062] Update package creation and distribution: Create update packages and distribute them to designated locations for users to download and install.

[0063] Update log records detailed information for each update, including the updated rule entries, update time, and updater, facilitating subsequent maintenance and management.

[0064] 103. Based on scheduling operation information, use scheduling deduction rules to perform switching operation deduction on scheduling instructions and generate switching operation sequences; Specifically, by combining scheduling operation information and topology technology, the switching operation sequence corresponding to the scheduling instruction is derived using scheduling deduction rules. Specific implementation steps may include: First, based on the equipment type of the operation object, the corresponding branch is found from the switching operation deduction rule base. Then, based on the operation type, initial operation state, and target operation state, the specific rule is located, the rule content is extracted, and based on the actual diagram data, the specific equipment is obtained using topology calculation, thereby generating a switching operation sequence that conforms to the scheduling instructions.

[0065] 104. Based on the preset power grid model and anti-misoperation rules, and combined with dispatch operation information, perform anti-misoperation judgment on the switching operation sequence, and output risk prediction results based on the anti-misoperation judgment.

[0066] In this embodiment, the error prevention judgment refers to verifying the electrical connectivity of the operation sequence based on the power grid topology model. Specifically, a depth-first search algorithm is used to traverse the device nodes to detect whether there is a risk of isolated power supply or loop closure.

[0067] In practical applications, the process of preventing misoperation mainly relies on the topology information of the power grid model. Based on the equipment status, the defined anti-misoperation rules are automatically calculated, and depth-first and breadth-first search methods are used to perform topology search and anti-misoperation judgment on switching operation items. Warnings are issued for operation items that violate the operation rules or pose a risk. The topology search process mainly studies the continuity of electrical equipment.

[0068] Specifically, after parsing, the dispatching instructions generate a set of operation information including equipment number and target status. The inference rule base filters matching rules according to equipment type; for example, circuit breaker tripping operations need to be associated with changes in the status of disconnecting switches. When generating the operation sequence, the operation steps are arranged according to the electrical connection order of the equipment to avoid operating disconnecting switches while they are energized. In the error prevention and judgment stage, the operation sequence is mapped to the power grid model to detect whether an unexpected topology structure is formed after execution, such as load loss or equipment overload.

[0069] In this embodiment, automated risk prediction under the control of scheduling instructions is achieved by executing the above-described method, without relying on massive historical data to build models; the operation rule base can be dynamically expanded and updated to adapt to changes in the power grid structure; topology simulation verification effectively identifies hidden risks that are difficult to detect by traditional methods, such as cascading failures caused by the coordinated operation of multiple devices, significantly improving the safety of power grid operation under complex conditions.

[0070] Example 2 Please see Figure 2 The second embodiment of the power grid operation and maintenance risk prediction method in this application includes: 201. Construct dispatch instructions from data generated from the power system's data interaction interface or dispatch instruction operation interface; Specifically, it interacts with the power system (OMS system) in real time through a standardized interface and automatically receives dispatch instructions issued by the OMS system. When the interface is abnormal or the OMS system is not integrated, it supports dispatchers to manually input dispatch order text through a graphical interface. It also supports batch importing dispatch order data from Excel, Word and other format files, and then extracting and generating dispatch instructions based on the dispatch data.

[0071] 202. Standardize the format of the scheduling instructions, and use NLP technology and named entity recognition technology to parse the processed scheduling instructions and extract the scheduling operation information. The scheduling operation information includes at least the operation object, operation type, operation initial state and target operation state. Understandably, the target of the operation is clearly indicated: which line, which device, or which area is the target of the operation.

[0072] Operation type: Specifies whether the operation is to start, stop, adjust, or perform other types of operations on the device.

[0073] Initial operating state: The state of the equipment before executing instructions, such as running, hot standby, cold standby, maintenance, etc.

[0074] Operation target state: The state that the device should reach after executing the command.

[0075] Additional information: such as charging operation, maintaining cold standby at certain intervals, etc.

[0076] Standardizing the scheduling instructions essentially involves converting instructions from different sources into a unified data format. This can be achieved using regular expression matching or template mapping methods to eliminate differences in instruction formats generated by different systems. The unified data format includes removing redundant characters, standardizing naming conventions, and verifying data integrity and accuracy.

[0077] Natural language processing technology refers to the segmentation, syntactic analysis, and semantic understanding of text instructions. Specifically, it can be implemented using a Transformer-based pre-trained model to parse operational elements in unstructured instructions. Named entity recognition technology refers to identifying specific categories of entity information from text. Specifically, it can be implemented using a bidirectional long short-term memory neural network combined with a conditional random field model to accurately extract key information such as device names and operational actions.

[0078] Specifically, the data interaction interface captures structured instruction data transmitted by the OMS system in real time, while the user interface synchronously collects manually input text instructions. The semi-structured instructions are split into fields using regular expression matching, and free text instructions are converted into a standard table structure using template mapping. The preprocessed instructions are then input into a natural language processing model for semantic parsing, identifying core elements such as operation verbs and device names. A named entity recognition model further extracts entity information such as device number and status parameters from the parsing results, ultimately forming a standardized set of operation information containing the operation object, operation type, and status parameters.

[0079] Furthermore, the standardization of the scheduling instructions includes: Use regular expressions or rule engines to filter invalid or erroneous data; Different processing methods are used for different types of redundant characters, such as merging consecutive spaces; Establish a naming lookup table to ensure the accuracy and consistency of equipment names and operation type naming.

[0080] The process of parsing the processed scheduling instructions using Natural Language Processing (NLP) and Named Entity Recognition (NENT) technologies includes: The dispatch text is segmented and part-of-speech tagging is performed using Chinese word segmentation algorithms (such as the Jieba algorithm or the HanLP algorithm); Named entity recognition technology (based on rules or machine learning methods) is used to extract entity information such as device name, operation type, initial state, and target state. Analyze entity relationships and construct a knowledge graph, then verify the data in conjunction with the power grid equipment directory.

[0081] 203. Based on the scheduling operation information, match the scheduling deduction rule that matches the scheduling instruction from the preset switching operation deduction rule base; In this embodiment, based on the operation object and the operation type, multiple candidate rules are retrieved from a preset switching operation deduction rule library; from the multiple candidate rules, the target deduction rule whose state elements are the initial state of the operation and the target state of the operation is selected as the scheduling deduction rule corresponding to the scheduling instruction; if no completely identical rule is found among the multiple candidate rules, the closest rule is selected from the multiple candidate rules as the target deduction rule using a similarity algorithm, and the differences in the target deduction rule are marked to prompt the operator for manual adjustment, thereby obtaining a scheduling deduction rule that matches the scheduling instruction.

[0082] Understandably, this similarity algorithm refers to a calculation method used to calculate the degree of matching between candidate rules and scheduling instructions. Specifically, it can be implemented using cosine similarity or edit distance algorithms. By quantifying the similarity between rule elements and operation information, the closest candidate rules are selected.

[0083] Candidate rules refer to the set of rules initially retrieved from the switching operation deduction rule base that are associated with the operation object and operation type. Specifically, they can be achieved through keyword matching or pattern recognition technology to narrow down the rule matching range.

[0084] The discrepancy refers to the rule content in the target deduction rules that is inconsistent with the scheduling operation information in terms of state elements or operation logic. This can be achieved through highlighting or annotation to guide operators to quickly locate adjustment points.

[0085] Specifically, when no candidate rule perfectly matches the scheduling instruction, the system calculates the similarity between the candidate rule and the initial and target states in the scheduling operation information. For example, it uses a cosine similarity algorithm to compare the vectorized representations of state elements and selects the rule with the highest similarity as the target deduction rule. Subsequently, the system automatically identifies state elements or operation steps in the target deduction rule that differ from the scheduling instruction. For example, by comparing equipment parameters or topological connections in the initial state of the operation, it generates difference markers and pushes them to the operation interface. Operators manually review the marked differences, such as modifying equipment state thresholds or adjusting the operation sequence, ultimately generating a deduction rule adapted to the current scheduling instruction. This achieves rapid generation of deduction rules adapted to the scheduling instruction through a collaborative mechanism of automated filtering and manual verification, even when the rule base is not fully covered. This reduces the risk of deduction interruption due to missing rules, reduces the workload of full-process manual intervention, and improves the accuracy and efficiency of switching operation deduction.

[0086] In another embodiment, a similarity algorithm is used to select the closest rule from multiple candidate rules as the target deduction rule, including: Calculate the similarity score between each candidate rule and the scheduling instruction in terms of operation object, operation type, operation initial state, and operation target state; The total similarity score is calculated by summing the results based on the weights assigned to each feature; the weights are set according to the actual situation. The candidate rule with the highest total similarity score is selected as the target deduction rule. If multiple candidate rules have the same and the highest total similarity scores, further screening can be performed based on other criteria (such as the historical usage frequency of the rule, expert recommendations, etc.), or the operator can be prompted to make a manual selection.

[0087] In practical applications, the similarity algorithm can specifically calculate cosine similarity or Jaccard similarity. The calculation process of the similarity algorithm includes: converting candidate rules and target rules into vector form; calculating the cosine similarity or Jaccard similarity between vectors; and selecting the closest rule as the target deduction rule according to a preset threshold.

[0088] 204. Based on the scheduling operation information, use the scheduling deduction rules to perform switching operation deduction on the scheduling instructions and generate a switching operation sequence; In this step, all operation objects in the scheduling operation information are sorted according to the scheduling deduction rules to obtain an object sequence; based on the initial operation state of each operation object in the scheduling operation information, the initial state of each operation object in the object sequence is set; based on the state change logic of each operation object, the object sequence after setting the initial state is deduced step by step until the final state of each operation object is the corresponding operation target state, and then a complete switching operation sequence is obtained.

[0089] The scheduling deduction rules refer to the set of constraints that dynamically adjust the execution order of operation objects. Specifically, they can be implemented using a priority algorithm based on the power grid topology to ensure that the execution order between operation objects conforms to the actual power grid operation logic.

[0090] The object sequence refers to the linear execution order generated according to the correlation between the operation objects. Specifically, it can be generated using a directed acyclic graph topological sorting algorithm to eliminate logical conflicts between operation steps.

[0091] The state change logic refers to the transition conditions that the operating object must follow from the initial state to the target state. Specifically, it can be defined using a finite state machine model to constrain the legality of state transitions.

[0092] Specifically, the process begins by analyzing the power grid topology to determine the dependencies between operational objects. Device nodes with upstream and downstream connections are then prioritized according to execution priority, forming a loop-free operation sequence. Initial operating parameters for each device are extracted from the scheduling operation information, such as the circuit breaker's open / closed status and the disconnector's position information, serving as the starting point for the simulation. During the simulation, the state transitions of each operational object are triggered sequentially according to the sequence. When a device's state change depends on the state results of other devices, the system automatically backtracks to previous steps for condition verification until the states of all devices meet the target requirements. For example, when switching a line from an operating state to a maintenance state, the simulation process will first disconnect the circuit breaker and then open the disconnector, verifying at each step whether adjacent devices are in a safe operating state. This progressive simulation mechanism ensures that the generated switching operation sequence conforms to equipment operating specifications and fully covers all necessary intermediate operation steps.

[0093] This application generates object sequences through dynamic topology sorting, automatically adapting to different power grid wiring methods. It also employs a state machine-driven step-by-step deduction mechanism to ensure that each operation step undergoes rigorous state condition verification, effectively eliminating the risk of state transitions caused by human experience-based misjudgments. Furthermore, this application automates the generation and logical self-verification of switching operation sequences, resolving the issue of missing operation steps caused by the inability of static rule bases to adapt to dynamic scheduling instructions in traditional deduction processes. The dynamic sorting mechanism of object sequences significantly reduces reliance on human experience, improving deduction efficiency to real-time response levels. Simultaneously, the step-by-step deduction process based on state change logic ensures state continuity between operation steps, avoiding the risk of equipment malfunctions due to state transitions, and providing a reliable guarantee for the safe operation of the power grid.

[0094] In another embodiment, after generating a switching operation sequence by performing switching operation deduction on the scheduling instructions based on the scheduling operation information and using the scheduling deduction rules, the method further includes: The switching operation sequence is initially verified, and the logic and completeness of the operation steps corresponding to the switching operation sequence are determined based on the verification results.

[0095] Specifically, the preliminary verification of the switching operation sequence includes: Check whether each step in the switching operation sequence complies with the operating specifications of the power grid equipment; Check whether each step in the switching operation sequence meets the electrical "five protections" rules; If any non-compliance with operating procedures or the "five preventions" rules is found, the scheduling instructions will be optimized and simulated.

[0096] In practical applications, a dual verification mechanism is constructed to optimize the operation sequence. After the switching operation sequence is generated, the verification module first loads a pre-set verification rule base, which contains basic logical constraints and typical operation modes for power grid operation. The verification algorithm parses the operation sequence step by step, simulating and deducing the execution conditions and subsequent effects of each operation step. For example, in the state transition verification stage, the system detects whether there are operation steps that violate equipment state transition rules; in the causal logic verification stage, the system identifies isolated steps lacking necessary pre-operations. For any anomalies detected, the system automatically generates a verification report and triggers a sequence reconstruction mechanism to complete the operation sequence by inserting missing steps or adjusting the operation order. This effectively solves the operational risks caused by logical inconsistencies and missing steps. This technical solution can proactively identify and correct abnormal steps that violate the basic logic of power grid operation during the operation sequence generation stage, ensuring that the execution conditions and subsequent effects of each operation step comply with power grid operation specifications. The application of the pre-verification mechanism avoids the risk of equipment malfunction due to incorrect operation steps and provides optimized input data for subsequent in-depth error prevention judgment, thus improving the overall safety and reliability of the power grid operation and maintenance system.

[0097] 205. Map the switching operation sequence to the preset power grid model to generate a topology change diagram for all operation objects; It should be noted that this power grid model refers to a digital representation of the connection relationships of power equipment based on graph theory. Specifically, it can be implemented using an adjacency matrix or adjacency list data structure to dynamically reflect the topological relationships during the process of equipment state changes.

[0098] The topology change diagram refers to a visual model of the state transition path formed by the operating object during the switching operation. Specifically, it can be stored and updated using a dynamic graph database to capture the chain reaction caused by changes in equipment state.

[0099] In this embodiment, the basic topology connection diagram of each operation object is determined based on a preset power grid model; the operation order between each operation object in the switching operation sequence is analyzed, the connection relationship between each operation object in the basic topology connection diagram is adjusted, and the topology change diagram of all operation objects is obtained.

[0100] 206. Based on the topology change graph and the preset anti-misoperation rules, the topology search of the switching operation sequence is performed using the depth-first search method and the breadth-first search method, and the risk prediction is made based on the search results.

[0101] In this step, based on the topology change diagram and the preset anti-misoperation rules (including electrical "five-prevention" rules and dispatch range expansion rules), the topology search of the switching operation sequence is performed using depth-first search and breadth-first search methods; risk prediction is made based on whether there are any violations of the anti-misoperation rules in the search results, and if so, the risk type, level and scope of impact are output, and early warning information is generated.

[0102] It should be noted that this depth-first search method refers to a search strategy that traverses vertically along the branches of the operation path to the terminal node and then backtracks. Specifically, it can be implemented using recursion or a stack structure to uncover deep-seated hidden dangers in a specific operation path.

[0103] The breadth-first search method refers to a strategy that expands the search scope horizontally according to the hierarchical order of operation steps. Specifically, it can be implemented using a queue structure to cover all possible branch paths.

[0104] In practical applications, after the switching operation sequence is mapped to the power grid model, the state changes of each operated object will trigger topology reconstruction, forming a topology change diagram that includes equipment connection status, electrical parameters, and operation steps. During the error prevention judgment process, a depth-first search algorithm is configured to vertically detect potential risks along the operation path, such as local overload problems caused by live-closing operations; a breadth-first search algorithm horizontally scans all associated equipment, such as detecting cross-regional power supply imbalances caused by erroneous switch opening operations. Depth and breadth limits are set during the search process. The depth limit can be set to twice the total number of operation steps to cover a reasonable operation range, and the breadth limit can be set to the three-level connection relationship of associated equipment to control computational complexity. Topology path node information is recorded in real time and pattern matched with entries in the error prevention rule base. When an operation path violating the error prevention rules is detected, a risk warning mechanism is automatically triggered. This effectively solves the problem of potential risk omissions caused by static models and single rule verification. The construction of a dynamic topology change diagram visualizes the associated changes in equipment status during the operation process, and the collaborative application of composite search algorithms ensures the simultaneous implementation of vertical hazard discovery and horizontal risk scanning. Depth-first search can accurately locate nodes of non-compliant operation within a specific operational path, while breadth-first search can promptly identify cascading risks across devices and regions. The synergy between the two significantly improves the comprehensiveness and accuracy of risk prediction. The real-time matching mechanism between anti-misoperation rules and topology paths enables the rapid identification of typical violations such as live-line operation and accidental switch disconnection. Operators can adjust their operational plans promptly based on risk level indicators, thereby preventing power grid failures.

[0105] In another embodiment, the step of performing a topology search on the switching operation sequence using depth-first search and breadth-first search methods based on the topology change graph and preset anti-misoperation rules, and making risk predictions based on the search results, includes: When initializing the search queue, set the depth and breadth limits for the search; Based on the depth and breadth constraints, the path and node information of each operation object in the topology transformation graph are searched and recorded; Using preset error prevention rules, the recorded path and node information is judged to determine whether it violates the error prevention rules, and risk prediction is made based on the judgment results.

[0106] The depth limit refers to constraining the maximum number of recursion levels during the search process. Specifically, it can be implemented using a preset integer threshold, such as setting the depth limit to 5 levels to prevent system resources from being exhausted due to infinite recursion.

[0107] The breadth constraint refers to limiting the number of scalable nodes at the same level. This can be achieved by using a queue capacity threshold, for example, processing a maximum of 100 nodes per level to avoid memory overflow.

[0108] The path and node information refer to the state change trajectory of the operation object in the topology change graph. Specifically, they can be stored using an adjacency list or matrix data structure, such as recording the line connection relationships associated with circuit breaker A during the process of closing and opening.

[0109] The process of defining the error prevention rules includes: Collect real-world cases and feedback information from power grid operations, and analyze the shortcomings of existing rules; Experts were organized to review the electrical "five protections" rules and the rules for expanding the dispatching scope to ensure the scientific validity and rationality of the rules; The rules that have passed the review will be recorded and archived in document form, and version management will be implemented for the rules, recording the modification history of the rules.

[0110] Specifically, during the initialization of the search queue, the depth limit is set to twice the maximum possible number of operation steps in the power grid model. For example, when a typical switching operation involves 10 steps, the depth limit is set to 20 levels. The breadth limit is dynamically adjusted according to the complexity of the power grid topology. For example, in a regional power grid with 500 nodes, the breadth limit is set to process no more than 50 nodes per level. During the search, the depth-first algorithm probes vertically along the main path of equipment state changes, such as tracing the cascading power outage range caused by circuit breaker operations; the breadth-first algorithm scans horizontally for parallel state changes of related equipment, such as detecting the load transfer of multiple transformers at the same voltage level. After the path information is recorded, the error prevention rules perform compliance checks on each path using a pattern matching algorithm, such as determining whether there are operation combinations that violate the N-1 safety criterion. For detected non-compliant paths, the system automatically generates a risk report containing the specific location of the violation and the relevant rule clauses.

[0111] In this embodiment, the risk prediction based on the discrimination result includes: If the judgment results show that an operation violates the error prevention rules, the risk level of the operation that violates the error prevention rules is determined based on the severity level of the potential risk; and a corresponding early warning message is generated based on the risk level.

[0112] The severity level of potential risks refers to a quantitative assessment system established based on dimensions such as the extent of equipment damage, the scope of power outages, and the impact on personal safety. Specifically, this can be achieved using a multi-indicator weighted scoring model based on scenario coefficients. By setting different weight coefficients, the differences in the impact of each dimension on the overall risk are reflected, thus providing data support for risk level classification. The weight coefficients are set based on experience. Risk level refers to dividing the severity of risks into multiple levels according to preset thresholds. This can be achieved using an interval division algorithm, where different threshold ranges correspond to different levels, achieving standardized classification of risk levels. Warning information refers to alert content automatically matched according to different risk levels. This can be achieved using a preset template library combined with dynamic parameter filling technology, ensuring information compatibility across different terminal devices through standardized information formats.

[0113] Specifically, when the error prevention judgment module detects an operation in the operation sequence that violates the error prevention rules, it first determines the scenario type of the operation and then determines the scenario coefficient based on the scenario type. Then, it calls the risk assessment model to perform a multi-dimensional analysis of the violation. For example, for an operation involving a high-voltage busbar, the system automatically calculates the number of users experiencing power outages, equipment maintenance costs, and emergency response time that the operation may result in. By calculating the weighted sum of these indicators and multiplying it by the scenario coefficient, a comprehensive risk value is obtained. This comprehensive risk value is then compared with a preset level threshold range, which is set based on experience. As a preferred embodiment of the invention, low risk: the final comprehensive risk value is between 0 and 50 points; Level 2 risk (medium risk): the final comprehensive risk value is between 50 and 80 points; high risk: the final comprehensive risk value is above 80 points.

[0114] Specifically, a scenario coefficient is determined based on the specific scenario in which the violation occurred (such as extreme weather conditions, equipment aging, peak grid load, etc.). The scenario coefficient reflects the degree of risk increase under different scenarios; its value is greater than 1, and the more complex or unfavorable the scenario, the larger the scenario coefficient.

[0115] Based on experience and actual power grid conditions, scenario factors are set for different scenarios. As a preferred embodiment of the present invention, examples of setting scenario factors are shown in Table 1.

[0116] Table 1. Examples of Scene Factor Classification and Benchmark Values

[0117] Based on the set scene factors, match the scene type of the current violation. For example, if it is "blizzard weather + old equipment", calculate the sum of the scene factors of the two scenes, and add 1 to get the scene coefficient of the current violation.

[0118] Furthermore, the risk assessment model calculates indicators including the number of users experiencing power outages, equipment maintenance costs, and emergency response time.

[0119] Specifically, the ratio of the number of users experiencing power outages to the total number of users is calculated and multiplied by 100 to obtain an indicator of the number of users experiencing power outages. Multiply the number of devices requiring repair by the corresponding repair cost and sum them up. Divide the sum by the average total cost threshold for devices at that voltage level (e.g., the average total cost threshold for 110kV devices is 5 million yuan), and then multiply by 100 to obtain the equipment repair cost index. Divide the estimated repair time by the standard repair time, and then multiply by 100 to obtain the repair response time index.

[0120] Based on experience, the weight coefficients of each indicator are dynamically set. As a preferred embodiment of the present invention, priority is given to ensuring power supply for people's livelihood, followed by controlling asset losses, and finally ensuring power restoration efficiency. Therefore, the weight coefficients of the indicators of the number of users experiencing power outages, equipment maintenance costs, and emergency repair response time are 0.5, 0.3, and 0.2, respectively.

[0121] The weighted sum of each indicator is calculated and then multiplied by the scenario coefficient to obtain the comprehensive risk value.

[0122] Finally, based on the risk level, the corresponding early warning template is invoked. For example, for level 2 risks, an early warning message containing a list of affected equipment, recommended handling measures, and a reverse path for switching operations is automatically generated and pushed to the dispatch terminal via message middleware. Potential risks are categorized into high, medium, and low levels based on their severity. Early warning methods are determined for different risk levels: high risks receive dual warnings via SMS and email, medium risks receive email warnings, and low risks receive graphical interface prompts. Early warning information including risk type, level, scope of impact, and recommended measures is generated and promptly sent to relevant personnel.

[0123] Furthermore, after generating the early warning information, the process also includes: determining the scope of the early warning information based on the type and scope of the risk; formatting the early warning information to make it clear and easy to understand; and recording the sending of the early warning information, including the sending time, recipients, and feedback.

[0124] In this embodiment, if the anti-misoperation rule is the "opening and closing of disconnecting switches under load" rule, the anti-misoperation verification process is as follows: Figure 3 As shown, its implementation process is as follows: 1) Whether the circuit breaker in this bay is in the closed position: Through topology search, find the circuit breaker in the bay where the currently operated equipment is located, and determine whether the circuit breaker is in the closed position; 2) Whether it is a double busbar disconnector: Determine whether the currently operating equipment is a busbar disconnector under a double busbar connection.

[0125] 3) Does a near-busbar switching loop exist? In a double-busbar, double-sectioned wiring configuration, the current disconnecting switch forms an equipotential loop through the bus tie interval of its respective busbar. This loop does not pass through the busbar section switches and appears to be formed nearby; we define this as a "near-busbar switching loop." 4) Does a remote busbar loop exist? In a double-busbar double-section connection, the current disconnecting switch does not pass through the bus tie interval of its own busbar, but instead forms an equipotential loop through the section interval of its own busbar and the bus tie and section switch on the other side. This loop is formed through the section switch, and visually it is a loop that goes around in a large circle. Here it is defined as a "remote busbar loop".

[0126] In another embodiment, after the error-prevention-based judgment outputs the risk prediction result, it further includes: If the risk prediction result is low risk, the scheduling instruction is executed based on the switching operation sequence, and the execution data generated during the execution process is recorded. The execution data includes the initial state and execution time of the instruction, the execution status of the operation steps and the changes in equipment status in real time, and the completion status and execution result of the instruction. After the scheduling instruction is executed, feedback information is generated according to the execution result and displayed visually via graphical interface, SMS or email.

[0127] Execution data refers to the structured operation records generated throughout the execution of scheduling instructions. This can be implemented using a distributed log collection system, which uses timestamps to mark the status changes of each operation node, forming a traceable data chain. The feedback information classification and generation mechanism matches preset feedback templates based on the positive and negative types of the execution results. This can be implemented using a rule engine, which automatically categorizes the execution results using predefined classification rules. Visualization displays convert the operation data into interactive graphical elements. This can be implemented using data visualization components, displaying the execution trajectory of the operation steps in a timeline format.

[0128] Specifically, when the system determines that a scheduling instruction has a low risk level, it automatically triggers the instruction execution process. During execution, the initial state of the operated object and the execution time are recorded as baseline parameters. The execution progress of the operation steps is dynamically tracked through a state machine, and changes in device status are collected in real time through sensor data. After the instruction is completed, the execution result is classified as successful, partially successful, or abnormal. The classification result triggers the generation of the corresponding feedback template. Feedback information is displayed as a timeline view of the entire operation process through a graphical interface. Abnormal results are pushed with alarm information via SMS or email, forming a closed-loop verification mechanism for the operation.

[0129] In another embodiment, after the risk prediction result is output based on the error prevention judgment, the method further includes: a scheduling instruction execution control step, wherein the scheduling instruction execution control step includes: Before the dispatcher issues an operation instruction, key information such as the operation object, operation type, and operation steps are extracted from the instruction, and real-time error prevention verification is performed in combination with the topology model and error prevention rules. If the verification passes, a prompt message is sent to the dispatcher allowing the command to be issued; if the verification fails, the dispatcher is prevented from issuing the command, and a detailed error reason report is generated and displayed to the dispatcher in graphical interface or text form. The execution process of operation instructions is tracked and recorded, including recording the initial state and execution time of the instructions, recording the execution status of operation steps and changes in equipment status in real time, and recording the completion status and execution results of the instructions. After the operation is completed, feedback information is generated according to the execution results and sent to the dispatcher via graphical interface, SMS or email.

[0130] Understandably, the real-time error prevention verification process includes: prioritizing verification rules and executing the verification of key rules first; handling conflicts according to preset conflict resolution strategies when conflicts occur during verification; and recording abnormal situations during the verification process for analysis and improvement.

[0131] After generating the feedback information, the process further includes: encrypting the feedback information to ensure its security; customizing different feedback content based on the role of the person receiving the feedback information; and tracking the sending of the feedback information to ensure its delivery.

[0132] In another embodiment, after the error-prevention-based judgment outputs the risk prediction result, the method further includes: a data storage and management step, wherein the data storage and management step includes: The data generated during the processes of order acquisition, data preprocessing, order parsing, operation sequence deduction, error prevention judgment, and operation instruction execution control are stored. Classify and manage the stored data, and create data indexes to facilitate quick querying and retrieval; Regularly back up stored data to prevent data loss; Data is categorized and stored according to its type (such as command data, equipment status data, operation sequence data, risk warning data, etc.), and different storage permissions are set for different types of data to ensure data security. Regularly clean up the stored data, delete expired or invalid data, and free up storage space; Perform quality checks on the stored data to ensure its accuracy and integrity.

[0133] It should be noted that the data quality inspection process includes: establishing data quality inspection standards and clarifying the acceptable range for each type of data; using automated data quality inspection tools for inspection; marking and processing the problematic data found, and recording the processing process and results.

[0134] In another embodiment, after the error-prevention-based judgment outputs the risk prediction result, the system further includes a system maintenance and update step, wherein the system maintenance and update step includes: Regularly monitor the performance of the power grid operation and maintenance risk prediction system, and check the system's operating status and response time; Based on changes in power grid structure and equipment upgrades, the system's rule base and topology model are updated and maintained. Perform security vulnerability scanning and patching on the system to ensure its security; Establish a system maintenance log to record the system maintenance time, maintenance content, and maintenance personnel information; Manage system updates by recording the content and timing of each update. Evaluate the effectiveness of system maintenance and updates, and make improvements based on the evaluation results.

[0135] The evaluation process for the effectiveness of system maintenance and updates includes: setting evaluation indicators, such as system response time, false alarm rate, and false negative rate; collecting actual data during system operation and comparing it with the evaluation indicators; analyzing the problems and deficiencies of the system based on the comparison results; formulating improvement measures and optimizing and upgrading the system.

[0136] Based on the previous embodiments, this embodiment automatically parses scheduling instructions, intelligently matches and extrapolates rules to generate operation sequences, and combines a dynamic error prevention and judgment mechanism. This solves the problems of traditional methods relying on human experience and insufficient static rule verification, thereby improving operational accuracy, reducing the risk of human error, and enhancing the safety of power grid operation.

[0137] Example 2 The above describes the power grid operation and maintenance risk prediction method in the embodiments of this application. The following describes the power grid operation and maintenance risk prediction device in the embodiments of this application. Please refer to [link / reference]. Figure 4 One embodiment of the power grid operation and maintenance risk prediction device in this application includes: The acquisition and parsing module 410 is used to acquire the dispatch instructions generated by the power system, and to perform feature parsing on the dispatch instructions to extract dispatch operation information; The matching module 420 is used to match the scheduling deduction rule that matches the scheduling instruction from a preset switching operation deduction rule library based on the scheduling operation information. The deduction module 430 is used to perform switching operation deduction on the scheduling instruction based on the scheduling operation information and using the scheduling deduction rules to generate a switching operation sequence. The prediction module 440 is used to make a prediction of the switching operation sequence based on the preset power grid model and the error prevention rules, combined with the scheduling operation information, and output the risk prediction result based on the error prevention judgment.

[0138] In this embodiment, the acquisition and analysis module 410 is specifically used for: Data generated from the power system's data interaction interface or dispatch instruction operation interface is used to construct dispatch instructions; The scheduling instructions are format-standardized, and NLP and named entity recognition technologies are used to parse the processed scheduling instructions to extract scheduling operation information. The scheduling operation information includes at least the operation object, operation type, operation initial state, and target operation state.

[0139] In this embodiment, the matching module 420 is specifically used for: Based on the operation object and the operation type, multiple candidate rules are retrieved from the preset switching operation deduction rule base; The target deduction rule, which selects the state elements as the initial state and the target state of the operation from multiple candidate rules, is used as the scheduling deduction rule corresponding to the scheduling instruction.

[0140] In this embodiment, the matching module 420 is further configured to: If no completely identical rule is found among the multiple candidate rules, a similarity algorithm is used to select the closest rule from the multiple candidate rules as the target deduction rule, and the differences in the target deduction rule are marked to prompt the operator for manual adjustment, so as to obtain a scheduling deduction rule that matches the scheduling instruction.

[0141] In this embodiment, the deduction module 430 is specifically used for: According to the scheduling deduction rules, all operation objects in the scheduling operation information are sorted to obtain an object sequence; Based on the initial state of each operation object in the scheduling operation information, the initial state of each operation object in the object sequence is set. Based on the state change logic of each operation object, the sequence of objects after setting the initial state is deduced step by step until the final state of each operation object is the corresponding operation target state, and then the complete switching operation sequence is obtained.

[0142] In this embodiment, the inference module 430 is further used for: The switching operation sequence is initially verified, and the logic and completeness of the operation steps corresponding to the switching operation sequence are determined based on the verification results.

[0143] In this embodiment, the prediction module 440 is specifically used for: The switching operation sequence is mapped to a preset power grid model to generate a topology change diagram of all operation objects; Based on the topology change graph and the preset anti-misoperation rules, the switching operation sequence is searched using depth-first search and breadth-first search methods, and risk prediction is made based on the search results.

[0144] Based on the previous embodiment, this embodiment describes in detail the specific functions of each module and the unit composition of some modules. Through the above modules, the specific functions of the original modules are refined, the operation of the power grid operation and maintenance risk prediction device is improved, its operational reliability is enhanced, and the actual logic between each step is clarified, thereby improving the practicality of the device.

[0145] Example 3 above Figure 4 The power grid operation and maintenance risk prediction device in this application embodiment is described in detail from the perspective of modular functional entities. The electronic equipment in this application embodiment is described in detail from the perspective of hardware processing.

[0146] Figure 5This is a schematic diagram of the structure of an electronic device 500 provided in an embodiment of this application. The electronic device 500 can vary significantly due to different configurations or performance characteristics. It may include one or more central processing units (CPUs) 510 (e.g., one or more processors) and a memory 520, and one or more storage media 530 (e.g., one or more mass storage devices) for storing application programs 533 or data 532. The memory 520 and storage media 530 can be temporary or persistent storage. The program stored in the storage media 530 may include one or more modules (not shown in the diagram), each module may include a series of request operations on the electronic device 500. Furthermore, the processor 510 may be configured to communicate with the storage media 530 and execute the series of request operations in the storage media 530 on the electronic device 500 to implement the steps of the aforementioned power grid operation and maintenance risk prediction method.

[0147] Electronic device 500 may also include one or more power supplies 540, one or more wired or wireless network interfaces 550, one or more input / output interfaces 560, and / or one or more operating systems 531, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 5 The illustrated electronic device structure does not constitute a limitation on the electronic device provided in this application. It may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0148] Example 4 This application also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores a request that, when the request is executed on a computer, causes the computer to perform the steps of the power grid operation and maintenance risk prediction method.

[0149] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system, device, or unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0150] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several requests to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0151] Example 5 This invention also proposes a power grid operation and maintenance risk prediction system, including an operation information acquisition module, a dispatching simulation rule matching module, a switching operation sequence generation module, and a risk prediction module: The operation information acquisition module acquires dispatch instructions generated by the power system, performs feature parsing on the dispatch instructions, and extracts dispatch operation information; wherein, the dispatch operation information includes the operation object, operation type, operation initial state, and target operation state; The scheduling simulation rule matching module, based on the scheduling operation information, matches the scheduling simulation rule that matches the scheduling instruction from a preset switching operation simulation rule library; the scheduling simulation rule refers to a set of constraints that adjust the execution order of the operation objects. The risk prediction module and the switching operation sequence generation module, based on the scheduling operation information, use the scheduling deduction rules to perform switching operation deduction on the scheduling instructions and generate a switching operation sequence; Based on the preset power grid model and anti-misoperation rules, the switching operation sequence is judged against misoperation in combination with the scheduling operation information, and the risk prediction result is output based on the result of the anti-misoperation judgment.

[0152] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for predicting power grid operation and maintenance risks, characterized in that, include: Obtain scheduling instructions, perform feature parsing on the scheduling instructions, and extract scheduling operation information; wherein, the scheduling operation information includes operation object, operation type, operation initial state, and target operation state; Based on the scheduling operation information, a scheduling deduction rule matching the scheduling instruction is matched from a preset switching operation deduction rule library; the scheduling deduction rule refers to a set of constraints that adjust the execution order of the operation objects. Based on the scheduling operation information, the scheduling deduction rules are used to perform switching operation deduction on the scheduling instructions to generate a switching operation sequence; Based on the preset power grid model and error prevention rules, the switching operation sequence is judged for error prevention in combination with the scheduling operation information. Based on the result of the error prevention judgment, a comprehensive risk value is calculated using a multi-index weighted scoring model based on scenario coefficients. The risk prediction result is output based on the comprehensive risk value.

2. The method for predicting power grid operation and maintenance risks according to claim 1, characterized in that: The scheduling instructions are converted into a unified data format. The converted scheduling instructions are then parsed using NLP and named entity recognition technologies to extract scheduling operation information.

3. The method for predicting power grid operation and maintenance risks according to claim 1, characterized in that: Based on the operation object and the operation type, multiple candidate rules are retrieved from the preset switching operation deduction rule base; From a plurality of candidate rules, a target deduction rule is selected whose state elements are the initial state of the operation and the target state of the operation, and the target deduction rule is used as the scheduling deduction rule corresponding to the scheduling instruction; If no completely identical rule is found among the multiple candidate rules, a similarity algorithm is used to select the closest rule from the multiple candidate rules as the target deduction rule, and adjustments are made according to the differences in the target deduction rule to obtain a scheduling deduction rule that matches the scheduling instruction.

4. The method for predicting power grid operation and maintenance risks according to claim 3, characterized in that: The candidate rules refer to the set of rules initially retrieved from the switching operation deduction rule base that are associated with the operation object and the operation type; The difference refers to the rule content in the target deduction rule that is inconsistent with the scheduling operation information in terms of state elements or operation logic. Adjustments are made based on the marked differences, including modifying device status thresholds or adjusting the operation sequence, to ultimately generate deduction rules adapted to the current scheduling instructions; Calculate the similarity scores between each candidate rule and the scheduling instruction on the operation object, operation type, operation initial state, and operation target state, respectively. Weights are assigned to the operation object, operation type, operation initial state, and operation target state, and the weighted sum of each similarity score is used to obtain the total similarity score. The candidate rule with the highest total similarity score is selected as the target deduction rule.

5. The method for predicting power grid operation and maintenance risks according to claim 1, characterized in that: By sorting all operation objects in the scheduling operation information according to the scheduling deduction rules, an object sequence is obtained; the object sequence refers to a linear execution order generated according to the correlation between operation objects. Based on the initial state of each operation object in the scheduling operation information, the initial state of each operation object in the object sequence is set. The sequence of objects after the initial state is set is deduced step by step until the final state of each operation object is the corresponding operation target state, thus obtaining a complete switching operation sequence; the state change logic refers to the transformation conditions that the operation object follows from the initial state of the operation to the operation target state.

6. The method for predicting power grid operation and maintenance risks according to claim 1, characterized in that: Before performing error prevention judgment, the switching operation sequence is preliminarily verified, including: Load a pre-set verification rule base, which includes basic logical constraints and typical operating modes for power grid operation; The switching operation sequence is analyzed step by step using the verification rule base, and the execution conditions for each operation step are simulated and deduced. When an abnormal situation occurs that does not conform to the verification rule base, a corresponding verification report is generated and a sequence reconstruction mechanism is triggered, including inserting missing steps or adjusting the operation order to improve the switching operation sequence.

7. The method for predicting power grid operation and maintenance risks according to claim 1, characterized in that: The method involves using a preset power grid model and error prevention rules, combined with the dispatch operation information, to perform error prevention judgment on the switching operation sequence. Based on the error prevention judgment result, a comprehensive risk value is calculated using a multi-index weighted scoring model based on scenario coefficients. The risk prediction result is then output based on the comprehensive risk value. Specifically, this includes: The power grid model is based on graph theory to construct the connection relationship of power equipment, and is implemented using an adjacency matrix or adjacency list data structure; The switching operation sequence is mapped to the power grid model to generate a topology change diagram of all operation objects in the switching operation sequence; The topology change diagram includes device connection status, electrical parameters, and operating procedures.

8. The method for predicting power grid operation and maintenance risks according to claim 7, characterized in that: Based on the topology change diagram and the anti-misoperation rules, a topology search is performed on the switching operation sequence using depth-first search and breadth-first search methods, and risk prediction is made based on the search results, specifically including: When initializing the search queue, set the depth and breadth limits for the search; Based on the depth and breadth constraints, the path and node information of each operation object in the topology transformation graph are searched and recorded; Using the aforementioned error prevention rules, the recorded path and node information is judged to determine whether it violates the error prevention rules, and a risk prediction result is output based on the result of the error prevention judgment.

9. The method for predicting power grid operation and maintenance risks according to claim 8, characterized in that: The path and the node information refer to the state change trajectory of the operation object in the topology change graph, which is stored using an adjacency list or matrix data structure; The definition of the error prevention rules specifically includes: Collect actual cases and feedback information during power grid operation, and record and archive the anti-misoperation rules; It also manages the version of the rules and records the rule modification history.

10. The method for predicting power grid operation and maintenance risks according to claim 8, characterized in that: The risk prediction based on the discrimination results includes: If the results of the assessment indicate that an operation violates the error prevention rules, the risk level of the operation violating the error prevention rules is determined based on the severity level of the potential risk; and a corresponding early warning message is generated based on the risk level. The process for obtaining the severity level of the potential risk is as follows: for the potential risk, a comprehensive risk value is calculated using a multi-index weighted scoring model based on scenario coefficients, and the comprehensive risk value is compared with a set level threshold range to obtain the risk level; The aforementioned early warning information refers to the warning content automatically matched according to different risk levels, including risk type, level, scope of impact, and recommended measures.

11. The method for predicting power grid operation and maintenance risks according to claim 10, characterized in that: The specific steps for calculating the comprehensive risk value using a multi-index weighted scoring model based on scenario coefficients are as follows: For different scenario types, set corresponding scenario factors based on experience and actual conditions; determine the specific scenario of the operation object that violates the error prevention rules, match the specific scenario with the different scenario types, sum the scenario factors corresponding to the matched scenario types, add the summation result to 1, and obtain the scenario coefficient. The multiple indicators include the number of users experiencing power outages, equipment maintenance costs, and emergency repair response time. Specifically, the ratio of the number of users experiencing power outages to the total number of users is calculated, multiplied by 100, to obtain the number of users experiencing power outages. The number of devices requiring repair is multiplied by their corresponding maintenance costs, summed, and then divided by the average total cost threshold for the equipment. This average total cost threshold is then multiplied by 100 to obtain the equipment maintenance cost indicator. The estimated emergency repair time is divided by the standard emergency repair time, and the result is multiplied by 100 to obtain the emergency repair response time indicator. Weighting coefficients are set for the power outage number, equipment maintenance cost, and emergency repair response time indicators. Based on the weighting coefficients, the indicators are weighted and summed. The sum is multiplied by the scenario coefficient to obtain the comprehensive risk value. The weighting coefficients are set based on the actual situation.

12. A power grid operation and maintenance risk prediction system utilizing the method of any one of claims 1-11, comprising an operation information acquisition module, a dispatching deduction rule matching module, a switching operation sequence generation module, and a risk prediction module, characterized in that: The operation information acquisition module acquires dispatch instructions generated by the power system, performs feature parsing on the dispatch instructions, and extracts dispatch operation information; wherein, the dispatch operation information includes the operation object, operation type, operation initial state, and target operation state; The scheduling simulation rule matching module, based on the scheduling operation information, matches the scheduling simulation rule that matches the scheduling instruction from a preset switching operation simulation rule library; the scheduling simulation rule refers to a set of constraints that adjust the execution order of the operation objects. The risk prediction module and the switching operation sequence generation module, based on the scheduling operation information, use the scheduling deduction rules to perform switching operation deduction on the scheduling instructions and generate a switching operation sequence; Based on the preset power grid model and anti-misoperation rules, the switching operation sequence is judged against misoperation in combination with the scheduling operation information, and the risk prediction result is output based on the result of the anti-misoperation judgment.

13. A device comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-11.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method according to any one of claims 1-11.

Images