Knowledge evolution and updating method in power transmission and distribution field knowledge base construction

Abstract

The application discloses a knowledge evolution and knowledge base updating method in the field of power transmission and distribution, and relates to the technical field of power systems and natural language processing. The method comprises the following steps: based on the obtained multi-source heterogeneous data, a dual driving mechanism of periodic monitoring and business time triggering is used to identify a knowledge updating time; based on the identified knowledge updating time, the semantic correlation between the newly added data and the existing data in the knowledge base is calculated. The application uses a dual driving mechanism of periodic monitoring and business event triggering to identify the knowledge updating time, which can not only ensure that the knowledge base is maintained regularly according to the preset period, but also timely respond to emergency business events such as equipment fault alarms, operation state changes and industry standard updates, realize the automatic identification of the knowledge updating time, and effectively solve the problems of knowledge updating lag and poor timeliness in the traditional manual maintenance mode.

Description

Technical Field

[0001] This invention relates to the fields of power systems and natural language processing technology, specifically to the knowledge evolution and knowledge base update methods in the construction of knowledge bases in the field of power transmission and distribution. Background Technology

[0002] With the rapid development of new power systems and smart grids, the power transmission and distribution sector has accumulated a wealth of professional texts, operation logs, inspection records, standards and specifications, and image data in business scenarios such as equipment operation and maintenance, fault diagnosis, dispatch management, and production command. Simultaneously, the demand for technologies such as large language models, intelligent question answering, knowledge augmentation generation, and intelligent knowledge services in the power industry is constantly growing. However, existing knowledge systems in the power transmission and distribution sector generally suffer from problems such as scattered knowledge sources, lagging knowledge updates, and insufficient knowledge structuring, resulting in significant deficiencies in existing intelligent systems regarding professional semantic understanding, business logic reasoning, and knowledge timeliness.

[0003] Most existing knowledge base systems adopt a static construction approach, lacking the ability to dynamically analyze the evolution of power transmission and distribution terminology, business rules, equipment operating status, and maintenance experience. This makes them ill-suited to the continuously evolving nature of knowledge in the power transmission and distribution field. Due to the frequent updates to industry standards, dispatching procedures, equipment status, and fault cases, traditional knowledge bases struggle to timely complete knowledge integration, conflict detection, and version iteration. This can easily lead to knowledge redundancy, knowledge obsolescence, and insufficient consistency, thus impacting the effectiveness of intelligent question answering, maintenance assistance, and production decision-making systems.

[0004] Furthermore, power transmission and distribution scenarios involve not only textual information but also a large amount of multimodal data, including equipment images, inspection photos, fault screenshots, and engineering drawings. Traditional knowledge management systems lack the ability to jointly analyze and correlate multimodal information, making it difficult to achieve collaborative understanding of text and graphics, equipment status identification, and knowledge correlation analysis in complex business scenarios. This limits the intelligent application capabilities of knowledge base systems in complex power transmission and distribution scenarios.

[0005] Therefore, there is an urgent need for a knowledge evolution and dynamic update system for knowledge base construction in the power transmission and distribution field. By integrating technologies such as dynamic knowledge updating, multi-source heterogeneous data fusion, knowledge evolution analysis, and multimodal intelligent reasoning, this system can achieve dynamic maintenance, intelligent updating, and efficient management of power transmission and distribution professional knowledge, thereby improving the construction efficiency, timeliness, and intelligent service capabilities of the knowledge base in the power transmission and distribution field. Summary of the Invention

[0006] The purpose of this invention is to provide a method for knowledge evolution and knowledge base updating in the construction of a knowledge base in the field of power transmission and distribution, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, this invention provides the following technical solution: a method for knowledge evolution and knowledge base updating in the construction of a knowledge base in the field of power transmission and distribution, comprising: Based on the acquired multi-source heterogeneous data, the timing of knowledge updates is identified through a dual-drive mechanism of periodic monitoring and business time triggering. Based on the identified knowledge update timing, the semantic correlation between the newly added data and the existing data in the knowledge base is calculated. Based on the comparison structure between the semantic relevance and the preset threshold, a knowledge update strategy is determined. The update strategy includes merging and supplementing existing knowledge items and updating the version, or storing the newly added data as an independent knowledge item in the knowledge base. Multidimensional evolutionary relationship modeling and version management are performed on knowledge entries in the knowledge base to track the trajectory of knowledge changes.

[0008] As a specific solution to the technical solution of this application, the business time triggering includes business status change time and / or business specification update time; The semantic relevance is determined based on the semantic similarity between the new data and the existing data in the vector space.

[0009] As a specific solution of the technical solution of this application, the business status change time includes at least one of equipment fault alarm, equipment operation status change and new operation and maintenance case entry. When any business change event is detected, the update process for the relevant knowledge item is automatically started. The business specification update time includes at least one of industry standard updates and scheduling procedure adjustments. When any business specification update event is detected, the update process for the relevant knowledge items is automatically initiated. The semantic similarity is calculated using the cosine similarity formula: ; in, This indicates the semantic correlation between newly added data and existing data. For the vector of newly added data, A vector of existing data in the knowledge base. For the dot product operation of two vectors, and The Euclidean norms of the newly added data vector and the existing data vector are used to determine the update strategy based on the comparison results of the semantic relevance and the preset threshold. When the semantic relevance is higher than the preset threshold, it is determined that the new data has a high degree of relevance with the existing knowledge entries. Supplementation, fusion and version update operations are performed on the existing knowledge entries to achieve knowledge redundancy removal and knowledge conflict control. When the semantic relevance is lower than a preset threshold, it is determined that the newly added data has a low relevance to existing knowledge entries. The newly added data is then stored in the knowledge base as a new independent knowledge entry to achieve knowledge expansion.

[0010] As a specific solution of the technical solution of this application, the multidimensional evolutionary relationship modeling includes modeling the inheritance relationship, conflict relationship and temporal evolution characteristics between knowledge items; The version association includes recording the version evolution path of knowledge entries to support dynamic tracking and historical tracing of knowledge changes.

[0011] As a specific solution to the technical solution of this application, the multidimensional evolutionary relationship modeling is implemented based on knowledge graphs and semantic association mechanisms, including: Based on knowledge graphs, the inheritance and conflict relationships between knowledge items are represented in a structured manner and stored in relation. Knowledge entities are represented by graph nodes and the evolutionary relationship type and association strength between knowledge entities are represented by graph edges. Based on the semantic association mechanism, the temporal evolution characteristics of knowledge items are modeled. By analyzing the semantic change trajectory of knowledge items in different time windows, the evolution trend and evolution law of knowledge content are identified. Combined with historical operation and maintenance cases and scheduling experience, the trend of knowledge change is predicted and analyzed. Based on the modeling structure of the knowledge graph and semantic association mechanism, incremental knowledge learning, knowledge effectiveness evaluation, and knowledge change trend prediction are realized to improve the adaptability of the knowledge base to complex business scenarios. The version association records the update history of knowledge items based on the knowledge version evolution path, including: the time type triggered by each update, the update timestamp, the differences in knowledge content before and after the update, and the update strategy type, so as to support dynamic tracking of knowledge changes, historical backtracking, and trend prediction analysis.

[0012] As a specific solution of the technical solution of this application, the preprocessing of the acquired multi-source heterogeneous data is performed before the knowledge update time is identified. The preprocessing includes data cleaning, format unification, noise filtering and structure conversion. Based on a deep learning feature extraction model, multimodal semantic feature extraction and encoding are performed on the preprocessed data to produce a knowledge data vector representation for semantic relevance calculation.

[0013] As a specific embodiment of the technical solution in this application, the deep learning feature extraction model includes a convolutional neural network module, a long short-term memory network module, and a connection-based temporal classification module, wherein: The convolutional neural network module is configured to extract local features from business text and image data in multi-source heterogeneous data, and generate local feature representations through multi-layer convolution and pooling operations. The Long Short-Term Memory (LSTM) network module is configured to capture the contextual temporal features of business text and the time-series features of device status. Based on the local feature representation, a gating mechanism is used to model long-distance dependencies and generate temporal feature representations. The connection time-series classification module is configured to achieve automatic matching and semantic mapping of unaligned text sequences. Based on the time-series feature representation, it aligns the variable-length input sequence with the target output sequence, generating the final knowledge data vector representation without prior sequence alignment annotation. Through the cascaded processing of the convolutional neural network module, long short-term memory network module, and connection-series classification module, the automatic recognition and encoding of business text, operational status information, and multimodal semantic features are achieved, thereby improving the accuracy of data recognition and knowledge encoding in complex business scenarios.

[0014] The knowledge evolution and knowledge base update system in the construction of a knowledge base in the power transmission and distribution field includes: The trigger monitoring module identifies knowledge update opportunities based on acquired multi-source heterogeneous data through a dual-drive mechanism of periodic monitoring and business event triggering. The correlation evaluation module calculates the semantic correlation between the newly added data and the existing data in the knowledge base based on the knowledge update timing identified by the trigger monitoring module. The update decision module, based on the comparison result of the semantic relevance calculated by the relevance evaluation module and the preset threshold, decides on the knowledge update strategy. The update strategy includes merging and supplementing existing knowledge items and updating the version, or adding new data as independent knowledge items and storing them in the knowledge base. The evolution analysis module is configured to perform multi-dimensional evolutionary relationship modeling and version management for knowledge entries in the knowledge base, in order to track the trajectory of knowledge changes.

[0015] As a specific solution of the technical solution of this application, the trigger monitoring module identifies the timing of knowledge updates based on monitoring business status change events and / or business specification update events; The correlation evaluation module determines the semantic correlation based on the semantic similarity calculation between the new data and the existing data in the vector space.

[0016] Compared with the prior art, the beneficial effects of the present invention are: This application identifies knowledge update opportunities through a dual-drive mechanism of periodic monitoring and business event triggering. This ensures that the knowledge base is maintained regularly according to a preset cycle, and also responds promptly to urgent business events such as equipment fault alarms, changes in operating status, and updates to industry standards. This achieves automated identification of knowledge update opportunities and effectively solves the problems of delayed and inefficient knowledge updates under traditional manual maintenance methods.

[0017] Meanwhile, by calculating the semantic correlation between new data and existing data in the knowledge base, and intelligently deciding on knowledge update strategies based on the comparison results of semantic correlation with preset thresholds, supplementary fusion and version updates can be performed when new data and existing knowledge items have a high correlation. When the correlation is low, the new data is stored in the knowledge base as an independent knowledge item, thereby effectively avoiding knowledge redundancy, achieving knowledge conflict control, and ensuring the continuous expansion of the knowledge base, improving the quality and coverage of the knowledge base.

[0018] By modeling and managing the multidimensional evolutionary relationships of knowledge entries in the knowledge base, and by representing inheritance and conflict relationships in a structured manner based on knowledge graphs, modeling temporal evolutionary features based on semantic association mechanisms, and recording the version evolution path of knowledge entries, the knowledge change trajectory can be fully tracked. This supports dynamic tracking, historical backtracking, and trend prediction analysis of knowledge changes, which helps improve the adaptability of the knowledge base to complex business scenarios.

[0019] By preprocessing multi-source heterogeneous data before identifying the timing of knowledge updates, and performing multimodal semantic feature extraction and encoding based on a deep learning feature extraction model that includes a convolutional neural network module, a long short-term memory network module, and a connection-time classification module, high-quality knowledge data vector representations can be generated. This effectively eliminates the impact of data noise and format differences, improves the accuracy of data recognition and knowledge encoding in complex business scenarios, and thus enhances the reliability of semantic correlation calculation. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the knowledge evolution and knowledge base update system structure in the construction of the knowledge base in the field of power transmission and distribution according to the present invention. Figure 2 This is a flowchart illustrating the knowledge evolution and knowledge base update method in the construction of the knowledge base in the field of power transmission and distribution according to the present invention. Figure 3 This is a flowchart illustrating another method for knowledge evolution and knowledge base update in the construction of a knowledge base in the field of power transmission and distribution, according to the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] The terms "first," "second," etc., in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. For example, the first cluster and the second cluster mentioned below belong to different clusters. It should be understood that such names can be used interchangeably 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" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or modules is not necessarily limited to those steps or modules explicitly listed, but may include other steps or modules not explicitly listed or inherent to these processes, methods, products, or devices. The division of modules in the embodiments of this application is merely a logical division. In actual applications, there may be other division methods. For example, multiple modules may be combined into or integrated into another system, or some features may be ignored or not performed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be through some interface, and the indirect coupling or communication connection between modules may be electrical or other similar forms. None of these are limited in the embodiments of this application. Furthermore, the modules or sub-modules described as separate components may or may not be physically separated, may or may not be physical modules, or may be distributed among multiple circuit modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the embodiments of this application.

[0023] To improve the timeliness and accuracy of knowledge bases in the power transmission and distribution field, and to achieve dynamic tracking of knowledge evolution trajectories, such as... Figure 1 As shown in the illustration, this application provides a system for knowledge evolution and knowledge base update in the construction of a knowledge base in the field of power transmission and distribution. The system includes a trigger monitoring module, a correlation evaluation module, an update decision module, and an evolution analysis module. Optionally, the system also includes a preprocessing module and a feature extraction module.

[0024] Option 1 In the first approach, a trigger monitoring module is used to identify knowledge update opportunities based on acquired multi-source heterogeneous data through a dual-drive mechanism of periodic monitoring and business event triggering; a correlation evaluation module is used to calculate the semantic correlation between newly added data and existing data in the knowledge base based on the knowledge update opportunities identified by the trigger monitoring module. The update decision module is used to determine the knowledge update strategy based on the comparison result of the semantic relevance calculated by the relevance evaluation module and the preset threshold. The update strategy includes merging and supplementing existing knowledge items and updating the version, or storing the new data as an independent knowledge item in the knowledge base. The evolution analysis module is used to perform multi-dimensional evolution relationship modeling and version management on the knowledge items in the knowledge base to track the trajectory of knowledge changes.

[0025] As one possible implementation method, a monitoring module is triggered, specifically used to monitor business status change events and / or business specification update events to identify the timing of knowledge updates.

[0026] As one possible implementation method, the correlation evaluation module is specifically used to determine the semantic correlation by calculating the semantic similarity between the new data and the existing data in the vector space.

[0027] As one possible implementation method, the update decision module is specifically used to determine that the new data has a high degree of correlation with the existing knowledge entries when the semantic correlation is higher than a preset threshold, and to perform supplementation, fusion and version update operations on the existing knowledge entries; or, when the semantic correlation is lower than a preset threshold, it is determined that the new data has a low degree of correlation with the existing knowledge entries, and the new data is stored as a new knowledge entry in the knowledge base.

[0028] As one possible implementation method, the evolutionary analysis module is specifically used to perform structured representation and relational storage of inheritance and conflict relationships between knowledge items based on knowledge graphs. Knowledge entities are represented by graph nodes, and the evolutionary relationship type and association strength between knowledge entities are represented by graph edges. Furthermore, the temporal evolutionary characteristics of knowledge items are modeled based on semantic association mechanisms.

[0029] As one possible implementation, the system also includes a preprocessing module for performing preprocessing on the acquired multi-source heterogeneous data before identifying the knowledge update timing. The preprocessing includes data cleaning, format unification, noise filtering, and structure transformation.

[0030] As one possible implementation, the system also includes a feature extraction module, which performs multimodal semantic feature extraction and encoding on the preprocessed data based on a deep learning feature extraction model to generate a knowledge data vector representation for semantic relevance calculation.

[0031] The second option In the second approach, a trigger monitoring module is used to identify knowledge update opportunities based on acquired multi-source heterogeneous data through a dual-drive mechanism of periodic monitoring and business event triggering; a correlation evaluation module is used to calculate the semantic correlation between newly added data and existing data in the knowledge base based on the actual knowledge update identified by the trigger monitoring module; an update decision module is used to decide on the knowledge update strategy based on the comparison result between the semantic correlation calculated by the correlation evaluation module and a preset threshold; and an evolutionary analysis module is used to perform multi-dimensional evolutionary relationship modeling and version management for knowledge items in the knowledge base.

[0032] As one possible implementation method, the business status change events that trigger the monitoring module to monitor include at least one of the following: equipment fault alarm, equipment operating status change, and new operation and maintenance case entry. When any business change event is detected, the update process for the relevant knowledge item is automatically started.

[0033] As one possible implementation method, the business specification update events that trigger the monitoring module to monitor include at least one of industry standard updates and scheduling procedure adjustments. When any task specification update event is detected, the update process for the relevant knowledge items is automatically initiated.

[0034] As one possible implementation method, the relevance evaluation module uses the cosine similarity formula to calculate semantic relevance: ; in, This indicates the semantic correlation between newly added data and existing data. For the vector of newly added data, A vector of existing data in the knowledge base. For the dot product operation of two vectors, and These are the Euclidean norms of the newly added data vector and the existing data vector, respectively.

[0035] As one possible implementation method, the update decision module determines the update strategy based on the comparison between semantic relevance and a preset threshold. When the semantic relevance is higher than the preset threshold, it is determined that the new data has a high degree of relevance with existing knowledge items, and the existing knowledge items are supplemented, merged and updated to achieve knowledge redundancy removal and conflict control. When the semantic relevance is lower than the preset threshold, it is determined that the new data has a low degree of relevance with existing knowledge items, and the new data is stored in the knowledge base as a new independent knowledge item to achieve knowledge expansion.

[0036] As one possible implementation method, the multidimensional evolutionary relationship modeling of the evolutionary analysis module includes modeling the inheritance relationship, conflict relationship and temporal evolution characteristics between knowledge items; version management includes recording the version evolution path of knowledge items to support dynamic tracking and historical backtracking of knowledge changes.

[0037] The following description, in conjunction with the accompanying drawings, introduces the knowledge evolution and knowledge base update methods in the construction of the power transmission and distribution knowledge base provided in this application. It is understood that in the embodiments of this application, the interactive execution entities are the trigger monitoring module, the correlation evaluation module, the update decision module, and the evolution analysis module. The meanings of the trigger monitoring module, the correlation evaluation module, the update decision module, and the evolution analysis module can be found in the preceding description.

[0038] Figure 2 This is a flowchart illustrating the knowledge evolution and knowledge base update method in the construction of a knowledge base in the power transmission and distribution field provided in this embodiment. The method includes the trigger monitoring module, correlation evaluation module, update decision module, and evolution analysis module, and comprises the following steps.

[0039] Step 101: Trigger the monitoring module to acquire multi-source heterogeneous data and monitor the timing of knowledge updates.

[0040] The trigger monitoring module is used to acquire multi-source heterogeneous data and identifies the timing of knowledge updates through a dual-drive mechanism of periodic monitoring and business event triggering.

[0041] Periodic monitoring refers to scanning multi-source heterogeneous data according to a preset time period (such as hourly, weekly, or daily) to detect whether there is new data or data changes. Business event triggering refers to immediately initiating the knowledge update process when a specific business event is detected.

[0042] Business events include business status change events and business specification update events. Business status change events include at least one of the following: equipment fault alarms, changes in equipment operating status, and the entry of new maintenance cases. Business specification update events include at least one of the following: industry standard updates and scheduling procedure adjustments.

[0043] This is a common practice where, when a transformer in a power transmission and distribution system triggers a fault alarm, the monitoring module detects the change in the business status and identifies the timing for a knowledge update. Alternatively, when the National Energy Administration releases a new power industry standard, the monitoring module detects the update event of the business specification and identifies the timing for a knowledge update.

[0044] It should be noted that step 101 is a step that is executed multiple times. Step 101 will be executed whenever the monitoring module detects a new business event or reaches the preset monitoring period.

[0045] As one implementation method, a data preprocessing step can also be performed between steps 101. The preprocessing module performs preprocessing on the acquired multi-source heterogeneous data. The preprocessing includes data cleaning, format unification, noise filtering, and structure conversion.

[0046] Based on the above scheme, by preprocessing multi-source heterogeneous data, noise and redundant information in the data can be eliminated, the data format can be unified, and a high-quality data foundation can be provided for subsequent semantic feature extraction and correlation calculation.

[0047] Step 102: The feature extraction module performs multimodal semantic feature extraction and encoding on the preprocessed data.

[0048] The feature extraction module performs multimodal semantic feature extraction and encoding on the preprocessed data based on the deep learning feature extraction module to generate a knowledge data vector representation for semantic relevance calculation.

[0049] The deep learning feature extraction model includes a convolutional neural network (CNN) module, a long short-term memory (LSTM) module, and a connectionist temporal classification (CTC) module.

[0050] The CNN module is configured to extract local features from business text and image data in multi-source heterogeneous data, and generate local feature representations through multi-layer convolution and pooling operations.

[0051] The LSTM module is designed to capture the contextual temporal features of business text and the event sequence features of device characteristic states. Based on local feature representation, it models long-distance dependencies through a gating mechanism to generate temporal feature representation.

[0052] The CTC module is configured to automatically match and semantically map unaligned text sequences. It aligns the input sequence with the target output sequence based on temporal feature representation, generating the final knowledge data vector representation without prior sequence alignment annotation.

[0053] By cascading CNN, LSTM, and CTC modules, the automatic recognition and encoding of business text, operational status information, and multimodal semantic features are achieved.

[0054] Based on the above scheme, multimodal semantic feature extraction and encoding of multi-source heterogeneous data through deep learning feature extraction models can generate high-quality knowledge data vector representations, which helps to improve the accuracy of subsequent semantic relevance calculation.

[0055] Step 103: The correlation evaluation module calculates the semantic correlation between the newly added data and the existing data in the knowledge base.

[0056] The correlation assessment module calculates the semantic correlation between newly added data and existing data in the knowledge base based on the knowledge update timing identified by the trigger monitoring module.

[0057] The semantic relevance is determined based on the semantic similarity between the new data and existing data in the vector space. The relevance evaluation module uses the cosine similarity formula to calculate the semantic relevance: ; in, This indicates the semantic correlation between newly added data and existing data. For the vector of newly added data, A vector of existing data in the knowledge base. For the dot product operation of two vectors, and These are the Euclidean norms of the newly added data vector and the existing data vector, respectively.

[0058] This step 103 can be executed multiple times. For example, the relevance assessment module can sequentially calculate the semantic relevance between the new data and multiple existing knowledge entries in the knowledge base to determine the existing knowledge entry most relevant to the new data.

[0059] For example, suppose the new data is a processing record of a transformer oil temperature abnormality alarm, and the knowledge base already contains a knowledge entry for "transformer fault handling". The correlation evaluation module calculates the semantic correlation between the new data and the existing knowledge entry. If the calculated semantic correlation is 0.85, it indicates that the new data and the existing knowledge entry have a high semantic correlation.

[0060] Based on the above scheme, by calculating the semantic correlation between new data and existing data, the degree of relevance between new data and existing knowledge items can be quantified, providing a basis for subsequent update strategy decisions.

[0061] Step 104: Update the decision module based on the comparison results of semantic relevance and preset threshold, and make a decision knowledge update strategy.

[0062] The update decision module determines the knowledge update strategy based on the comparison between the semantic relevance calculated by the relevance evaluation module and the preset threshold.

[0063] The preset threshold can be set according to actual business needs, such as 0.7 or 0.8.

[0064] When the semantic relevance is higher than a preset threshold, the update decision module determines that the new data has a high degree of relevance to the existing knowledge entries, and performs supplementary fusion and version update operations on the existing knowledge entries to achieve knowledge redundancy removal and conflict control.

[0065] When the semantic relevance is lower than a preset threshold, the update decision module determines that the new data has a low relevance to existing knowledge entries and stores the new data as a new independent knowledge entry in the knowledge base to achieve knowledge expansion.

[0066] For example, assume the preset threshold is 0.75. If the semantic correlation degree calculated in step 103 is 0.85, which is higher than the preset threshold of 0.75, then the decision update module decides to supplement, integrate and update the version of the "transformer fault handling" knowledge item, integrate the newly added oil temperature abnormal alarm handling record into the knowledge item, and update the version number.

[0067] If the semantic relevance calculated in step 103 is 0.45, which is lower than the preset threshold of 0.75, the update decision module will decide to store the newly added data as a new independent knowledge entry in the knowledge base, such as creating a knowledge entry for "new type of transformer anomaly diagnosis method".

[0068] As an implementation method, the update decision module can also record the update strategy type when performing knowledge update operations, including the fusion supplement type or the new entry type, so as to facilitate subsequent version management and evolution analysis.

[0069] Based on the above scheme, by comparing semantic relevance with a preset threshold, the knowledge update strategy can be intelligently decided, which can avoid knowledge redundancy and achieve continuous knowledge expansion, thus helping to improve the quality and coverage of the knowledge base.

[0070] Step 105: The evolutionary analysis module performs multi-dimensional evolutionary relationship modeling and version management on the knowledge entries in the knowledge base.

[0071] The evolutionary analysis module performs multi-dimensional evolutionary relationship modeling and version management on knowledge entries in the knowledge base to track the trajectory of knowledge changes.

[0072] Among them, multidimensional evolutionary relationship modeling includes modeling the inheritance relationship, conflict relationship and temporal evolution characteristics between knowledge items.

[0073] As an implementation method, the evolutionary analysis module uses a knowledge graph to structurally represent and store the inheritance and conflict relationships between knowledge items. Knowledge entities are represented by graph nodes, and the evolutionary relationship type and association strength between knowledge entities are represented by graph edges.

[0074] As an implementation method, the evolutionary analysis module models the temporal evolution characteristics of knowledge items based on the semantic association mechanism. By analyzing the semantic change trajectory of knowledge items in different time windows, it identifies the evolutionary trend and evolutionary law of knowledge content, and combines historical operation and maintenance cases and scheduling experience to predict and analyze the trend of knowledge change.

[0075] As an implementation method, the evolutionary analysis module records the update history of knowledge items based on the knowledge version evolution path, including: the time type triggered by each update, the update timestamp, the difference in knowledge content before and after the update, and the update strategy type, so as to support dynamic tracking of knowledge changes, historical backtracking, and trend prediction analysis.

[0076] Based on the above scheme, by modeling the multidimensional evolutionary relationship of knowledge items and managing their versions, the trajectory of knowledge change can be fully recorded, supporting dynamic tracking, historical backtracking and trend prediction of knowledge, which helps to improve the adaptability of the knowledge base to complex business scenarios.

[0077] Optionally, after step 105 above, steps 106 to 107 may also be performed.

[0078] Step 106: Based on the modeling results of knowledge graphs and semantic association mechanisms, the evolutionary analysis module realizes incremental knowledge learning, knowledge effectiveness evaluation, and prediction of knowledge change trends.

[0079] The evolutionary analysis module, based on the modeling results of knowledge graphs and semantic association mechanisms, enables incremental knowledge learning, knowledge effectiveness evaluation, and prediction of knowledge change trends.

[0080] Among these, incremental knowledge learning refers to continuously updating and expanding the knowledge graph based on new data to improve the coverage of the knowledge base. Knowledge effectiveness assessment refers to regularly evaluating the timeliness and accuracy of knowledge items and marking outdated or invalid knowledge. Knowledge change trend prediction refers to predicting the future direction of knowledge change based on historical evolution data, providing a basis for the forward-looking updating of the knowledge base.

[0081] Based on the above scheme, the intelligence level of the knowledge base can be continuously improved through knowledge increment learning, knowledge effectiveness assessment and knowledge change trend prediction, so that it can better adapt to the complex and ever-changing business scenarios in the power transmission and distribution field.

[0082] Step 107: The evolution analysis module outputs a knowledge evolution analysis report.

[0083] The evolution analysis module generates and outputs a knowledge evolution analysis report, which includes the overall evolution trend of the knowledge base, frequently updated knowledge items, knowledge conflict hotspots, and predictions of future knowledge demand.

[0084] Based on the above scheme, by outputting knowledge evolution analysis reports, decision support can be provided to knowledge base administrators to help them optimize the knowledge base structure and update strategies.

[0085] As an implementation method, this embodiment is applicable to complex business scenarios in the power transmission and distribution field. It can use a dual-drive mechanism to achieve intelligent identification of knowledge update timing and use semantic correlation calculation to achieve automatic decision-making on update strategies. Application scenarios of this embodiment include, for example, automatically triggering the update process of relevant knowledge entries when equipment failure occurs in the power transmission and distribution network, and promptly incorporating fault handling experience into the knowledge base; and automatically identifying affected knowledge entries and updating them when industry standards are updated, ensuring the compliance and timeliness of the knowledge base.

[0086] Based on the above scheme, the trigger monitoring module identifies knowledge update opportunities through a dual-drive mechanism of periodic monitoring and business event triggering. The correlation evaluation module calculates the semantic correlation between new data and existing data. The update decision module intelligently decides on knowledge update strategies based on the comparison results of semantic correlation and preset thresholds. The evolution analysis module performs multi-dimensional evolutionary relationship modeling and version management for knowledge items. This enables automated and intelligent updates of the knowledge base, effectively tracks the trajectory of knowledge changes, and improves the adaptability of the knowledge base to complex business scenarios. In this method, a high-quality knowledge data vector representation is generated through a deep learning feature extraction model, and the degree of knowledge correlation is quantified through semantic similarity calculation, avoiding the subjectivity and uncertainty of manual judgment, thus preventing bias in knowledge update decisions. Furthermore, due to the use of multi-dimensional evolutionary relationship modeling and version management, this method can completely record the evolutionary history of knowledge, support dynamic tracking and historical backtracking of knowledge, and has excellent traceability and interpretability.

[0087] Figure 3 This is a flowchart illustrating another method for knowledge evolution and knowledge base updating in the construction of a knowledge base in the field of power transmission and distribution, provided as an embodiment of this application. The trigger monitoring module, correlation evaluation module, update decision module, and evolution analysis module in this embodiment can be... Figure 1 The method comprises a trigger monitoring module, a correlation assessment module, an update decision module, and an evolutionary analysis module. The steps include:

[0088] Step 201: Configure periodic monitoring parameters and business event triggering rules for the trigger monitoring module.

[0089] The monitoring module is configured with periodic monitoring parameters, including the monitoring period (e.g., hourly, daily) and the range of monitoring data. It also configures business event triggering rules, including the type and priority of business status change events, and the source and identification method of business specification update events.

[0090] As an implementation method, the trigger monitoring module can also be configured with whitelists and blacklists for business events. Events in the whitelist are processed first, while events in the blacklist are filtered to avoid invalid triggers.

[0091] Step 202: Trigger the monitoring module to perform periodic monitoring and detect changes in multi-source heterogeneous data.

[0092] The trigger monitoring module performs periodic scans of multi-source heterogeneous data according to the configured monitoring cycle to detect whether there are any changes such as new data, data modification, or data deletion.

[0093] As one implementation method, the trigger monitoring module can record the timestamp, type, and content of data changes during periodic monitoring to form a data change log for subsequent evolutionary analysis.

[0094] Step 203: The monitoring module is triggered to determine whether a business event has been detected.

[0095] The trigger monitoring module monitors the occurrence of business events in real time and determines whether a business event trigger is detected outside of periodic monitoring.

[0096] If a business event is detected, proceed to step 204; if no business event is detected, continue with the periodic monitoring in step 202.

[0097] Step 204: The monitoring module is triggered to identify the knowledge update timing and notify the correlation evaluation module to start semantic correlation calculation.

[0098] When the trigger monitoring module detects data changes through periodic monitoring or detects a knowledge update requirement through business events, it identifies the timing of the knowledge update and sends a notification to the correlation assessment module to initiate the semantic correlation calculation process.

[0099] For specific implementation details of steps 205 to 208, please refer to the aforementioned... Figure 2 The descriptions of steps 103 to 105 in the method embodiment are not repeated here.

[0100] Based on the above solution, by configuring periodic monitoring parameters and business event triggering rules, the timing of knowledge updates can be flexibly identified. This ensures both regular maintenance of the knowledge base and timely response to urgent business events, thereby improving the timeliness and relevance of knowledge base updates.

[0101] The above mainly describes the solution provided by the embodiments of this application from the perspective of the interaction between the trigger monitoring module, the correlation evaluation module, the update decision module, and the evolution analysis module. It can be understood that, in order to achieve the above functions, each module may include the corresponding hardware structure and / or software module for executing each function.

[0102] In the case of integrated units, possible exemplary devices involved in the embodiments of this application include: a knowledge base updating device, which may include a processing unit and a communication unit. The processing unit is used to control and manage the operations of the knowledge base updating device. The communication unit is used to support communication between the knowledge base updating device and other devices. Optionally, the communication unit, also referred to as a transceiver unit, may include a receiving unit and / or a sending unit, respectively used to perform receiving and sending operations. The knowledge base updating device may also include a storage unit for storing the program code and / or data of the knowledge base updating device.

[0103] The knowledge base update device can be the trigger monitoring module, correlation evaluation module, update decision module, or evolution analysis module in the above embodiments. The processing unit can support the knowledge base update device in executing the actions of the corresponding modules in the above method embodiments.

[0104] In the first scheme, the knowledge base update device is used to perform the aforementioned Figure 2 In the method embodiments, the operations are performed by the trigger monitoring module, the correlation evaluation module, the update decision module, or the evolution analysis module. Specifically, the processing unit is used to identify the timing of knowledge updates based on the acquired multi-source heterogeneous data through a dual-drive mechanism of periodic monitoring and business event triggering; calculate the semantic correlation between the newly added data and the existing data in the knowledge base; decide on the knowledge update strategy based on the comparison result of the semantic correlation and the preset threshold; and perform multi-dimensional evolutionary relationship modeling and version management on the knowledge items in the knowledge base.

[0105] As one possible implementation method, the processing unit is specifically used to monitor business status change events and / or business specification update events in order to identify the timing of knowledge updates.

[0106] As one possible implementation method, the processing unit is specifically used to calculate the semantic similarity between the new data and the existing data in the vector space to determine the semantic relevance.

[0107] As one possible implementation method, the processing unit is specifically used to determine that the new data has a high degree of correlation with the existing knowledge entries when the semantic correlation is higher than a preset threshold, and to perform supplementation, fusion and version update operations on the existing knowledge entries; or, when the semantic correlation is lower than the preset threshold, determine that the new data has a low degree of correlation with the existing knowledge entries, and to store the new data as a new independent knowledge entry in the knowledge base.

[0108] As one possible implementation method, the processing unit is also used to perform structured representation and relational storage of inheritance and conflict relationships between knowledge items based on knowledge graphs; to model the temporal evolution characteristics of knowledge items based on semantic association mechanisms; and to record the update history of knowledge items based on knowledge version evolution paths.

[0109] This application embodiment can divide the knowledge base updating device into functional units according to the above method example. For example, each function can be divided into a separate functional unit, or two or more functions can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0110] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended embodiments and their equivalents.

Claims

1. A method for knowledge evolution and knowledge base updating in the construction of a knowledge base in the field of power transmission and distribution, characterized in that, include: Based on the acquired multi-source heterogeneous data, the timing of knowledge updates is identified through a dual-drive mechanism of periodic monitoring and business time triggering. Based on the identified knowledge update timing, the semantic correlation between the newly added data and the existing data in the knowledge base is calculated. Based on the comparison structure between the semantic relevance and the preset threshold, a knowledge update strategy is determined. The update strategy includes merging and supplementing existing knowledge items and updating the version, or storing the newly added data as an independent knowledge item in the knowledge base. Multidimensional evolutionary relationship modeling and version management are performed on knowledge entries in the knowledge base to track the trajectory of knowledge changes.

2. The method for knowledge evolution and knowledge base updating in the construction of a knowledge base in the field of power transmission and distribution according to claim 1, characterized in that, The business time triggering includes business status change time and / or business specification update time; The semantic relevance is determined based on the semantic similarity between the new data and the existing data in the vector space.

3. The method for knowledge evolution and knowledge base updating in the construction of a knowledge base in the field of power transmission and distribution according to claim 2, characterized in that, The business status change time includes at least one of equipment fault alarm, equipment operation status change and new operation and maintenance case entry. When any business change event is detected, the update process for the relevant knowledge item is automatically started. The business specification update time includes at least one of industry standard updates and scheduling procedure adjustments. When any business specification update event is detected, the update process for the relevant knowledge items is automatically initiated. The semantic similarity is calculated using the cosine similarity formula: ； in, This indicates the semantic correlation between newly added data and existing data. For the vector of newly added data, A vector of existing data in the knowledge base. For the dot product operation of two vectors, and The Euclidean norms of the newly added data vector and the existing data vector are used to determine the update strategy based on the comparison results of the semantic relevance and the preset threshold. When the semantic relevance is higher than the preset threshold, it is determined that the new data has a high degree of relevance with the existing knowledge entries. Supplementation, fusion and version update operations are performed on the existing knowledge entries to achieve knowledge redundancy removal and knowledge conflict control. When the semantic relevance is lower than a preset threshold, it is determined that the newly added data has a low relevance to existing knowledge entries. The newly added data is then stored in the knowledge base as a new independent knowledge entry to achieve knowledge expansion.

4. The method for knowledge evolution and knowledge base updating in the construction of a knowledge base in the field of power transmission and distribution according to claim 1, characterized in that, The multidimensional evolutionary relationship modeling includes modeling the inheritance relationship, conflict relationship, and temporal evolution characteristics between knowledge items; The version association includes recording the version evolution path of knowledge entries to support dynamic tracking and historical tracing of knowledge changes.

5. The method for knowledge evolution and knowledge base updating in the construction of a knowledge base in the field of power transmission and distribution according to claim 4, characterized in that, The multidimensional evolutionary relationship modeling is implemented based on knowledge graphs and semantic association mechanisms, including: Based on knowledge graphs, the inheritance and conflict relationships between knowledge items are represented in a structured manner and stored in relation. Knowledge entities are represented by graph nodes and the evolutionary relationship type and association strength between knowledge entities are represented by graph edges. Based on the semantic association mechanism, the temporal evolution characteristics of knowledge items are modeled. By analyzing the semantic change trajectory of knowledge items in different time windows, the evolution trend and evolution law of knowledge content are identified. Combined with historical operation and maintenance cases and scheduling experience, the trend of knowledge change is predicted and analyzed. Based on the modeling structure of the knowledge graph and semantic association mechanism, incremental knowledge learning, knowledge effectiveness evaluation, and knowledge change trend prediction are realized to improve the adaptability of the knowledge base to complex business scenarios. The version association records the update history of knowledge items based on the knowledge version evolution path, including: the time type triggered by each update, the update timestamp, the differences in knowledge content before and after the update, and the update strategy type, so as to support dynamic tracking of knowledge changes, historical backtracking, and trend prediction analysis.

6. The method for knowledge evolution and knowledge base updating in the construction of a knowledge base in the field of power transmission and distribution according to claim 1, characterized in that, Before the knowledge update time is identified, the acquired multi-source heterogeneous data is preprocessed, including data cleaning, format unification, noise filtering and structure conversion. Based on a deep learning feature extraction model, multimodal semantic feature extraction and encoding are performed on the preprocessed data to produce a knowledge data vector representation for semantic relevance calculation.

7. The method for knowledge evolution and knowledge base updating in the construction of a knowledge base in the field of power transmission and distribution according to claim 6, characterized in that, The deep learning feature extraction model includes a convolutional neural network module, a long short-term memory network module, and a connection-based temporal classification module, wherein: The convolutional neural network module is configured to extract local features from business text and image data in multi-source heterogeneous data, and generate local feature representations through multi-layer convolution and pooling operations. The Long Short-Term Memory (LSTM) network module is configured to capture the contextual temporal features of business text and the time-series features of device characteristic states. Based on the local feature representation, a gating mechanism is used to model long-distance dependencies and generate temporal feature representations. The connection time-series classification module is configured to achieve automatic matching and semantic mapping of unaligned text sequences. Based on the time-series feature representation, it aligns the variable-length input sequence with the target output sequence, generating the final knowledge data vector representation without prior sequence alignment annotation. Through the cascaded processing of the convolutional neural network module, long short-term memory network module, and connection-time classification module, the automatic recognition and encoding of business text, running status information, and multimodal semantic features are achieved, thereby improving the accuracy of data recognition and knowledge encoding in complex business scenarios.

8. A knowledge evolution and knowledge base update system for the construction of a knowledge base in the field of power transmission and distribution, characterized in that... include: The trigger monitoring module identifies knowledge update opportunities based on acquired multi-source heterogeneous data through a dual-drive mechanism of periodic monitoring and business event triggering. The correlation evaluation module calculates the semantic correlation between the newly added data and the existing data in the knowledge base based on the knowledge update timing identified by the trigger monitoring module. The update decision module, based on the comparison result of the semantic relevance calculated by the relevance evaluation module and the preset threshold, decides on the knowledge update strategy. The update strategy includes merging and supplementing existing knowledge items and updating the version, or adding new data as independent knowledge items and storing them in the knowledge base. The evolution analysis module is configured to perform multi-dimensional evolutionary relationship modeling and version management for knowledge entries in the knowledge base, in order to track the trajectory of knowledge changes.

9. The knowledge evolution and knowledge base update system in the construction of a knowledge base in the field of power transmission and distribution according to claim 8, characterized in that, The trigger monitoring module identifies the timing of knowledge updates based on monitoring events that change the business status and / or events that update business specifications. The correlation evaluation module determines the semantic correlation based on the semantic similarity calculation between the new data and the existing data in the vector space.

Images