Standard semantic knowledge base retrieval method for power equipment

Abstract

The invention provides a power equipment-oriented standard semantic knowledge base retrieval method, which comprises the following steps of: acquiring power equipment standard data, analyzing text information of the power equipment standard data, establishing semantic association among standard terms, equipment attributes and key indexes, and forming an equipment-standard-index knowledge graph; standard terms are calculated through a semantic vectorization technology, and matching calculation of the standard and equipment suitability is carried out based on a vector retrieval method; and according to the standard historical data and the deep learning model, carrying out evolution analysis on the standard terms, and providing suitability recommendation of the equipment standard. By establishing the semantic association between the equipment and the standard terms, efficient and accurate standard suitability recommendation and evolution analysis are realized, and efficient and accurate equipment standard suitability recommendation is provided through a semantic technology and a deep learning model.

Description

Technical Field

[0001] This invention relates to the field of power equipment management and intelligent retrieval technology, and in particular to a standard semantic knowledge base retrieval method for power equipment. Background Technology

[0002] With the continuous development of information technology and big data analysis, the management and maintenance of power equipment are gradually becoming more intelligent. Against this backdrop, the management and application of power equipment standards have become an important means to improve the performance and maintenance efficiency of power equipment. Power equipment standards involve numerous technical clauses, covering the technical parameters, operating requirements, and safety specifications of the equipment. Traditional standard retrieval methods usually rely on keyword-based matching, but these methods often lack flexibility and accuracy, especially when the standard content is complex, numerous clauses are made, and constantly updated. With the increasing variety of equipment and technical requirements, how to effectively manage and utilize power equipment standards has become an urgent problem to be solved. In recent years, standard retrieval methods based on semantic technology and artificial intelligence have gradually attracted widespread attention, especially the application of semantic vectorization technology and deep learning models, which have provided new ideas and technical paths for the retrieval and suitability assessment of power equipment standards.

[0003] However, existing power equipment standard retrieval technologies still have shortcomings in several aspects. Traditional standard retrieval methods rely heavily on keyword matching of text, which cannot effectively understand the deep semantic relationships between standard clauses, resulting in low accuracy of retrieval results. In existing technologies, many standard retrieval systems fail to effectively utilize the multidimensional attribute information of power equipment and cannot establish accurate semantic associations between equipment and standard clauses. Therefore, they are difficult to meet the standard adaptation needs of complex power equipment in practical applications. Although some methods attempt to combine deep learning models for evolutionary analysis and adaptation recommendations of standard clauses, these methods usually lack comprehensive utilization of historical standard data and fail to fully consider the evolutionary trends of standard clauses between different versions. Therefore, existing technologies cannot fully leverage the advantages of big data analysis and artificial intelligence technologies to achieve efficient and accurate standard adaptation recommendations and evolutionary analysis. Especially when facing complex and ever-changing power equipment standard environments, there is still considerable room for improvement. Summary of the Invention

[0004] This invention provides a standard semantic knowledge base retrieval method for power equipment. By establishing semantic associations between equipment and standard clauses, it achieves efficient and accurate standard compatibility recommendations and evolutionary analysis. Furthermore, it provides efficient and accurate equipment standard compatibility recommendations through semantic technology and deep learning models.

[0005] The first aspect of this invention provides a standard semantic knowledge base retrieval method for power equipment, comprising the following steps:

[0006] Acquire power equipment standard data, parse the text information of the power equipment standard data, establish semantic relationships between standard clauses, equipment attributes and key indicators, and form a knowledge graph of equipment-standards-indicators;

[0007] The standard clauses are calculated using semantic vectorization technology, and the standard and device compatibility are matched and calculated based on vector retrieval methods.

[0008] Based on historical standard data and deep learning models, we conduct evolutionary analysis on standard clauses and provide recommendations on equipment standard compatibility.

[0009] Furthermore, the parsing of the text information of the power equipment standard data includes the following steps:

[0010] Data from national standards, enterprise standards, international standards, and equipment technical manuals can be obtained through API data interfaces, web crawling technology, or manual input, and unstructured data can be converted into structured text.

[0011] The scanned standard document is parsed using OCR technology, and formatted using regular expressions to store the text hierarchically according to chapters, clauses, and sub-clauses.

[0012] Furthermore, establishing semantic relationships between standard clauses, device attributes, and key metrics includes the following steps:

[0013] Based on the BERT+CRF entity recognition method, the equipment name, technical parameters and key indicators in the standard clauses are automatically extracted and stored in the database;

[0014] A knowledge graph of devices, standards, and metrics is constructed using the Neo4j graph database, and the hierarchical relationships between standard clauses are parsed to perform semantic reasoning.

[0015] Furthermore, the calculation of standard clauses using semantic vectorization technology includes the following steps:

[0016] The standard clause text is converted into a vector representation using the Sentence-BERT model, and the generated vectors are stored in the FAISS vector database for similarity calculation.

[0017] Calculate the cosine similarity between standard terms and device descriptions, and optimize the accuracy of standard retrieval through vector similarity matching.

[0018] Furthermore, the standard and device compatibility matching calculation includes the following steps:

[0019] By using the FAISS indexing technique, an approximate nearest neighbor ANN search structure is constructed, and large-scale standard vector matching is performed.

[0020] By setting a similarity threshold, the system filters out the standard terms that best match the device description and returns the corresponding search results.

[0021] Furthermore, the evolutionary analysis of the standard clauses includes the following steps:

[0022] The textual changes of different versions of the standard clauses are calculated based on Levenshtein distance, and key change points are extracted based on TF-IDF.

[0023] We use an LSTM+Attention time series model to predict future standard change trends and provide suggestions for standard version upgrades.

[0024] Furthermore, the provision of device standard compatibility recommendations includes the following steps:

[0025] The Transformer algorithm is used to calculate the semantic similarity of different standard terms, and combined with the collaborative filtering algorithm, personalized recommendations are provided based on the user's historical query records.

[0026] The applicability score of the standard is calculated by combining equipment operation data.

[0027] The second aspect of this invention provides a standard semantic knowledge base retrieval system for power equipment, including a historical data analysis module for acquiring standard data of power equipment, parsing the text information of the standard data of power equipment, establishing semantic associations between standard clauses, equipment attributes and key indicators, and forming a knowledge graph of equipment-standards-indicators;

[0028] The standard evolution analysis module is used to calculate standard clauses through semantic vectorization technology and to perform standard and device compatibility matching calculations based on vector retrieval methods.

[0029] The compatibility recommendation module is used to perform evolutionary analysis on standard clauses based on historical standard data and deep learning models, and to provide compatibility recommendations for equipment standards.

[0030] A third aspect of the present invention provides a computer device, comprising:

[0031] Memory, transceiver, processor, and bus system;

[0032] The memory is used to store programs;

[0033] The processor is used to execute the program in the memory, including the standard semantic knowledge base retrieval method for power equipment described above.

[0034] The bus system is used to connect the memory and the processor to enable communication between the memory and the processor.

[0035] A fourth aspect of the present invention provides a readable storage medium storing computer-readable instructions, characterized in that, when executed by a processor, the computer-readable instructions implement the steps of the standard semantic knowledge base retrieval method for power equipment described above.

[0036] As can be seen from the above technical solutions, the present invention has the following advantages:

[0037] The present invention provides a standard semantic knowledge base retrieval method for power equipment, which collects standard data of power equipment, parses the text information, establishes semantic relationships between standard clauses, equipment attributes, and key indicators, and forms a knowledge graph of equipment-standards-indicators. This achieves efficient parsing and structured storage of standard data. Semantic vectorization technology is used to calculate standard clauses, and a vector retrieval method is used to perform standard-equipment compatibility matching calculations, which improves the efficiency and accuracy of standard retrieval, reduces manual intervention and error rate. Combining historical standard data and deep learning models, the method performs evolutionary analysis of standard clauses and provides equipment standard compatibility recommendations, helping equipment manufacturers adapt to standard changes in advance and reducing compliance risks. The present invention achieves better results in standard data parsing, standard compatibility matching, standard evolution prediction, and personalized recommendations.

[0038] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from an examination of the following, or may be learned from the practice of the invention. Attached Figure Description

[0039] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0040] The terms “first,” “second,” “third,” “fourth,” etc. (if present) 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. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “corresponding to,” and any variations thereof, are intended to cover 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.

[0041] Example 1

[0042] The implementation method in this embodiment can be implemented in a system, on a server, or on a terminal; no specific limitation is made. The method in this application will be described from the perspective of system implementation below. As shown in the figure, a standard semantic knowledge base retrieval method for power equipment includes the following steps:

[0043] Acquire power equipment standard data, parse the text information of the power equipment standard data, establish semantic relationships between standard clauses, equipment attributes and key indicators, and form a knowledge graph of equipment-standards-indicators;

[0044] The standard clauses are calculated using semantic vectorization technology, and the standard and device compatibility are matched and calculated based on vector retrieval methods.

[0045] Based on historical standard data and deep learning models, we conduct evolutionary analysis on standard clauses and provide recommendations on equipment standard compatibility.

[0046] To verify the beneficial effects of this invention, scientific demonstration was conducted through economic benefit calculations and simulation experiments.

[0047] First, national standards, enterprise standards, international standards, and equipment technical manuals related to power equipment were collected through API data interfaces, web crawling, and manual input. OCR technology was then used to parse the scanned standard documents. All text data was converted into structured text and stored hierarchically by chapter, clause, and sub-clause. Next, the BERT+CRF entity recognition method was used to automatically extract equipment names, technical parameters, and key indicators from the standard clauses and store them in the database. Then, a knowledge graph of equipment, standards, and indicators was constructed based on the Neo4j graph database to parse the hierarchical relationships between standard clauses and perform semantic reasoning to ensure the correct association of standard clauses. In the model calculation stage, the Sentence-BERT model was used to convert the standard clause text into vector representations. The generated vectors were stored in the FAISS vector database, vector matching calculations were performed, and cosine similarity was used to improve the accuracy of the standard representation. The accuracy of the retrieval was improved. Furthermore, FAISS indexing technology was employed, using an Approximate Nearest Neighbor (ANN) search structure to quickly match a large number of standard vectors. After setting a similarity threshold, the standard terms that best matched the device description were selected. Combining historical standard data with a deep learning model, the textual changes of different versions of standard terms were analyzed by calculating Levenshtein distance, and key change points were extracted based on TF-IDF technology. An LSTM+Attention time series model was used to predict future standard change trends and provide upgrade suggestions for standard versions. To provide personalized recommendations, a Transformer model was used to calculate the semantic similarity between different standard terms. Combined with a collaborative filtering algorithm, standard-fit recommendations were provided for devices based on user historical query records, and the recommendation weights were dynamically adjusted based on device operating data. See Table 1 for the experimental data recording and analysis.

[0048] Table 1 Experimental Data Recording Table

[0049]

[0050] The data in the experimental results table demonstrates the practical effectiveness of this invention in power equipment standard retrieval and compatibility recommendation. The table shows the accuracy performance after performing standard clause information extraction, vectorization calculation, similarity matching, and deep learning prediction using technologies such as BERT+CRF, Sentence-BERT, FAISS, and LSTM+Attention. Looking at the "BERT+CRF Extraction Accuracy (%)" column, the accuracy stabilizes between 85% and 93% with different standard clauses processed, proving that this entity recognition method can efficiently and accurately extract equipment names, technical parameters, and key indicators from standard clauses, ensuring data accuracy and completeness. In the "Sentence-BERT Vectorization Accuracy (%)" column, experimental data also shows that through vectorization, the vector representation accuracy of standard clauses is between 78% and 87%, improving the matching degree between standard clauses and verifying that the model can effectively convert text information into vector representations with high semantic expressive power. The data in the "FAISS Matching Accuracy (%)" column shows that through FAISS for standard clause similarity calculation, the matching accuracy is between 76% and 85%, indicating that FAISS's approximate nearest neighbor search structure can effectively improve the matching accuracy between standard clauses and device descriptions, thereby optimizing search results. The results in the "Deep Learning Prediction Accuracy (%)" column also show that through LSTM... The LSTM+Attention model performs time-series prediction of standard clauses, with a prediction accuracy consistently between 81% and 91%, indicating that the model can effectively predict the evolution trend of standard clauses and provide accurate predictions of future changes. In the "Adaptability Recommendation Matching Rate (%)" column, the accuracy of equipment standard adaptation recommendation is good, between 83% and 92%, indicating that the invention can provide accurate standard recommendations for equipment and can dynamically adjust according to the equipment's operating status. Finally, the data in the "Prediction Trend Accuracy (%)" column further validates the effectiveness of the deep learning model in standard change prediction, with an accuracy range of 79% to 89%, demonstrating the high performance of the LSTM+Attention model in time-series prediction. Efficiency: By comparing the shortcomings of existing technologies, this invention demonstrates innovation and advantages in the automatic extraction of standard clauses, semantic matching, adaptability recommendation, and standard evolution analysis. Traditional methods often suffer from low processing efficiency and low matching accuracy when faced with massive amounts of standard clauses. However, this invention improves the accuracy of retrieval and recommendation by combining deep learning models, graph databases, and vectorization technology. In particular, the standard clause association and semantic reasoning functions based on the Neo4j graph database make the adaptability retrieval of device standards more intelligent and accurate. The combination of FAISS index and LSTM+Attention model enables the matching and prediction of standard clauses to be performed quickly and accurately, improving work efficiency and accuracy.Finally, through the recommendation mechanism of Transformer and collaborative filtering algorithms, personalized device standard compatibility recommendations are provided to users, further improving the practicality and adaptability of this invention.

[0051] Example 2

[0052] The difference between this embodiment and Embodiment 1 is that the parsing of the text information of the power equipment standard data includes the following steps:

[0053] Data from national standards, enterprise standards, international standards, and equipment technical manuals can be obtained through API data interfaces, web crawling technology, or manual input, and unstructured data can be converted into structured text.

[0054] The scanned standard document is parsed using OCR technology, and formatted using regular expressions to store the text hierarchically according to chapters, clauses, and sub-clauses.

[0055] The semantic association involves using the BERT+CRF entity recognition method to automatically extract equipment names, technical parameters, and key indicators from standard clauses and store them in a database. A knowledge graph of equipment, standards, and indicators is constructed based on the Neo4j graph database, and the hierarchical relationships between standard clauses are parsed to perform semantic reasoning.

[0056] This step, by collecting standard data on power equipment and parsing text information, achieves the conversion from unstructured text to structured data. Specifically, it employs various methods, including API (Application Programming Interface), web scraping, and manual input, to obtain national standards, enterprise standards, international standards, and equipment technical manuals. This data is typically unstructured and needs to be parsed and converted into structured text. The introduction of OCR (Optical Character Recognition) technology, especially for processing scanned standard documents, automates this conversion process, reduces the need for manual intervention, and improves the efficiency and accuracy of data conversion. The use of regular expressions further optimizes text formatting, ensuring that the text is stored hierarchically according to chapters, clauses, and sub-clauses. This improves the operability and retrieval of subsequent data processing. BERT (Bidirectional Encoder-Representation Transformer) + CRF (Conditional Random Field) entity recognition technology is introduced to automatically extract equipment names, technical parameters, and key indicators from standard clauses, forming clear data relationships and storing this information in a database. Combined with the Neo4j graph database, a knowledge graph of equipment, standards, and indicators is established. This process not only transforms equipment standard data into operable information but also stores the hierarchical relationships between equipment standards through the graph database, supporting subsequent semantic reasoning. The construction of the knowledge graph improves the retrieval efficiency and intelligence level of equipment standard data, automatically associating different devices with applicable standards, thereby providing equipment manufacturers with more accurate and efficient standard query services.

[0057] Example 3

[0058] The difference between this embodiment and Embodiment 2 is that establishing the semantic association between standard clauses, device attributes, and key indicators includes the following steps:

[0059] Based on the BERT+CRF entity recognition method, the equipment name, technical parameters and key indicators in the standard clauses are automatically extracted and stored in the database;

[0060] A knowledge graph of devices, standards, and metrics is constructed using the Neo4j graph database, and the hierarchical relationships between standard clauses are parsed to perform semantic reasoning.

[0061] A preferred scheme for constructing a knowledge graph of equipment, standards, and indicators based on the Neo4j graph database specifically includes: acquiring equipment-related standard document data, including national standards, enterprise standards, international standards, and equipment technical manuals, using API data interfaces, web crawling technology, or manual input to acquire and convert them into structured text data; parsing the text information in the standard documents, using OCR technology to parse scanned documents, and using regular expressions to hierarchically store the standard clauses by chapter, clause, and sub-clause; performing entity recognition based on the BERT+CRF model, automatically extracting equipment names, technical parameters, and key indicators from the standard clauses, and storing them in the database; and using Neo4j graph data. The library constructs a knowledge graph of equipment, standards, and indicators, creating nodes and defining edges according to the relationships between equipment, standards, and indicators, generating a graph data model of equipment, standard clauses, and key indicators. It parses the hierarchical relationships between standard clauses, constructs a semantic reasoning model between standard clauses, and executes reasoning queries in the Neo4j graph database. It recursively analyzes the hierarchical structure and dependencies of standard clauses to obtain more accurate equipment standard compatibility assessment results, revealing the interdependencies and connections between standard clauses. Finally, it uses graph algorithms in the graph database to perform matching derivations between equipment and standard clauses, and utilizes the query language of the graph database to calculate and optimize the compatibility between equipment and standards.

[0062] Example 4

[0063] The difference between this embodiment and Embodiment 3 is that the calculation of the standard clauses using semantic vectorization technology includes the following steps:

[0064] The standard clause text is converted into a vector representation using the Sentence-BERT model, and the generated vectors are stored in the FAISS vector database for similarity calculation.

[0065] Calculate the cosine similarity between standard terms and device descriptions, and optimize the accuracy of standard retrieval through vector similarity matching.

[0066] Semantic vectorization technology was employed, using the Sentence-BERT model to transform standard clause text into vector representations. The generated vectors were then stored in the FAISS (Facebook AI Similarity Search) vector database. This utilized pre-trained models from natural language processing to convert text from traditional symbolic representations into high-dimensional vectors that machine learning models could process. This transformation preserved the implicit semantic information in the text, enabling machines to better understand the content. By calculating the cosine similarity between standard clauses and device descriptions, precise matching between the two was achieved. Cosine similarity is a common method for measuring the similarity between two vectors in a vector space, effectively quantifying the semantic relevance between standard clauses and device descriptions. In large-scale standard data, traditional text matching methods often suffer from low efficiency and poor accuracy; however, vectorization technology allows for efficient and precise matching within the vector space. Accurate matching is achieved, thereby improving the precision and speed of retrieval. The ANN (Approximate Nearest Neighbor) search structure of the FAISS vector database further accelerates the matching process between standard clauses and device descriptions. The introduction of FAISS technology enables the rapid identification of the most relevant standard clauses for the device, even in large amounts of standard data, reducing the time and resource consumption of the retrieval process in traditional methods. Through semantic vectorization and vector retrieval methods, the accuracy and efficiency of matching between standard clauses and device descriptions are improved, avoiding the inefficiency and mismatch problems in traditional matching methods. This effectively enhances the automation and intelligence of standard retrieval, enabling the rapid provision of relevant standard recommendations in large-scale data and improving the calculation accuracy of device standard compatibility.

[0067] Example 5

[0068] The difference between this embodiment and embodiment four is that the standard and device compatibility matching calculation includes the following steps:

[0069] By using the FAISS indexing technique, an approximate nearest neighbor ANN search structure is constructed, and large-scale standard vector matching is performed.

[0070] By setting a similarity threshold, the system filters out the standard terms that best match the device description and returns the corresponding search results.

[0071] The standard and device compatibility matching calculations include using FAISS indexing technology, constructing an approximate nearest neighbor ANN search structure, and performing fast matching of large-scale standard vectors.

[0072] A preferred approach for setting a similarity threshold specifically involves constructing an efficient Approximate Nearest Neighbor (ANN) search structure using FAISS technology to store and index vector representations of standard clauses. These vector representations are derived by transforming the standard clause text using semantic vectorization techniques. The transformation process includes vectorizing the standard clause text using a Sentence-BERT model to obtain a high-dimensional vector representation. The vector of each standard clause is stored in the FAISS index structure. FAISS supports efficient retrieval operations based on vector similarity, ensuring fast matching in large-scale datasets. During retrieval, the textual information of the device description is received and transformed using the same semantic vectorization technique to generate a semantic vector representation of the same dimension as the standard clause vectors. The device description vector is then matched with the standard clause vectors in the FAISS index. Specifically, FAISS performs approximate nearest neighbor retrieval based on a similarity metric (such as cosine similarity) in the vector space. The process involves selecting several standard clause vectors that are closest to the device description vector; determining a similarity threshold, which is used to filter out the standard clauses that best match the device description; the similarity threshold is determined as follows: by analyzing the similarity distribution of each retrieval based on the historical matching results of standard clauses and device descriptions, the average similarity and standard deviation between the standard clauses and the device description are calculated; a dynamic threshold is set, which is adjusted according to the similarity distribution. Under normal circumstances, the threshold is set as the average similarity plus a certain multiple of the standard deviation to ensure that the retrieval results have a high degree of matching; when the similarity value between the device description vector and the standard clause vector is greater than or equal to the dynamic threshold, the standard clause is considered to have a high degree of fit with the device description and can be returned as a retrieval result; based on the set similarity threshold, the standard clauses that best match the device description are filtered out, and the retrieval results of these clauses are returned to provide standard-device fit matching and optimize the accuracy and efficiency of standard retrieval.

[0073] Example 6

[0074] The difference between this embodiment and Embodiment 5 is that the evolutionary analysis of the standard clauses includes the following steps:

[0075] The textual changes of different versions of the standard clauses are calculated based on Levenshtein distance, and key change points are extracted based on TF-IDF.

[0076] We use an LSTM+Attention time series model to predict future standard change trends and provide suggestions for standard version upgrades.

[0077] A preferred scheme for calculating textual changes in different versions of standard clauses includes: calculating these changes by comparing each clause individually using the Levenshtein distance algorithm to identify types of changes such as insertion, deletion, and replacement of characters; extracting keywords and their weights from the standard clauses using the TF-IDF algorithm, and identifying key change points by comparing changes in keywords across different versions; ranking the extracted change points and evaluating their impact on the evolution of the standard clauses based on their significance and frequency to generate standard clause change trend data; performing time-series analysis of the evolution trend using an LSTM model, employing an attention mechanism to enhance the weights of important change points, and predicting possible future changes to the standard clauses based on historical data; and providing upgrade suggestions for the standard version based on the prediction results, along with relevant adaptation adjustment guidance.

[0078] We use an LSTM+Attention time series model to predict future standard change trends and provide suggestions for standard version upgrades.

[0079] A preferred approach for predicting future standard change trends includes: performing time-series processing on historical standard clause change data, converting textual changes in each version of the standard clause into time-series data; using an LSTM model to model this time-series data, capturing the long-term dependencies of historical standard clause changes; introducing an Attention mechanism on top of the LSTM model, assigning higher weights to key change points in the model, thereby focusing on the parts that have a significant impact on future standard change trends; after completing model training, using the trained LSTM+Attention model to predict future standard clause changes, outputting data on future standard change trends; combining the prediction results to generate an evolution scheme for future standard clauses, and providing version upgrade suggestions for standard clauses based on this scheme.

[0080] Example 7

[0081] The difference between this embodiment and embodiment six is ​​that the provision of device standard compatibility recommendations includes the following steps:

[0082] The Transformer algorithm is used to calculate the semantic similarity of different standard terms, and combined with the collaborative filtering algorithm, personalized recommendations are provided based on the user's historical query records.

[0083] The applicability score of the standard is calculated by combining equipment operation data.

[0084] It should be noted that providing device standard compatibility recommendations involves using Transformer to calculate the semantic similarity of different standard clauses and combining it with collaborative filtering algorithms to provide personalized recommendations based on the user's historical query records.

[0085] By combining equipment operation data, the applicability score of the standard is calculated, and the suitability of the recommended standard is improved by dynamically adjusting the recommendation weight.

[0086] A preferred approach to calculating the applicability score of a standard includes: collecting data during equipment operation, including operating status, historical maintenance records, and performance indicators; matching the equipment operating data with the technical requirements in the standard clauses to extract the standard clauses relevant to equipment operation; using data preprocessing techniques to clean and normalize the equipment operating data to ensure consistency and accuracy; calculating the equipment's adaptability score under each standard clause based on the applicable indicators in the standard clauses, with the score calculated based on the difference between the equipment's actual operating data and the standard requirements; weighting the calculation results by combining historical operating data and current operating status to comprehensively evaluate the equipment's adaptability to the standard clauses; outputting the equipment's applicability score under different standard clauses, and generating a recommended fit for the standard clauses based on the score results.

[0087] An optimal solution for generating recommended standard clauses based on scoring results specifically includes: calculating the suitability score between the device and each standard clause based on the device's operating data and historical standard clause matching records; analyzing the semantic similarity between the device description and the content of the standard clauses; using a Transformer model to represent the semantics of the standard clauses, calculating the semantic similarity between each standard clause and the device description, and generating a vector representation of the standard clauses; further optimizing the recommended standard clauses using a collaborative filtering algorithm based on the device's historical usage records and the standard usage of other similar devices; dynamically evaluating the device's suitability by combining the device's real-time operating data and the standard clause suitability score to determine the current optimal standard clause; providing personalized standard suitability recommendations for the device based on the suitability score and dynamic evaluation results, outputting suggested standard clauses, and adjusting the recommendations according to device needs and operating status.

[0088] This study combines historical standard data and deep learning models to analyze the evolution of standard clauses and provide recommendations for device standard compatibility. It calculates textual changes across different versions of standard clauses using Levenshtein distance (edit distance) and extracts key change points based on TF-IDF (Term Frequency-Inverse Document Frequency). This analytical method helps device manufacturers and standard setters promptly identify standard changes by comparing textual differences between different versions, effectively mitigating compliance risks arising from standard changes. Levenshtein distance, as a measure of text edit distance, quantifies the magnitude of changes between standard clauses, revealing the specific content of clause changes. Furthermore, by combining an LSTM+Attention time series model, it predicts future standard change trends. By processing time-series data, the memory network can capture long-term dependencies in standard clause changes, while the attention mechanism enhances the model's focus on key time nodes and changing trends. This deep learning model can predict future standard changes and provide upgrade suggestions for standard versions, helping equipment manufacturers adapt to standard changes in advance and reduce compliance issues caused by failure to update standards in a timely manner. On this basis, through the Transformer model and collaborative filtering algorithm, personalized standard recommendations are provided to users. The Transformer model ensures a high degree of matching between the recommended standards and the device description by calculating the semantic similarity of standard clauses. The collaborative filtering algorithm, combined with the user's historical query records, can provide personalized recommendations based on the user's needs, further improving the accuracy of the recommendation system.

[0089] Example 8

[0090] A standard semantic knowledge base retrieval system for power equipment includes a historical data analysis module for acquiring standard data of power equipment, parsing the text information of the standard data of power equipment, establishing semantic relationships between standard clauses, equipment attributes and key indicators, and forming a knowledge graph of equipment-standards-indicators;

[0091] The standard evolution analysis module is used to calculate standard clauses through semantic vectorization technology and to perform standard and device compatibility matching calculations based on vector retrieval methods.

[0092] The compatibility recommendation module is used to perform evolutionary analysis on standard clauses based on historical standard data and deep learning models, and to provide compatibility recommendations for equipment standards.

[0093] Example 9

[0094] A computer device, comprising:

[0095] Memory, transceiver, processor, and bus system;

[0096] The memory is used to store programs;

[0097] The processor is used to execute the program in the memory, including the standard semantic knowledge base retrieval method for power equipment described above.

[0098] The bus system is used to connect the memory and the processor to enable communication between the memory and the processor.

[0099] Example 10

[0100] A readable storage medium storing computer-readable instructions, characterized in that, when executed by a processor, the computer-readable instructions implement the steps of the standard semantic knowledge base retrieval method for power equipment described above.

[0101] In summary, the standard semantic knowledge base retrieval method for power equipment provided by this invention collects standard data of power equipment, parses the text information, establishes semantic relationships between standard clauses, equipment attributes, and key indicators, and forms a knowledge graph of equipment-standards-indicators. This achieves efficient parsing and structured storage of standard data. Semantic vectorization technology is used to calculate standard clauses, and a vector retrieval method is used to perform standard-equipment compatibility matching calculations, improving the efficiency and accuracy of standard retrieval, reducing manual intervention and error rates. Combining historical standard data with deep learning models, evolutionary analysis of standard clauses is performed, and equipment standard compatibility recommendations are provided, helping equipment manufacturers adapt to standard changes in advance and reducing compliance risks. This invention achieves better results in standard data parsing, standard compatibility matching, and standard evolution prediction and personalized recommendations.

[0102] It is understood that those skilled in the art can combine various implementation methods in the above embodiments under the guidance of the above examples to obtain technical solutions with multiple implementation methods.

[0103] Those skilled in the art will recognize that the units of the various examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of the invention.

[0104] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical functional division. In actual implementation, there may be other division methods, such as multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored.

[0105] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0106] 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 the present invention, 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 instructions 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 the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0107] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A standard semantic knowledge base retrieval method for power equipment, characterized in that, Includes the following steps: Acquire power equipment standard data, parse the text information of the power equipment standard data, establish semantic relationships between standard clauses, equipment attributes and key indicators, and form a knowledge graph of equipment-standards-indicators; The standard clauses are calculated using semantic vectorization technology, and the standard and device compatibility are matched and calculated based on vector retrieval methods. Based on historical standard data and deep learning models, we conduct evolutionary analysis on standard clauses and provide recommendations on equipment standard compatibility.

2. The standard semantic knowledge base retrieval method for power equipment according to claim 1, characterized in that, The parsing of text information from standard data for power equipment includes the following steps: Data from national standards, enterprise standards, international standards, and equipment technical manuals can be obtained through API data interfaces, web crawling technology, or manual input, and unstructured data can be converted into structured text. The scanned standard document is parsed using OCR technology, and formatted using regular expressions to store the text hierarchically according to chapters, clauses, and sub-clauses.

3. The standard semantic knowledge base retrieval method for power equipment according to claim 1, characterized in that, Establishing semantic relationships between standard clauses, equipment attributes, and key indicators includes the following steps: Based on the BERT+CRF entity recognition method, the equipment name, technical parameters and key indicators in the standard clauses are automatically extracted and stored in the database; A knowledge graph of devices, standards, and metrics is constructed using the Neo4j graph database, and the hierarchical relationships between standard clauses are parsed to perform semantic reasoning.

4. The standard semantic knowledge base retrieval method for power equipment according to claim 1, characterized in that, The calculation of standard clauses using semantic vectorization technology includes the following steps: The standard clause text is converted into a vector representation using the Sentence-BERT model, and the generated vectors are stored in the FAISS vector database for similarity calculation. Calculate the cosine similarity between standard terms and device descriptions, and optimize the accuracy of standard retrieval through vector similarity matching.

5. The standard semantic knowledge base retrieval method for power equipment according to claim 1, characterized in that, The standard and equipment compatibility matching calculation includes the following steps: By using the FAISS indexing technique, an approximate nearest neighbor ANN search structure is constructed, and large-scale standard vector matching is performed. By setting a similarity threshold, the system filters out the standard terms that best match the device description and returns the corresponding search results.

6. The standard semantic knowledge base retrieval method for power equipment according to claim 1, characterized in that, The evolutionary analysis of the standard clauses includes the following steps: The textual changes of different versions of the standard clauses are calculated based on Levenshtein distance, and key change points are extracted based on TF-IDF. We use an LSTM+Attention time series model to predict future standard change trends and provide suggestions for standard version upgrades.

7. The standard semantic knowledge base retrieval method for power equipment according to claim 1, characterized in that, The provision of device standard compatibility recommendations includes the following steps: The Transformer algorithm is used to calculate the semantic similarity of different standard terms, and combined with the collaborative filtering algorithm, personalized recommendations are provided based on the user's historical query records. The applicability score of the standard is calculated by combining equipment operation data.

8. A standard semantic knowledge base retrieval system for power equipment, characterized in that, It includes a historical data analysis module, which is used to acquire power equipment standard data, parse the text information of power equipment standard data, establish semantic relationships between standard clauses, equipment attributes and key indicators, and form a knowledge graph of equipment-standards-indicators; The standard evolution analysis module is used to calculate standard clauses through semantic vectorization technology and to perform standard and device compatibility matching calculations based on vector retrieval methods. The compatibility recommendation module is used to perform evolutionary analysis on standard clauses based on historical standard data and deep learning models, and to provide compatibility recommendations for equipment standards.

9. A computer device, characterized in that, include: Memory, transceiver, processor, and bus system; The memory is used to store programs; The processor is configured to execute a program in the memory, including executing a standard semantic knowledge base retrieval method for power equipment as described in any one of claims 1 to 7; The bus system is used to connect the memory and the processor to enable communication between the memory and the processor.

10. A readable storage medium storing computer-readable instructions, characterized in that, When the computer-readable instructions are executed by the processor, they implement the steps of the standard semantic knowledge base retrieval method for power equipment as described in any one of claims 1 to 7.

Images