A power equipment whole-process data service layered classification sharing method and system based on a data middle platform

Abstract

The application discloses a power equipment whole-process data service layered classification sharing method and system based on a data middle station, which comprises the following steps: extracting and quantifying equipment fine semantic similarity, whole life cycle scene overlapping degree and power grid hierarchical professional distance constraint coefficient; iteratively training a special logistic regression model by using real ambiguous coding samples to obtain optimal feature weights and bias values, and generating a coding correlation probability formula; and finally outputting a device unified semantic view according to the probability to realize whole-process data semantic connection. Compared with the short board of the prior art, which depends on general matching, has no power-specific features, is prone to false or missed matching, and only realizes physical aggregation, the application creatively combines three-dimensional fusion features with a special probability model, adapts to power semantic ambiguous scenes, greatly improves the correlation accuracy, does not need manual maintenance, and realizes intelligent matching of ambiguous codes.

Description

Technical Field

[0001] This application relates to the field of power grid data processing, specifically to a hierarchical classification and sharing method and system for power equipment full-process data services based on a data middle platform. Background Technology

[0002] The entire lifecycle data of core primary equipment in the power grid (such as main transformers, circuit breakers, and reactors) is scattered across multiple departments and heterogeneous business systems, including the ECP material system, e-infrastructure system, PMS operation and maintenance system, and online power grid planning system, totaling 16 source systems. Because the coding systems, parameter naming rules, and semantic definitions of each system are independent, semantic ambiguity issues such as synonyms, homonyms, and inconsistent parameter definitions are prevalent. This leads to difficulties in cross-system coding association of equipment, severe data silos, and the inability to achieve end-to-end data connectivity and accurate sharing. This hinders the implementation of advanced business applications such as equipment health assessment, family defect investigation, and accurate material forecasting.

[0003] In existing technologies, the power industry typically employs enterprise-level data platforms to achieve equipment data aggregation and sharing. A typical solution is exemplified by patent document CN112163039A, titled "Data Resource Standardization Management System Based on Enterprise-Level Data Platform Analysis Domain," which discloses a data resource standardization management system based on an enterprise-level data platform analysis domain. This system adopts a "two-zone, four-layer" architecture, achieving data aggregation from multiple business systems through a data synchronization and replication component, and outputting data externally via standardized interfaces, thus realizing cross-domain data sharing to a certain extent.

[0004] However, existing technologies still have the following key shortcomings:

[0005] The data matching method is crude and cannot adapt to the semantic ambiguity scenarios of power equipment coding. Existing technologies only use general keywords and fixed rule matching methods to achieve data association, without designing an adaptation mechanism for the specific ambiguity characteristics of synonyms and synonyms of power grid equipment parameters. It lacks power professional scenario constraints and domain prior knowledge, resulting in a high rate of mismatch and missed match in cross-system coding, and failing to achieve semantic-level data association.

[0006] The lack of a power-specific multi-dimensional feature fusion and matching mechanism results in persistent cross-disciplinary data semantic silos. Existing technologies have not built a feature system adapted to power grid operations, have not distinguished the close and distant relationships between power disciplines, and cannot accurately identify the potential relationships between multi-source heterogeneous codes of equipment. They can only achieve physical data aggregation, but cannot achieve semantic integration, making it difficult to fundamentally solve the problem of data breakpoints throughout the entire equipment process.

[0007] In summary, existing power data sharing technologies are ill-suited to address industry pain points such as semantic ambiguity in cross-disciplinary coding of power grid equipment and difficulty in accurately associating heterogeneous codes. There is an urgent need for a targeted technical solution to achieve semantic-level connectivity and hierarchical classification for accurate sharing of equipment data throughout the entire process. Summary of the Invention

[0008] In view of the above problems, this application provides a hierarchical classification and sharing method and system for power equipment full-process data services based on a data middle platform, which is used to realize the connection and accurate sharing of full-process data of core primary equipment of the power grid.

[0009] To achieve the above objectives, the inventors provide a hierarchical and categorized sharing method for power equipment full-process data services based on a data platform. This method is used to process multi-source heterogeneous coded data of core primary equipment in the power grid, realizing data connectivity and hierarchical and categorized sharing throughout the equipment's entire lifecycle. The method includes the following steps:

[0010] S1: Extract device-specific refined semantic similarity from the multi-source heterogeneous encoded data. Overlap of scenarios in the closed-loop lifecycle of equipment , power grid hierarchical professional distance constraint coefficient Three core, multi-dimensional related features were identified and quantized and vectorized.

[0011] S2: Training and constructing the association function for a dedicated logistic regression probability model for power equipment. Real ambiguous coding samples from the entire lifecycle of core primary equipment in the power grid are collected as the training set. Using the three core association feature vectors as input and whether the coding is associated as a binary classification label, the dedicated logistic regression probability model is iteratively trained. The optimal feature weights w1, w2, and w3, and the bias value b, adapted to the ambiguous scenarios of power grid equipment, are obtained through convergence and optimization. Substituting these optimal parameters into the standard form of logistic regression, the formula for calculating the association probability of equipment coding is obtained:

[0012] ;

[0013] S3: Construction of a unified semantic view for equipment: Based on the coding association probability output by the calculation formula of the equipment coding association probability, a unified coding semantic view for the entire life cycle of the core primary equipment of the power grid is generated to realize the semantic-level connection of equipment data throughout the entire process.

[0014] Furthermore, it also includes verification and iterative optimization steps:

[0015] Key indicators for statistical coding matching include accuracy, false matching rate, and missed matching rate.

[0016] If the matching accuracy is lower than the preset threshold, the ambiguous coding pairs are re-extracted based on the verification samples to supplement the training set, the dedicated logistic regression probability model is iteratively retrained, and the optimal feature weights w1, w2, w3 and bias value b are updated.

[0017] Based on the verification scene matching effect, the initial value and fine-tuning step size of the dynamic association probability threshold are fine-tuned for each scene, and the optimal matching parameter combination is solidified.

[0018] Furthermore, in step S1, the equipment-specific refined semantic similarity extraction includes: limiting five categories of core power-specific parameter fields—equipment type, voltage level, rated electrical parameters, structural attributes, and power industry standard definitions—as the sole basis for semantic matching, filtering out invalid and interfering fields; and using a Word2Vec model finely tuned and optimized from the power equipment professional corpus to generate a specific semantic vector, calculating and outputting a high-precision semantic similarity coefficient. .

[0019] Furthermore, step S1, the extraction of the overlap of closed-loop scenarios throughout the equipment lifecycle, includes: selecting six types of core business documents—equipment purchase orders, factory test reports, handover test records, maintenance and repair work orders, defect ledgers, and project acceptance documents—to construct a set of core business scenarios corresponding to each code; calculating the overlap of scenario sets corresponding to different codes using the Jaccard coefficient, and outputting the scenario overlap. .

[0020] Furthermore, in step S1, the assignment of professional distance constraint coefficients for power grid hierarchical division includes: setting the distance coefficient for equipment-operation and maintenance to 0.9, the distance coefficient for materials-procurement to 0.8, and the distance coefficient for planning-infrastructure to 0.7; forcibly setting the distance coefficients for non-equipment-related professional fields such as finance and human resources to 0; and dynamically assigning professional distance constraint coefficients based on the division of professional domains according to the multi-identifier linkage system. .

[0021] Furthermore, the model training samples in step S2 include: more than 10,000 manually verified cross-system valid association coding pairs as positive samples, and more than 5,000 power equipment-independent heterogeneous coding pairs as negative samples.

[0022] A hierarchical and categorized data sharing system for the entire process of power equipment, based on a data middle platform, is used to process multi-source heterogeneous coded data of core primary equipment in the power grid, realizing data connectivity and hierarchical and categorized sharing throughout the entire lifecycle of the equipment, including:

[0023] Data extraction and quantization module: used to extract device-specific refined semantic similarity from the multi-source heterogeneous encoded data. Overlap of scenarios in the closed-loop lifecycle of equipment , power grid hierarchical professional distance constraint coefficient Three core, multi-dimensional related features were identified and quantized and vectorized.

[0024] Modeling and Training Module: Used for training a dedicated logistic regression probability model and constructing association functions for power equipment. It collects real ambiguous coding samples from the entire lifecycle of core primary equipment in the power grid as the training set. Using the three core association feature vectors as input and whether the coding is associated as a binary classification label, it iteratively trains the dedicated logistic regression probability model. The model converges and optimizes to obtain the optimal feature weights w1, w2, w3, and bias value b suitable for ambiguous scenarios involving power grid equipment. Substituting these optimal parameters into the standard form of logistic regression, the formula for calculating the association probability of equipment coding is obtained.

[0025] ;

[0026] Execution module: Used for constructing a device-specific unified semantic view. Based on the coding association probability output by the device coding association probability calculation formula, it generates a device-specific unified coding semantic view for the entire life cycle of the core primary equipment of the power grid, realizing semantic-level connectivity of data throughout the entire process of the device.

[0027] Unlike existing technologies, the above technical solution targets multi-source heterogeneous coding data of core primary equipment in the power grid that spans systems and disciplines. First, it extracts three core association features: equipment-specific refined semantic similarity, equipment lifecycle closed-loop scenario overlap, and power grid hierarchical professional distance constraint coefficient, and completes quantification and vectorization processing. Then, it collects real ambiguous coding samples from the entire lifecycle of core primary equipment in the power grid and iteratively trains a dedicated logistic regression probability model. The model converges and optimizes to obtain the optimal feature weights w1, w2, w3 and bias value b, and substitutes them into the standard form of logistic regression to obtain the equipment coding association probability calculation formula. Finally, based on the coding association probability output by the formula, it generates a unified semantic view specific to the equipment, realizing semantic-level connectivity of equipment data throughout the entire process. Compared with existing technologies that only use general keywords, fixed rule matching, lack power-specific feature fusion mechanisms, suffer from serious coding mismatches and missed matches, and can only physically aggregate without semantic integration, this invention, through its original three-dimensional fusion feature system for power equipment, dedicated ambiguous sample training probability model, and quantifiable association probability calculation, accurately adapts to semantic ambiguity scenarios in power coding, greatly improves coding association accuracy, eliminates the need for manual maintenance of matching rules, achieves adaptive intelligent matching of ambiguous codes, completely breaks down cross-professional data semantic silos, and fills the gap in accurate association technology for heterogeneous codes of power equipment.

[0028] The above description of the invention is merely an overview of the technical solution of this application. In order to enable those skilled in the art to better understand the technical solution of this application and to implement it based on the description and drawings, and to make the above-mentioned objectives and other objectives, features and advantages of this application easier to understand, the following description is provided in conjunction with the specific embodiments and drawings of this application. Attached Figure Description

[0029] The accompanying drawings are only used to illustrate the principles, implementation methods, applications, features, and effects of specific embodiments of the present invention and other related contents, and should not be considered as limitations on this application.

[0030] In the accompanying drawings of the instruction manual:

[0031] Figure 1 A flowchart illustrating the hierarchical and categorized sharing method for full-process data services of power equipment based on a data middle platform, as described in the specific implementation method;

[0032] Figure 2 This is a flowchart illustrating the verification and iterative optimization steps described in the specific implementation method;

[0033] Figure 3 This is a block diagram of the hierarchical and classified sharing system for full-process data services of power equipment based on a data middle platform, as described in the specific implementation method.

[0034] The reference numerals used in the above figures are explained as follows:

[0035] 100. A hierarchical and categorized data sharing system for the entire process of power equipment based on a data platform; 101. Data extraction and quantification module; 102. Modeling and training module; 103. Execution module; 104. Verification and iterative optimization module; Detailed Implementation

[0036] To illustrate the possible application scenarios, technical principles, implementable specific solutions, and achievable objectives and effects of this application in detail, the following description, in conjunction with the listed specific embodiments and accompanying drawings, provides a detailed explanation. The embodiments described herein are merely illustrative of the technical solutions of this application and are therefore intended to limit the scope of protection of this application.

[0037] In this document, the term "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The term "embodiment" appearing in various places throughout the specification does not necessarily refer to the same embodiment, nor does it specifically limit its independence or connection with other embodiments. In principle, in this application, as long as there are no technical contradictions or conflicts, the technical features mentioned in each embodiment can be combined in any way to form corresponding implementable technical solutions.

[0038] Unless otherwise defined, the technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the use of related terms herein is merely for the purpose of describing particular embodiments and is not intended to limit this application.

[0039] In the description of this application, the term "and / or" is used to describe the logical relationship between objects, indicating that three relationships can exist. For example, A and / or B means: A exists, B exists, and A and B exist simultaneously. Additionally, the character " / " in this document generally indicates that the preceding and following objects have an "or" logical relationship.

[0040] In this application, terms such as “first” and “second” are used only to distinguish one entity or operation from another, and do not necessarily require or imply any actual quantity, hierarchy or order relationship between these entities or operations.

[0041] Without further limitations, the use of terms such as “comprising,” “including,” “having,” or other similar open-ended expressions in this application is intended to cover non-exclusive inclusion, which does not exclude the presence of additional elements in a process, method, or product that includes the stated elements, such that a process, method, or product that includes a list of elements may include not only those defined elements but also other elements not expressly listed, or elements inherent to such a process, method, or product.

[0042] In this application, expressions such as "greater than", "less than", and "exceeding" are understood to exclude the stated number; expressions such as "above", "below", and "within" are understood to include the stated number. Furthermore, in the description of the embodiments of this application, "multiple" means two or more (including two), and similar expressions related to "multiple" are also understood in this way, such as "multiple groups" and "multiple times", unless otherwise explicitly specified.

[0043] In the description of the embodiments of this application, the space-related expressions used, such as "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "vertical," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential," indicate the orientation or positional relationship based on the orientation or positional relationship shown in the specific embodiments or drawings. They are only for the purpose of describing the specific embodiments of this application or for the reader's understanding, and do not indicate or imply that the device or component referred to must have a specific position, a specific orientation, or be constructed or operated in a specific orientation. Therefore, they should not be construed as limitations on the embodiments of this application.

[0044] The processor described in the embodiments of this application can be implemented by hardware, firmware, software, or a combination thereof. It can be a circuit, one or more of an application-specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field-programmable gate array (FPGA), a central processing unit (CPU), a controller, a microcontroller, or a microprocessor. It also includes other physical, biological, or chemical structures that can implement the same or equivalent functions as the processors listed above, such as biological neurons, quantum computing units, DNA computing units, etc., so that the processor can execute some or all of the steps in the computer program or method involved in the various embodiments of this application, or any combination of the steps mentioned therein.

[0045] The computer program involved in the embodiments can be stored in a computer device readable storage medium, which includes, but is not limited to, disks, magnetic tapes, magnetic cards, floppy disks, flash memory, optical disks, optical cards, read-only memory (ROM), random access memory (RAM), erasable programmable ROM (EPROM), and electrically erasable programmable ROM (EEPROM), etc., and also includes other biological, physical, or chemical structures that can achieve the same or equivalent functions as the storage media listed above, such as DNA, RNA, proteins, and other units with information storage capabilities. In specific embodiments, the storage medium involved can be one of the above-mentioned media types, or a combination of the above-mentioned media types. In different embodiments, the computer program involved in the embodiments can be centrally stored in a single medium, or distributed and stored in multiple media. The memory containing the computer device readable storage medium can be non-volatile memory or random access memory. These computer device readable storage media can be built into the device, or can be connected to the device involved in the embodiments as an external device or part of an external device. In some embodiments, the memory having a computer device readable storage medium is deployed locally; in other embodiments, the memory may be deployed remotely from the processor, for example, as a network-attached memory accessed via RF circuitry or an external port and a communication network, wherein the communication network may be the Internet, one or more intranets, a local area network (LAN), a wide area network (WLAN), a storage area network (SAN), or a suitable combination thereof, as long as computer device access to the memory is enabled. Furthermore, the computer program involved in the embodiments may be stored in plaintext / ciphertext form, or it may be designed as training data, integrated and recombined through model training and implicitly stored in the parameter states of a deep neural network or other machine learning model.

[0046] Please see Figure 1 This embodiment provides a hierarchical and categorized sharing method for full-process data services of power equipment based on a data middleware platform. This embodiment can be applied to the full life-cycle data processing scenario of core primary equipment in the power grid, targeting six major business departments of the power grid and involving sixteen source-end business systems, specifically including: ECP material system, e-infrastructure system, PMS operation and maintenance system, online power grid planning system, etc., and covering core primary equipment of the power grid such as main transformers, circuit breakers, and reactors.

[0047] In related technologies, the main approach is to use general keywords and fixed rules for matching. This approach cannot adapt to the specific scenarios of semantic ambiguity, synonymy, and homonymy in cross-system coding of core primary equipment of the power grid. As a result, the mismatch and missed match rate of cross-system coding of core primary equipment of the power grid is high. It is impossible to accurately identify the potential associations of multi-source heterogeneous coding of core primary equipment of the power grid. It can only achieve physical data aggregation, but cannot complete semantic connection. The problem of data breakpoints throughout the entire process of core primary equipment of the power grid cannot be solved at the root.

[0048] This embodiment of the hierarchical classification and sharing method for full-process data services of power equipment based on a data platform constructs a dedicated three-dimensional fusion feature system for core primary equipment of the power grid. It extracts three core association features: refined semantic similarity specific to core primary equipment, overlap of closed-loop scenarios throughout the entire lifecycle of core primary equipment, and distance constraint coefficients of power grid hierarchical professional systems, and completes quantification and vectorization processing. It collects real ambiguous coding samples from the entire lifecycle of core primary equipment, iteratively trains a dedicated logistic regression probability model, converges and optimizes to obtain the optimal feature weights w1, w2, w3 and bias value b, and substitutes them into the standard form of logistic regression to obtain the equipment coding association probability calculation formula. Based on the coding association probability output by the formula, a dedicated unified semantic view for core primary equipment of the power grid is generated. On this basis, practical verification and iterative optimization are carried out, the feature extraction process is refined, and adaptive dynamic threshold matching is configured to achieve semantic-level connectivity and hierarchical classification and accurate sharing of data throughout the entire process of core primary equipment of the power grid. This significantly improves the coding association accuracy of core primary equipment of the power grid, eliminates the need for manual maintenance of matching rules, achieves adaptive intelligent matching of ambiguous codes, completely breaks down cross-professional data semantic silos, and fills the technical gap in accurate association of heterogeneous codes of core primary equipment of the power grid in the power industry.

[0049] like Figure 1 As shown, the hierarchical and categorized data sharing method for the entire process of power equipment based on a data platform in this embodiment includes the following steps:

[0050] Step S1: Multi-dimensional correlation feature extraction and vectorization processing.

[0051] For the coding of core primary equipment in the power grid that spans multiple systems and disciplines, three core association features are extracted: refined semantic similarity specific to core primary equipment in the power grid, overlap of closed-loop scenarios throughout the entire life cycle of core primary equipment in the power grid, and distance constraint coefficient of power grid hierarchical disciplines. These features are then quantified and vectorized.

[0052] The refined semantic similarity of core primary equipment in the power grid is determined by using five categories of core power-specific parameters—equipment type, voltage level, rated electrical parameters, structural attributes, and power industry standard definitions—as the sole basis for semantic matching. Invalid and interfering fields such as project names and construction units are filtered out. A Word2Vec model, finely tuned and optimized from a power equipment professional corpus, is used to generate exclusive semantic vectors, and the similarity coefficient is calculated. .

[0053] Overlap of Closed-Loop Scenarios Throughout the Core Primary Equipment Lifecycle of the Power Grid: Six types of core business documents are selected, including equipment purchase orders, factory test reports, handover test records, operation and maintenance work orders, defect ledgers, and project acceptance documents. A set of core business scenarios corresponding to each code is constructed. The overlap of scenario sets corresponding to different codes is calculated using the Jaccard coefficient to obtain the degree of scenario overlap. .

[0054] Distance constraint coefficients for different professional categories within the power grid: These coefficients are set based on the correlation characteristics of power grid business domains. The distance coefficient for equipment-maintenance is 0.9, for materials-procurement is 0.8, and for planning-infrastructure is 0.7. The distance coefficients for non-equipment-related professional categories such as finance and human resources are forcibly set to 0. Professional domains are divided according to a multi-identifier linkage system, and distance weights are dynamically assigned. .

[0055] Step S2: Training the logistic regression probability model specific to power equipment and constructing the correlation function.

[0056] Over 10,000 manually verified cross-system valid association code pairs were collected as positive samples, and over 5,000 power equipment-independent heterogeneous code pairs were collected as negative samples to form a training set of real ambiguous codes throughout the entire lifecycle of core primary equipment in the power grid. Using three core association feature vectors as input and whether the codes are associated as binary classification labels, a dedicated logistic regression probability model was iteratively trained. The optimal feature weights w1, w2, w3, and bias value b, adapted to the ambiguous scenarios of power grid equipment, were obtained through convergence optimization. Substituting the optimal feature weights w1, w2, w3, and bias value b into the standard form of logistic regression, the formula for calculating the association probability of equipment codes was obtained.

[0057] ;

[0058] in, Refined semantic similarity specifically for core primary equipment in the power grid. The overlap of closed-loop scenarios throughout the entire lifecycle of core primary equipment in the power grid. This is the professional distance constraint coefficient for power grid hierarchical classification. The calculation logic of this formula is as follows: First, the features of the three dimensions (semantic similarity, scene consistency, and file attribute similarity) are linearly combined with the bias value through weighted summation to obtain a comprehensive feature score; then, the linear score is mapped to the [0,1] interval through the Sigmoid (logistic regression) function, and the output result is the association probability of two sets of codes for the same device. The weights w1, w2, and w3 are obtained through training with real ambiguous samples of power equipment, which can adaptively adjust the contribution of each dimension of features to the association judgment. The bias value b is used to correct the offset of the overall probability, and finally realizes an accurate and quantitative evaluation of the cross-system heterogeneous coding association relationship of power grid equipment.

[0059] In this implementation, instead of using a general public dataset to train the model, we specifically collect real ambiguous coding pairs from the entire lifecycle of core primary equipment in the power grid (main transformers, circuit breakers, reactors, etc.) as training samples. We select 10,000+ manually verified cross-system valid association coding pairs as positive samples and 5,000+ power equipment-independent heterogeneous coding pairs as negative samples. Using the unique three-dimensional exclusive feature vector of this invention as the sole input and whether the coding is associated as a binary classification label, we iteratively train an exclusive power coding logistic regression classifier. The optimal feature weights w1, w2, w3 and bias value b are obtained by convergence and optimization to adapt to the ambiguous scenarios of power grid equipment. Among them: w1: semantic similarity feature weight, used to control the degree of influence of the semantic matching degree of the coded text on the association probability; w2: scenario consistency feature weight, used to control the degree of influence of the matching degree of the coding business scenario on the association probability; w3: file attribute similarity feature weight, used to control the degree of influence of the consistency of equipment file attributes on the association probability.

[0060] Step S3: Construct a unified semantic view for core primary equipment of the power grid.

[0061] Based on the coding association probability output by the equipment coding association probability calculation formula, a unique unified coding semantic view covering only the entire life cycle of core primary equipment of the power grid is generated, realizing semantic-level connectivity of data of core primary equipment of the power grid throughout the entire process.

[0062] The data platform is an enterprise-level core data platform that aggregates and integrates data from various professional source-end business systems of the power grid, undertaking data standardization processing, integrated aggregation, and shared service output. It serves as the core carrier for data processing, storage, and sharing in this invention. The entire process / lifecycle of equipment covers the complete business cycle of core primary equipment such as power grid transformers, circuit breakers, and reactors, from planning and design, material procurement, engineering construction, operation and maintenance to decommissioning and scrapping, encompassing 5 major business domains, 31 business processes, and 65 business links. The proprietary logistic regression probability model is uniquely designed for multi-source heterogeneous coding ambiguity scenarios of core primary equipment in the power grid. It achieves ruleless, adaptive, and low-mismatch cross-system intelligent coding association through three-dimensional fusion of power equipment-specific refined semantic features, equipment lifecycle closed-loop scenario features, and power grid hierarchical professional distance constraint features, combined with a probability algorithm trained on proprietary samples. This is a unique and innovative technology in the industry.

[0063] This embodiment abandons the existing technology's general text matching coarse mode, indiscriminate scene matching method, and unconstrained matching logic, and creates a unique three-dimensional fusion feature system dedicated to power equipment: it limits feature extraction to core power-specific parameter fields and core business documents, and proposes for the first time a power grid hierarchical professional distance constraint mechanism. It uses real ambiguous coding samples from core primary equipment of the power grid to train a dedicated logistic regression probability model, replacing traditional fixed rules with quantifiable association probabilities, thus thoroughly solving the problem of ambiguous matching of core primary equipment coding from the feature level, algorithm level, and constraint level.

[0064] In this embodiment, the hierarchical classification and sharing method for full-process data services of power equipment based on a data platform comprehensively characterizes the correlation probability of cross-system codes of core primary equipment in the power grid from three core correlation features: semantic dimension, scenario dimension, and professional constraint dimension. Then, it learns the correlation patterns of real ambiguous codes of core primary equipment in the power grid through a dedicated logistic regression probability model, and outputs accurate and quantifiable correlation probabilities. Finally, it constructs a unified semantic view based on high-probability correlation mapping relationships, breaking down semantic barriers between cross-system codes.

[0065] Compared to existing technologies that rely solely on general keywords and fixed rule matching, lack a power-specific feature fusion mechanism, suffer from severe coding mismatches and missed matches, and can only achieve physical aggregation without semantic integration, this embodiment, through its unique three-dimensional fusion feature system, dedicated ambiguous sample training probability model, and quantifiable association probability calculation, accurately adapts to semantic ambiguity scenarios in power coding. The accuracy of coding association for core primary equipment in the power grid is significantly improved. It eliminates the need for manual maintenance of matching rules, achieves adaptive intelligent matching of ambiguous codes, completely breaks down cross-professional data semantic silos, and fills the gap in accurate association technology for heterogeneous coding of core primary equipment in the power grid.

[0066] like Figure 2As shown in this embodiment, the hierarchical classification and sharing method for power equipment full-process data services based on a data platform further includes verification and iterative optimization steps:

[0067] S41: Practical Verification: Select preset data sharing scenarios for core primary equipment of the power grid, such as equipment ledger filing, test data association, family defect investigation, material demand forecasting, and project performance management, to conduct practical verification and statistically analyze the core indicators of code matching accuracy, mismatch rate, and missed match rate of core primary equipment of the power grid.

[0068] S42: Model Iteration and Retraining: If the coding accuracy of the core primary equipment of the power grid is lower than the preset threshold (e.g., 95%), the ambiguous coding pairs of the core primary equipment of the power grid are re-extracted based on the verification samples to supplement the training set, the dedicated logistic regression probability model is iteratively retrained, and the optimal feature weights w1, w2, w3 and bias value b are updated.

[0069] S43: Threshold parameter solidification: Based on the verification scene matching effect, fine-tune the initial value and fine-tuning step size of the dynamic association probability threshold for each scene, and solidify the optimal matching parameter combination.

[0070] In this embodiment, adaptive threshold matching and device-specific unified semantic view construction are adopted. First, a scenario-specific initial association probability threshold is preset for power equipment (e.g., the scenario-specific initial association probability threshold is 0.7). Based on the differences in the type, voltage level, and professional domain of the core primary equipment of the power grid, the threshold is fine-tuned by 0.02 to 0.05 according to the scenario category. For scenarios with low matching confidence, the threshold is automatically raised and for scenarios with high ambiguity matching, the threshold is automatically lowered. High-probability effective association code pairs are dynamically selected, and low-probability invalid matching data is filtered out to complete adaptive threshold matching.

[0071] This invention abandons the general fixed threshold matching method. Based on the ambiguous characteristics of power grid equipment data, it presets a dynamic and adjustable association probability threshold of 0.7 specific to power equipment scenarios. It then performs batch intelligent filtering and matching of heterogeneous codes across systems for core primary equipment in the entire network, retaining only high-probability valid association mappings and filtering low-probability invalid matching data. Ultimately, it generates a dedicated unified coding semantic view covering only the entire lifecycle of core primary equipment in the power grid, accurately solving the problem of semantic silos in equipment data. This provides unique and accurate underlying semantic support for the seamless integration of equipment data throughout its entire lifecycle, eliminating the problem of global data redundancy.

[0072] like Figure 3As shown, in another embodiment, a hierarchical classification and sharing system 100 for power equipment full-process data services based on a data platform is provided. This system is used to process multi-source heterogeneous coded data of core primary equipment in the power grid, and to realize the interconnection and hierarchical classification and sharing of equipment data throughout its entire life cycle. The power equipment full-process data service hierarchical classification and sharing system 100 based on a data platform includes: a data extraction and quantification module 101, a modeling and training module 102, and an execution module 103.

[0073] Data extraction and quantization module 101: used to extract device-specific refined semantic similarity from the multi-source heterogeneous encoded data. Overlap of scenarios in the closed-loop lifecycle of equipment , power grid hierarchical professional distance constraint coefficient Three core, multi-dimensional related features were identified and quantized and vectorized.

[0074] Modeling and Training Module 102: Used for training a dedicated logistic regression probability model and constructing association functions for power equipment. It collects real ambiguous coding samples from the entire lifecycle of core primary equipment in the power grid as the training set. Using the three core association feature vectors as input and whether the coding is associated as a binary classification label, it iteratively trains the dedicated logistic regression probability model. The model converges and optimizes to obtain the optimal feature weights w1, w2, w3, and bias value b suitable for ambiguous scenarios involving power grid equipment. Substituting these optimal parameters into the standard form of logistic regression, the formula for calculating the association probability of equipment coding is obtained.

[0075] ;

[0076] in, Refined semantic similarity specifically for core primary equipment in the power grid. The overlap of closed-loop scenarios throughout the entire lifecycle of core primary equipment in the power grid. This is the professional distance constraint coefficient for power grid hierarchical classification. The calculation logic of this formula is as follows: First, the features of the three dimensions (semantic similarity, scene consistency, and file attribute similarity) are linearly combined with the bias value through weighted summation to obtain a comprehensive feature score; then, the linear score is mapped to the [0,1] interval through the Sigmoid (logistic regression) function, and the output result is the association probability of two sets of codes for the same device. The weights w1, w2, and w3 are obtained through training with real ambiguous samples of power equipment, which can adaptively adjust the contribution of each dimension of features to the association judgment. The bias value b is used to correct the offset of the overall probability, and finally realizes an accurate and quantitative evaluation of the cross-system heterogeneous coding association relationship of power grid equipment.

[0077] Execution module 103: Used for constructing a device-specific unified semantic view. Based on the coding association probability output by the device coding association probability calculation formula, it generates a device-specific unified coding semantic view for the entire life cycle of the core primary equipment of the power grid, realizing semantic-level connectivity of the entire process data of the equipment.

[0078] The equipment-specific refined semantic similarity extraction includes: using five categories of core power-specific parameter fields—equipment type, voltage level, rated electrical parameters, structural attributes, and power industry standard definitions—as the sole basis for semantic matching, filtering out invalid and interfering fields; and generating exclusive semantic vectors using a Word2Vec model finely tuned and optimized from a power equipment professional corpus, calculating and outputting a high-precision semantic similarity coefficient. .

[0079] The equipment lifecycle closed-loop scenario overlap extraction includes: selecting six types of core business documents, namely equipment purchase orders, factory test reports, handover test records, operation and maintenance work orders, defect ledgers, and project acceptance documents, to construct a set of core business scenarios corresponding to each code; calculating the overlap of scenario sets corresponding to different codes using the Jaccard coefficient, and outputting the scenario overlap. .

[0080] This embodiment comprehensively characterizes the correlation possibilities of cross-system codes for core primary equipment in the power grid from three core correlation features: semantic, scenario, and professional constraint dimensions. Then, a dedicated logistic regression probability model is used to learn the correlation patterns of real ambiguous codes in core primary equipment, outputting accurate and quantifiable correlation probabilities. Finally, a unified semantic view is constructed based on high-probability correlation mapping relationships, breaking down semantic barriers between cross-system codes. Compared to existing technologies, the correlation accuracy of core primary equipment codes is significantly improved, eliminating the need for manual maintenance of matching rules, achieving adaptive intelligent matching of ambiguous codes, completely breaking down cross-professional data semantic silos, and filling the technological gap in accurate correlation of heterogeneous codes for core primary equipment in the power grid.

[0081] like Figure 3 As shown in this embodiment, the power equipment full-process data service hierarchical classification and sharing system 100 based on the data middle platform also includes a verification and iterative optimization module 104.

[0082] The verification and iterative optimization module 104 is used to statistically analyze the core indicators of coding matching accuracy, mismatch rate, and missed match rate. If the matching accuracy is lower than the preset threshold, ambiguous coding pairs are re-extracted from the verification samples to supplement the training set, and the dedicated logistic regression probability model is iteratively retrained to update the optimal feature weights w1, w2, w3 and bias value b. Based on the matching effect of the verification scenario, the initial value and fine-tuning step size of the dynamic association probability threshold are fine-tuned for each scenario to solidify the optimal matching parameter combination.

[0083] Finally, it should be noted that although the above embodiments have been described in the text and drawings of this application, this should not limit the scope of patent protection of this application. Any technical solutions that are based on the essential concept of this application and utilize the content described in the text and drawings of this application, resulting in equivalent structural or procedural substitutions or modifications, as well as the direct or indirect application of the technical solutions of the above embodiments to other related technical fields, are all included within the scope of patent protection of this application.

Claims

1. A hierarchical and categorized data sharing method for the entire process of power equipment data based on a data middle platform, used to process multi-source heterogeneous coded data of core primary equipment in the power grid, realizing data connectivity and hierarchical and categorized sharing throughout the entire lifecycle of the equipment, characterized in that... Includes the following steps: S1: Extract device-specific refined semantic similarity from the multi-source heterogeneous encoded data. Overlap of scenarios in the closed-loop lifecycle of equipment , power grid hierarchical professional distance constraint coefficient Three core, multi-dimensional related features were identified and quantized and vectorized. S2: Training and constructing the association function for a dedicated logistic regression probability model for power equipment. Real ambiguous coding samples from the entire lifecycle of core primary equipment in the power grid are collected as the training set. Using the three core association feature vectors as input and whether the coding is associated as a binary classification label, the dedicated logistic regression probability model is iteratively trained. The optimal feature weights w1, w2, and w3, and the bias value b, adapted to the ambiguous scenarios of power grid equipment, are obtained through convergence and optimization. Substituting these optimal parameters into the standard form of logistic regression, the formula for calculating the association probability of equipment coding is obtained: ; S3: Construction of a unified semantic view for equipment: Based on the coding association probability output by the calculation formula of the equipment coding association probability, a unified coding semantic view for the entire life cycle of the core primary equipment of the power grid is generated to realize the semantic-level connection of equipment data throughout the entire process.

2. The method according to claim 1, characterized in that, It also includes verification and iterative optimization steps: Key indicators for statistical coding matching include accuracy, false matching rate, and missed matching rate. If the matching accuracy is lower than the preset threshold, the ambiguous coding pairs are re-extracted based on the verification samples to supplement the training set, the dedicated logistic regression probability model is iteratively retrained, and the optimal feature weights w1, w2, w3 and bias value b are updated. Based on the verification scene matching effect, the initial value and fine-tuning step size of the dynamic association probability threshold are fine-tuned for each scene, and the optimal matching parameter combination is solidified.

3. The method according to claim 1, characterized in that, Step S1 involves refining the semantic similarity extraction of equipment-specific features. This includes using five core power-specific parameter fields—equipment type, voltage level, rated electrical parameters, structural attributes, and power industry standard definitions—as the sole basis for semantic matching, filtering out invalid and interfering fields. A Word2Vec model, finely tuned and optimized using a power equipment professional corpus, is then used to generate a dedicated semantic vector, and a high-precision semantic similarity coefficient is calculated and output. .

4. The method according to claim 1, characterized in that, Step S1, the extraction of the overlap of closed-loop scenarios throughout the equipment lifecycle, includes: selecting six types of core business documents: equipment purchase orders, factory test reports, handover test records, maintenance and repair work orders, defect ledgers, and project acceptance documents; constructing a set of core business scenarios corresponding to each code; calculating the overlap of scenario sets corresponding to different codes using the Jaccard coefficient; and outputting the scenario overlap. .

5. The method according to claim 1, characterized in that, Step S1 involves assigning distance constraint coefficients to different professional disciplines within the power grid hierarchy. This includes setting the distance coefficient for equipment-maintenance to 0.9, materials-procurement to 0.8, and planning-infrastructure to 0.7; setting the distance coefficients for non-equipment-related disciplines such as finance and human resources to 0; and dynamically assigning distance constraint coefficients based on the multi-identifier linkage system to divide professional domains. .

6. The method according to claim 1, characterized in that, The model training samples in step S2 include: more than 10,000 manually verified cross-system valid association coding pairs as positive samples, and more than 5,000 power equipment-independent heterogeneous coding pairs as negative samples.

7. A hierarchical and categorized data sharing system for the entire process of power equipment based on a data middle platform, used to process multi-source heterogeneous coded data of core primary equipment in the power grid, realizing data connectivity and hierarchical and categorized sharing throughout the entire lifecycle of equipment, characterized in that... include: Data extraction and quantization module: used to extract device-specific refined semantic similarity from the multi-source heterogeneous encoded data. Overlap of scenarios in the closed-loop lifecycle of equipment , power grid hierarchical professional distance constraint coefficient Three core, multi-dimensional related features were identified and quantized and vectorized. Modeling and Training Module: Used for training a dedicated logistic regression probability model and constructing association functions for power equipment. It collects real ambiguous coding samples from the entire lifecycle of core primary equipment in the power grid as the training set. Using the three core association feature vectors as input and whether the coding is associated as a binary classification label, it iteratively trains the dedicated logistic regression probability model. The model converges and optimizes to obtain the optimal feature weights w1, w2, w3, and bias value b suitable for ambiguous scenarios involving power grid equipment. Substituting these optimal parameters into the standard form of logistic regression, the formula for calculating the association probability of equipment coding is obtained. ; Execution module: Used for constructing a device-specific unified semantic view. Based on the coding association probability output by the device coding association probability calculation formula, it generates a device-specific unified coding semantic view for the entire life cycle of the core primary equipment of the power grid, realizing semantic-level connectivity of data throughout the entire process of the device.

8. The system according to claim 7, characterized in that, Also includes: Verification and Iterative Optimization Module: Used to statistically analyze core metrics such as coding matching accuracy, false matching rate, and missed matching rate; If the matching accuracy is lower than the preset threshold, the ambiguous coding pairs are re-extracted based on the verification samples to supplement the training set, the dedicated logistic regression probability model is iteratively retrained, and the optimal feature weights w1, w2, w3 and bias value b are updated. Based on the verification scene matching effect, the initial value and fine-tuning step size of the dynamic association probability threshold are fine-tuned for each scene, and the optimal matching parameter combination is solidified.

9. The system according to claim 7, characterized in that, The equipment-specific refined semantic similarity extraction includes: using five categories of core power-specific parameter fields—equipment type, voltage level, rated electrical parameters, structural attributes, and power industry standard definitions—as the sole basis for semantic matching, filtering out invalid and interfering fields; and generating exclusive semantic vectors using a Word2Vec model finely tuned and optimized from a power equipment professional corpus, calculating and outputting a high-precision semantic similarity coefficient. .

10. The system according to claim 7, characterized in that, The equipment lifecycle closed-loop scenario overlap extraction includes: selecting six types of core business documents, namely equipment purchase orders, factory test reports, handover test records, operation and maintenance work orders, defect ledgers, and project acceptance documents, to construct a set of core business scenarios corresponding to each code; calculating the overlap of scenario sets corresponding to different codes using the Jaccard coefficient, and outputting the scenario overlap. .

Images