Relay protection setting value checking method, device and equipment based on multi-feature matching

Abstract

The invention discloses a relay protection constant value checking method, device and equipment based on multi-feature matching, and relates to the technical field of safe and stable operation of a power grid, and the method comprises the steps: calculating a plurality of similarities between an interrogation measured value set from a relay protection device and each constant value item in a standard constant value list; weighted fusion is carried out on a plurality of similarities of word forms, word orders, sentence lengths, semantics and component relations to obtain comprehensive similarities, an optimal allocation algorithm is combined, a unique matching item is searched in a standard constant value list for each constant value item in an interrogated constant value set, a matching pair set is formed, and the matching pair set is used for matching the word forms, the word orders, the sentence lengths, the semantics and the component relations. And performing depth comparison on each pair of constant value items in the matching pair set to obtain a constant value comparison result. According to the method, false alarms caused by naming differences are reduced through multi-feature matching, so that the accuracy of a constant value comparison result is improved.

Description

Technical Field

[0001] This invention relates to the field of power grid safe and stable operation technology, specifically to a method, device and equipment for verifying relay protection settings based on multi-feature matching. Background Technology

[0002] Relay protection devices are automated safety devices installed in various substations, power plants, and power users within a power system. Their core function is to monitor the real-time operating status of connected power equipment. When a fault (such as a short circuit, grounding, or overload) or abnormal operating condition is detected in the power system, the device can automatically, quickly, and selectively issue commands to isolate the faulty component from the power system by driving a circuit breaker to trip, thereby minimizing damage to the power equipment and ensuring the stable operation of the non-faulty parts of the power system.

[0003] The accuracy and reliability of the operation of relay protection devices are crucial, and this is directly determined by the correctness of their internally preset protection settings (such as threshold values ​​for parameters like current, voltage, and time). Existing technologies include some setting verification algorithms, whose basic workflow involves retrieving settings from the device, importing standard setting sheets, performing comparisons, and outputting the comparison results.

[0004] However, existing setting verification algorithms generally rely on the premise of "exact match of the name string of the setting item" for numerical comparison. Once there is a slight difference in the naming of the same function between the field device and the standard setting sheet (such as "overcurrent stage I" and "instantaneous current protection"), the algorithm cannot identify it as the same item, resulting in comparison failure or reporting a large number of "false differences". Summary of the Invention

[0005] The technical problem to be solved by the present invention is how to improve the accuracy of setting verification results. The purpose is to provide a relay protection setting verification method, device and equipment based on multi-feature matching, which solves the above-mentioned problem.

[0006] This invention is achieved through the following technical solution:

[0007] In a first aspect, the present invention provides a method for verifying relay protection settings based on multi-feature matching, comprising:

[0008] Obtain the set of call values ​​from the relay protection device;

[0009] Calculate multiple similarities between the recall value set and each value item in the standard value sheet; the multiple similarities include value word form similarity, value word order similarity, value sentence length similarity, value semantic similarity, and value component relationship similarity;

[0010] The multiple similarities of each pair of fixed-value items are weighted and fused to obtain the comprehensive similarity of each pair of fixed-value items;

[0011] A similarity matrix is ​​established based on the comprehensive similarity of all terms with fixed values;

[0012] Using the similarity matrix as input, the optimal allocation algorithm is used to find a unique matching item in the standard value list for each value item in the recall value set, forming a set of matching pairs, and marking the value items with a comprehensive similarity lower than a preset threshold as unmatched items;

[0013] A deep comparison is performed on each pair of fixed-value items in the matching pair set to obtain the fixed-value comparison result.

[0014] Optionally, the formula for calculating the fixed-value word form similarity is as follows:

[0015] ;

[0016] Wherein, A represents the first value item in the set of determination values; B represents the second value item in the standard value sheet; This represents the word form similarity between the first fixed-value item A and the second fixed-value item B; This indicates the number of similar words contained in the second fixed-value item A and the second fixed-value item B; This represents the number of words in the first fixed-value term A; This indicates the number of words in the second fixed-value term B.

[0017] Optionally, the formula for calculating the fixed-value word order similarity is as follows:

[0018] ;

[0019] Wherein, A represents the first value item in the set of determination values; B represents the second value item in the standard value sheet; This indicates the similarity of the word order of the first fixed value item A and the second fixed value item B. The maximum number of inversions in the natural number sequence representing the number of similar words in the first fixed-value term A and the second fixed-value term B; It represents the number of inversions of the natural number sequence formed by the positional order of similar words in the first fixed-value term A and the second fixed-value term B.

[0020] Optionally, the formula for calculating the fixed sentence length similarity is as follows:

[0021] ;

[0022] Wherein, A represents the first value item in the set of determination values; B represents the second value item in the standard value sheet; This indicates the similarity of the fixed-value sentence length between the first fixed-value item A and the second fixed-value item B; This represents the number of words in the first fixed-value term A; represents the number of words in the second fixed-value term B; abs is the absolute value function.

[0023] Optionally, the formula for calculating the fixed-value semantic similarity is as follows:

[0024] ;

[0025] Wherein, A represents the first value item in the set of determination values; B represents the second value item in the standard value sheet; This represents the semantic similarity between the first fixed-value item A and the second fixed-value item B; m is the number of fixed-value features contained in the first fixed-value item A; and n is the number of fixed-value features contained in the second fixed-value item B. Indicates the first fixed value term A. The maximum semantic similarity between each fixed-value feature and each fixed-value feature in the second fixed-value term B; This indicates the second fixed value term B. The maximum semantic similarity between each fixed-value feature and each fixed-value feature in the first fixed-value item A.

[0026] Optionally, the formula for calculating the similarity of the fixed-value component relationships is as follows:

[0027] ;

[0028] Wherein, A represents the first value item in the set of determination values; B represents the second value item in the standard value sheet; This indicates the similarity of the value component relationship between the first value item A and the second value item B; For the first fixed value item A fixed-value characteristic; For the second fixed value term, the first There are 1 fixed-value feature; n is the number of fixed-value features contained in the first fixed-value item or the second fixed-value item.

[0029] Optionally, the step of performing a deep comparison on each pair of fixed-value items in the matching pair set to obtain the fixed-value comparison result includes:

[0030] For any two constant terms in a pair of constant terms, perform the following operations:

[0031] If at least one of the two fixed-value items is a state quantity, then according to the predefined state semantic mapping rules, the textual description of the state quantity is translated into the corresponding logical value, and the translated logical values ​​are compared to see if they are equal.

[0032] If both of the specified values ​​are numerical values, then compare the units of the two specified values.

[0033] When the units of the two fixed values ​​are the same, compare whether the values ​​of the two fixed values ​​are equal;

[0034] When the units of the two fixed values ​​are inconsistent, the built-in unit conversion library is called to convert the values ​​of the two fixed values ​​to the same unit of measurement, and the converted values ​​are compared to see if they are equal.

[0035] Optionally, after performing a deep comparison on each pair of fixed-value items in the matching pair set to obtain the fixed-value comparison result, the method further includes:

[0036] Based on the comparison results of the set values ​​and the unmatched items, a structured verification report is generated; the verification report includes consistent items, inconsistent items, the unmatched items, and the reasons for the differences in the inconsistent items.

[0037] Secondly, the present invention provides a relay protection setting verification device based on multi-feature matching, comprising:

[0038] The setting value acquisition module is used to acquire the set of reference values ​​from the relay protection device;

[0039] The similarity calculation module is used to calculate multiple similarities between the recall value set and each value item in the standard value sheet; the multiple similarities of each pair of value items are weighted and fused to obtain the comprehensive similarity of each pair of value items; the multiple similarities include value word form similarity, value word order similarity, value sentence length similarity, value semantic similarity, and value component relationship similarity;

[0040] The matrix construction module is used to build a similarity matrix based on the comprehensive similarity of all pairs of fixed-value items;

[0041] The fixed-value matching module is used to take the similarity matrix as input, adopt the optimal allocation algorithm, find a unique matching item in the standard fixed-value list for each fixed-value item in the recall set of fixed-value items, form a set of matching pairs, and mark the fixed-value items with a comprehensive similarity lower than a preset threshold as unmatched items.

[0042] The fixed value comparison module is used to perform a deep comparison of each pair of fixed value items in the matching pair set to obtain the fixed value comparison result.

[0043] Thirdly, the present invention provides a computer device, the computer device including a processor, a memory and a computer program stored in the memory, wherein the computer program is executed by the processor to implement the relay protection setting verification method with multi-feature matching as described in any of the first aspects.

[0044] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0045] This application provides a relay protection setting verification method based on multi-feature matching. By employing a similarity calculation model that integrates multiple features such as word form, word order, sentence length, semantics, and component relationships, and combined with an optimal allocation algorithm, it can effectively identify and handle situations where the names of setting items are not completely consistent (such as "overcurrent stage I" and "instantaneous current protection"). This greatly reduces false alarms and matching failures caused by naming differences, thereby improving the accuracy and reliability of the verification results. Since this method is based on text analysis and semantic understanding, rather than simple string or format matching, it can adapt well to the format differences of setting sheets exported by relay protection devices from different manufacturers and models, as well as subtle changes in the standard setting sheet format, making it more universally applicable. Attached Figure Description

[0046] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0047] Figure 1 A flowchart illustrating the relay protection setting verification method based on multi-feature matching provided in this application embodiment;

[0048] Figure 2 This is a schematic diagram of the structure of the relay protection setting verification device based on multi-feature matching provided in the embodiments of this application. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0050] Existing relay protection setting verification algorithms have significant limitations, specifically as follows:

[0051] 1. Comparison of machinery: Existing technologies mostly use the simple premise of "complete matching of setting item name strings" for numerical comparison. Once there are slight differences in the naming of the same function between the field device and the standard setting sheet (such as "overcurrent stage I" and "instantaneous current protection"), it cannot be identified as the same item, resulting in comparison failure or a large number of "false differences".

[0052] 2. Lack of semantic understanding: Existing technologies cannot understand the physical meaning and contextual relationships of fixed values, and cannot handle situations such as unit conversion (e.g., kA and A) and different fixed value types (e.g., Boolean values ​​for "input / output", "0 / 1" encoding, etc.).

[0053] 3. High requirements for document format: Existing technologies are highly sensitive to file format, data sorting, and additional text descriptions, requiring standard value sheets to have a highly standardized format, which is not very universal.

[0054] Therefore, embodiments of this application provide a relay protection setting verification method based on multi-feature matching. Please refer to... Figure 1 The following is a flowchart illustrating the multi-feature matching relay protection setting verification method provided in this application embodiment. Figure 1 The multi-feature matching relay protection setting verification method is introduced.

[0055] S101. Obtain the set of call values ​​from the relay protection device.

[0056] In the specific implementation process, the setting items to be verified are extracted from the on-site relay protection devices, generating a structured set of reference setting items. Each setting item includes the following:

[0057] 1. Name: Functional descriptive text for the setting item, such as "Overcurrent Stage I Setting" or "Zero Sequence Current Protection".

[0058] 2. Value: The actual set value of the fixed value item, which can be a floating-point number (such as 5.15), an integer (such as 10), or a discrete state value (such as 1).

[0059] 3. Unit: The physical unit corresponding to the numerical value, such as "A", "kA", "s", etc.

[0060] 4. Type: Used to distinguish the nature of the set value, mainly including numerical type (such as current and time set values) and status type (such as the "input / output" of the function pressure plate).

[0061] S102. Calculate the similarity between the set of predicted values ​​and each value item in the standard value sheet.

[0062] In practical implementation, the standard setting sheet refers to a normative technical document formally issued by the power system's operation and management department or professional calculation department after rigorous theoretical calculations and approval processes, based on the real-time operation mode of the power grid, primary equipment parameters, relay protection setting procedures, and relevant industry standards. The standard setting sheet is the sole authoritative basis for setting relay protection devices, and includes multiple setting items, each containing the name, value, unit, and type of the setting item.

[0063] Multiple similarity metrics include fixed-value word form similarity, fixed-value word order similarity, fixed-value sentence length similarity, fixed-value semantic similarity, and fixed-value component relationship similarity. Each similarity metric is described below.

[0064] (1) Fixed-value word form similarity: used to measure the degree of overlap in the word surface structure of the names of two fixed-value items.

[0065] ;

[0066] Where A represents the first value item in the set of measured values; B represents the second value item in the standard value sheet; This indicates the word form similarity between the first fixed-value term A and the second fixed-value term B. This indicates the number of similar words contained in the second fixed-value term A and the second fixed-value term B; This indicates the number of words in the first fixed-value term A; This indicates the number of words in the second fixed-value item B. In the embodiments of this application, fixed-value items with completely identical names or most identical words can be effectively matched through fixed-value word similarity.

[0067] (2) Fixed value word order similarity: used to measure the degree of consistency in the order of common words in the names of two fixed value items.

[0068] ;

[0069] Where A represents the first value item in the set of measured values; B represents the second value item in the standard value sheet; This indicates the similarity of the word order of the first fixed-value term A and the second fixed-value term B. The maximum number of inversions in a natural number sequence representing the number of similar words in the first fixed-value term A and the second fixed-value term B; It represents the number of inversions of the natural number sequence formed by the positional order of similar words in the first fixed term A and the second fixed term B.

[0070] For example, the first setting item A: "Overcurrent delay protection setting," and the second setting item B: "Delayed overcurrent protection enable / disable." Both setting items A and B contain four similar words: "overcurrent," "delayed," and "protection," and the natural number sequence is {4,3,2,1}. What is the maximum number of inversions in this natural number sequence? Starting from index 1, the natural number sequence {1,2,3,4} formed by the positions of these four similar words in the first fixed-value term A, and the natural number sequence {3,2,1,4} formed by the positions of these four similar words in the first fixed-value term B, are the number of inversions in this natural number sequence. .

[0071] In this embodiment of the application, fixed-value word order similarity is used to identify fixed-value items that have the same word composition but different order, resulting in different or different meanings, thereby avoiding incorrect matching.

[0072] (3) Fixed sentence length similarity: used to measure the closeness of two fixed terms from the macro perspective of the overall text length.

[0073] ;

[0074] Where A represents the first value item in the set of measured values; B represents the second value item in the standard value sheet; This indicates the similarity of the fixed-value sentence lengths between the first fixed-value item A and the second fixed-value item B. This indicates the number of words in the first fixed-value term A; represents the number of words in the second constant term B; abs is the absolute value function.

[0075] In this embodiment, fixed sentence length similarity is used as a fast filtering feature, which can significantly reduce the overall similarity between fixed items with significantly different name lengths, thereby improving matching efficiency and accuracy.

[0076] (4) Semantic similarity of fixed values: used to measure the degree of similarity in the intrinsic meaning of the names of two fixed values.

[0077] ;

[0078] Where A represents the first value item in the set of measured values; B represents the second value item in the standard value sheet; This represents the semantic similarity between the first fixed-value item A and the second fixed-value item B; m is the number of fixed-value features contained in the first fixed-value item A; and n is the number of fixed-value features contained in the second fixed-value item B. Indicates the first fixed value term A. The maximum semantic similarity between each fixed-value feature and each fixed-value feature in the second fixed-value term B; This indicates the second fixed value term B. The maximum semantic similarity between each fixed-value feature and each fixed-value feature in the first fixed-value term A.

[0079]

[0080]

[0081] in, , ,…, The fixed-value characteristics included in the first fixed-value term A; , ,…, S represents the fixed-value features contained in the second fixed-value term B; S(,) represents the semantic similarity; m is the number of fixed-value features contained in the first fixed-value term A; n is the number of fixed-value features contained in the second fixed-value term B; max is the maximum value function.

[0082] For example, if the fixed-value characteristics of the first fixed-value term A include name characteristic A1, number characteristic A2, unit characteristic A3, and type characteristic A4, then m=4; if the fixed-value characteristics of the second fixed-value term B include name characteristic B1, number characteristic B2, unit characteristic B3, and type characteristic B4, then n=4.

[0083] In the embodiments of this application, the problem of matching synonyms, near-synonyms, and different expressions of professional terms can be effectively solved by using fixed-value semantic similarity.

[0084] (5) Similarity of fixed-value components: used to treat the names of two fixed-value items as a whole and measure the consistency of the distribution and direction of their overall feature vector in the vector space.

[0085] ;

[0086] Where A represents the first value item in the set of measured values; B represents the second value item in the standard value sheet; This indicates the similarity of the value component relationships between the first fixed-value term A and the second fixed-value term B. For the first fixed value term A fixed-value characteristic; For the second fixed value term There are 1 fixed-value features; n is the number of fixed-value features contained in the first or second fixed-value term.

[0087] In the embodiments of this application, by using the similarity of the relationship between the fixed-value components, the similarity of the text can be captured from a more global and deeper perspective, which can make up for the shortcomings of simple word or semantic comparison, comprehensively judge the similarity of the two fixed-value items in the overall expression, and further improve the robustness of matching.

[0088] S103. Weighted fusion of multiple similarities for each pair of fixed-value terms to obtain the comprehensive similarity for each pair of fixed-value terms.

[0089] In the specific implementation process, the formula for calculating the overall similarity is as follows:

[0090]

[0091] in, The comprehensive similarity between the first and second fixed-value terms. These are the weighted coefficients for fixed-value word form similarity, fixed-value sentence length similarity, fixed-value word order similarity, fixed-value semantic similarity, and fixed-value component relationship similarity, respectively.

[0092] In this embodiment, configurable weight coefficients conforming to domain logic are introduced to scientifically integrate the text feature similarity across five dimensions. This fusion strategy amplifies the contribution of key features (such as semantics) while also taking into account the correction and filtering effects of other features. This allows the final comprehensive similarity to accurately and intelligently quantify the degree of matching between two fixed-value terms in actual function, laying a solid foundation for subsequent optimal matching based on the similarity matrix.

[0093] In one possible embodiment, the weighting coefficients Configuration can be performed based on the experience of those skilled in the art. Preferably, And satisfy .

[0094] In another possible embodiment, to further optimize the verification accuracy, the weight coefficients can also be determined by supervised machine learning methods, specifically including: constructing a historical fixed-value dataset containing a large number of manually labeled matching relationships; using a comprehensive similarity calculation model as the basis, with matching accuracy or F1 score as the objective function; and using optimization methods such as grid search and genetic algorithms to find the optimal combination of weight coefficients that maximizes the objective function.

[0095] S104. Based on the comprehensive similarity of all terms with fixed values, establish a similarity matrix.

[0096] In the specific implementation process, each fixed value item in the set of determined values ​​S is traversed. and standard setting value setting sheet Each fixed term in For each pair of constant terms Calculate the overall similarity This value is then filled into the position of the i-th row and j-th column of the similarity matrix S, i.e. The similarity matrix is ​​used to indicate the strength of the matching association between each value in the recall set and each value in the standard set.

[0097] S105. Using the similarity matrix as input, the optimal allocation algorithm is adopted to find a unique matching item in the standard value list for each value item in the recall value set, forming a matching pair set, and marking the value items with a comprehensive similarity lower than the preset threshold as unmatched items.

[0098] In the specific implementation process, unmatched items are divided into redundant items in the field and missing items in the standard. Redundant items in the field refer to the value items that exist in the recall measurement value set but are not defined in the standard value set; missing items in the standard value set refer to the value items that are required in the standard value set but are missing in the recall measurement value set.

[0099] In one possible implementation, the optimal allocation algorithm is a greedy algorithm, with the following specific steps:

[0100] Iterate through each fixed value in the set of called values ​​S. For each fixed value item Perform the following operations:

[0101] S1.1, Traverse each value item in the standard value sheet T. ;

[0102] S1.2 Extracting from the similarity matrix With each Overall similarity ;

[0103] S1.3, Selection and The second fixed-value term with the highest overall similarity As its only matching item, it forms a candidate matching pair. .

[0104] S1.4 Determine whether the highest comprehensive similarity is greater than or equal to the preset threshold.

[0105] If so, then confirm. If a match is valid, then the match is terminated. Release, will Marked as a non-match.

[0106] After iterating through each value in the recall value set S, recall all unmatched values ​​in the recall value set S. Mark as redundant items on site, and include all items in the standard setting sheet T that are not marked with any redundancy. Matching Mark as a standard missing item.

[0107] In the embodiments of this application, the greedy algorithm has relatively low computational complexity, is suitable for application scenarios with high real-time requirements, does not require the maintenance of complex intermediate state data, and has a small memory footprint.

[0108] While greedy algorithms can achieve good matching results in most cases, in many... With the same In complex scenarios where all elements have high overall similarity, a globally optimal matching result may not be obtained. Therefore, in another possible implementation, the optimal allocation algorithm is the Hungarian algorithm, with the following specific steps:

[0109] S2.1 Construct the cost matrix C based on the similarity matrix M;

[0110] Each element satisfies ; Let be the element in the i-th row and j-th column of the cost matrix C. Let K be the element in the i-th row and j-th column of the similarity matrix M; K is a preset constant that satisfies .

[0111] S2.2 Initialize the top label vectors u and v.

[0112] For each fixed value term in the set of determined values ​​S... ,set up ;

[0113] For each setting item in the standard setting sheet T ,set up .

[0114] S2.3, For each fixed value term The optimal matching is found by iteratively performing operations to find augmenting paths and adjust the top label vectors.

[0115] S2.4 Output a set of matching pairs, which satisfies the condition that the sum of the overall similarity of all matching pairs is the maximum value.

[0116] S2.5 For each matching pair in the matching pair set, if the overall similarity is greater than the preset threshold, the matching pair is retained; if the overall similarity is less than the preset threshold, the matching pair is removed, and the two fixed-value items of the matching pair are marked as unmatched items.

[0117] S2.6 Mark all unmatched values ​​in the recall measurement set as redundant items in the field, and mark all unmatched values ​​in the standard value sheet as missing items in the standard.

[0118] In the embodiments of this application, the Hungarian algorithm ensures the acquisition of the global optimal solution through rigorous matrix transformation and iterative optimization, effectively handling the complex situation where multiple fixed-value terms compete for the same matching target.

[0119] S106. Perform a deep comparison on each pair of fixed-value items in the matching pair set to obtain the fixed-value comparison result.

[0120] In one possible embodiment, for any two of the setpoints in a pair of setpoints, the following operation is performed:

[0121] If at least one of the two fixed-value items is a state variable, then according to the predefined state semantic mapping rules, the textual description of the state variable is translated into the corresponding logical value, and the translated logical values ​​are compared to see if they are equal. If both fixed-value items are numerical, then the units of the two fixed-value items are compared. If the units of the two fixed-value items are the same, then the numerical values ​​of the two fixed-value items are compared to see if they are equal. If the units of the two fixed-value items are different, then the built-in unit conversion library is called to perform unit conversion, converting the numerical values ​​of the two fixed-value items to the same unit of measurement, and the converted numerical values ​​are compared to see if they are equal.

[0122] In the specific implementation process, for any pair of fixed-value items, the types of the two fixed-value items are first determined. If at least one of the two fixed-value items is a state variable, a state variable comparison process is triggered. According to predefined state semantic mapping rules, the textual description of the state variable is translated into the corresponding logical value, and the consistency of the translated logical values ​​is compared. If the logical values ​​are consistent, the two fixed-value items are recorded as consistent items; if the logical values ​​are inconsistent, the two fixed-value items are recorded as inconsistent items, and the reason for the difference is marked as "inconsistent state variable values". The state semantic mapping rules include: mapping "Enter", "On", and "ON" to the logical value 1 or "True"; and mapping "Exit", "Close", and "OFF" to the logical value 0 or "False".

[0123] If both constants are numeric, a numeric comparison process is triggered. The units of the two constants are compared; if the units are the same, the numerical values ​​of the two constants are directly compared for equality; if the units are different, the built-in unit conversion library (e.g., ...) is called. Perform unit conversion to convert the values ​​of the two constants to the same unit of measurement; based on the conversion result, compare whether the values ​​of the two constants are equal. If the values ​​are equal, record the two constants as consistent; if the values ​​are not equal, record the two constants as inconsistent, and indicate the reason for the difference as "inconsistent values".

[0124] In this embodiment, the built-in unit conversion library automatically processes fixed values ​​in different unit systems (such as kA and A, s and ms). After intelligent conversion, an equivalent comparison is performed, avoiding the misjudgment of correct fixed values ​​due to unit differences and reducing unnecessary manual verification. Through predefined state semantic mapping rules, it can intelligently understand and translate state quantities in different expressions (such as recognizing "entering" and "1", and "exiting" and "0" as equivalent), achieving a deep understanding of the semantics of fixed values ​​and resolving misjudgments caused by inconsistent state expressions.

[0125] In one possible embodiment, after performing a deep comparison of each pair of fixed-value items in the matching pair set to obtain the fixed-value comparison result, the method further includes:

[0126] Based on the comparison results and the unmatched items, a structured verification report is generated. The verification report includes the verified consistent items, inconsistent items, unmatched items, and the reasons for the differences in inconsistent items.

[0127] In the specific implementation process, consistent items refer to fixed-value items whose values ​​and statuses are completely consistent after in-depth comparison. Inconsistent items refer to fixed-value items where differences are found during in-depth comparison. Consistent items can be displayed in a collapsed manner. By default, the specific content is hidden in the collapsed state, and only the statistical quantity is displayed. Users can expand to view the full details as needed, effectively maintaining the simplicity and readability of the report interface. Inconsistent items are displayed in a highlighted manner (such as using prominent colors, icon markings, etc.), and the standard value (the value of the fixed-value item in the standard fixed-value sheet), the field value (the value of the fixed-value item in the recall measurement set), the difference value (the difference between the standard value and the field value), and the reason for the difference (such as inconsistent values, inconsistent units, inconsistent status descriptions, etc.). Unmatched items are displayed in two categories: redundant field items and missing standard items.

[0128] In this embodiment, the verification results are displayed in a three-level information hierarchy, making the verification conclusions clear at a glance. By clearly marking the reasons for the differences and distinguishing different types of mismatched items, sufficient basis is provided for professional review. By highlighting inconsistencies, maintenance personnel can quickly locate the essence of the problem and focus on in-depth analysis of a few abnormal results, greatly improving the efficiency and quality of value verification work.

[0129] In summary, this application provides a relay protection setting verification method based on multi-feature matching, which has the following advantages compared to the prior art:

[0130] 1. Significantly improved accuracy and reliability of verification

[0131] By integrating a similarity calculation model that incorporates multi-dimensional features such as word form, word order, sentence length, semantics, and component relationships, and combining it with a global optimal allocation algorithm, this application can effectively identify and match value items that are not completely identical in name but have the same function. This fundamentally overcomes the mechanical defects of traditional string-based exact matching, significantly reduces false alarms of "spurious differences," and greatly improves the accuracy and reliability of value verification.

[0132] 2. It achieves deep semantic understanding and intelligent judgment.

[0133] With its built-in unit conversion library, the system can automatically handle fixed values ​​in different unit systems, achieving intelligent conversion and equivalence comparison, thus avoiding misjudgments caused by unit differences. Through predefined state semantic mapping rules, it can understand and translate state quantities in different representations, solving the verification problem caused by differences in state quantity representations. These mechanisms enable this application to go beyond simple numerical comparisons, achieving more intelligent equivalence judgments.

[0134] 3. Improved operation and maintenance efficiency

[0135] The verification report uses a three-level information display (collapsed consistent items, highlighted inconsistent items, and categorized unmatched items), enabling operators to quickly focus on key issues without having to manually sift through massive amounts of data, greatly improving the decision-making efficiency and problem location speed of maintenance personnel.

[0136] 4. Strong robustness and high adaptability

[0137] The method provided in this application is insensitive to the format, sorting, and representation of the input data. Its core lies in text analysis and semantic understanding, rather than format matching. Therefore, it can be applied to relay protection devices of different manufacturers and models, as well as setting sheets with different formats exported from different management systems, without adaptation or adjustment. It demonstrates strong universality (strong robustness) and solves the pain points of existing technologies that have excessively high requirements for standardized formats and lack universality.

[0138] 5. The level of automation and intelligence has been comprehensively improved.

[0139] From automatic detection, intelligent matching, and semantic comparison to report generation, the entire process is automated, minimizing manual intervention and reducing the risk of human error. This provides a complete and reliable technical solution for intelligent operation and maintenance of relay protection setting verification.

[0140] Based on the same inventive concept, please refer to Figure 2 This application also provides a relay protection setting verification device based on multi-feature matching, the device comprising:

[0141] The setting value acquisition module is used to acquire the set of reference values ​​from the relay protection device;

[0142] The similarity calculation module is used to calculate multiple similarities between the recall value set and each value item in the standard value sheet; the multiple similarities of each pair of value items are weighted and fused to obtain the comprehensive similarity of each pair of value items; the multiple similarities include value word form similarity, value word order similarity, value sentence length similarity, value semantic similarity, and value component relationship similarity;

[0143] The matrix construction module is used to build a similarity matrix based on the comprehensive similarity of all pairs of fixed-value items;

[0144] The fixed-value matching module is used to take the similarity matrix as input, use the optimal allocation algorithm to find a unique matching item in the standard fixed-value list for each fixed-value item in the called fixed-value set, form a set of matching pairs, and mark the fixed-value items with a comprehensive similarity lower than a preset threshold as unmatched items.

[0145] The fixed value comparison module is used to perform a deep comparison of each pair of fixed value items in the matching pair set to obtain the fixed value comparison result.

[0146] Optionally, the formula for calculating fixed-value word form similarity is as follows:

[0147] ;

[0148] Where A represents the first value item in the set of measured values; B represents the second value item in the standard value sheet; This indicates the word form similarity between the first fixed-value term A and the second fixed-value term B. This indicates the number of similar words contained in the second fixed-value term A and the second fixed-value term B; This indicates the number of words in the first fixed-value term A; This indicates the number of words in the second fixed-value term B.

[0149] Optionally, the formula for calculating fixed-value word order similarity is as follows:

[0150] ;

[0151] Where A represents the first value item in the set of measured values; B represents the second value item in the standard value sheet; This indicates the similarity of the word order of the first fixed-value term A and the second fixed-value term B. The maximum number of inversions in a natural number sequence representing the number of similar words in the first fixed-value term A and the second fixed-value term B; It represents the number of inversions of the natural number sequence formed by the positional order of similar words in the first fixed term A and the second fixed term B.

[0152] Optionally, the formula for calculating the similarity of a fixed sentence length is as follows:

[0153] ;

[0154] Where A represents the first value item in the set of measured values; B represents the second value item in the standard value sheet; This indicates the similarity of the fixed-value sentence lengths between the first fixed-value item A and the second fixed-value item B. This indicates the number of words in the first fixed-value term A; represents the number of words in the second constant term B; abs is the absolute value function.

[0155] Optionally, the formula for calculating fixed-value semantic similarity is as follows:

[0156] ;

[0157] Where A represents the first value item in the set of measured values; B represents the second value item in the standard value sheet; This represents the semantic similarity between the first fixed-value item A and the second fixed-value item B; m is the number of fixed-value features contained in the first fixed-value item A; and n is the number of fixed-value features contained in the second fixed-value item B. Indicates the first fixed value term A. The maximum semantic similarity between each fixed-value feature and each fixed-value feature in the second fixed-value term B; This indicates the second fixed value term B. The maximum semantic similarity between each fixed-value feature and each fixed-value feature in the first fixed-value term A.

[0158] Optionally, the formula for calculating the similarity of constant-value component relationships is as follows:

[0159] ;

[0160] Where A represents the first value item in the set of measured values; B represents the second value item in the standard value sheet; This indicates the similarity of the value component relationships between the first fixed-value term A and the second fixed-value term B. For the first fixed value term A fixed-value characteristic; For the second fixed value term There are 1 fixed-value features; n is the number of fixed-value features contained in the first or second fixed-value term.

[0161] Optionally, the fixed value comparison module is specifically used for:

[0162] For any two constant terms in a pair of constant terms, perform the following operations:

[0163] If at least one of the two fixed-value items is a state variable, then according to the predefined state semantic mapping rules, the textual description of the state variable is translated into the corresponding logical value, and the translated logical values ​​are compared to see if they are equal.

[0164] If both constants are of type numeric, then compare the units of the two constants.

[0165] When the units of two constants are the same, compare whether the values ​​of the two constants are equal;

[0166] When the units of two fixed values ​​are inconsistent, the built-in unit conversion library is called to convert the units to the same unit of measurement, and the converted values ​​are compared to see if they are equal.

[0167] Optionally, the device may include a report generation module, which is used for:

[0168] After performing a deep comparison of each pair of fixed-value items in the matching pair set and obtaining the fixed-value comparison results, a structured verification report is generated based on the fixed-value comparison results and the unmatched items. The verification report includes consistent items, inconsistent items, unmatched items, and the reasons for the differences in inconsistent items.

[0169] It should be noted that each module in the multi-feature matching relay protection setting verification device in this embodiment corresponds one-to-one with each step in the multi-feature matching relay protection setting verification method in the aforementioned embodiment. Therefore, the specific implementation of this embodiment can refer to the implementation of the aforementioned multi-feature matching relay protection setting verification method, and will not be repeated here.

[0170] Based on the same inventive concept, this application also provides a computer device, which includes a processor, a memory, and a computer program stored in the memory. The computer program is executed by the processor to implement the aforementioned multi-feature matching relay protection setting verification method.

[0171] Based on the same inventive concept, this application also provides a computer storage medium storing a computer program, which is executed by a processor to implement the aforementioned multi-feature matching relay protection setting verification method.

[0172] In some embodiments, the computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface memory, optical disk, or CD-ROM; or it may be a device including one or any combination of the above-mentioned memories. The computer may be a variety of computing devices, including smart terminals and servers.

[0173] In some embodiments, executable instructions may take the form of a program, software, software module, script, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as a standalone program or as a module, component, subroutine, or other unit suitable for use in a computing environment.

[0174] As an example, executable instructions may, but do not necessarily, correspond to files in the file system. They may be stored as part of a file that holds other programs or data, for example, in one or more scripts in a HyperText Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple co-located files (e.g., a file that stores one or more modules, subroutines, or code sections).

[0175] As an example, executable instructions can be deployed to execute on a single computing device, or on multiple computing devices located in one location, or on multiple computing devices distributed across multiple locations and interconnected via a communication network.

[0176] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0177] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0178] The above specific embodiments further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for verifying relay protection settings based on multi-feature matching, characterized in that, include: Obtain the set of call values ​​from the relay protection device; Calculate multiple similarities between the recall value set and each value item in the standard value sheet; the multiple similarities include value word form similarity, value word order similarity, value sentence length similarity, value semantic similarity, and value component relationship similarity; The multiple similarities of each pair of fixed-value items are weighted and fused to obtain the comprehensive similarity of each pair of fixed-value items; A similarity matrix is ​​established based on the comprehensive similarity of all terms with fixed values; Using the similarity matrix as input, the optimal allocation algorithm is used to find a unique matching item in the standard value list for each value item in the recall value set, forming a set of matching pairs, and marking the value items with a comprehensive similarity lower than a preset threshold as unmatched items; A deep comparison is performed on each pair of fixed-value items in the matching pair set to obtain the fixed-value comparison result.

2. The relay protection setting verification method based on multi-feature matching according to claim 1, characterized in that, The formula for calculating the fixed-value word form similarity is as follows: ； Wherein, A represents the first value item in the set of determination values; B represents the second value item in the standard value sheet; This represents the word form similarity between the first fixed-value item A and the second fixed-value item B; This indicates the number of similar words contained in the second fixed-value item A and the second fixed-value item B; This represents the number of words in the first fixed-value term A; This indicates the number of words in the second fixed-value term B.

3. The relay protection setting verification method based on multi-feature matching according to claim 1, characterized in that, The formula for calculating the fixed-value word order similarity is as follows: ； Wherein, A represents the first value item in the set of determination values; B represents the second value item in the standard value sheet; This indicates the similarity of the word order of the first fixed value item A and the second fixed value item B. The maximum number of inversions in the natural number sequence representing the number of similar words in the first fixed-value term A and the second fixed-value term B; It represents the number of inversions of the natural number sequence formed by the positional order of similar words in the first fixed-value term A and the second fixed-value term B.

4. The relay protection setting verification method based on multi-feature matching according to claim 1, characterized in that, The formula for calculating the fixed sentence length similarity is as follows: ； Wherein, A represents the first value item in the set of determination values; B represents the second value item in the standard value sheet; This indicates the similarity of the fixed-value sentence length between the first fixed-value item A and the second fixed-value item B; This represents the number of words in the first fixed-value term A; represents the number of words in the second fixed-value term B; abs is the absolute value function.

5. The relay protection setting verification method based on multi-feature matching according to claim 1, characterized in that, The formula for calculating the fixed-value semantic similarity is as follows: ； Wherein, A represents the first value item in the set of determination values; B represents the second value item in the standard value sheet; This represents the semantic similarity between the first fixed-value item A and the second fixed-value item B; m is the number of fixed-value features contained in the first fixed-value item A; and n is the number of fixed-value features contained in the second fixed-value item B. Indicates the first fixed value term A. The maximum semantic similarity between each fixed-value feature and each fixed-value feature in the second fixed-value term B; This indicates the second fixed value term B. The maximum semantic similarity between each fixed-value feature and each fixed-value feature in the first fixed-value item A.

6. The relay protection setting verification method based on multi-feature matching according to claim 1, characterized in that, The formula for calculating the similarity of the fixed-value component relationships is as follows: ； Wherein, A represents the first value item in the set of determination values; B represents the second value item in the standard value sheet; This indicates the similarity of the value component relationship between the first value item A and the second value item B; For the first fixed value item A fixed-value characteristic; For the second fixed value term, the first There are 1 fixed-value feature; n is the number of fixed-value features contained in the first fixed-value item or the second fixed-value item.

7. The relay protection setting verification method based on multi-feature matching according to claim 1, characterized in that, The step of performing a deep comparison of each pair of fixed-value items in the matching pair set to obtain the fixed-value comparison result includes: For any two constant terms in a pair of constant terms, perform the following operations: If at least one of the two fixed-value items is a state quantity, then according to the predefined state semantic mapping rules, the textual description of the state quantity is translated into the corresponding logical value, and the translated logical values ​​are compared to see if they are equal. If both of the specified values ​​are numerical values, then compare the units of the two specified values. When the units of the two fixed values ​​are the same, compare whether the values ​​of the two fixed values ​​are equal; When the units of the two fixed values ​​are inconsistent, the built-in unit conversion library is called to convert the values ​​of the two fixed values ​​to the same unit of measurement, and the converted values ​​are compared to see if they are equal.

8. The relay protection setting verification method based on multi-feature matching according to claim 1, characterized in that, After performing a deep comparison on each pair of fixed-value items in the matching pair set to obtain the fixed-value comparison result, the method further includes: Based on the comparison results of the set values ​​and the unmatched items, a structured verification report is generated; the verification report includes consistent items, inconsistent items, the unmatched items, and the reasons for the differences in the inconsistent items.

9. A relay protection setting verification device based on multi-feature matching, characterized in that, include: The setting value acquisition module is used to acquire the set of reference values ​​from the relay protection device; The similarity calculation module is used to calculate multiple similarities between the recall value set and each value item in the standard value sheet; the multiple similarities of each pair of value items are weighted and fused to obtain the comprehensive similarity of each pair of value items; the multiple similarities include value word form similarity, value word order similarity, value sentence length similarity, value semantic similarity, and value component relationship similarity; The matrix construction module is used to build a similarity matrix based on the comprehensive similarity of all pairs of fixed-value items; The fixed-value matching module is used to take the similarity matrix as input, adopt the optimal allocation algorithm, find a unique matching item in the standard fixed-value list for each fixed-value item in the recall set of fixed-value items, form a set of matching pairs, and mark the fixed-value items with a comprehensive similarity lower than a preset threshold as unmatched items. The fixed value comparison module is used to perform a deep comparison of each pair of fixed value items in the matching pair set to obtain the fixed value comparison result.

10. A computer device, comprising a processor, a memory, and a computer program stored in the memory, wherein the computer program is executed by the processor to implement the relay protection setting verification method for multi-feature matching as described in any one of claims 1-8.

Images