A secondary virtual loop automation configuration and online checking system and method suitable for an island new energy power grid

Abstract

This invention provides an automated configuration and online verification system and method for secondary virtual loops in isolated new energy power grids. The method includes the following steps: (1) automatic extraction and dimensionality reduction modeling of underlying communication features; (2) intelligent fault-tolerant matching based on dynamic programming and KNN; (3) anti-malfunction interception and holographic topology reconstruction; (4) bidirectional difference set comparison verification based on high-dimensional joint keywords; and (5) anti-tampering underlying self-check based on CRC32 hardware-level fingerprint. This invention completely solves the problem of adaptation and disconnection of heterogeneous devices, provides millisecond-level anti-tampering security verification, and significantly improves the fast iteration capability and operational reliability of the secondary system of isolated microgrids.

Description

Technical Field

[0001] This invention belongs to the field of automation control technology for secondary systems of smart grids and new energy power plants, specifically a system and method for automated configuration and online verification of secondary virtual loops applicable to isolated new energy power grids. Background Technology

[0002] Driven by the global energy transition and the "dual carbon" goal, new islanded renewable energy grids, primarily based on photovoltaic, wind power, and energy storage, are developing rapidly. In islanded operation mode, the grid's frequency and voltage are highly susceptible to sudden load changes, and the system as a whole exhibits characteristics of "low inertia and weak damping." This places extremely high demands on the response speed and configuration accuracy of the secondary systems (protection and control devices) within the substation. The secondary systems of smart substations have completely abandoned traditional physical hard wiring, relying heavily on digital substation configuration description (SCD) files to achieve object-oriented substation event (GOOSE) and sampled measurement value (SMV) message communication between intelligent electronic devices (IEDs).

[0003] However, existing secondary systems and devices face the following core challenges that urgently need to be addressed when dealing with isolated renewable energy grids:

[0004] Traditional programming models are time-consuming and cannot adapt to the "rapid iteration" that leads to wind and solar curtailment: Currently, the configuration of secondary equipment in new energy power plants relies heavily on the low-level hard coding of senior engineers, often taking several months from strategy design to debugging. This long-cycle model is simply unable to meet the "multi-scenario, rapid iteration" operational needs of new energy, which may cause serious curtailment of solar and wind power and waste of clean energy.

[0005] Heterogeneous equipment exhibits extremely poor compatibility, leading to high custom development costs: Existing measurement, control, and protection devices mostly employ fixed logic designs. Isolated renewable energy power plants often connect to heterogeneous IED devices from multiple manufacturers. Due to the lack of mandatory underlying constraints, the naming of virtual terminals from different manufacturers is highly inconsistent. Traditional rule-based lookup table methods or precise string matching methods are completely unable to identify such heterogeneous strings, resulting in repeated custom development for each new device model. This is not only costly but also prone to errors and omissions in manual wiring.

[0006] Large response latency and lack of multi-dimensional online verification and anti-tampering mechanisms: Currently, the response latency of most embedded secondary devices is generally above 50ms, posing a safety risk under extreme operating conditions. During actual operation or renovation and commissioning of substations, construction personnel often privately modify the operating configuration of on-site equipment (such as leaving test leads or omitting critical interlocking circuits). Traditional methods typically involve line-by-line text comparison of massive amounts of XML code, with a computational complexity as high as O(N). 2This process is time-consuming and highly susceptible to code formatting issues. Furthermore, existing systems lack a data anti-counterfeiting mechanism at the physical storage level, making it impossible to effectively prevent cybersecurity risks such as malicious modification of configuration files. Summary of the Invention

[0007] To address the shortcomings of the existing technologies, this invention provides an automated configuration and online verification system and method for secondary virtual loops applicable to isolated new energy power grids, thereby solving the problems of poor compatibility of heterogeneous devices, low efficiency of traditional manual comparison, and lack of underlying anti-tampering mechanisms in the existing technologies.

[0008] The technical solution provided by this invention is a method for automated configuration and online verification of secondary virtual loops suitable for isolated renewable energy power grids, comprising the following steps:

[0009] (1) Automatic extraction and dimensionality reduction modeling of underlying communication features

[0010] Read and parse the configuration description file of the intelligent electronic device, extract the communication access block and external reference information, and generate a two-dimensional data matrix containing the feature attributes of the sender and receiver by dimensionality reduction;

[0011] (2) Dynamic programming calculation and multidimensional feature construction

[0012] A distance matrix is ​​established using a dynamic programming algorithm to calculate the Levenshtein edit distance between the sender description string and the receiver description string to be connected, and a multi-dimensional test feature vector is constructed by combining logical node prefix attributes, etc.

[0013] (3) Intelligent fault-tolerant matching and locking based on KNN

[0014] The multidimensional test feature vector is input into a KNN machine learning model pre-trained using historical legitimate data to calculate the posterior confidence probability of the legitimate connection of the virtual terminal combination; when the confidence is greater than a preset safety threshold, a virtual connection relationship is automatically established, and a quadratic virtual loop holographic topology is generated based on a directed graph algorithm; combinations with confidence below the safety threshold are forcibly intercepted to prevent errors.

[0015] (4) Bidirectional difference set comparison and verification based on high-dimensional joint keywords

[0016] The reference information in the standard configuration file and the actual on-site running configuration memory is extracted to construct a dimension-reduced joint key. By performing a bidirectional difference comparison operation on the two sets of joint keys, the abnormal feature sets of missing links and illegally added links are accurately isolated.

[0017] (5) Anti-tampering self-test based on CRC32 hardware fingerprint.

[0018] The data streams of the standard configuration and the field operation configuration are subjected to cyclic redundancy check polynomial hash operation to generate hexadecimal digital fingerprints. When the fingerprints do not match, the underlying tampering alarm is triggered and the instruction to lock the relevant protection devices is issued.

[0019] Furthermore, the feature attributes extracted in step (1) include at least: the sending device name (iedName), the sending logical node prefix (prefix), the receiving device name, and the receiving logical node class (lnClass), so as to achieve efficient flattening of complex tree-like XML structures.

[0020] Furthermore, the Levenshtein edit distance in step (2) is calculated by the state transition equation and is the sum of the minimum single-character insertion, deletion or replacement computational costs required to convert the sending string into the receiving string. The sum of costs is used to quantify and overcome heterogeneous naming deviations of different equipment manufacturers for the same electrical signal.

[0021] Furthermore, step (2) includes:

[0022] After obtaining the extracted terminal description strings, let the sending string be 'a' with length 'm' and the receiving string be 'b' with length 'n'. Use dynamic programming to establish the distance matrix D. a,b The state transition equation for (i,j) is as follows:

[0023] ,

[0024] Where, variable i: character position pointer of the sending string, representing the index of the i-th character of the first string (let's call it a); variable j: character position pointer of the receiving string, representing the index of the j-th character of the second string (b); variable C: cost penalty score for character substitution, the substitution cost constant in the state transition equation;

[0025] After obtaining the edit distance through the state transition equation, a one-dimensional test feature vector X is constructed by combining it with the Boolean matching degree of the logical node prefix. test。

[0026] Furthermore, in step (3), the feature vector is input into the KNN model trained from the historical standard SCD file. Let K be the number of nearest neighbor samples. The model selects the K nearest samples by calculating the Euclidean distance and outputs the posterior confidence probability P of classifying the connection as a valid connection.

[0027]

[0028] Where I is the indicator function and y is the classification label, the system automatically decides whether to establish a connection or execute a latching interception by setting a critical safety threshold.

[0029] Furthermore, the bidirectional difference comparison operation logic in step (4) includes:

[0030] Subtract the field-running joint keyword set from the standard configuration joint keyword set, determine the resulting disjoint elements as missing links, and mark them as fatal hidden dangers that cause the protection device to fail to operate;

[0031] By subtracting the standard configuration set of joint keywords from the set of joint keywords used in the field operation, the resulting disjoint elements are identified as unauthorized illegal links and marked as potential safety violations that could cause the protection device to malfunction.

[0032] Furthermore, step (4) includes:

[0033] Network topology information is mapped to a highly abstract joint key, in the format: Sending IED_Sending Node -> Receiving IED_Receiving Node.

[0034] Let S be the set of legal keywords obtained by parsing a standard SCD file. std The current keyword set obtained by parsing the real-time CCD configuration file read from the physical memory of the field device is S. rt The system identifies potential problems by performing extremely fast set subtraction:

[0035] Collection of potential hazards related to non-operation:

[0036]

[0037] Collection of potential hazards for accidental operation:

[0038] .

[0039] Furthermore, step (5) includes:

[0040] To verify whether the file has been tampered with by hackers or malicious programs at the byte level, the system introduces the Ethernet standard CRC32 generator polynomial G(x):

[0041] ,

[0042] Where x represents a data bit placeholder in the underlying binary bit stream of the computer, its exponent represents the position index from right to left in the binary sequence, and the coefficients of each term of the polynomial (implicitly 1 or 0) represent the binary value at that position.

[0043] The system will S std With S rt The byte stream is divided by this polynomial to obtain a 32-bit checksum fingerprint hash. CRCThe system performs strict inequality checks:

[0044] ,

[0045] If the above inequality holds, then physical-level hardware locking is triggered.

[0046] Another technical solution provided by this invention: an automated configuration and online verification system for secondary virtual loops suitable for isolated renewable energy power grids, implementing the automated configuration and online verification method described above, including:

[0047] Data source layer: Stores configuration description files for smart electronic devices, including standard SCD files or IED capability description files;

[0048] The algorithm processing layer includes an XML and DOM dimensionality reduction engine, an NLP and KNN matching engine, and a bidirectional difference verification engine. The XML and DOM dimensionality reduction engine skips redundant display tags in XML, traverses the document object model tree, extracts the core attributes of the sender and receiver from the external reference tags under the smart electronic device nodes, and reconstructs and reduces the dimensionality to a one-dimensional or two-dimensional structured feature matrix. The NLP and KNN matching engine calculates the Levenshtein edit distance measurement and inputs it into a KNN classifier pre-trained using a standard substation SCD sample library. The bidirectional difference verification engine constructs a high-dimensional joint key and performs mathematical set hash comparison to replace the underlying line-by-line error checking.

[0049] Application execution layer: used for automatic error correction and establishing connections to generate directed graph holographic secondary virtual circuits, outputting classified alarm logs, and performing self-checks to prevent computational modifications at the underlying running logic. When fingerprints do not match, it forcibly locks out protection.

[0050] The beneficial effects of this invention are:

[0051] (1) Imparting "semantic fault tolerance" capability to the system: This invention breaks through the fatal limitation of "precise string matching" in traditional protection configuration software, and for the first time applies the edit distance of NLP and machine learning models to the power secondary system. It not only solves the problem of broken links caused by non-standard naming of heterogeneous equipment, but also compresses the configuration iteration cycle of new equipment from several months to hours, greatly reducing the curtailment of wind and solar power caused by the lag in strategy adjustment.

[0052] (2) It greatly improves the data processing efficiency of computer systems: It abandons the inefficient XML large text line-by-line comparison mechanism and pioneers the technology of mapping multi-dimensional electrical configurations into dimension-reduced joint keywords. Through direct mathematical set difference operation, it compresses the error checking time of tens of thousands of nodes in complex substations from hours to milliseconds, which meets the technical requirements of improving computing power by optimizing the internal data structure of computers.

[0053] (3) A three-dimensional security protection network combining macro and micro levels was constructed: macroscopically, the test interference line was intercepted through the algorithm model (to prevent accidental operation), and microscopically, the underlying bytes were locked through CRC32 fingerprint verification (to prevent tampering). This two-layer architecture fundamentally eliminates the secondary logic risks of isolated micronetworks and makes up for the shortcomings of traditional devices with large response delays. Attached Figure Description

[0054] Figure 1 This is a schematic diagram of the three-layer overall software architecture of the system of this invention;

[0055] Figure 2 It is a flowchart of intelligent matching and error prevention topology based on multidimensional features and KNN algorithm;

[0056] Figure 3 It is a KNN-based heterogeneous terminal bipartite graph matching and error prevention interception topology graph;

[0057] Figure 4 It is a diagram of set comparison based on high-dimensional joint keywords and underlying CRC anti-tampering mechanism;

[0058] Figure 5 This is the final directed graph interface of the secondary virtual circuit of the intelligent substation generated in the embodiment. Detailed Implementation

[0059] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0060] An automated configuration and online verification method for secondary virtual loops applicable to isolated renewable energy power grids is presented. The method consists of two core stages: "automated configuration" and "online consistency verification." Specifically, it includes the following steps:

[0061] Step 1: Automatic extraction and dimensionality reduction modeling of underlying communication features

[0062] The system automatically reads standard SCD files or IED capability description (ICD) files, skips redundant display tags in XML, traverses the Document Object Model (DOM) tree, and directly and accurately extracts the core attributes of the sender and receiver (including iedName, prefix, lnClass, etc.) from the external reference (ExtRef) tags under the intelligent electronic device (IED) node, and reconstructs and reduces them to a one-dimensional or two-dimensional structured feature matrix.

[0063] Step 2: Dynamic Programming Calculation and Multidimensional Feature Construction

[0064] For heterogeneous terminal strings named by different manufacturers, a dynamic programming algorithm is used to calculate the Levenshtein edit distance between the sending and receiving strings. This edit distance is then combined with the Boolean matching degree (0 or 1) of the logical node prefix to construct a multi-dimensional test feature vector.

[0065] The data is then input into a KNN classifier pre-trained using a standard substation SCD sample library to calculate the posterior confidence probability of a valid connection for the virtual terminal block combination. When the confidence level exceeds a preset safety threshold, the system determines it as a valid match and automatically establishes a virtual connection.

[0066] Step 3: Anti-malfunction interception and holographic topology reconstruction

[0067] The multidimensional test feature vector is input into a K-Nearest Neighbor (KNN) machine learning model pre-trained using historical legitimate data to calculate the posterior confidence probability of the legitimate connection of the virtual terminal combination. When the confidence is greater than a preset safety threshold, a virtual connection is automatically established. For interfering terminal combinations (such as unauthorized test lines) whose KNN classifier output confidence is lower than the safety threshold, the system performs error prevention blocking, forcibly intercepts and marks them as isolated nodes. For legitimate connections, a visualized holographic topology of a directed graph (Digraph) with the sending end as the source node and the receiving end as the target node is automatically generated using computer graph theory algorithms.

[0068] Step 4: Bidirectional difference set comparison and verification based on high-dimensional joint keywords

[0069] During the verification phase, the system extracts configuration features from the standard design baseline (SCD) and the memory image (CCD) of the field IED device, and extracts reference information from the standard configuration file and the actual field operation configuration memory to construct dimensionality-reduced joint keywords. By performing a bidirectional difference comparison operation on the two sets of joint keywords, the abnormal feature sets of missing links and illegally added links are accurately isolated. Using the bidirectional difference operation rules in mathematical set theory, the standard set is subtracted from the real-time set to extract "missing link (no-operation risk)", and the real-time set is subtracted from the standard set to extract "illegally added link (maloperation risk)", thereby generating accurate classification diagnosis alarms.

[0070] Step 5: Anti-tampering self-test based on CRC32 hardware-level fingerprint

[0071] Cyclic Redundancy Check (CRC) hash polynomial operations are performed on the underlying byte streams of the two sets of joint keywords to generate unique hexadecimal digital fingerprints. The system performs online cyclic verification. When a mismatch is detected between the standard fingerprint and the real-time running fingerprint, it is determined that the underlying operating logic has been illegally tampered with. The highest level alarm is immediately issued, and electrical commands to lock the relevant protection devices are forcibly sent.

[0072] The Levenshtein edit distance in step 2 is calculated using the state transition equation. It is the sum of the minimum single-character insertion, deletion, or replacement computational costs required to convert the sending string into the receiving string. This sum of costs is used to quantify and overcome heterogeneous naming discrepancies for the same electrical signal from different equipment manufacturers.

[0073] Levenshtein distance calculation model for heterogeneous strings

[0074] After obtaining the extracted terminal description strings, let the sending string be 'a' (length m) and the receiving string be 'b' (length n). Use dynamic programming to establish the distance matrix D. a,b (i,j), its state transition equation is as follows:

[0075] (1)

[0076] Where, variable i: character position pointer of the sending string, representing the index of the i-th character of the first string (let's call it a); variable j: character position pointer of the receiving string, representing the index of the j-th character of the second string (b); variable C: cost penalty score for character substitution, the substitution cost constant in the state transition equation.

[0077] After calculating the edit distance using the above formula, and combining it with the Boolean matching degree of the logical node prefixes (1 for identical, 0 for different), a one-dimensional test feature vector X is constructed. test .

[0078] In step 3, the feature vector is input into the KNN model trained using historical standard SCD files. Let K be the number of nearest neighbors. The model selects the K nearest neighbors by calculating the Euclidean distance and outputs the posterior confidence probability P of classifying the line as a valid connection (classification label y=1).

[0079] (2)

[0080] Where I is the indicator function. By setting a critical safety threshold, the system can automatically decide whether to establish a connection or execute a latching interception.

[0081] The bidirectional difference set comparison operation logic in step 4 includes:

[0082] Subtract the field-running joint keyword set from the standard configuration joint keyword set, determine the resulting disjoint elements as missing links, and mark them as fatal hidden dangers that cause the protection device to fail to operate;

[0083] By subtracting the standard configuration set of joint keywords from the set of joint keywords used in the field operation, the resulting disjoint elements are identified as unauthorized illegal links and marked as potential safety violations that could cause the protection device to malfunction.

[0084] Traditional line-by-line text validation tends to break down computationally when faced with tens of thousands of configuration codes. In step 4, network topology information is mapped to highly abstract union keys (format: Sending IED_Sending Node -> Receiving IED_Receiving Node).

[0085] Let S be the set of legal keywords obtained by parsing the standard SCD file issued by the design institute. std The current keyword set obtained by parsing the real-time CCD configuration file read from the physical memory of the field device is S. rt The system identifies potential problems by performing extremely fast set subtraction:

[0086] Collection of potential hazards related to non-operation:

[0087] (3)

[0088] Collection of potential hazards for accidental operation:

[0089] (4)

[0090] In step 5, to verify whether the file has been tampered with by hackers or malicious programs at the byte level, the system introduces the Ethernet standard CRC32 generator polynomial G(x):

[0091] (5)

[0092] Where x represents a data bit placeholder in the underlying binary bit stream of the computer, its exponent represents the position index from right to left in the binary sequence, and the coefficients of each term of the polynomial (implicitly 1 or 0) represent the binary value at that position.

[0093] The system will S std With S rt The byte stream is divided by this polynomial to obtain a 32-bit checksum fingerprint hash. CRC The system performs strict inequality checks:

[0094] (6)

[0095] If the above inequality holds, then physical-level hardware locking is triggered.

[0096] Embodiments of the present invention also provide an automated configuration and online verification system for secondary virtual loops suitable for isolated renewable energy power grids, implementing the automated configuration and online verification method described above, including:

[0097] Data source layer: Stores configuration description files for smart electronic devices, including standard SCD files or IED capability description files;

[0098] The algorithm processing layer includes an XML and DOM dimensionality reduction engine, an NLP and KNN matching engine, and a bidirectional difference verification engine. The XML and DOM dimensionality reduction engine skips redundant display tags in XML, traverses the document object model tree, extracts the core attributes of the sender and receiver from the external reference tags under the smart electronic device nodes, and reconstructs and reduces the dimensionality to a one-dimensional or two-dimensional structured feature matrix. The NLP and KNN matching engine calculates the Levenshtein edit distance measurement and inputs it into a KNN classifier pre-trained using a standard substation SCD sample library. The bidirectional difference verification engine constructs a high-dimensional joint key and performs mathematical set hash comparison to replace the underlying line-by-line error checking.

[0099] Application execution layer: used for automatic error correction and establishing connections to generate directed graph holographic secondary virtual circuits, outputting classified alarm logs, and performing self-checks to prevent computational modifications at the underlying running logic. When fingerprints do not match, it forcibly locks out protection.

[0100] Verification using real-world test data:

[0101] In a practical application of a 100% renewable energy island power station, due to the different manufacturers of the purchased equipment, there are many heterogeneous naming conventions. The system extracts some test data and performs calculations using the above mathematical model. The results are shown in Table 1.

[0102] Table 1. Test Data for Virtual Terminal Feature Extraction and AI Fault Tolerance Judgment

[0103]

[0104] As can be seen from Table 1 and the system's bipartite graph firewall topology, the system not only accurately identified and tolerated heterogeneous data with missing or extra letters, generating correct connections, but also triggered forced interception with extremely low confidence (0.12) when faced with interference lines left over from construction, such as sample 04, without generating any false topology, thus ensuring the purity of the system.

[0105] Verification based on actual hazard interception data:

[0106] During the commissioning phase of the aforementioned stations, the verification engine of this invention was used for automatic comparison, and the output diagnostic logic and system control actions are shown in Table 2.

[0107] Table 2. Results of Bidirectional Difference Verification for Multi-Source Configuration File Consistency

[0108]

[0109] As can be seen from Table 2, this invention, through dimensionality reduction set operation, instantly and highly targetedly classifies ordinary configuration "inconsistencies" into two major categories that threaten power grid security: "refusal to operate" and "false operation." Combined with the underlying hexadecimal fingerprint mismatch, it directly executes the physical interlocking defense line, enabling the digital asset security of smart substations to reach a new level.

[0110] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention without departing from the spirit and scope of the present invention.

Claims

1. A method for automated configuration and online verification of secondary virtual loops suitable for isolated renewable energy power grids, characterized in that, Includes the following steps: (1) Automatic extraction and dimensionality reduction modeling of underlying communication features Read and parse the configuration description file of the intelligent electronic device, extract the communication access block and external reference information, and generate a two-dimensional data matrix containing the feature attributes of the sender and receiver by dimensionality reduction; (2) Dynamic programming calculation and multidimensional feature construction A distance matrix is ​​established using a dynamic programming algorithm to calculate the Levenshtein edit distance between the sender description string and the receiver description string to be connected, and a multi-dimensional test feature vector is constructed by combining logical node prefix attributes, etc. (3) Intelligent fault-tolerant matching and locking based on KNN The multidimensional test feature vector is input into a KNN machine learning model pre-trained using historical legitimate data to calculate the posterior confidence probability of the legitimate connection of the virtual terminal combination; when the confidence is greater than a preset safety threshold, a virtual connection relationship is automatically established, and a quadratic virtual loop holographic topology is generated based on a directed graph algorithm; combinations with confidence below the safety threshold are forcibly intercepted to prevent errors. (4) Bidirectional difference set comparison and verification based on high-dimensional joint keywords The reference information in the standard configuration file and the actual on-site running configuration memory is extracted to construct a dimension-reduced joint key. By performing a bidirectional difference comparison operation on the two sets of joint keys, the abnormal feature sets of missing links and illegally added links are accurately isolated. (5) Anti-tampering low-level self-test based on CRC32 hardware fingerprint The data streams of the standard configuration and the field operation configuration are subjected to cyclic redundancy check polynomial hash operation to generate hexadecimal digital fingerprints. When the fingerprints do not match, the underlying tampering alarm is triggered and the instruction to lock the relevant protection devices is issued.

2. The method for automated configuration and online verification of secondary virtual loops applicable to isolated renewable energy power grids according to claim 1, characterized in that, The feature attributes extracted in step (1) include at least: the name of the sending device, the prefix of the sending logical node, the name of the receiving device, and the class of the receiving logical node.

3. The method for automated configuration and online verification of secondary virtual loops applicable to isolated renewable energy power grids according to claim 1, characterized in that, The Levenshtein edit distance in step (2) is calculated by the state transition equation and is the sum of the minimum single-character insertion, deletion or replacement computational costs required to convert the sending string into the receiving string. The sum of costs is used to quantify and overcome heterogeneous naming deviations of different equipment manufacturers for the same electrical signal.

4. The method for automated configuration and online verification of secondary virtual loops applicable to isolated renewable energy power grids according to claim 1, characterized in that, Step (2) includes: After obtaining the extracted terminal description strings, let the sending string be 'a' with length 'm' and the receiving string be 'b' with length 'n'. Use dynamic programming to establish the distance matrix D. a,b The state transition equation for (i,j) is as follows: ， Where: variable i: character position pointer of the sending string, representing the index of the i-th character of the first string; variable j: character position pointer of the receiving string, representing the index of the j-th character of the second string; variable C: cost penalty score for character substitution, the substitution cost constant in the state transition equation; After obtaining the edit distance through the state transition equation, a one-dimensional test feature vector X is constructed by combining it with the Boolean matching degree of the logical node prefix. test。 5. The method for automated configuration and online verification of secondary virtual loops applicable to isolated renewable energy power grids according to claim 1, characterized in that, In step (3), the feature vector is input into the KNN model trained from the historical standard SCD file. Let K be the number of nearest neighbor samples. The model selects the K nearest samples by calculating the Euclidean distance and outputs the posterior confidence probability P of classifying the connection as a valid connection. ， Where I is the indicator function and y is the classification label, the system automatically decides whether to establish a connection or execute a latching interception by setting a critical safety threshold.

6. The method for automated configuration and online verification of secondary virtual loops applicable to isolated renewable energy power grids according to claim 1, characterized in that, The bidirectional difference comparison operation logic in step (4) includes: Subtract the field-running joint keyword set from the standard configuration joint keyword set, determine the resulting disjoint elements as missing links, and mark them as fatal hidden dangers that cause the protection device to fail to operate; By subtracting the standard configuration set of joint keywords from the set of joint keywords used in the field operation, the resulting disjoint elements are identified as unauthorized illegal links and marked as potential safety violations that could cause the protection device to malfunction.

7. The method for automated configuration and online verification of secondary virtual loops applicable to isolated renewable energy power grids according to claim 1, characterized in that, Step (4) includes: Network topology information is mapped to a highly abstract joint key, in the format: Sending IED_Sending Node -> Receiving IED_Receiving Node. Let S be the set of legal keywords obtained by parsing a standard SCD file. std The current keyword set obtained by parsing the real-time CCD configuration file read from the physical memory of the field device is S. rt The system identifies potential problems by performing extremely fast set subtraction: Collection of potential hazards related to non-operation: ， Collection of potential hazards for accidental operation: 。 8. The method for automated configuration and online verification of secondary virtual loops applicable to isolated renewable energy power grids according to claim 7, characterized in that, Step (5) includes: To verify whether the file has been tampered with by hackers or malicious programs at the byte level, the system introduces the Ethernet standard CRC32 generator polynomial G(x): ， Where x represents a data bit placeholder in the computer's underlying binary bit stream. The system will S std With S rt The byte stream is divided by this polynomial to obtain a 32-bit checksum fingerprint hash. CRC The system performs strict inequality checks: ， If the above inequality holds, then physical-level hardware locking is triggered.

9. An automated configuration and online verification system for secondary virtual loops suitable for isolated renewable energy power grids, implementing the automated configuration and online verification method as described in any one of claims 1-8, characterized in that, include: Data source layer: Stores configuration description files for smart electronic devices, including standard SCD files or IED capability description files; Algorithm processing layer: includes XML and DOM dimensionality reduction engine, NLP and KNN matching engine and bidirectional difference verification engine. The XML and DOM dimensionality reduction engine is used to skip redundant display tags in XML, traverse the document object model tree, extract the core attributes of the sender and receiver in the external reference tags under the smart electronic device node, and reconstruct and reduce the dimensionality to a one-dimensional or two-dimensional structured feature matrix. The NLP and KNN matching engine is used to calculate the Levenshtein edit distance measurement, which is then input into a KNN classifier pre-trained using a standard substation SCD sample library; the bidirectional difference verification engine is used to construct a high-dimensional joint key to perform mathematical set hash comparison instead of the underlying line-by-line error checking. Application execution layer: used for automatic error correction and establishing connections to generate directed graph holographic secondary virtual circuits, outputting classified alarm logs, and performing self-checks to prevent computational modifications at the underlying running logic. When fingerprints do not match, it forcibly locks out protection.

Images