Semantic Recognition-Based Configuration Transfer Method for Industrial Control Systems
By generating standard intermediate structures and reverse parsing rules, the problem of difficulty in recognizing real control semantics in the configuration migration of industrial control systems is solved, and the high fault tolerance and independent closed-loop verification of the migration results are achieved, ensuring the reliability and security of the migration process.
Patent Information
- Application Number
- CN202611022224.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-10
- Publication Date
- 2026-11-10
AI Technical Summary
In the process of configuring and migrating existing industrial control systems, the difficulties in recognizing real control semantics and the lack of independent verification of migration results lead to problems such as variable matching errors, distortion of function block behavior, and omission of topology connections during the migration process, which affect the reliability and security of the project.
By acquiring the source control system files and generating an audit snapshot containing hash fingerprints and timestamps, parsing it into a standard intermediate structure, inferring and filling missing fields of variable objects, constructing the source control logic topology diagram, and generating target configuration files based on the target system configuration rule base, the reverse parsing rules are used for verification, and multi-dimensional consistency verification and risk assessment are performed to ensure the accuracy and security of the migration results.
It achieves accurate identification and unified parsing of the source control logic topology, avoids functional distortion and connection omissions during the migration process, ensures high fault tolerance and independent closed-loop verification of the migration results, and improves the reliability and security of the migration process.
Smart Images

Figure CN123070113A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of industrial automation and industrial control system configuration migration technology, specifically to a method for industrial control system configuration migration based on semantic recognition. Background Technology
[0002] With the increasing demand for industrial control system upgrades, heterogeneous platform replacements, and continuous operation and maintenance of existing projects, industrial control system configuration migration has become an important technical aspect in the process industry automation field. The smooth switching and continuous operation of control projects between different vendors and platforms rely on the extraction, understanding, conversion, and verification of the source control system configuration data. In the configuration migration process of industrial control systems, existing methods often rely on manually reviewing point tables, logic diagrams, alarm configurations, and human-machine interface files, and then rebuilding them item by item according to the rules of the target platform. However, the configuration files of the source control system are usually scattered, have inconsistent formats, and commonly suffer from non-standard variable naming, missing fields, fragmented cross-file associations, and functional block definitions that exceed the preset similarity threshold. This makes it difficult to accurately identify the true control semantics and logical relationships during the migration process. Especially in complex industrial scenarios such as thermal power generation, if only variable names or static mapping rules are relied upon for conversion, problems such as variable matching errors, distortion of functional block behavior, omission of topology connections, and inconsistencies between alarms and human-machine bindings are likely to occur, thus affecting the reliability and safety of the project after migration. Unified parsing of multi-source configuration files of existing source control systems can ensure the quality of industrial control system migration and ensure the safe operation of critical control logic. Summary of the Invention
[0003] The purpose of this invention is to provide a semantic recognition-based method for industrial control system configuration migration, addressing the following technical problems: Existing industrial control system configuration migration methods suffer from significant shortcomings in addressing the difficulty of recognizing real control semantics and the lack of independent verification of migration results after target engineering generation. There is an urgent need to propose a method that can uniformly parse and construct a logical topology reflecting real control relationships and achieve closed-loop risk management based on reverse parsing and multi-dimensional equivalence verification. The purpose of this invention can be achieved through the following technical solutions: The system acquires the input / output variable table, control logic file, human-machine interface (HMI) file, alarm configuration file, and communication address table of the source control system. It generates an audit snapshot containing a copy of the original file, file hash fingerprint, and timestamp, and parses it into a standard intermediate structure according to preset parsing rules. Based on variable name, variable description, variable unit, range, input / output type, communication address, HMI binding relationship, alarm configuration relationship, and logical connection relationship, it infers and fills in missing fields of variable objects in the standard intermediate structure, and determines the corresponding equipment category, physical quantity type, signal direction, and safety level of the variable object. Based on input triggering, output response, internal parameters, scan cycle, delay, and state holding method, the function block objects in the source control system are unified into standard control function blocks, and the consistency between the ports of the standard control function blocks and the connected variable objects is checked; a source control logic topology diagram is constructed based on variable objects, standard control function blocks, connection relationship objects, human-machine interface objects, and alarm objects, and an integrity check is performed on the source control logic topology diagram. Based on a pre-established target control system configuration rule library, a target configuration file is generated based on nodes and edges that have passed integrity checks. The target control system configuration rule library includes at least an object attribute mapping table specific to the target platform, underlying syntax generation rules, and hardware resource allocation constraints. A reverse parsing rule, different from the generation rule used to generate the target configuration file, is used to parse the target configuration file to construct a target control logic topology diagram. Establish node mapping, edge mapping, and attribute mapping relationships between the source control logic topology diagram and the target control logic topology diagram, and perform variable consistency verification, topology consistency verification, functional consistency verification, timing consistency verification, and external binding consistency verification; determine the risk based on the difference information obtained from the verification, and export the target configuration project file after passing the risk determination.
[0004] Preferably, the process of acquiring and parsing the source control system file includes: generating file hash fingerprints, project batch numbers, and audit snapshots for the received comma-separated files, Extensible Markup Language files, structured text files, or binary intermediate files; parsing according to the field extraction rules, syntax parsing rules, and object mapping rules corresponding to the file format to obtain a unified standard intermediate structure; assigning a unique object identifier to each object in the standard intermediate structure and binding it to the source file location, parsing rule version, and association basis; when source conflicts, missing fields, or duplicate objects are found in subsequent processing, tracing back to the original file based on the unique object identifier and triggering re-parsing.
[0005] Preferably, in the process of inferring, filling, and judging the variable objects, the comprehensive matching score of the variable objects is obtained by weighted summation of the variable name matching score, description matching score, topological location matching score, unit and range matching score, and human-machine interface or alarm association relationship matching score; Each matching score is converted to a value between zero and one according to a preset normalization rule. The comprehensive matching score is the sum of the products of each normalized matching score and the corresponding weight coefficient. The weight coefficient is a preset non-negative number and the sum of the weight coefficients is one. Among them, the variable name matching score is determined based on a preset industry business terminology list. The preset industry business terminology list includes at least equipment category words, physical quantity words, variable abbreviations, synonyms, and alias mapping relationships.
[0006] Preferably, the variable objects are divided into automatic migration objects, review migration objects, pending confirmation placeholder objects, and blocked migration objects based on the comprehensive matching score; when the comprehensive matching score is greater than or equal to the first threshold, it is classified as an automatic migration object; when the comprehensive matching score is less than the first threshold but greater than or equal to the second threshold, it is classified as a review migration object; when the comprehensive matching score is less than the second threshold but greater than or equal to the third threshold, it is classified as a pending confirmation placeholder object; when the comprehensive matching score is less than the third threshold, it is classified as a blocked migration object. Among them, the first threshold, the second threshold, and the third threshold are all preset score thresholds with values between zero and one, and the first threshold is greater than the second threshold, and the second threshold is greater than the third threshold; the placeholder object to be confirmed does not participate in the generation of target project variable configuration, and the migration object is blocked from triggering the pending processing mark.
[0007] Preferably, the unified processing of the function block object does not depend on the name of the function block in the source system, but is identified and unified into the corresponding standard control function block based on the input-output timing relationship, internal parameters and state holding method; after unified processing, cross-checking is performed on the physical quantity type, direction, unit and safety level of the input and output ports of the standard control function block and the connected variable object, and variable re-determination or re-parsing is performed according to the conflict source determined by the cross-check.
[0008] Preferably, the source control logic topology diagram includes a node set, an edge set, and an attribute set. The node set includes variable nodes, function block nodes, human-machine interface nodes, and alarm nodes. The edge set includes data flow edges, control flow edges, display binding edges, and alarm triggering edges. The attribute set includes at least unit, range, scan cycle, delay, polarity, priority, and security level. The integrity check includes at least isolated node check, disconnection check, floating port check, duplicate definition check, abnormal loop check, alarm passive point binding check, and human-machine interface variable binding check. Furthermore, each node and each edge is assigned an object type identifier, timing attribute, and association basis.
[0009] Preferably, the target configuration file is generated only based on nodes and edges that have passed variable object processing, function block unification, and integrity checks; the reverse parsing adopts a reverse parsing rule different from the generation rule to construct a target control logic topology diagram for the target configuration file, and retains the generation rule number, target platform object identifier, and association basis for each node.
[0010] Preferably, the consistency verification of execution variables, topology, function, timing and external binding includes: variable consistency verification, used to check whether the device category, physical quantity type, signal direction and security level of the variables before and after mapping are consistent; topology consistency verification, used to check whether the connection relationship between the variable node in the source control logic topology diagram and another variable node through the function block node exists in the target control logic topology diagram after mapping. Functional consistency check is used to check whether the mapped function blocks have the same output response, state transition relationship and parameter constraints under the same input conditions; timing consistency check is used to check whether the scan cycle difference of the corresponding nodes does not exceed the preset timing difference threshold; external binding consistency check is used to check whether the alarm level and alarm limit of the alarm object and the display attributes and operation permissions of the human-machine interface object are consistent.
[0011] Preferably, the risk assessment is carried out in a graded manner based on the security level, logical category, and scope of impact of the difference information, including: when there is a difference in tripping logic, blocking the export of the target project and generating a review task; when the number of high-impact differences in the regulation loop or sequential control logic reaches a preset threshold, blocking the export of subsequent projects until the simulation verification is completed; when the difference is a difference in human-machine interface color, layout, or annotation, marking it as a display-type difference; and updating the review results to the industry business thesaurus, function block rule base, target platform rule base, and risk rule base.
[0012] Compared with existing technologies, this method generates audit snapshots containing hash fingerprints and unique identifiers for multi-source files, and infers the semantics of variable objects based on weighted summation of multi-dimensional features, and classifies variables into intermediate structures for admission standards. This solves the difficulty of semantic recognition caused by the scattered formats and non-standard naming of source configuration files, and improves the accuracy of real control semantic parsing and the traceability of process data. This method eliminates the reliance on static names of function blocks and controls function blocks according to a unified standard based on timing relationships, internal parameters, and state maintenance methods. At the same time, it constructs a source control logic topology diagram containing nodes such as variables and human-machine interfaces and performs integrity checks, effectively avoiding connection omissions such as functional distortion, disconnection, and isolated nodes during migration, and ensuring the integrity and reliability of control relationships. After generating the target configuration file, this method uses independent reverse parsing rules to construct the target topology map and performs consistency verification with the source topology map, covering multiple dimensions such as variables, functions, and timing. Combined with a graded risk judgment and interception mechanism based on the security level of the difference information, it overcomes the defects of hidden errors in traditional one-way conversion and achieves high fault tolerance and independent closed-loop verification of the migration results. Attached Figure Description
[0013] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0014] Figure 1 A flowchart illustrating a semantic recognition-based industrial control system configuration migration method according to an embodiment of this disclosure is shown. Detailed Implementation
[0015] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.
[0016] like Figure 1 As shown, the semantic recognition-based industrial control system configuration migration method includes: S100: Obtain the input / output variable table, control logic, human-machine interface screen, alarm configuration and communication address file of the source system, generate an audit snapshot containing the original copy, hash fingerprint and timestamp, and parse it into a standard intermediate structure according to the preset parsing rules; S200, based on variable name, description, unit, range, input / output type, communication address, human-machine interface binding, alarm configuration and logical connection relationship, infers and fills the missing fields of variable objects in the standard intermediate structure, and determines its equipment category, physical quantity type, signal direction and safety level; S300, based on input trigger, output response, internal parameters, scan cycle, delay and state holding method, unifies the source system function block object into a standard control function block, and checks the consistency between its port and connected variable object. S400 constructs a source control logic topology graph containing nodes and edges based on variable objects, standard control function blocks, connection relationship objects, human-machine interface objects, and alarm objects, and performs integrity checks. S500 generates a target configuration file based on nodes and edges that have passed integrity checks, according to a preset target system configuration rule library. It then uses a reverse parsing rule, which is different from the rule used to generate the target configuration file, to parse the target configuration file in order to construct a target control logic topology diagram. S600 establishes the node, edge, and attribute mapping relationship between the source control logic topology diagram and the target control logic topology diagram, performs consistency checks on variables, topology, functions, timing, and external bindings, and obtains difference information. S700 performs risk assessment based on the discrepancy information obtained from the verification, and exports the target configuration file after passing the risk assessment.
[0017] The system acquires the multi-source configuration files of the source control system, constructs the source control logic topology diagram and generates the target configuration file based on parsing, semantic recognition, behavior normalization and graphing operations; it then establishes the target control logic topology diagram through reverse parsing, performs equivalence verification and exports the target configuration engineering file based on risk assessment. Among them, multi-source configuration files refer to the underlying data set containing the control logic and human-machine interface of thermal power plants, such as boiler control logic diagrams, turbine input / output point tables, and alarm configuration files; This embodiment introduces a control logic topology graph and netlist equivalence verification mechanism to extract control logic and perform migration under the constraint of incomplete configuration data; Through the above steps, the configuration of heterogeneous industrial control systems is migrated based on the control logic topology diagram and equivalent verification, and a verification mechanism is provided to support the traceability of migration results. In this embodiment, S100 includes: S110 generates file hash fingerprints, project batch numbers, and audit snapshots for received comma-separated value formats, Extensible Markup Language formats, structured text, or binary intermediate files; S120, according to the field extraction rules, syntax parsing rules and object mapping rules corresponding to the file format, parses to obtain a unified standard intermediate structure; S130 assigns a unique object identifier to each object in the standard intermediate structure and binds it to the source file location, parsing rule version, and associated basis; S140: When source conflicts, missing fields, or duplicate objects are found in S200 to S400, backtrack to the original file based on the unique object identifier and trigger re-parsing.
[0018] In S200, the variable name matching score is obtained based on the preset industry business vocabulary, and combined with the description matching score, topology location matching score, unit and range matching score, and human-machine interface or alarm association matching score, the comprehensive matching score of the variable object is determined. The variable name matching score is determined based on a preset industry business terminology list, which includes at least equipment category terms, physical quantity terms, variable abbreviations, synonyms, and alias mapping relationships.
[0019] S200 classifies variable objects into automatic migration objects, review migration objects, pending confirmation placeholder objects, and blocked migration objects based on the comprehensive matching score. When the overall matching score is greater than or equal to the first threshold, it is classified as an automatic migration object; When the comprehensive matching score is less than the first threshold but greater than or equal to the second threshold, it is classified as a transfer object for review. When the overall matching score is less than the second threshold but greater than or equal to the third threshold, it is classified as a placeholder object to be confirmed. When the overall matching score is less than the third threshold, the object is classified as a blocked migration object; Among them, the first threshold, the second threshold, and the third threshold are all preset score thresholds with values between zero and one, and the first threshold is greater than the second threshold, and the second threshold is greater than the third threshold; the placeholder object to be confirmed does not participate in the generation of target configuration file variable configuration, and the migration object is blocked from triggering the pending processing mark.
[0020] Obtain a copy of the original file from the source control system. Based on the extraction and parsing rules corresponding to the file format, standardize and convert the original file copy to generate a standard intermediate structure containing a unique object identifier. The standard intermediate structure refers to the underlying data carrier in a unified format, such as a variable object or function block object in the JavaScript object notation format. It not only contains object properties but also binds the source file location and parsing rule version. The system assigns a unique object identifier to each object and establishes a traceability chain; when an anomaly is detected in a subsequent stage, the system traces back to the original file based on the identifier to trigger re-parsing. Based on the above embodiments, variable objects in the standard intermediate structure are obtained, and missing fields of the variable objects are inferred and filled according to multi-source cross features to determine the comprehensive matching score of the variable objects; the comprehensive matching score is determined by weighted summation of multi-dimensional feature matching scores and corresponding weights to characterize the comprehensive semantic confidence of the variable objects; Among them, the variable name matching score is used to measure the character similarity between the variable name and the preset industry business vocabulary; the description matching score is used to evaluate the vector similarity of the annotation text; the topology location matching score is used to reflect the contextual structure characteristics of the variable in the control logic topology diagram, such as a variable with an unknown name being connected to the process variable port of the proportional-integral-derivative module; the unit and range matching score is used to verify the physical rationality of the engineering range; and the human-machine interface / alarm association matching score is used to verify the correspondence between the human-machine interface display and the alarm limit setting. Each of the above matching scores is multiplied by its corresponding weighting coefficient, for example, set to 0.30, 0.25, 0.20, 0.15 and 0.10 respectively; after calculation, each of the above matching scores is converted into a value between zero and one according to a preset normalization rule; For example, for variable name matching score, the character edit distance between the input name and the corresponding standard words in the preset industry business thesaurus is calculated, and the normalized score is equal to 1 minus the ratio of the edit distance to the maximum length of the two; for topology location matching score, the score is assigned according to the matching level of the connection relationship. If the connected functional block type and port attributes are completely consistent, it is recorded as 1, partially consistent as 0.5, and no consistent features as 0. For the unit and range matching score, if the source range is completely contained within the target allowable range, it is recorded as 1; otherwise, the score is calculated based on the overlap interval ratio. The overlap interval ratio is calculated by dividing the length of the intersection interval between the source range and the target system allowable range by the length of the union interval. For the description matching score, to avoid the inability to perform matrix operations due to feature vector dimension mismatch. A global feature dictionary containing all standard process description terms is constructed. The term frequency-inverse document frequency algorithm is used to segment and weight the annotation text to extract the top M keyword set. This keyword set is then mapped to the global feature dictionary to generate an N-dimensional sparse term frequency vector with a dimension that is strictly consistent with the total number of words N in the global feature dictionary. For feature word dimensions that do not match the keyword set, their component values are padded with zeros. The cosine similarity is calculated based on the dot product of the corresponding elements to obtain the similarity score between zero and one. For the HMI / alarm association matching score, the HMI graphic element identifier and alarm limit attribute bound to the source system variable are extracted and compared with the preset target system's allowed configuration object set. If the HMI display address mapping and alarm upper and lower limits match, it is recorded as 1; if only partially matched, it is recorded as 0.5; and if there is no association configuration at all, it is recorded as 0. Through the above logical flow and business decomposition, the abstract multi-dimensional features are transformed into clear and verifiable quantitative indicators. When the naming standardization in the project falls below the preset standardization threshold, i.e., when the average name hit rate of global variable objects in the preset industry business thesaurus is less than 50%, a weight adaptive adjustment mechanism is triggered: the name matching weight is dynamically reduced by 0.15, and the deducted weight is proportionally distributed and added to the topology location matching weight and the human-machine interface / alarm association relationship matching weight. The specific adjustment formula is as follows:
[0021] in, The weight value is reduced. and These are the initial and adjusted weights of the target's assigned features, respectively. The sum of the initial weights for all accepted features. For multiplication operators, The system adaptively lowers the weight of name matching and correspondingly increases the weight of topology and association configuration in order to determine the physical semantics of variables; Based on the calculated comprehensive matching score, the variable objects are divided into automatic migration objects, review migration objects, pending confirmation placeholder objects, and blocked migration objects. When the comprehensive matching score is greater than or equal to the first threshold, it is classified as an automatic migration object. When the comprehensive matching score is less than the first threshold but greater than or equal to the second threshold, it is classified as a review migration object, which is allowed to participate in generation but has an increased priority for subsequent sampling inspection. When the comprehensive matching score is less than the second threshold but greater than or equal to the third threshold, it is classified as a pending confirmation placeholder object, which only enters the topology map as a placeholder and does not participate in target generation. When the comprehensive matching score is less than the third threshold, it is classified as a blocked migration object, which directly triggers the pending processing mark. This embodiment utilizes the cross-completion of multi-source configuration files and the multi-dimensional confidence quantification evaluation mechanism to coordinately adjust the weights of various features, thereby achieving accurate identification and hierarchical access of variable objects. In this embodiment, in S300, the function block object is identified based on the input-output timing relationship, internal parameters and state holding method, and is unified into the corresponding standard control function block; After unified processing, cross-checks are performed on the input and output ports of the standard control function block and the physical quantity type, direction, unit, and safety level of the connected variable objects. Based on the conflict sources determined by the cross-checks, variable re-determination or re-parsing is performed.
[0022] The source control logic topology diagram constructed by S400 includes a set of nodes, a set of edges, and a set of attributes. The set of nodes includes variable nodes, function block nodes, human-machine interface nodes, and alarm nodes. The set of edges includes data flow edges, control flow edges, display binding edges, and alarm triggering edges. The set of attributes includes at least units, range, scan cycle, delay, polarity, priority, and security level. Integrity checks include at least the following: isolated node checks, disconnection checks, floating port checks, duplicate definition checks, abnormal loop checks, alarm passive point binding checks, and human-machine interface variable binding checks. Furthermore, each node and each side is assigned an object type identifier, a timing attribute, and a relationship basis.
[0023] In S500, the generation of target configuration files is based solely on nodes and edges that have been processed by S200, unified by S300, and checked by S400. The reverse analysis uses a different reverse analysis rule than the generation rule to construct a target control logic topology diagram for the target configuration file, and retains the generation rule number, target platform object identifier and association basis for each node.
[0024] Consistency checks are performed on execution variables, topology, functions, timing, and external bindings, including: Variable consistency check is used to check whether the device category, physical quantity type, signal direction and security level of variables are consistent before and after mapping. Topology consistency check is used to check whether the connection relationship between a variable node in the source control logic topology diagram and another variable node through a function block node exists in the target control logic topology diagram after the corresponding mapping. Functional consistency verification is used to check whether the mapped function blocks have the same output response, state transition relationship and parameter constraints under the same input conditions. Timing consistency check is used to check whether the scan cycle difference of the corresponding node does not exceed the preset timing difference threshold. External binding consistency check is used to check whether the alarm level and alarm limit of the alarm object, as well as the display attributes and operation permissions of the human-machine interface object, are consistent.
[0025] In S700, differential information is handled in a tiered manner based on its security level, logical category, and scope of impact, including: When there are differences in tripping logic, block the export of the target configuration file and generate a review task; When the number of high-impact differences in the regulation loop or sequential control logic reaches a preset threshold, the subsequent export of the target configuration file is blocked until dynamic simulation verification is completed and no logical conflict is confirmed. When the differences are in human-computer interface colors, layouts, or annotations, they are marked as display differences. The review results will be updated in the industry business terminology, function block rule base, target platform rule base, and risk rule base.
[0026] The system acquires functional block objects from the source control system. Based on the input-output timing relationship and internal parameter characteristics, it extracts behavioral features from the functional block objects to determine the corresponding standard control function blocks. A standard control function block refers to a general-purpose control unit with defined logical behavior and decoupled from a specific hardware vendor, such as a standard timer module with specific input triggering and delay response. Based on a behavior-driven feature extraction mechanism, the modules are uniformly normalized according to physical behavioral characteristics such as starting timing after an input state change and setting the output after the timing reaches a set value. In the specific feature extraction process, the input pin triggering conditions, output pin response behavior and internal parameter constraints of the source functional block are constructed into a one-dimensional behavioral feature vector. Specifically, the triggering mode, such as rising edge or falling edge, is mapped to a discrete integer value, the response delay time is mapped to a standardized millisecond value, and the internal state retention is mapped to a binary identifier value, thereby constructing the various dimensional components of the vector. Before calculating the distance, range normalization is performed on each dimension component of the behavior feature vector to map all eigenvalues to a dimensionless interval of zero to one. The weighted squared Euclidean distance between this normalized behavior feature vector and the baseline behavior vectors of each standard control function block in the target control system configuration rule base is then calculated. The specific formula is as follows:
[0027] in, For the weighted squared Euclidean distance For feature dimension, The index number is the feature dimension. and The first The source system behavioral feature components and the target system baseline behavioral feature components after dimensionality normalization. The preset weight coefficients for each dimension of features are set, for example, the trigger mode weight is set to 0.5, the response delay time weight is set to 0.3, and the internal state preservation weight is set to 0.2, while satisfying the normalization constraint. The calculated weighted squared Euclidean distance is used as the similarity distance. The standard control function block with the smallest distance that meets the physical constraints such as port data type matching and maximum allowable delay covering the source system delay parameter is selected as the normalized target object to avoid incomplete execution actions caused by unfounded matching. At the same time, cross-checks are performed on the input and output ports of the standard control function block and the physical quantity type and direction of the connected variable object, and conflicting items with similar behavior but inconsistent context are returned for re-parsing; The above-mentioned approved variable objects and standard control function blocks are obtained. Based on the connection relationship and attribute configuration, the source control logic topology diagram is constructed and an integrity check is performed. The integrity check covers the identification of structural defects such as isolated nodes, broken connections and abnormal loops. When generating the target system configuration, the target configuration file is generated only based on the nodes and edges that have passed the previous steps and is output to the offline isolation sandbox area. The offline isolation sandbox area refers to an independent computing environment that is physically or logically isolated from the current production system network. It is used to perform secure static parsing and dynamic timing simulation analysis on the generated target configuration engineering files without affecting actual production. To avoid logical vulnerabilities caused by the generator's own verification deviation, a reverse mapping table is used to extract the nodes and connection relationships in the target configuration file from the preset target system bottom-level tags to the standard intermediate structured lightweight data exchange format fields. The nodes and connection relationships in the target configuration file are then converted into a unified format of bottom-level data carriers for parsing and reconstructing the target control logic topology. Establish the node and edge mapping relationship between the source control logic topology graph and the target control logic topology graph, and perform equivalence verification covering variables, topology, functions, timing and external bindings; when performing functional consistency verification, apply standardized test stimuli to the functional blocks before and after mapping, such as applying a step trigger signal to the input of a standard delay module, and collect the response sequences of the state flipping of the output ports of the source system functional blocks and the target system functional blocks under a unified test clock source. When comparing the output state transition relationship and parameter constraints of the two, a preset timing tolerance window is introduced, for example, allowing a jitter deviation of ±5 milliseconds. When the output state transition logic of the two is consistent and the difference in response delay falls within this timing tolerance window, they are judged to be functionally consistent. As a further limitation, the verification and comparison of parameter constraints specifically involves checking whether the tuning parameters in the source system function block, such as the dead zone threshold or upper and lower limit values, fall completely within the allowable configuration range of the corresponding parameters of the target system standard control function block. If there is an out-of-bounds or over-limit phenomenon, extreme value truncation processing is triggered to correct the parameters, and a parameter out-of-bounds verification alarm is generated as closed-loop feedback. In timing consistency verification, the scan cycle difference of the corresponding nodes does not exceed the preset timing difference threshold. For example, for the core trip protection logic, this timing difference threshold is strictly less than the absolute value of the preset output jitter tolerance window. If the scheduling cycle difference of the target system exceeds this limit, it is determined to be inequivalent and migration is rejected. Based on the difference information obtained from the equivalence verification, risk classification is determined according to the safety level and the scope of impact. When there is a difference in tripping logic, such as an inconsistency in the main fuel tripping logic, the forced termination mechanism is immediately triggered to block the export of the target project and generate a manual review task. When the high impact difference of the regulation loop reaches the preset number threshold, the export is blocked until the simulation verification is completed. Finally, the confirmation conclusions of manual review are written back to update the rule base and business vocabulary; in the actual deployment of the upgrade and transformation of the control system of thermal power plants, this system runs as an independently deployed offline analysis platform; after the new source data flows into the access module, the system automatically triggers parsing and semantic confidence calculation to form a variable list for hierarchical access; the system completes the normalization of functional blocks based on behavioral characteristics and builds the source control logic topology diagram in memory; After the integrity check is passed, the system automatically generates the target configuration file and places it in the isolation sandbox; the system starts reverse analysis and netlist equivalence verification. Once timing limit exceedance or high-risk interlock protection logic difference is found, the system immediately triggers forced interception on the interface and generates a work order with traceable verification information; the operator performs manual approval and rule correction based on the work order, and the corrected data flows back into the system to trigger a new round of verification. Only after all high-risk differences are completely eliminated and all verifications pass can the system unlock export permissions and output the final target configuration project file that can be downloaded and run. Through the above steps, closed-loop verification and risk management of control logic can be achieved based on behavior normalization, reverse analysis and multi-dimensional equivalence verification.
[0028] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.
Claims
1. A configuration migration method for industrial control systems based on semantic recognition, characterized in that, include: S100: Obtain the input / output variable table, control logic, human-machine interface screen, alarm configuration and communication address file of the source system, generate an audit snapshot containing the original copy, hash fingerprint and timestamp, and parse it into a standard intermediate structure according to the preset parsing rules; S200, based on variable name, description, unit, range, input / output type, communication address, human-machine interface binding, alarm configuration and logical connection relationship, infers and fills the missing fields of variable objects in the standard intermediate structure, and determines its equipment category, physical quantity type, signal direction and safety level; S300, based on input trigger, output response, internal parameters, scan cycle, delay and state holding method, unifies the source system function block object into a standard control function block, and checks the consistency between its port and connected variable object. S400 constructs a source control logic topology graph containing nodes and edges based on variable objects, standard control function blocks, connection relationship objects, human-machine interface objects, and alarm objects, and performs integrity checks. S500: Based on the preset target system configuration rule library, a target configuration file is generated based on the nodes and edges that have passed the integrity check. The target configuration file is then parsed using a reverse parsing rule that is different from the rule used to generate the target configuration file, so as to construct a target control logic topology diagram. S600, establish the node, edge and attribute mapping relationship between the source control logic topology graph and the target control logic topology graph, perform consistency verification of variables, topology, function, timing and external binding and obtain difference information; S700, perform risk assessment based on the difference information obtained from the verification, and export the target configuration file after passing the risk assessment.
2. The method for configuration migration of industrial control systems based on semantic recognition according to claim 1, characterized in that, S100 includes: S110 generates file hash fingerprints, project batch numbers, and audit snapshots for received comma-separated value formats, Extensible Markup Language formats, structured text, or binary intermediate files; S120, according to the field extraction rules, syntax parsing rules and object mapping rules corresponding to the file format, parses to obtain a unified standard intermediate structure; S130 assigns a unique object identifier to each object in the standard intermediate structure and binds it to the source file location, parsing rule version, and associated basis; S140: When source conflicts, missing fields, or duplicate objects are found in S200 to S400, backtrack to the original file based on the unique object identifier and trigger re-parsing.
3. The method for configuration migration of industrial control systems based on semantic recognition according to claim 1, characterized in that, In step S200, the variable name matching score is obtained based on a preset industry business terminology, and combined with the description matching score, topology location matching score, unit and range matching score, and human-machine interface or alarm association matching score, the comprehensive matching score of the variable object is determined. The variable name matching score is determined based on a preset industry business terminology list, which includes at least equipment category terms, physical quantity terms, variable abbreviations, synonyms, and alias mapping relationships.
4. The method for configuration migration of industrial control systems based on semantic recognition according to claim 3, characterized in that, S200 classifies variable objects into automatic migration objects, verification migration objects, pending confirmation placeholder objects, and blocked migration objects based on the comprehensive matching score. When the overall matching score is greater than or equal to the first threshold, it is classified as an automatic migration object; When the comprehensive matching score is less than the first threshold but greater than or equal to the second threshold, it is classified as a review migration object; When the overall matching score is less than the second threshold but greater than or equal to the third threshold, it is classified as a placeholder object to be confirmed. When the overall matching score is less than the third threshold, the object is classified as a blocked migration object; Among them, the first threshold, the second threshold and the third threshold are all preset score thresholds with values between zero and one, and the first threshold is greater than the second threshold, and the second threshold is greater than the third threshold; the placeholder object to be confirmed does not participate in the generation of target configuration file variable configuration, and the migration object is blocked from triggering the pending processing mark.
5. The method for configuration migration of industrial control systems based on semantic recognition according to claim 1, characterized in that, In S300, the function block object is identified based on the input-output timing relationship, internal parameters and state holding method, and then unified into the corresponding standard control function block. After unified processing, cross-checks are performed on the input and output ports of the standard control function block and the physical quantity type, direction, unit, and safety level of the connected variable object. Based on the conflict source determined by the cross-check, variable re-determination or re-parsing is performed.
6. The method for configuration migration of industrial control systems based on semantic recognition according to claim 1, characterized in that, The source control logic topology diagram constructed by the S400 includes a set of nodes, a set of edges, and a set of attributes. The set of nodes includes variable nodes, function block nodes, human-machine interface nodes, and alarm nodes. The set of edges includes data flow edges, control flow edges, display binding edges, and alarm triggering edges. The set of attributes includes at least units, ranges, scan cycles, delays, polarities, priorities, and security levels. The integrity check includes at least the following: isolated node check, disconnection check, floating port check, duplicate definition check, abnormal loop check, alarm passive point binding check, and human-machine interface variable binding check. Furthermore, each node and each side is assigned an object type identifier, a timing attribute, and a relationship basis.
7. The method for configuration migration of industrial control systems based on semantic recognition according to claim 1, characterized in that, In S500, the target configuration file is generated only based on nodes and edges that have been processed by S200, unified by S300, and checked by S400. The reverse parsing uses a reverse parsing rule different from the generation rule to construct a target control logic topology diagram for the target configuration file, and retains the generation rule number, target platform object identifier and association basis for each node.
8. The method for configuration migration of industrial control systems based on semantic recognition according to claim 1, characterized in that, The consistency verification of execution variables, topology, functions, timing, and external bindings includes: Variable consistency check is used to check whether the device category, physical quantity type, signal direction and security level of variables are consistent before and after mapping. Topology consistency check is used to check whether the connection relationship between a variable node in the source control logic topology graph and another variable node via a function block node exists in the target control logic topology graph after a corresponding mapping. Functional consistency verification is used to check whether the mapped function blocks have the same output response, state transition relationship and parameter constraints under the same input conditions. Timing consistency check is used to check whether the scan cycle difference of the corresponding node does not exceed the preset timing difference threshold. External binding consistency check is used to check whether the alarm level and alarm limit of the alarm object, as well as the display attributes and operation permissions of the human-machine interface object, are consistent.
9. The method for configuration migration of industrial control systems based on semantic recognition according to claim 1, characterized in that, In S700, the differential information is handled in a tiered manner based on its security level, logical category, and scope of impact, including: When there are differences in tripping logic, block the export of the target configuration file and generate a review task; When the number of high-impact differences in the regulation loop or sequential control logic reaches a preset threshold, the subsequent export of the target configuration file is blocked until dynamic simulation verification is completed and no logical conflict is confirmed. When the differences are in human-computer interface colors, layouts, or annotations, they are marked as display differences. The review results will be updated in the industry business terminology, function block rule base, target platform rule base, and risk rule base.
