MFT system automatic configuration method and system based on intelligent algorithm

Abstract

The embodiment of the invention provides an MFT system automatic configuration method and system based on an intelligent algorithm, and belongs to the technical field of automatic control. The method comprises the following steps: preprocessing data, and executing value feature and mode extraction; constructing an MFT domain knowledge graph based on the extracted value features and modes, and optimizing a logic structure of an MFT system by using a genetic algorithm based on the MFT domain knowledge graph so as to generate an optimal logic structure by defining a fitness function; based on a reinforcement learning technology and the optimal logic structure, constructing an intelligent decision model by learning historical operation data; and creating a digital twinborn model of the target MFT system based on the intelligent decision model, verifying the configuration of the MFT system based on the digital twinborn model of the target MFT system, and importing the configuration of the MFT system passing the verification into a practical project. According to the scheme, the manual workload and the error probability are remarkably reduced, and high efficiency, intelligence and automation of the MFT system configuration process are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of automatic control, and particularly relates to an automatic configuration method for an MFT system based on intelligent algorithms and an automatic configuration system for an MFT system based on intelligent algorithms. Background Art

[0002] The MFT (Master Fuel Trip) system is a core component in the boiler safety control system of a thermal power generation unit, mainly used for detecting multiple key safety parameters during the operation of the boiler. Once an abnormality is detected, the fuel supply is immediately cut off to prevent boiler explosion or major equipment accidents. Traditional MFT systems mostly rely on manual configuration work for multiple subsystems such as control logic, screens, and databases on the DCS platform. There are many logic points, the structure is complex, the dependence on manual work is high, it is easy to make mistakes during the configuration process, and the efficiency of the verification link is low, affecting the system configuration quality and safety.

[0003] Current automatic configuration methods mainly perform logic splicing through template reuse and semi-automatic scripting tools, but still require a large amount of manual intervention. For example, key links such as setting trip conditions, adjusting logic block parameters, and generating interlocking screens still rely on technicians to complete according to experience, with low efficiency and prone to problems such as incorrect measurement point configuration and inappropriate algorithm selection. In addition, existing solutions lack systematic optimization means and intelligent decision-making mechanisms for logic configuration, and it is difficult to meet the requirements of large-scale units for high reliability, rapid deployment, and intelligent operation and maintenance.

[0004] Especially in the context of frequent new construction or renovation projects, how to improve the configuration efficiency, accuracy, and intelligent level of the MFT system has become an urgent technical problem to be solved. Therefore, there is an urgent need for a new type of automatic configuration method for the MFT system that integrates intelligent algorithms, knowledge modeling, and simulation verification to achieve full-process automation and intelligence from logic design to deployment verification, thereby enhancing the safety and engineering implementation efficiency of the power plant control system. Summary of the Invention

[0005] The purpose of the embodiments of the present invention is to provide an automatic configuration method and system for an MFT system based on intelligent algorithms to at least solve the problems of low efficiency and easy error in the configuration process of the MFT system in the prior art.

[0006] To achieve the above object, a first aspect of the present invention provides an automatic configuration method for an MFT system based on intelligent algorithms, the method comprising: collecting historical operation data of a target MFT system, preprocessing the data, and performing value feature and pattern extraction; constructing an MFT domain knowledge graph based on the extracted value features and patterns, and based on the MFT domain knowledge graph, using a genetic algorithm to optimize the logical structure of the MFT system to generate an optimal logical structure by defining a fitness function; constructing an intelligent decision-making model based on reinforcement learning technology and the optimal logical structure through learning of historical operation data; creating a digital twin model of the target MFT system based on the intelligent decision-making model, and performing verification on the configuration of the MFT system based on the digital twin model of the target MFT system, and importing the configuration of the MFT system that passes the verification into the actual project.

[0007] Optionally, using a genetic algorithm to optimize the logical structure of the MFT system includes: defining a chromosome coding method, representing the logical structure of the MFT system as a gene sequence; designing a fitness function including a logical complexity index, a response time index, and / or a security index; generating a new generation of population through selection, crossover, and mutation operations; and iterating the evolutionary process until an optimal logical structure that meets the requirements is found.

[0008] Optionally, constructing an intelligent decision-making model based on reinforcement learning technology and the optimal logical structure through learning of historical operation data includes: defining a state space and an action space of the MFT system; designing a reward function to encourage correct tripping decisions and punish false tripping and missed tripping; training a decision-making model using a deep Q network or a policy gradient algorithm; and improving learning efficiency and model stability through an experience replay mechanism.

[0009] Optionally, the steps for constructing the MFT domain knowledge graph include: collecting domain knowledge and expert experience information related to the MFT system as initial information; extracting entities, relationships, and attributes from the initial information, and constructing a basic structure of the knowledge graph based on the extracted entities, relationships, and attributes; defining entity types and relationship types in the basic structure of the knowledge graph to obtain an ontology model; and setting an inference mechanism based on rules and semantic reasoning for knowledge in the ontology model to obtain the MFT domain knowledge graph.

[0010] Optionally, the steps for creating the digital twin model include: establishing a geometric model and a physical model of the MFT system as an initial digital model; constructing a data exchange interface for the initial digital model to achieve data synchronization between the physical system and the digital model; performing simulation engine construction based on the initial digital model after completing the construction of the data exchange interface to simulate the operating state of the MFT system under various working conditions; and constructing a visualization interface to obtain the corresponding digital twin model.

[0011] Optionally, based on reinforcement learning technology and the optimal logical structure, an intelligent decision-making model is constructed by learning historical operation data, including: optimizing the trip decision logic of the MFT system, and the corresponding optimization steps include: identifying historical trip events and marking valid trips and false trips; extracting relevant factors and patterns leading to valid trips to construct corresponding training samples; constructing a trip prediction model based on the constructed training samples and machine learning algorithms; configuring an adaptive threshold adjustment mechanism based on the trip prediction model to dynamically adjust the trip threshold according to the operating state.

[0012] Optionally, the configuration of the MFT system is verified based on the digital twin model of the target MFT system, including: based on a preset test scenario library, sequentially performing tests for each preset test scenario, and determining performance parameters under each preset test scenario based on the test results as the regular test results; performing a fault injection test to obtain performance parameters under each fault type as the fault test results; generating a test report based on the configuration corresponding to the regular test results and the configuration corresponding to the fault test results, and performing configuration correction of the MFT system when the test report does not meet the expectations until a test report that meets the expectations is generated.

[0013] Optionally, the configuration of the MFT system that passes the verification is imported into the actual project, including: generating a standardized configuration file for the configuration of the MFT system that passes the verification; wherein, the standardized configuration file includes a logic diagram, a screen, and / or a database; performing a configuration consistency check on the standardized configuration file, and after the consistency check passes, transmitting the configuration file to the target system based on a secure channel to perform configuration deployment; in response to the completion signal of the configuration deployment execution step, performing affiliated result verification, and outputting a corresponding deployment completion indication signal after the verification passes.

[0014] The second aspect of the present invention provides an automatic configuration system for an MFT system based on an intelligent algorithm, the system includes: a collection unit for collecting historical operation data of the target MFT system, preprocessing the data, and performing value feature and pattern extraction; a processing unit for constructing an MFT domain knowledge graph based on the extracted value features and patterns, and using a genetic algorithm to optimize the logical structure of the MFT system based on the MFT domain knowledge graph to generate an optimal logical structure by defining a fitness function; a model construction unit for constructing an intelligent decision-making model by learning historical operation data based on reinforcement learning technology and the optimal logical structure; a deployment unit for creating a digital twin model of the target MFT system based on the intelligent decision-making model, verifying the configuration of the MFT system based on the digital twin model of the target MFT system, and importing the configuration of the MFT system that passes the verification into the actual project.

[0015] On the other hand, the present invention provides a computer-readable storage medium, on which instructions are stored, and when the instructions run on a computer, the computer is caused to execute the above-mentioned automatic configuration method of the MFT system based on intelligent algorithms.

[0016] Through the above technical solutions, the solution of the present invention collects and analyzes the historical operation data of the MFT system, extracts key features and operation modes, and realizes an in-depth understanding of the system behavior. Based on the knowledge graph and genetic algorithm, the automatic optimization of the MFT logic structure can be carried out, and the rationality of the configuration logic and the system performance can be improved. The reinforcement learning technology is introduced to construct an intelligent decision-making model, making the tripping decision more accurate and adaptive, and significantly reducing the risks of false tripping and missed tripping. The digital twin technology is combined to simulate and verify the configuration results, ensuring that the configuration meets the requirements of various operation and fault scenarios before deployment, thereby improving the correctness of the configuration and the reliability of system deployment, significantly reducing the manual workload and error probability, and realizing the high efficiency, intelligence and automation of the MFT system configuration process.

[0017] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent specific embodiments section. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings are used to provide a further understanding of the embodiments of the present invention, and constitute a part of the specification, and are used to explain the embodiments of the present invention together with the following specific embodiments, but do not constitute a limitation to the embodiments of the present invention. In the drawings: Figure 1 is a flowchart of the steps of the automatic configuration method of the MFT system based on intelligent algorithms provided by an embodiment of the present invention; Figure 2 is a system structure diagram of the automatic configuration system of the MFT system based on intelligent algorithms provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] The following will describe in detail the specific embodiments of the present invention with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0020] Figure 1 is a method flowchart of the automatic configuration method of the MFT system based on intelligent algorithms provided by an embodiment of the present invention. As Figure 1 shown, an embodiment of the present invention provides an automatic configuration method of the MFT system based on intelligent algorithms, and the method includes: Step S10: Collect the historical operation data of the target MFT system, preprocess the data, and perform value feature and pattern extraction.

[0021] Specifically, first, comprehensively collect the historical operation data of the target MFT system. The historical data includes but is not limited to real-time values of measurement points, alarm records, trip event logs, operating condition parameters, maintenance operation records, etc. These data usually come from a distributed control system (DCS) or a historical database server, and have characteristics such as large volume, complex structure, and inconsistent data quality. To ensure the accuracy of subsequent modeling and analysis, preprocessing operations need to be performed on the collected raw data. Specifically, it includes steps such as outlier removal, missing value filling, signal denoising, time alignment, and normalization. Among them, outlier removal can be achieved based on statistical analysis or density detection algorithms to remove abnormal points that may be caused by sensor failures or communication interruptions; missing value filling can use the moving average method or interpolation algorithms for completion; signal denoising can use wavelet transform or filter methods to eliminate the influence of high-frequency noise and improve data quality.

[0022] After completing data cleaning, further extract value features and perform pattern recognition on the historical data. The so-called value features refer to variables highly relevant to the MFT trip decision and their evolution trends, such as key parameters like fuel flow, furnace negative pressure, and flame signals; while pattern extraction uses data mining techniques such as clustering analysis, principal component analysis (PCA), and correlation analysis to identify typical operating patterns, fault evolution paths, and trip precursor rules existing in the historical data.

[0023] Through the above feature extraction and pattern recognition processes, not only can it provide structured entity and attribute information for the subsequent construction of the knowledge graph, but also serve as an important reference basis for optimizing the MFT logic structure. This automated analysis method based on historical data breaks through the limitations of traditional reliance on manual experience judgment and significantly improves the comprehensiveness of system cognition and data-driven capabilities.

[0024] Based on the solution of the present invention, this process can accurately extract data features with decision-making value for the trip logic design of the MFT system, significantly reducing the workload and misjudgment rate of manual analysis; at the same time, the operating patterns extracted based on objective data have high repeatability and system adaptability, laying a solid data foundation for the training of subsequent intelligent models, system logic optimization, and digital twin construction, thereby improving the scientificity, intelligence, and reliability of the entire MFT system configuration plan.

[0025] Step S20: Construct an MFT domain knowledge graph based on the extracted value features and patterns, and based on the MFT domain knowledge graph, use a genetic algorithm to optimize the logic structure of the MFT system to generate an optimal logic structure by defining a fitness function.

[0026] Specifically, the genetic algorithm is used to optimize the logical structure of the MFT system, including: defining the chromosome encoding method to represent the logical structure of the MFT system as a gene sequence; designing a fitness function including logical complexity index, response time index, and / or security index; generating a new generation of population through selection, crossover, and mutation operations; and iterating the evolutionary process until the optimal logical structure that meets the requirements is found.

[0027] Furthermore, based on the reinforcement learning technology and the optimal logical structure, an intelligent decision-making model is constructed by learning historical operation data, including: defining the state space and action space of the MFT system; designing a reward function to encourage correct tripping decisions and punish false tripping and missed tripping; training the decision-making model using a deep Q-network or policy gradient algorithm; and improving the learning efficiency and model stability through an experience replay mechanism.

[0028] In the embodiment of the present invention, a knowledge graph in the MFT field is constructed based on the value features and typical patterns extracted from the operation data of the MFT system. The knowledge graph is a means of knowledge modeling that expresses the key objects, attributes, and their relationships in the MFT system in the form of a graph structure, aiming to structurally and semantically represent the knowledge in the field and support subsequent logical structure optimization and intelligent decision-making.

[0029] Specifically, the construction of the knowledge graph first extracts entities (such as measurement points, logic blocks, protection conditions, tripping methods, etc.), attributes (such as measurement point types, tripping levels, number of associated conditions), and relationships between entities (such as dependency relationships, trigger relationships, interlock relationships, etc.) from MFT system engineering data, expert experience, and existing project templates. On this basis, an ontology model is defined, including entity category hierarchies, relationship types, constraint rules, etc. Then, semantic reasoning technology is used to expand the knowledge graph, for example, to deduce hidden dependencies or linkage logics based on existing relationships, further enhancing the integrity and expressive power of the knowledge graph.

[0030] The constructed knowledge graph not only serves as the basis for system structure modeling but also drives the optimization of the MFT logical structure. In the present invention, based on the structural information and semantic constraints of the knowledge graph, the genetic algorithm is combined to globally optimize and search the logical structure of the MFT system to generate an optimal configuration plan that meets multiple constraint conditions.

[0031] The specific process of using the genetic algorithm for optimization includes the following steps: 1) Chromosome encoding: Structurally encode the logical structure in the MFT system (such as tripping logic chains, protection block arrangements, signal combination methods, etc.) and convert it into a gene sequence representation. For example, each tripping condition is represented by a set of binary or integer-type codes to represent its structure type (single point, two-out-of-two, etc.), the measurement point numbers used, the logic block type, etc.

[0032] 2) Fitness function design: Combine the actual engineering requirements and system performance requirements to construct a comprehensive fitness function. The fitness function may include: Logical complexity metrics: Such as the number of logic blocks, wiring complexity, etc., which reflect the implementation difficulty; Response time metrics: Calculate the signal transmission path length based on the logical structure to evaluate the response delay; Safety metrics: Evaluate the tripping accuracy rate, false tripping probability, etc. under specific working conditions.

[0033] 3) Population evolution: Initialize multiple logical structure individuals as the initial population, and generate a new generation of structures through selection, crossover, and mutation operations, evolving generation by generation to optimize the fitness value. Selection can use methods such as roulette wheel, tournament, etc., the crossover method can be single-point / multi-point crossover, and the mutation operation controls the perturbation of key parameters.

[0034] 4) Termination condition judgment: When the number of iterations reaches the set upper limit or the optimal fitness value converges, terminate the evolution process, and output the current optimal logical structure as the optimization result.

[0035] Through the above genetic algorithm optimization process, it is possible to automatically search for a logical structure that takes into account safety, complexity, and response performance among a large number of possible logical configurations, effectively avoiding the blind spots and local optimum problems of manual design.

[0036] Furthermore, the construction steps of the MFT domain knowledge graph include: collecting domain knowledge and expert experience information related to the MFT system as the initial information; extracting entities, relationships, and attributes from the initial information, and constructing the basic structure of the knowledge graph based on the extracted entities, relationships, and attributes; defining entity types and relationship types in the basic structure of the knowledge graph to obtain an ontology model; setting an inference mechanism based on rules and semantic reasoning for the ontology model to obtain the MFT domain knowledge graph.

[0037] In the embodiments of the present invention, to achieve semantic understanding and knowledge reuse of the logical structure of the MFT system, an MFT domain knowledge graph is first constructed. The construction process includes multiple stages, and the first is the collection of initial knowledge. By systematically sorting out engineering materials, control logic design documents, standardized templates, trip rule specifications, etc. related to the MFT system, and combining the experience knowledge of front-line engineers and experts, an initial knowledge set that coexists in structured and unstructured forms is formed.

[0038] Next, based on natural language processing and rule mining algorithms, entities, relationships, and attributes of the above information are extracted. For example, "protection conditions", "tripping measurement points", "control modules", "logic blocks", etc. are abstracted as entities, and the corresponding attributes may include measurement point types, signal sources, circuits to which they belong, etc.; relationships may include semantic connections such as "dependency", "trigger", "inclusion", "priority", etc. By constructing triples (entity - relationship - entity), the basic structure of the knowledge graph is formed.

[0039] On the basic structure, an ontology model is further defined, including entity categories (such as tripping points, logic units, module interfaces, etc.) and relationship types (such as causal relationships, trigger sequences, etc.), forming a unified semantic framework and classification system.

[0040] To enhance the reasoning ability of the knowledge graph, a rule base and a semantic reasoning mechanism are also introduced. Based on the rules defined by Description Logic, automatic judgments such as "if the tripping condition depends on any two of the three measurement points being over - limit, then its relationship is of the 'two - out - of - three' type" can be realized, so as to complete or verify the missing information in the logical structure.

[0041] Based on the solution of the present invention, this knowledge graph transforms the knowledge expression of the MFT system from traditional static documents into a structured and computable graph model, supporting the automatic generation of logic configurations, semantic verification, and intelligent retrieval. Through the semantic reasoning mechanism, potential logical errors and redundant configurations can also be discovered, thereby improving the rationality, standardization, and reuse efficiency of the system's logical structure, and providing a solid knowledge foundation for subsequent logic optimization and intelligent decision - making.

[0042] Step S30: Based on reinforcement learning technology and the optimal logical structure, an intelligent decision - making model is constructed by learning historical operation data.

[0043] Specifically, the tripping decision logic of the MFT system is optimized, and the corresponding optimization steps include: identifying historical tripping events and marking valid trippings and false trippings; extracting relevant factors and patterns leading to valid trippings to construct corresponding training samples; constructing a tripping prediction model based on the constructed training samples and machine learning algorithms; configuring an adaptive threshold adjustment mechanism based on the tripping prediction model to dynamically adjust the tripping threshold according to the operating state.

[0044] In the embodiment of the present invention, to further improve the response ability of the MFT system and the intelligence level of tripping decisions, based on reinforcement learning technology combined with the optimal logical structure, an intelligent decision - making model is constructed by learning historical operation data to realize the dynamic optimization and adaptive adjustment of tripping strategies.

[0045] Specifically, it is first necessary to comprehensively identify and classify the tripping events during the historical operation of the MFT system. The tripping event data usually comes from historical alarm records, operation logs, DCS real-time databases, and accident tracing systems. Through event matching and time series analysis, tripping behaviors can be classified into two categories: "valid tripping" and "false tripping". Among them, valid tripping means that the system correctly triggers a trip under the condition of real safety risks; false tripping is an unnecessary trip caused by logic errors, measurement point failures, or critical jitters in the system. This classification result, as the training data label, is directly related to the accuracy of model training.

[0046] On this basis, further extract the key features and evolution patterns that lead to valid tripping from the original operation data. The feature extraction process includes but is not limited to: Multi-variable signal trend analysis (such as the change rate of temperature, pressure, and flow); Joint index construction (such as the linkage trend between furnace negative pressure and flame signal); Mutation detection and early warning pattern recognition (such as statistical deviation analysis based on a sliding window).

[0047] Through the above methods, a training sample is constructed by combining the "feature sequence" before tripping and the event label, which is used to train the intelligent prediction model. Subsequently, based on the constructed training data, machine learning and reinforcement learning techniques are used to construct a tripping prediction model. Among them, the reinforcement learning model takes the current operating state of the MFT system as input and outputs a control strategy of tripping or maintaining. In the present invention, it is preferably to use a Deep Q Network (DQN) or a policy gradient algorithm (such as Proximal Policy Optimization, PPO) for policy modeling. These models learn and iterate through the "rewards" of the tripping behavior results to strengthen the preference for correct tripping behaviors and punish false tripping or missed tripping behaviors.

[0048] In particular, an experience replay mechanism is introduced to store the historical decision-making process as a training sample pool and repeatedly utilize it during training to improve the sample efficiency and the convergence stability of the model. At the same time, the state space can be dimensionally reduced, such as using principal component analysis (PCA) or convolutional neural network (CNN) to extract time series features to improve the generalization ability of the model. To improve the adaptability of the model in actual operation, an adaptive tripping threshold adjustment mechanism is also introduced in the present invention. Based on the output probability of the prediction model, the current operating trend of the system, and the historical error distribution, the tripping judgment threshold is dynamically adjusted. For example, when the system volatility increases or the measurement points fluctuate frequently, the tripping threshold is increased to avoid false tripping; when a high-risk feature pattern is detected, the threshold is appropriately reduced to trigger a tripping response in advance, realizing real-time dynamic optimization of the tripping strategy.

[0049] Step S40: Create a digital twin model of the target MFT system based on the intelligent decision-making model, and perform verification on the configuration of the MFT system based on the digital twin model of the target MFT system. Import the configuration of the MFT system that passes the verification into the actual project.

[0050] Specifically, the steps for creating the digital twin model include: establishing the geometric model and physical model of the MFT system as the initial digital model; constructing a data exchange interface for the initial digital model to achieve data synchronization between the physical system and the digital model; performing the construction of the simulation engine based on the initial digital model after the construction of the data exchange interface to simulate the operating state of the MFT system under various working conditions; and constructing a visualization interface to obtain the corresponding digital twin model.

[0051] Furthermore, performing verification on the configuration of the MFT system based on the digital twin model of the target MFT system includes: based on the preset test scenario library, sequentially perform tests for each preset test scenario, and determine the performance parameters under each preset test scenario based on the test results as the regular test results; conduct a fault injection test to obtain the performance parameters under each fault type as the fault test results; generate a test report based on the configuration corresponding to the regular test results and the configuration corresponding to the fault test results, and perform configuration correction on the MFT system when the test report does not meet the expectations until a test report that meets the expectations is generated.

[0052] Furthermore, importing the configuration of the MFT system that passes the verification into the actual project includes: generating a standardized configuration file for the configuration of the MFT system that passes the verification; wherein, the standardized configuration file includes a logic diagram, a screen, and / or a database; perform a configuration consistency check on the standardized configuration file, and after the consistency check passes, transmit the configuration file to the target system based on a secure channel to perform configuration deployment; in response to the completion signal of the configuration deployment execution step, perform the verification of the allocation result, and output the corresponding deployment completion indication signal after the verification passes.

[0053] In the embodiment of the present invention, the digital twin model of the MFT system is a mapping and dynamic simulation of the real control system in the virtual space, covering multiple dimensions such as structural modeling, physical modeling, operation logic simulation, data interaction interface, and human-machine visualization. The construction process mainly includes the following steps: 1) Establish the geometric model and physical model: First, according to the actual control structure, card layout, measuring point position, and process connection relationship of the MFT system of the target unit, construct the geometric model and physical model of the system. The geometric model is used to restore the equipment layout and logical topology structure, and the physical model simulates the internal operating behavior of the system, including the trip trigger mechanism, signal flow transmission delay, relay protection response, etc., to ensure the physical authenticity of the simulation process.

[0054] 2) Build a data exchange interface: To achieve real-time / quasi-real-time data synchronization between the physical MFT system and the digital model, a standardized data exchange interface is built. The interface can be based on protocols such as OPC (OLE for Process Control), Modbus, MQTT, or a custom API to enable two-way data interaction of measured point values, alarm status, logical variables, etc. This step ensures that the digital model can accurately receive on-site data and at the same time output the simulation results to the control platform or the upper-level visualization system.

[0055] 3) Simulation engine integration and operating state simulation: After completing the model structure and data interface, deploy a simulation engine to drive the behavior simulation of the MFT system under various operating conditions. The simulation engine needs to support functions such as multi-condition operation simulation, real-time operation of logic blocks, and state transition analysis, and be able to simulate the trigger paths of different trip conditions, trip response times, and multi-signal comprehensive judgment processes.

[0056] 4) Build a visual interaction interface: To facilitate the monitoring and interaction of the simulation process by configuration engineers and verification personnel, build a graphical visualization interface for the digital twin model. The interface should support functions such as real-time data display, trip process animation, alarm playback, and signal trend charts to enhance interactivity and intuitiveness.

[0057] After completing the above steps, the digital twin model can be used as a platform for MFT system configuration verification and optimization, with the ability to pre-verify the effect of logic configuration and visually analyze the system response.

[0058] Furthermore, based on the digital twin model constructed above, perform a systematic verification test on the MFT system configuration. The verification process mainly includes the following links: 1) Preset test scenario test (conventional test): Pre-establish a test scenario library covering typical working conditions, including scenarios such as normal startup, load fluctuation, stable operation, and signal loss. The system runs these scenarios in the digital model in sequence and monitors the accuracy of logical judgment, the timeliness of trip response, and the reliability of redundant channels under each scenario. Extract performance parameters (such as trip time, interlock response, signal delay, etc.) under each scenario as conventional test data.

[0059] 2) Fault injection test: Introduce various possible faults into the digital twin model through a simulator, such as measured point drift, signal mutation, logical conflict, etc., to perform perturbation tests on the system and evaluate the fault tolerance, false trip probability, and recovery mechanism performance of the configuration in extreme situations. These results form a "fault test result set".

[0060] 3) Test result analysis and configuration correction: Summarize the regular test results and fault test results into a test report to determine whether each configuration item meets the expected design standards. If logical redundancy in the configuration, judgment delay exceeding the limit, or misjump logic is found, the system will mark the problem points, roll back to the logic optimization module for correction, regenerate the logic diagram, and return to the simulation test process again to form a self-feedback closed-loop verification mechanism until all test results meet the standards.

[0061] After the logic of the MFT system is verified through digital twin simulation, it enters the final deployment stage. The deployment process includes the following steps: 1) Generate a standardized configuration file: The system exports the verified logic diagram, control panel, and database parameters as a standardized configuration file. The configuration file has a clear structure and a standardized naming convention, supporting direct docking with the engineering tools of the DCS system to avoid manual editing errors.

[0062] 2) Perform configuration consistency check: Before importing the configuration, the system will automatically perform consistency verification on information such as logical references, measurement point naming, and module connections in the file to ensure that there are no cross errors or logical conflicts in each part.

[0063] 3) Secure deployment and download: The configuration deployment is transmitted through a secure encrypted channel to ensure that the configuration file is not tampered with or lost during transmission. The deployment process connects to the engineering software of the target system to complete the download and activate the new logic.

[0064] 4) Deployment verification and confirmation feedback: After the download is completed, the system will read the running status signals and compare them with the standard logic responses to verify the correctness of the deployment results. When all configuration functionality and consistency tests pass, the system will output a "deployment completed" indication signal and generate a deployment log archive to provide data support for later traceability and auditing.

[0065] Figure 2 It is the system structure diagram of the automatic configuration system of the MFT system based on intelligent algorithms provided by an embodiment of the present invention. As Figure 2As shown in the figure, an automatic configuration system for an MFT system based on an intelligent algorithm is provided in an embodiment of the present invention. The system includes: a collection unit, configured to collect historical operation data of a target MFT system, preprocess the data, and perform value feature and pattern extraction; a processing unit, configured to construct an MFT domain knowledge graph based on the extracted value features and patterns, and based on the MFT domain knowledge graph, use a genetic algorithm to optimize the logical structure of the MFT system, so as to generate an optimal logical structure by defining a fitness function; a model construction unit, configured to construct an intelligent decision-making model based on reinforcement learning technology and the optimal logical structure through learning of historical operation data; a deployment unit, configured to create a digital twin model of the target MFT system based on the intelligent decision-making model, and perform verification on the configuration of the MFT system based on the digital twin model of the target MFT system, and import the configuration of the MFT system that passes the verification into the actual project. An embodiment of the present invention also provides a computer-readable storage medium, on which instructions are stored, and when the instructions run on a computer, the computer is caused to execute the above-mentioned automatic configuration method for an MFT system based on an intelligent algorithm.

[0066] Those skilled in the art can understand that all or part of the steps in the method of the above embodiments can be completed by instructing relevant hardware through a program. The program is stored in a storage medium, including several instructions for causing a single-chip microcomputer, a chip, or a processor to execute all or part of the steps of the method described in various embodiments of the present invention. The foregoing storage medium includes: various media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk, or an optical disc that can store program codes.

[0067] The optional embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details in the above embodiments. Within the scope of the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical solutions of the embodiments of the present invention, and these simple modifications all fall within the protection scope of the embodiments of the present invention. In addition, it should be noted that, in the above specific embodiments, the various specific technical features described can be combined in any appropriate manner without conflict. To avoid unnecessary repetition, the embodiments of the present invention will not separately describe various possible combination methods.

[0068] In addition, any combination can be made between various different embodiments of the present invention as long as it does not violate the idea of the embodiments of the present invention, and it should also be regarded as the content disclosed in the embodiments of the present invention.

Claims

1. An automatic configuration method for an MFT system based on intelligent algorithms, characterized in that, The method includes: Collecting historical operation data of the target MFT system, preprocessing the data, and performing value feature and pattern extraction; Constructing an MFT domain knowledge graph based on the extracted value features and patterns, and based on the MFT domain knowledge graph, using a genetic algorithm to optimize the logical structure of the MFT system to generate an optimal logical structure by defining a fitness function; Constructing an intelligent decision-making model based on reinforcement learning technology and the optimal logical structure through learning from historical operation data; Creating a digital twin model of the target MFT system based on the intelligent decision-making model, and performing verification on the configuration of the MFT system based on the digital twin model of the target MFT system, and importing the configuration of the MFT system that passes the verification into the actual project.

2. The method according to claim 1, wherein Using a genetic algorithm to optimize the logical structure of the MFT system includes: Defining a chromosome encoding method to represent the logical structure of the MFT system as a gene sequence; Designing a fitness function including logical complexity metrics, response time metrics, and / or security metrics; Generating a new generation of population through selection, crossover, and mutation operations; Iterating the evolutionary process until an optimal logical structure that meets the requirements is found.

3. The method according to claim 1, wherein Constructing an intelligent decision-making model based on reinforcement learning technology and the optimal logical structure through learning from historical operation data includes: Defining the state space and action space of the MFT system; Designing a reward function to encourage correct tripping decisions and punish false tripping and missed tripping; Training the decision-making model using a deep Q-network or policy gradient algorithm; Improving learning efficiency and model stability through an experience replay mechanism.

4. The method according to claim 1, wherein The steps for constructing the MFT domain knowledge graph include: Collecting domain knowledge and expert experience information related to the MFT system as initial information; Extracting entities, relationships, and attributes from the initial information, and constructing the basic structure of the knowledge graph based on the extracted entities, relationships, and attributes; Defining entity types and relationship types in the basic structure of the knowledge graph to obtain an ontology model; Setting an inference mechanism based on rules and semantic reasoning for the ontology model to obtain the MFT domain knowledge graph.

5. The method according to claim 1, wherein The steps for creating the digital twin model include: Establishing a geometric model and a physical model of the MFT system as an initial digital model; Constructing a data exchange interface for the initial digital model to achieve data synchronization between the physical system and the digital model; Performing simulation engine construction based on the initial digital model after completing the construction of the data exchange interface to simulate the operating state of the MFT system under various working conditions; Constructing a visualization interface to obtain the corresponding digital twin model.

6. The method according to claim 1, wherein Constructing an intelligent decision-making model based on reinforcement learning technology and the optimal logical structure through learning from historical operation data includes: Optimizing the tripping decision logic of the MFT system, and the corresponding optimization steps include: Identifying historical tripping events and marking valid tripping and false tripping; Extracting relevant factors and patterns leading to valid tripping to construct corresponding training samples; Constructing a tripping prediction model based on the constructed training samples and machine learning algorithms; Configure an adaptive threshold adjustment mechanism based on the trip prediction model to dynamically adjust the trip threshold according to the operating state.

7. The method according to claim 1, wherein Perform verification on the configuration of the MFT system based on the digital twin model of the target MFT system, including: Based on a preset test scenario library, sequentially execute each preset test scenario for testing, and determine the performance parameters under each preset test scenario based on the test results as the regular test results; Conduct a fault injection test to obtain the performance parameters under each fault type as the fault test results; Generate a test report based on the configuration corresponding to the regular test results and the configuration corresponding to the fault test results, and perform configuration correction on the MFT system when the test report does not meet the expectations until a test report that meets the expectations is generated.

8. The method according to claim 1, wherein Import the configuration of the MFT system that has passed the verification into the actual project, including: Generate a standardized configuration file for the configuration of the MFT system that has passed the verification; where The standardized configuration file includes logic diagrams, screens, and / or databases; Perform a configuration consistency check on the standardized configuration file, and after the consistency check passes, transmit the configuration file to the target system based on a secure channel to perform configuration deployment; In response to the completion signal of the configuration deployment execution step, perform affiliated result verification, and output the corresponding deployment completion indication signal after the verification passes.

9. An automatic configuration system for an MFT system based on an intelligent algorithm, characterized in that, The system includes: An acquisition unit for acquiring the historical operation data of the target MFT system, preprocessing the data, and performing value feature and pattern extraction; A processing unit for constructing an MFT domain knowledge graph based on the extracted value features and patterns, and based on the MFT domain knowledge graph, using a genetic algorithm to optimize the logical structure of the MFT system to generate an optimal logical structure by defining a fitness function; A model construction unit for constructing an intelligent decision-making model based on the reinforcement learning technology and the optimal logical structure through learning the historical operation data; A deployment unit for creating a digital twin model of the target MFT system based on the intelligent decision-making model, performing verification on the configuration of the MFT system based on the digital twin model of the target MFT system, and importing the configuration of the MFT system that has passed the verification into the actual project.

10. A computer-readable storage medium, characterized in that, Instructions are stored on the computer-readable storage medium, which when run on a computer cause the computer to execute the method for automatic configuration of the MFT system based on an intelligent algorithm described in any one of claims 1-8.

Images