High-speed railway multi-specialty safety control strategy generation method

Abstract

The invention discloses a high-speed railway multi-specialty safety control strategy generation method, and relates to the technical field of high-speed railway safety control. According to the method, firstly, data collection and arrangement are carried out, key knowledge elements are extracted, risk identification is carried out, and an operation security knowledge graph is constructed; secondly, constructing an intelligent decision logic tree based on the security risk mode to realize intelligent decision of the security risk mode; then, a high-speed railway operation safety control knowledge rule base is constructed, and function and safety risk oriented safety control mode design is achieved; and finally, unifying each subject element, and realizing a multi-specialty collaborative safety control strategy by adopting a cooperative game and a collaborative evolution framework strategy under multi-specialty constraint. By constructing the knowledge framework, establishing the safety control mode library and realizing the collaborative safety control strategy, the management level and the risk control capability of the operation safety of the high-speed railway are comprehensively improved, and a powerful guarantee is provided for the safety operation of the high-speed railway.

Description

Technical Field

[0001] The invention relates to the technical field of high-speed railway safety control, and in particular to a method for generating a multi-professional safety control strategy for a high-speed railway. Background Art

[0002] With the rapid expansion of high-speed rail networks, the speed and frequency of trains are increasing, and the complexity and risks that come with it are also increasing accordingly. Ensuring the safety of high-speed rail systems has become a major challenge for railway operators. Traditional safety assurance measures mainly rely on expert experience and rule systems, which are effective in dealing with known problems, but are unable to cope with increasingly complex and diverse safety risks.

[0003] As an advanced form of information representation, knowledge graph can associate structured and unstructured information in scattered data sources to form a comprehensive knowledge system. It can not only capture the specific knowledge in various professional fields, but also reveal the implicit relationship between different elements, making cross-domain collaborative work possible.

[0004] Although knowledge graphs have been applied in many industries, their potential has not yet been fully tapped in the field of high-speed railway safety control. Current applications are still in their infancy and lack a complete solution to achieve effective integration and intelligent use of knowledge in various professional fields in the high-speed railway system. Summary of the invention

[0005] The purpose of this invention is to propose a method for generating multi-professional safety control strategies for high-speed railways, which comprehensively improves the management level and risk control capability of high-speed railway operation safety by constructing a knowledge framework, realizing railway operation monitoring, setting up multiple safety pipelines, building an intelligent decision-making unit, establishing a safety control pattern library, and realizing collaborative safety control strategy generation.

[0006] To achieve the above purpose, the present invention proposes a method for generating a multi-professional safety control strategy for high-speed railways, and the specific steps are as follows:

[0007] S1. Identify multi-professional risk sources and build an operational safety knowledge graph;

[0008] S2. Construct an intelligent decision logic tree based on security risk model;

[0009] S3. Design of security control mode oriented to functions and security risks;

[0010] S4. Generation of safety control strategies through multi-disciplinary collaboration.

[0011] Preferably, in S1, the specific steps of multi-professional risk source identification and construction of operation safety knowledge graph are as follows:

[0012] S11. Collect and organize data to form a standard specification database and accident case database;

[0013] S12, Extraction of key knowledge elements and identification of risk sources, extracting entities and concepts from various risk sources involved in the accident to form an entity list, identifying entity pairs and their relationships, and constructing triples of entities, relationships, and entity forms;

[0014] S13. Constructing an operation security knowledge graph based on the entity list and the triples;

[0015] S14. Based on statistical analysis and graph neural networks, the structure and parameters of the accident evolution network are iteratively optimized; through data analysis and machine learning, the potential laws and patterns in the accident evolution process are discovered.

[0016] Preferably, in S11, the data are historical detection data, accident report data, and national standard data, and the collected data are classified, coded and organized to form a standard specification library; through accident investigation reports, news reports and academic research, accident cases are sorted out and analyzed in detail to form an accident case library.

[0017] Preferably, the specific method for establishing the standard specification library and the accident case library is as follows:

[0018] S111. Establish an energy change model to analyze the energy change process of the accident;

[0019] S112. Establish an operational safety accident analysis model based on energy conservation and mutation topological space to analyze the energy transfer characteristics of high-speed railway system accidents;

[0020] S113. Analyze the characteristic changes of parameters and state behaviors of risk sources during the occurrence, diffusion and accidents of the system operation, and study the evolution mechanism and evolution conditions of operation safety accidents through the high-speed railway system operation monitoring parameters and subsystem mechanism models;

[0021] S114. Construct a multi-dimensional hierarchical professional domain conceptual model based on the knowledge system structure content and characteristics;

[0022] S115. Construct a mixed-granularity standard specification conceptual model based on the standard specification’s own structure and usage requirements;

[0023] S116. Based on the risk source knowledge requirements of accident analysis, a risk source conceptual model with multi-agent association is constructed;

[0024] S117. Model the hierarchical relationship between concepts based on the knowledge classification system architecture to form a knowledge structure model.

[0025] Preferably, in S12, the specific steps of extracting key knowledge elements are as follows:

[0026] S121. Extract different types of entity knowledge elements using manual extraction, rule-based extraction, and deep learning-based entity recognition methods;

[0027] S122, extracting relational knowledge elements and identifying entity attributes by using rule-based relation extraction, mapping relationship-based transformation, entity co-occurrence-based relation extraction, and machine learning-based relation extraction methods;

[0028] S123. A knowledge fusion algorithm based on representation learning and cluster analysis is used to fuse descriptive information about the same entity from multiple sources.

[0029] Preferably, in S2, accident statistical analysis is automatically constructed according to statistical analysis indicators, accident visualization analysis based on association paths is used to achieve a comprehensive analysis of the high-speed railway operation safety knowledge map, obtain the risk source-accident evolution propagation path, determine the top event, bottom event and the association between the two, take the accident as the top event and connect multiple branches through general AND or OR gates to form a logical tree structure, systematically and logically display the accident, and form a risk source-accident safety risk model;

[0030] According to the safe operation rules and conditions of high-speed trains, the safe operation range of risk sources is simulated and analyzed to obtain the warning line of risk sources; based on the real-time operation monitoring data of high-speed trains, similarity analysis, cluster analysis and parameter overlimit analysis are used to determine whether the risk source is within the warning line; according to the bottom event triggering situation, the logical relationship of the decision logic tree is analyzed from bottom to top, the top event triggering situation is determined, and intelligent decision-making is achieved on what safety risk mode the high-speed train is in.

[0031] Preferably, in S3, the safety control mode design steps are as follows:

[0032] S31. Calculation of high-speed railway operation safety domain;

[0033] S32, judging the running status of high-speed train;

[0034] S33. Construct a knowledge rule base for high-speed railway operation safety control;

[0035] S34. Construction and matching of safety control pattern library.

[0036] Preferably, in S31, the monitoring parameters are divided into multiple safety pipelines related to speed, inclination and visibility according to the operation characteristics and safety risk mode of the high-speed railway, and the pipeline deviation is calculated according to the designed warning line and stop line. The pipeline deviation calculation formula is as follows:

[0037]

[0038] Among them, δ gr is the high-speed railway operation deviation data, δ cl is the measurement deviation data, δ xs is the display error, δ sy is the transmission system delay error, δ sm is the mathematical model error; k1 and k2 are the warning and parking pipeline coefficients.

[0039] Preferably, in S34, a typical safety control mode set is constructed based on Fisher optimal segmentation method, K-means, and mean shift-related clustering methods; the actual domain, set domain, and public domain are defined, and risk characteristics and safety control mode characteristics are mapped to the actual domain and the set domain respectively. The public domain is determined by interlacing area, and the amount of information of the matching degree under each characteristic indicator is calculated to achieve the initial selection of the mode; the 2-additive Choquet integral operator is used to aggregate the total amount of information of the characteristic indicator weights and correlation coefficients to select the optimal safety control mode.

[0040] Preferably, in S4, the specific steps of generating the security control strategy are as follows:

[0041] S41. Unify the elements of each discipline, and express the safety control requirements, control objectives and control methods of each discipline of high-speed railway in a unified form of constraints. i =[X i1, X i2, ...,X im ] represents the local quantity of each discipline, and the constraint formula is as follows:

[0042] Find:X;

[0043] minF=(f1(X),f2(X),...,f p (X));

[0044]

[0045] Where X is the safety control parameter vector of high-speed railway, f p (X) represents the pth safety control target, F is the set vector of safety control targets, g1(X), g2(X) and g l (X) represent the first, second and lth inequality constraints respectively, G(X) represents the set vector composed of inequality constraints, h1(X), h2(X) and h k (X) represents the 1st, 2nd and kth equality constraints respectively, and H(X) represents the set vector composed of equality constraints;

[0046] S42. The calculation formula for cooperative game under multi-professional constraints is as follows:

[0047]

[0048] Among them, f i is the objective function of the ith major, ω i is the target weight of the ith major, and n is an integer;

[0049] S43. Adopt a collaborative evolutionary framework strategy to solve constrained optimization problems. First, decompose the variable dimension of the large-scale problem into multiple low-dimensional sub-component groups, and then enable the general optimization algorithm to calculate each sub-component group in distributed parallel optimization, and finally merge the solutions of all sub-problems into a global optimal solution to the original problem.

[0050] Therefore, the present invention proposes a method for generating a multi-professional safety control strategy for high-speed railways, and its beneficial effects are as follows:

[0051] (1) The present invention comprehensively improves the management level and risk control capability of high-speed railway operation safety by constructing a knowledge framework, realizing railway operation monitoring, establishing multiple safety pipelines, building an intelligent decision-making unit, establishing a safety control model library, and realizing collaborative safety control strategy generation, thus providing a strong guarantee for the safe operation of high-speed railways.

[0052] (2) The present invention realizes the comprehensive identification of multi-professional risk sources and the structured organization of knowledge by constructing a knowledge graph of high-speed railway operation safety; realizes the rapid and accurate judgment of safety risk patterns by constructing an intelligent decision-making logic tree of safety risk patterns; provides targeted safety control strategies by designing safety control patterns oriented to functions and safety risks; realizes effective collaboration among multiple professions by uniformly expressing the safety control requirements of each profession and adopting cooperative game and constraint optimization algorithms.

[0053] (3) The present invention ensures the accuracy of data interaction and the efficiency of information processing by comprehensively using a variety of data extraction, fusion and reasoning technologies.

[0054] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is the overall technical circuit diagram of a method for generating a multi-professional safety control strategy for a high-speed railway according to the present invention;

[0056] Figure 2 It is a logic decision tree of train speeding accidents in a high-speed railway multi-professional safety control strategy generation method of the present invention;

[0057] Figure 3The invention discloses a logic decision tree for track abnormality accidents in a method for generating a multi-professional safety control strategy for a high-speed railway. DETAILED DESCRIPTION

[0058] In order to make the technical solutions, advantages and purposes of the present invention clearer, the technical solutions of the embodiments of the present invention are clearly and completely described below. The described embodiments are part of the embodiments of the present invention, rather than all the embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the protection scope of this application.

[0059] Unless otherwise defined, technical or scientific terms used in the present invention shall have the common meanings understood by one having ordinary skills in the field to which the present invention belongs.

[0060] like Figure 1 As shown, the present invention provides a method for generating a multi-professional safety control strategy for a high-speed railway, and the specific steps are as follows:

[0061] 1. Build a knowledge framework

[0062] 1.1. Data collection and organization: Data sources include historical test data, accident report data, and national standards data. Data can be collected through official channels, professional databases, libraries and other related channels. The collected standards are classified, coded and organized to form a standard specification library. Data can be collected through accident investigation reports, news reports, academic research and other channels. Accident cases are sorted out and analyzed in detail to form an accident case library.

[0063] 1.2. Extraction of key knowledge elements and identification of risk sources: Identify, extract and integrate key information from diverse data sources; use knowledge fusion algorithms to integrate descriptive information about the same entity from multiple sources to address issues such as different knowledge sources, uneven knowledge quality, knowledge duplication, and ambiguity or uncertainty in the relationship between knowledge; knowledge fusion algorithms include representation learning, cluster analysis and other technologies. Representation learning technology can map entities and relationships into low-dimensional vector space to calculate the similarity or distance between them, which helps solve the problems of knowledge duplication and ambiguity; cluster analysis technology can classify similar entities or relationships into one category, thereby integrating information from different data sources and improving the accuracy and completeness of the knowledge graph.

[0064] Risk source identification, extracting specific entities or concepts from accident-related data. These entities can be physical objects (such as trains, tracks, signal equipment, etc.) or abstract concepts (such as regulations, operating procedures, etc.). These entities or concepts are organized into an entity list, and the relationships between these entities are identified. These relationships describe how entities interact or influence each other, which is the key to understanding the mechanism and evolution of accidents. By identifying entity pairs and their relationships, entities, relationships, and entity triples can be formed, and various relationships between entities can be identified, such as "train-running-track", "signal system-control-train", etc. These relationships describe the interactions and dependencies between entities.

[0065] 1.3. Construct an operational safety knowledge graph: Based on the above entity list and triples, construct a high-speed railway operational safety knowledge graph. The knowledge graph is a graphical knowledge representation method that can intuitively display the complex relationships between entities and facilitate knowledge query, reasoning and analysis. During the construction process, technologies such as graph databases can be used to store and manage knowledge graphs.

[0066] 1.4. Risk evolution law: Based on statistical analysis and graph neural network-related reasoning methods, the structure and parameters of the accident evolution network are iteratively optimized. Through data analysis and machine learning technology, the potential laws and patterns in the accident evolution process are discovered.

[0067] 2. Railway operation monitoring

[0068] Multiple factors of railway operation are monitored. EMU data monitoring focuses on the real-time operating parameters of the train, such as speed, acceleration, braking force, traction, temperature, vibration, etc. These data can be collected in real time through various sensors and monitoring equipment installed on the train; the electrical system monitors signals, communications, power supply, etc.; engineering data monitoring focuses on the geometric dimensions of the track (such as gauge, level, height), the status of the track structure (such as fasteners, sleepers, and roadbed), etc.

[0069] 3. Multiple safety channels

[0070] According to the operating characteristics of high-speed railways, the monitoring parameters can be divided into multiple safety channels according to the safety risk mode, such as speed (corresponding to the safety risk mode of train speeding), inclination (corresponding to the safety risk mode of train overturning), visibility (corresponding to the safety risk mode of train collision), etc.

[0071] Suppose a high-speed train is running at 300km / h on a straight line, and the speed limit of the line is 310km / h. If the train suddenly accelerates to 320km / h, the system will immediately detect the overspeed and trigger the corresponding safety response mechanism;

[0072] Alternatively, when a high-speed train passes through a curve with a small radius, if the train's tilt angle exceeds a preset safety threshold (exceeding a certain lateral acceleration limit), the system will detect the potential risk of overturning and take corresponding safety measures, such as slowing down or providing additional stability support.

[0073] Or, in foggy weather, a high-speed train's visibility is significantly reduced. If the system detects that visibility falls below a safe threshold, it will trigger an alarm and may require the train to slow down or stop until weather conditions improve or take other safety measures to ensure safe travel.

[0074] 4. Intelligent decision-making unit

[0075] The intelligent decision-making unit is composed of a knowledge framework and is triggered by multiple safety channels. The intelligent decision-making unit supports multiple safety risk modes, each of which corresponds to a decision logic tree. Safety risk modes can be train speeding risk mode, track abnormality risk mode, signal system failure risk mode, severe weather risk mode, etc. Figure 2 As shown in the figure, in the train overspeed risk mode, the decision logic tree takes the train overspeed accident as the top event, and connects multiple bottom events and sub-events through AND or OR gates. For example, when the train speed is too fast, it may be caused by signal system failure, driver operation error, or train automatic driving system failure. The intelligent decision unit can analyze the triggering of these bottom events and sub-events based on real-time monitoring data, trace the root cause of the speeding accident from bottom to top, and take corresponding safety control measures. Figure 3 As shown in the figure, in the track anomaly risk mode, the decision logic tree also takes the accident as the top event, and connects multiple bottom events and sub-events through logical branches. For example, track breakage may be caused by insufficient maintenance, material quality problems, or extreme weather. The intelligent decision unit can monitor the track status in real time, analyze the triggering of bottom events and sub-events, and promptly warn and take necessary maintenance measures to prevent the occurrence of track anomaly accidents.

[0076] The decision logic tree comprehensively and visually displays the accident cognitive structure in the form of an accident portrait. Through statistical analysis, correlation path analysis and other means, it realizes a comprehensive analysis of the high-speed railway operation safety knowledge map, thereby determining the evolutionary propagation path between risk sources and accidents, and providing a scientific basis for safety decision-making.

[0077] Accident portrait is a visual representation of the accident cognitive structure of the intelligent decision-making unit. It contains key elements such as the basic information of the accident, the risk sources involved, and the accident evolution path. Through the accident portrait, relevant personnel can intuitively understand the overall picture of the accident and grasp the core characteristics of the accident.

[0078] Risk source identification is the basis for building a decision logic tree. The intelligent decision-making unit extracts risk source information related to accidents from the high-speed railway operation safety knowledge map, including physical objects (such as trains, tracks, signal equipment, etc.) and abstract concepts (such as rules and regulations, operating procedures, etc.). These risk sources constitute the bottom events of the decision logic tree.

[0079] The decision logic tree is a method to represent accidents and their risk sources in the form of a logical structure. It takes the accident as the top event and connects multiple branches through "AND or OR gates" to form a "rooted and leafed" tree structure. Each branch represents a risk source or a combination of risk sources, and there is a clear correlation between them and the accident.

[0080] When constructing the decision logic tree, the intelligent decision unit first automatically constructs an accident statistical analysis model based on statistical analysis indicators, and conducts in-depth mining and analysis of accident data in the high-speed railway operation safety knowledge map. Then, based on the accident visualization analysis method of the associated path, the intelligent decision unit further determines the evolutionary propagation path between the risk source and the accident. Finally, these paths are expressed in the form of a logic tree to form a complete risk source-accident safety risk model.

[0081] The application of decision logic tree in intelligent decision-making unit is mainly reflected in two aspects: one is the bottom-up risk source tracing, and the other is the top-down accident reasoning.

[0082] Bottom-up risk source tracing: Relevant personnel can start from the lowest level of the decision logic tree and gradually move upward to view the impact of each risk source on the accident evolution process.

[0083] Top-down accident reasoning: Starting from the first layer of the decision logic tree, reasoning layer by layer about the causes or key factors that lead to the top event (accident). By analyzing the content of each branch layer by layer, the basic event or bottom event that led to the accident is finally found.

[0084] 5. Security Control Mode Library

[0085] 5.1. Calculation of the safety domain of high-speed railway operation. According to the operating characteristics of high-speed railways, the monitoring parameters can be divided into various safety channels according to the safety risk mode, such as speed (corresponding to the safety risk mode of train speeding), inclination (corresponding to the safety risk mode of train overturning), visibility (corresponding to the safety risk mode of train conflict). The channel deviation is calculated according to the designed warning line and stop line. The calculation formula of the channel deviation is as follows:

[0086]

[0087] Among them, δ gr is the high-speed railway operation deviation data; δ clFor measurement deviation data, high-precision measurement deviation is used when calculating the alarm pipeline, and low-precision measurement deviation is used when calculating the emergency stop pipeline; δ xs is the display error, δ sy is the transmission system delay error, δ sm is the mathematical model error; k1, k2 are the calculation coefficients of the warning and parking pipelines; δ xs 2 +δ sy 2 +δ sm 2 When calculating the deviations of various pipelines, the value is taken as a constant.

[0088] 5.2 Judgment of high-speed train running status

[0089] Taking the speed parameter exceeding the limit as an example, the change of the speed parameter is mainly related to the change of the power system parameter. When the speed exceeds the warning line, first calculate whether the actual speed processing result is larger or smaller than the theoretical result; at the same time, process the influence of the corresponding power system, power supply system and other parameters on the train running speed. When the influence trend is consistent with the change trend of the high-speed train speeding deviation, it is determined that the speed exceeds the limit as the real state, otherwise the judgment conclusion of the speed exceeding the limit is cancelled.

[0090] 5.3. Construction of security control rule library

[0091] Aiming at several typical safety risk modes in the operation of high-speed railways, we analyze the processing time and the line capacity of risk sections under different historical safety risk modes, extract the key factors of high-speed railway operation safety, and build a knowledge rule base for high-speed railway operation safety control.

[0092] 6. Construction and matching of safety control mode library

[0093] Based on clustering methods such as Fisher optimal segmentation, K-means, and mean shift, a typical safety control mode set is constructed. The actual domain, set domain, and public domain are defined, and the risk characteristics (risk probability, loss consequences, etc.) and safety control mode characteristics (risk processing time, concentrated passing capacity of risk area lines, etc.) are mapped to the actual domain and set domain respectively, and the public domain is determined by the interweaving area. The amount of information of the matching degree under each characteristic indicator is calculated to realize the preliminary selection of the mode; the 2-additive Choquet integral operator is used to aggregate the total amount of information such as the characteristic indicator weight and correlation coefficient to select the optimal safety control mode.

[0094] Specifically, the construction of a safety control pattern library is the basis of the matching process. The library contains typical safety control strategies designed for different safety risk patterns. In order to build this library, it is first necessary to collect and analyze a large amount of historical accident data, safety regulations, and expert experience to clarify various safety risk patterns and their corresponding control strategies. Then, cluster analysis methods (such as Fisher optimal segmentation, K-means, mean shift, etc.) are used to classify these control strategies to form a typical safety control pattern set.

[0095] Fisher optimal segmentation is a clustering method for ordered samples. It segments samples based on their similarities, so that the intra-group differences are minimized and the inter-group differences are maximized. In the construction of the safety control model library, control strategies with similar risk characteristics and processing strategies can be grouped together to form a typical safety control model.

[0096] K-means clustering is a commonly used unsupervised learning method that divides data points into K clusters so that the data points within each cluster are as similar as possible, while the data points between different clusters are as different as possible. In the construction of the safety control mode library, the K value (i.e., the expected number of safety control modes) can be set, and then the control strategies can be divided into K clusters through K-means clustering, each cluster representing a typical safety control mode.

[0097] Mean shift clustering is a density-based clustering method that does not require the number of clusters to be specified in advance, but automatically determines the number and shape of clusters based on the density distribution of data points. In the construction of a safety control pattern library, mean shift clustering can help discover those control strategy clusters with high density and irregular shapes, thereby building a safety control pattern library more flexibly.

[0098] The matching process of security control mode is to match the risk characteristics monitored in real time with the typical modes in the security control mode library and select the most appropriate security control strategy. This process includes defining the actual domain, setting the domain and the public domain, calculating the amount of information of the matching degree, and using the 2-additive Choquet integral operator to select the optimal mode.

[0099] Actual domain: represents the risk characteristics monitored in real time, such as risk probability, loss consequences, etc.

[0100] Setting domain: represents the control strategy characteristics in the safety control mode library, such as risk processing time, concentrated passing capacity of risk area lines, etc.

[0101] The similarity between the actual risk signature and the control strategy signature can be quantified by mapping the risk signature and the control strategy signature to the actual domain and the set domain, respectively, and calculating their interleaved area to determine the common domain.

[0102] After determining the common domain, calculate the amount of information about the matching degree under each characteristic index. Perform quantitative comparisons on the characteristic values ​​in the actual domain and the set domain, such as calculating Euclidean distance, cosine similarity, etc. By calculating the amount of information about the matching degree, each security control mode can be preliminarily screened to exclude those modes that differ greatly from the actual risk characteristics.

[0103] On the basis of preliminary screening, the 2-additiveChoquet integral operator is used to further aggregate the total amount of information such as characteristic index weights and correlation coefficients to select the optimal safety control mode. The 2-additiveChoquet integral operator is a nonlinear integral operator that can consider the interaction between characteristic indicators and assign different weights to different indicators. By calculating the 2-additiveChoquet integral value of each safety control mode, the mode with the highest integral value can be selected as the optimal safety control strategy.

[0104] 7. Generate collaborative security control strategies

[0105] 7.1 Unification of elements across disciplines

[0106] The safety control requirements, control objectives and control methods of various disciplines of high-speed railway are uniformly expressed in the form of constraints. i =[X i1 ,X i2 ,...,X im ] represents the local quantity of each discipline:

[0107] Find: X;

[0108] minF=(f1(X),f2(X),...,f p (X));

[0109]

[0110] Where X is the safety control parameter vector of high-speed railway, f p (X) represents the pth safety control target, F is the set vector of safety control targets. Since the maximization target can be transformed into the minimization target, the minimization method is used here to represent the safety control target of high-speed railway. l (X) is the lth inequality constraint, G(X) represents the set vector of inequality constraints, h k (X) is the kth equality constraint, H(X) represents the set vector composed of equality constraints; G(X) and H(X) can be transformed into each other by adding auxiliary variables.

[0111] 7.2 Cooperative Game under Multi-professional Constraints

[0112] The mathematical model of multi-professional safety control coordination is defined, and its calculation formula is as follows:

[0113]

[0114] Among them, f i is the objective function of the ith major, ω i is the target weight of the i-th profession; after processing the safety control target in the formula by cooperative game, the multi-objective constraint network model of collaborative control becomes a single-objective constraint optimization problem. Therefore, the safety control can be obtained by designing the corresponding constraint optimization algorithm to solve the model.

[0115] 7.3 Solving Constrained Optimization Problems

[0116] In order to effectively solve large-scale optimization problems, a collaborative evolution framework strategy is proposed. First, the variable dimension of the large-scale problem is decomposed into multiple low-dimensional sub-component groups, such as using PCA to analyze the variable dimension in the large-scale problem, and arranged according to the contribution of the variable dimension. The decision variables with large contribution are divided into one group, the decision variables with small contribution are divided into other groups, and the decision variables with very small contribution are ignored. In this way, the variable dimensions are not only grouped according to the contribution rules, but also a part of the variable dimensions with very small contribution is simplified, which plays a role in dimensionality reduction; then the general optimization algorithm can be distributed and parallel to optimize the calculation of each sub-component group, such as the exocellular particle swarm intelligent algorithm, which has better search reliability; finally, the solutions of all sub-problems are merged into a global optimal solution for the original problem, so as to achieve the purpose of optimizing and solving the original problem.

[0117] Aiming at the fact that the collaborative particle swarm algorithm is prone to fall into local optimality when solving large-scale optimization problems, two strategies are adopted to improve the collaborative particle swarm algorithm: one is to use learning automata to mine the coupling relationship between problem sub-dimensions and put sub-dimensions with similar relationships into a sub-population; the other is to adaptively divide the sub-population into two parts, and assign new inertia weights to particles with poor optimization effects, so as to make them jump out of the local optimality and accelerate the convergence of the algorithm.

[0118] Therefore, the present invention provides a method for generating multi-professional safety control strategies for high-speed railways. By constructing a knowledge framework, realizing railway operation monitoring, setting up multiple safety pipelines, building an intelligent decision-making unit, establishing a safety control pattern library, and realizing collaborative safety control strategy generation, the management level and risk control capabilities of high-speed railway operation safety are comprehensively improved, providing a strong guarantee for the safe operation of high-speed railways.

[0119] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for generating a multi-professional safety control strategy for high-speed railways, characterized in that: Here are the steps: S1. Identify multi-professional risk sources and build an operational safety knowledge graph; S2. Construct an intelligent decision logic tree based on security risk model; S3. Design of security control mode oriented to functions and security risks; S4. Generation of safety control strategies through multi-disciplinary collaboration.

2. A method for generating a high-speed railway multi-professional safety control strategy according to claim 1, characterized in that: In S1, the specific steps for multi-professional risk source identification and operation safety knowledge graph construction are as follows: S11. Collect and organize data to form a standard specification database and accident case database; S12, Extraction of key knowledge elements and identification of risk sources, extracting entities and concepts from the risk sources involved in the accident to form an entity list, and constructing triples of entities, relationships, and entity forms by identifying entity pairs and their relationships; S13. Constructing an operation security knowledge graph based on the entity list and the triples; S14. Based on statistical analysis and graph neural networks, the structure and parameters of the accident evolution network are iteratively optimized; through data analysis and machine learning, the potential laws and patterns in the accident evolution process are discovered.

3. The method for generating a high-speed railway multi-professional safety control strategy according to claim 2 is characterized in that: In S11, the data include historical detection data, accident report data, and national standard data, and the collected data are classified, coded and organized to form a standard specification library; through accident investigation reports, news reports and academic research, accident cases are sorted out and analyzed in detail to form an accident case library.

4. A method for generating a high-speed railway multi-professional safety control strategy according to claim 3, characterized in that: The specific method for establishing the standard specification library and the accident case library is as follows: S111. Establish an energy change model to analyze the energy change process of the accident; S112. Establish an operational safety accident analysis model based on energy conservation and mutation topological space to analyze the energy transfer characteristics of high-speed railway system accidents; S113. Analyze the characteristic changes of parameters and state behaviors of risk sources during the occurrence, diffusion and accidents of the system operation, and study the evolution mechanism and evolution conditions of operation safety accidents through the high-speed railway system operation monitoring parameters and subsystem mechanism models; S114. Construct a multi-dimensional hierarchical professional domain conceptual model based on the knowledge system structure content and characteristics; S115. Construct a mixed-granularity standard specification conceptual model based on the standard specification’s own structure and usage requirements; S116. Based on the risk source knowledge requirements of accident analysis, a risk source conceptual model with multi-agent association is constructed; S117. Model the hierarchical relationship between concepts based on the knowledge classification system architecture to form a knowledge structure model.

5. The method for generating a high-speed railway multi-professional safety control strategy according to claim 2 is characterized in that: In S12, the specific steps of extracting key knowledge elements are as follows: S121. Extract different types of entity knowledge elements using manual extraction, rule-based extraction, and deep learning-based entity recognition methods; S122, extracting relational knowledge elements and identifying entity attributes by using rule-based relation extraction, mapping relationship-based transformation, entity co-occurrence-based relation extraction, and machine learning-based relation extraction methods; S123. A knowledge fusion algorithm based on representation learning and cluster analysis is used to fuse descriptive information about the same entity from multiple sources.

6. The method for generating a high-speed railway multi-professional safety control strategy according to claim 1 is characterized in that: In S2, accident statistical analysis is automatically constructed based on statistical analysis indicators, and accident visualization analysis based on association paths realizes a comprehensive analysis of the high-speed railway operation safety knowledge map, obtains the risk source-accident evolution propagation path, determines the top event, bottom event and the association between the two, takes the accident as the top event and connects multiple branches through general AND or OR gates to form a logical tree structure, systematically and logically displays the accident, and forms a risk source-accident safety risk model; According to the high-speed train safety operation rules and conditions, the safe operation range of the risk source is simulated and analyzed to obtain the warning line of the risk source; Based on the real-time operation monitoring data of high-speed trains, similarity analysis, cluster analysis and parameter over-limit analysis are used to determine whether the risk source is within the warning line; based on the bottom event triggering situation, the logical relationship of the decision logic tree is analyzed from bottom to top to determine the top event triggering situation, so as to achieve intelligent decision-making on what safety risk mode the high-speed train is in.

7. The method for generating a high-speed railway multi-professional safety control strategy according to claim 1 is characterized in that: In S3, the safety control mode design steps are as follows: S31. Calculation of high-speed railway operation safety domain; S32, judging the running status of high-speed train; S33. Construct a knowledge rule base for high-speed railway operation safety control; S34. Construction and matching of safety control pattern library.

8. A method for generating a high-speed railway multi-professional safety control strategy according to claim 7, characterized in that: In S31, according to the operation characteristics and safety risk model of high-speed railways, the monitoring parameters are divided into multiple safety pipelines such as speed, inclination and visibility. The pipeline deviation δ is calculated according to the designed warning line and stop line. The pipeline deviation calculation formula is as follows: Among them, δ gr is the high-speed railway operation deviation data, δ cl is the measurement deviation data, δ xs is the display error, δ sy is the transmission system delay error, δ sm is the mathematical model error; k1 and k2 are the pipeline coefficients for warning and parking.

9. The method for generating a high-speed railway multi-professional safety control strategy according to claim 7 is characterized in that: In S34, a typical safety control mode set is constructed based on Fisher optimal segmentation method, K-means, and mean shift-related clustering methods; the actual domain, set domain, and public domain are defined, and the risk characteristics and safety control mode characteristics are mapped to the actual domain and the set domain respectively. The public domain is determined by the interlaced area, and the amount of information on the matching degree under each characteristic indicator is calculated to achieve the initial selection of the mode; the 2-additive Choquet integral operator is used to aggregate the total amount of information of the characteristic indicator weights and correlation coefficients to select the optimal safety control mode.

10. The method for generating a high-speed railway multi-professional safety control strategy according to claim 1, characterized in that: In S4, the specific steps for generating security control policies are as follows: S41. Unify the elements of each discipline, and express the safety control requirements, control objectives and control methods of each discipline of high-speed railway in a unified form of constraints. i =[X i1 ,X i2, ...,X im ] represents the local quantity of each discipline, and the constraint formula is as follows: Find:X; minF=(f1(X),f2(X),...,f p (X)); Where X is the safety control parameter vector of high-speed railway, f p (X) represents the pth security control target, F is the set vector of security control targets, g l (X) is the lth inequality constraint, G(X) represents the set vector of inequality constraints, h k (X) is the kth equality constraint, H(X) represents the set vector composed of equality constraints; S42. The calculation formula for cooperative game under multi-professional constraints is as follows: Among them, f i is the objective function of the ith major, ω i is the target weight of the ith major, and n is an integer; S43. Adopt a collaborative evolutionary framework strategy to solve constrained optimization problems. First, decompose the variable dimension of the large-scale problem into multiple low-dimensional sub-component groups, and then enable the general optimization algorithm to calculate each sub-component group in distributed parallel optimization, and finally merge the solutions of all sub-problems into a global optimal solution to the original problem.

Images