A method and device for industrial control equipment identification based on knowledge graph

By building a knowledge graph and training protocol type recognition model, the problem of difficult to identify industrial control equipment types in the existing technology that do not support detection information feedback is solved, high accuracy, automation and real-time equipment recognition is achieved, and industrial control systems are improved. Safety and operation and maintenance efficiency.

CN118779764BActive Publication Date: 2024-11-15CHINA ELECTRONICS CORP 6TH RES INST
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411260511.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-10
Publication Date
2024-11-15
Estimated Expiration
2044-09-10

AI Technical Summary

Technical Problem

The prior art is difficult to effectively identify the types of industrial control equipment that do not support detection information feedback, resulting in insufficient security detection and risk identification capabilities of industrial control networks.

Method used

Using a knowledge graph-based method, by acquiring and preprocessing cross-industry industrial control equipment data, a knowledge graph that characterizes the entity, attributes and relationships of the equipment, and training a protocol type recognition model, identifying protocol types in a real network environment, and determining in-depth processing. Device type, infer unknown device type.

Benefits of technology

It improves the accuracy and efficiency of industrial control equipment identification, enhances cross-industry adaptability, improves the level of automation and intelligence, supports identification in complex environments, reflects equipment changes in real time, improves system security and reduces operation and maintenance costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118779764B_ABST
    Figure CN118779764B_ABST
Patent Text Reader

Abstract

The present application provides a method and device for identifying industrial control equipment based on a knowledge graph, the method comprising: first, collecting and preprocessing cross-industry industrial control equipment data, and building a knowledge graph containing equipment entities, attributes and relationship information based on this; then, using the network communication data in these data to train a protocol type recognition model; in a real industrial control network environment, by capturing network traffic data and using the model to identify the protocol type, the device type is preliminarily determined in combination with the correlation between the industrial control protocol and the device; finally, with the help of the knowledge graph, the possible or unknown device types are deeply processed to uniquely determine the possible type of device and infer the unknown type of device; the present application can identify the type of device that does not support detection information feedback in the industrial control network.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of machine learning, and in particular to a method and device for identifying industrial control equipment based on a knowledge graph. Background Art

[0002] With the continuous advancement of industrial intelligence, the originally closed environment of industrial control networks has become more and more open, and the network devices in industrial control systems face more risks. In recent years, the security issues of industrial control networks have received more and more attention. In order to more effectively implement the security detection of industrial control networks and enhance the ability of industrial control networks to resist risks, the identification of device types in the network is the basis.

[0003] However, the existing technology is usually based on feedback data after the industrial control equipment receives the detection packet. When some devices set relevant rules or policies and do not provide feedback, it will be difficult to identify the type of industrial control equipment. Summary of the invention

[0004] In view of this, an embodiment of the present application provides an industrial control equipment identification method and device based on a knowledge graph, which can identify the types of equipment that does not support detection information feedback in the industrial control network.

[0005] The technical solution of the embodiment of the present application is implemented as follows:

[0006] In a first aspect, an embodiment of the present application provides an industrial control equipment identification method based on a knowledge graph, the method comprising:

[0007] Acquire cross-industry industrial control equipment data, and pre-process the cross-industry industrial control equipment data;

[0008] Constructing a knowledge graph based on the preprocessed cross-industry industrial control equipment data, wherein the knowledge graph represents all equipment entities, attributes, and relationship information;

[0009] Training and building a protocol type recognition model based on network communication data in the cross-industry industrial control equipment data;

[0010] Obtain network traffic data in a real industrial control network environment;

[0011] Identify the protocol type in the network traffic data based on the protocol type identification model, and determine a first device type and a second device type through the association between the industrial control protocol and the specific device, wherein the first device type represents a determined device type, and the second device type represents a possible device type or an unknown device type;

[0012] The second device type is deeply processed based on the knowledge graph to determine a unique type of possible device types and to infer unknown device types.

[0013] In a second aspect, an embodiment of the present application further provides an industrial control equipment identification device based on a knowledge graph, the device comprising:

[0014] A data processing module, used to obtain cross-industry industrial control equipment data and pre-process the cross-industry industrial control equipment data;

[0015] A graph construction module, used to construct a knowledge graph based on the preprocessed cross-industry industrial control equipment data, wherein the knowledge graph represents all equipment entities, attributes and relationship information;

[0016] A model training module, used for training and building a protocol type recognition model based on the network communication data in the cross-industry industrial control equipment data;

[0017] Traffic acquisition module, used to obtain network traffic data in a real industrial control network environment;

[0018] a type determination module, configured to identify the protocol type in the network traffic data based on the protocol type identification model, and determine a first device type and a second device type through an association between an industrial control protocol and a specific device, wherein the first device type represents a determined device type, and the second device type represents a possible device type or an unknown device type;

[0019] A deep processing module is used to perform deep processing on the second device type based on the knowledge graph, determine a unique type of possible device types, and infer unknown device types.

[0020] In the third aspect, an embodiment of the present application also provides an electronic device, comprising: a processor, a storage medium and a bus, wherein the storage medium stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the storage medium through the bus, and the processor executes the machine-readable instructions to execute the industrial control equipment identification method based on the knowledge graph as described in any one of the first aspects.

[0021] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the industrial control equipment identification method based on the knowledge graph as described in any one of the first aspects is executed.

[0022] The embodiments of the present application have the following beneficial effects:

[0023] (1) Improving recognition accuracy and efficiency: By building a cross-industry knowledge graph, a wealth of device entity, attribute, and relationship information is integrated, making it possible to use more comprehensive data for in-depth analysis and reasoning during the device recognition process, thereby significantly improving recognition accuracy and efficiency.

[0024] (2) Enhanced cross-industry adaptability: The method is not limited to a single industry, but can collect and process industrial control equipment data across industries, construct a knowledge graph with strong versatility, and enhance the applicability and portability of the method in different industries.

[0025] (3) High degree of automation and intelligence: It can automatically extract key device attributes, build relationship types, train protocol type recognition models, and identify protocol types in real time under a real industrial control network environment, significantly improving the level of automation and intelligence in industrial control equipment recognition.

[0026] (4) Support device identification in complex environments: In a real industrial control network environment, it can cope with complex network traffic data, effectively identify unknown or possible device types through deep data mining and pattern matching, and improve the recognition ability in complex environments.

[0027] (5) Real-time and dynamic: By capturing and analyzing network traffic data in real time and combining it with the dynamic update of the knowledge graph, it can reflect the changes in equipment in the industrial control network in real time, providing timely and accurate information support for security monitoring and troubleshooting.

[0028] (6) Improve system security: Accurately identifying the type of industrial control equipment helps to timely discover and prevent potential security threats, and improve the security and stability of industrial control systems. By deeply mining the relationships and interaction patterns between devices, potential security vulnerabilities and attack paths can also be discovered, providing strong support for the formulation of security strategies.

[0029] (7) Reduce operation and maintenance costs: Through automated and intelligent identification methods, the reliance on manual intervention is reduced, and the operation and maintenance costs are reduced. At the same time, accurate identification of equipment types also helps to optimize resource allocation, improve equipment utilization and overall operation and maintenance efficiency.

[0030] In summary, this knowledge graph-based industrial control equipment identification method has shown significant beneficial effects in improving identification accuracy, enhancing cross-industry adaptability, improving automation and intelligence levels, supporting complex environment identification, improving system security, and reducing operation and maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0032] Figure 1 It is a flowchart of steps S101-S106 provided in an embodiment of the present application;

[0033] Figure 2 It is a flowchart of steps S201-S203 provided in an embodiment of the present application;

[0034] Figure 3 It is a flowchart of steps S301-S302 provided in an embodiment of the present application;

[0035] Figure 4 It is a flowchart of steps S401-S403 provided in an embodiment of the present application;

[0036] Figure 5 It is a flowchart of steps S501-S503 provided in an embodiment of the present application;

[0037] Figure 6 It is a flowchart of steps S601-S604 provided in an embodiment of the present application;

[0038] Figure 7 It is a flowchart of steps S701-S703 provided in an embodiment of the present application;

[0039] Figure 8 It is a structural schematic diagram of an industrial control equipment identification device based on a knowledge graph provided in an embodiment of the present application;

[0040] Fig. 9 It is a schematic diagram of the composition structure of the electronic device provided in the embodiment of the present application. DETAILED DESCRIPTION

[0041] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the drawings in the present application only serve the purpose of explanation and description and are not used to limit the scope of protection of the present application. In addition, it should be understood that the schematic drawings are not drawn in real proportion. The flowchart used in this application shows the operations implemented according to some embodiments of the present application. It should be understood that the operations of the flowchart can be implemented out of sequence, and the steps without logical context can be reversed in order or implemented simultaneously. In addition, those skilled in the art can add one or more other operations to the flowchart under the guidance of the content of the present application, or remove one or more operations from the flowchart.

[0042] In the following description, reference is made to “some embodiments”, which describe a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0043] In addition, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0044] In the following description, the terms "first\second\third" involved are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It can be understood that "first\second\third" can be interchanged with a specific order or sequence where permitted, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0045] It should be noted that the term "comprising" will be used in the embodiments of the present application to indicate the existence of the features declared thereafter, but does not exclude the addition of other features.

[0046] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0047] See also Figure 1 , Figure 1is a flow chart of steps S101-S106 of the industrial control equipment identification method based on knowledge graph provided in the embodiment of the present application, which will be combined with Figure 1 Steps S101-S106 are shown for explanation.

[0048] In step S101, cross-industry industrial control equipment data is obtained, and the cross-industry industrial control equipment data is preprocessed;

[0049] This step first uses literature research and field visits to analyze typical industrial control system architectures across industries, and comprehensively collects key data such as equipment and network communications, covering information such as equipment type, attributes, functions, connection relationships, communication protocols, and business traffic. Subsequently, the data is strictly cleaned to eliminate duplications, errors, and inconsistencies to ensure data quality and lay a data foundation for subsequent steps.

[0050] In step S102, a knowledge graph is constructed based on the preprocessed cross-industry industrial control equipment data, wherein the knowledge graph represents all equipment entities, attributes and relationship information.

[0051] This step uses natural language processing technology to identify device entities and their attribute information, and defines relationship types such as physical, logical and business connections. The rule matching method is used to extract the connectivity relationship between devices from the data, and all device entities, attributes and relationship information are integrated to form a unified knowledge graph.

[0052] In step S103, a protocol type recognition model is trained and constructed based on the network communication data in the cross-industry industrial control equipment data.

[0053] This step is based on the collected network message data of known protocols, extracting key feature information such as port number, transport layer and application layer header information, byte sequence of payload content, etc., using deep learning algorithm to obtain high-level and abstract feature representation, and training a protocol type recognition model.

[0054] In step S104, network traffic data is obtained in a real industrial control network environment.

[0055] In this step, in a real industrial control network environment, you configure the mirroring function of the router or switch, send a copy of the traffic of the selected port to the specified mirror port, and use a packet capture tool to connect to the mirror port to capture network traffic data.

[0056] In step S105, the protocol type in the network traffic data is identified based on the protocol type identification model, and the first device type and the second device type are determined through the association between the industrial control protocol and the specific device, wherein the first device type represents a determined device type, and the second device type represents a possible device type or an unknown device type.

[0057] This step identifies the protocol type in the network traffic based on the classification model trained in step S103. For typical industrial control protocols, the strong correlation between them and specific devices is used to accurately identify some target device types. For other protocols, weak correlation is used to preliminarily identify multiple possible device types to form a candidate set.

[0058] In step S106, the second device type is deeply processed based on the knowledge graph to determine a unique type of possible device types, and to infer unknown device types.

[0059] This step is based on step S105 and relies on the knowledge graph for in-depth mining. By retrieving the associated links and feature interaction patterns of known device types in the knowledge graph, combined with the actual interaction data in the network traffic, and using advanced algorithms of graph reasoning and pattern matching, the unique device type is determined and the inference of unknown protocol device types is achieved.

[0060] In some embodiments, see Figure 2 , Figure 2 It is a flow chart of steps S201-S203 provided in an embodiment of the present application. The acquisition of cross-industry industrial control equipment data can be achieved through steps S201-S203, which will be explained in combination with each step.

[0061] In step S201, the overall architecture of different industrial control networks is determined, and statistical processing is performed on the devices included in each layer of the overall architecture to construct an alternative set for the result of industrial control device identification.

[0062] In step S202, data collection is performed on the devices included in each layer to determine basic properties of different devices, wherein the data collection includes at least one of technical parameter collection, performance indicator collection, and functional feature collection.

[0063] In step S203, the connection relationship and communication protocol between devices in different industrial control networks are determined, and the business traffic is collected, wherein the business traffic includes the data transmission volume and transmission rate, data type, transmission mode and time characteristics, the data type includes at least one of sensor data, control commands, status information, and log records, the transmission mode includes at least one of periodic transmission, event-triggered transmission and real-time streaming data, and the time characteristics include at least one of transmission frequency, peak time and data timeliness.

[0064] Furthermore, the preprocessing of the cross-industry industrial control equipment data includes:

[0065] The cross-industry industrial control equipment data is processed by at least one of the following: cleaning, sorting, conversion, and standardization.

[0066] Furthermore, the method further comprises:

[0067] The preprocessed data is organized logically and functionally and stored in a matching database.

[0068] As an example, steps S201-S203 specifically include:

[0069] 1) Typical industrial control network architecture: Industrial control networks usually include five layers: field equipment layer, field control layer, process monitoring layer, production management layer, and enterprise resource layer. By analyzing industrial control networks in different industries, the overall architecture of the network can be sorted out.

[0070] 2) Equipment type statistics: The equipment included in different layers of the industrial control network is relatively fixed. Taking the field control layer as an example, it usually includes remote terminal units (RTUs), distributed control systems (DCSs), programmable logic controllers (PLCs), etc. By statistically analyzing the equipment included in each layer of the network architecture, an alternative set can be constructed for the equipment type identification results.

[0071] 3) Attribute and function collection: The quality of the collection of attributes and functions of devices at all levels will directly affect the inference of device types. Attribute and function collection includes technical parameter collection, such as the sensitivity, accuracy, and detection range of sensors, the driving ability and response speed of actuators, and the processing power and flexibility of PLC and DCS systems; performance indicator collection such as the response time of sensors, the control accuracy of actuators, the logical processing speed of PLCs, and the stability of DCS systems; functional characteristics collection such as whether sensors support multiple detection types and whether actuators can adapt to different working environments. The various types of information collected provide detailed information for the construction of each device point in the knowledge graph, which helps to depict the basic attributes of the equipment and help to deeply understand its role and importance in industrial control systems.

[0072] 4) Analysis of connection relationships and communication protocols: In industrial control networks, the connection relationships and communication protocols between devices are the key to achieving efficient and stable control. In order to build a knowledge graph, it is first necessary to comprehensively collect detailed information on these connection relationships and communication protocols. Common communication protocols include Modbus, IEC-104, DNP3, etc. These protocols can not only adapt to changing production needs, but also have good scalability and compatibility. During the collection process, it is necessary to record in detail the characteristics of each device, such as the type of communication protocol supported, data transmission rate, transmission distance, and connection method with other devices. In addition, it is also necessary to collect characteristic information of the connection topology between devices, including master-slave relationships, network topology, etc. Collecting connection relationships and communication protocols helps the knowledge graph to build interactions and dependencies between devices, providing a solid foundation for device type identification.

[0073] 5) Business traffic collection: Business traffic collection not only includes the amount and rate of data transmission, but also involves the type, transmission mode and time characteristics of data. Data types include sensor data, control commands, status information, log records, etc. Transmission modes include periodic transmission, event-triggered transmission or real-time streaming data. Time characteristics include transmission frequency, peak time and timeliness of data. By collecting business traffic, in addition to providing detailed information on the communication mode and data exchange between devices in the knowledge graph, it also helps to build interactions and dependencies between devices in the knowledge graph, providing support for device type identification and the formation of candidate sets.

[0074] 6) Data preprocessing: In order to make the collected raw data meet the needs of subsequent analysis and model recognition, the data needs to be cleaned and organized, which involves data conversion and standardization. This mainly includes removing noise, duplicates and inconsistencies in the data, converting the data from the original format to a format suitable for analysis and processing, such as standardizing numerical data into a unified dimension and unit; standardizing the data to eliminate differences between different data sources and devices; labeling the data type to clarify the type and purpose of the data; extracting key features from the raw data, such as the peak value, average value, and change trend of the data. Finally, in order to improve the processing efficiency of the model, the data needs to be reduced in dimensionality to reduce the complexity and redundancy of the data.

[0075] 7) Data storage and management: In the construction of the knowledge graph, in order to ensure the persistence and accessibility of data, this method needs to store and manage the collected data, which involves data organization, storage, maintenance and protection. First, the collected and pre-processed data needs to be organized according to logic and function. Secondly, the appropriate database type should be selected for storage, and transaction management, data backup and recovery mechanisms should be used to ensure the consistency and integrity of the data during the storage process. Finally, data access control strategies should be implemented to strengthen data protection and ensure that only authorized users can access sensitive data.

[0076] In some embodiments, see Figure 3 , Figure 3 It is a flow chart of steps S301-S302 provided in an embodiment of the present application. The construction of a knowledge graph based on the preprocessed cross-industry industrial control equipment data can be achieved through steps S301-S305, which will be explained in combination with each step.

[0077] In step S301, the name of each industrial control device is automatically acquired and identified based on named entity recognition.

[0078] In step S302, key attributes of the industrial control device are extracted from the device description and related documents of the industrial control device, and the key attributes are associated with the corresponding industrial control device, wherein the key attributes include at least one of technical parameters, performance indicators, and functional characteristics.

[0079] In step S303, the relationship types between industrial control devices are identified and defined, wherein the relationship types include at least one of physical connection, logical connection and business connection, and the relationship types are classified, and the classification categories include at least one of direct connection, indirect connection and dependency relationship, so as to be represented in the knowledge graph.

[0080] In step S304, the connectivity relationship between industrial control devices is extracted based on rule matching, and the connectivity relationship is converted into an edge in the knowledge graph, wherein the rule is determined based on device type, communication protocol, and connection mode;

[0081] In step S305, the device entities, attributes and relationship information of all industrial control devices are integrated into the knowledge graph. The connectivity of the knowledge graph is determined by checking whether there is a path between any two nodes, represented by the connection matrix A. The density D of the graph of the knowledge graph is the possible maximum value of the number of edges E of the graph and the number of edges in the complete graph. The ratio is:

[0082]

[0083] Where N is the total number of nodes in the graph.

[0084] As an example, steps S301-S305 specifically include:

[0085] 1) Entity recognition: Use named entity recognition technology to automatically obtain and identify information such as network device names, and combine rules and statistical methods to improve the accuracy and efficiency of recognition.

[0086] 2) Attribute information extraction: Extract key attributes of the device from the device description and related documents, such as technical parameters, performance indicators, functional characteristics, etc. Associate these attribute information with the corresponding device entity to form a detailed attribute description of the device entity.

[0087] 4) Relationship type definition: Identify and define the relationship types between devices, such as physical connection, logical connection, and business connection. Physical connection refers to the actual electrical or mechanical connection between devices; logical connection refers to the relationship between devices in the control logic; business connection involves the functional association of devices in the production process. And classify the relationships into different categories, such as direct connection, indirect connection, dependency, etc., so that they can be clearly represented in the knowledge graph.

[0088] 5) Application of rule matching method: The rule matching method is used to extract the connectivity relationship between devices from the data. These rules can be based on device type, communication protocol or connection mode, etc. Through pattern recognition and matching algorithms, the connectivity relationship between devices is automatically identified and converted into edges in the knowledge graph.

[0089] 6) Knowledge graph construction: Integrate all device entities, attributes, and relationship information into a unified knowledge graph. Ensure the consistency and completeness of information. Graph theory indicators such as graph connectivity and density can be used to evaluate the structure of the knowledge graph, handle redundancy and conflicts in the data, and ensure that the relationship and attribute information in the knowledge graph are accurate. The connectivity of the graph can be determined by checking whether there is a path between any two nodes. If such a path exists, the graph is connected and represented by the adjacency matrix A. The density D of the graph is the ratio of the number of edges E of the graph to the possible maximum number of edges in the complete graph (N(N-1)) / 2, and the formula is: D=2E / (N(N-1)), where N is the total number of nodes in the graph.

[0090] In some embodiments, see Figure 4 , Figure 4 It is a flow chart of steps S401-S403 provided in an embodiment of the present application. The training and construction of a protocol type identification model based on the network communication data in the cross-industry industrial control equipment data can be achieved through steps S401-S403, which will be explained in combination with each step.

[0091] In step S401, based on the collected network message data of known protocols, key features in the network message are extracted, wherein the key features include at least one of the port number, transport layer protocol, application layer protocol, IP address, message length, and byte sequence of the payload content.

[0092] In step S402, a deep learning model is constructed based on the key features, wherein the deep learning model is used to identify the protocol type of message data, and the deep learning model is a hybrid model of three models: convolutional neural network CNN, recurrent neural network RNN, and long short-term memory network LSTM.

[0093] In step S403, the trained deep learning model is used as the protocol type recognition model and deployed into the target industrial control system to identify the protocol type in the network message in real time, and the recognition result is integrated into the knowledge graph to update the communication protocol information between devices.

[0094] As an example, steps S401-S403 specifically include:

[0095] 1) Feature extraction: Based on the collected network message data of known protocols, key features in the network message are extracted, including but not limited to port number, transport layer protocol (such as TCP, UDP), application layer protocol (such as HTTP, FTP), IP address, message length, byte sequence of payload content, etc.

[0096] 2) Model training: A deep learning model is built based on the characteristics of network message data. In order for the model to be able to identify the protocol type of the message data, the model needs to be able to process structured information, serialized information, and time-series information. Therefore, a hybrid model of three models, namely, convolutional neural network (CNN), recurrent neural network (RNN), and long short-term memory (LSTM), is used in this method for prediction and recognition.

[0097] Convolutional neural networks (CNNs) use a set of learnable filters (or convolution kernels) to slide over input data through convolutional layers to extract local features. Secondly, these features are transformed nonlinearly through activation functions, such as Sigmoid (Formula 1-1), to enhance the expressive power of the model.

[0098] (Formula 1-1)

[0099] Next, the features are downsampled through pooling layers (such as maximum pooling) to reduce the spatial dimension of the data while increasing the invariance to data displacement. Then, after multiple convolution and pooling operations, the features are flattened and passed through a fully connected layer for classification or regression tasks. Finally, the network weights are optimized through backpropagation algorithms and gradient descent, so that the network can automatically learn the complex mapping relationship from the original input to the target output. Through this hierarchical feature extraction and abstraction, CNN can effectively identify and classify patterns and objects in the data.

[0100] Recurrent Neural Network (RNN) is a deep learning model specifically designed to process sequence data. Its core working principle is to capture the temporal dependencies in the sequence. RNN forms an internal state or memory by applying the same weight matrix to each time step of the sequence and taking the output of the previous time step as an additional input of the current time step, so that the previous information can be passed to the subsequent time.

[0101] t-1, t, t+1 represent time series. X represents the input sample. St represents the memory of the sample at time t, as shown in (Equation 1-2).

[0102] (Formula 1-2)

[0103] W represents the input weight, U represents the weight of the sample input at this moment, and V represents the output sample weight.

[0104] At t = 1, the initialization input is generally , randomly initialize W,U,V, and perform the following (Formula 1-3) calculation:

[0105]

[0106]

[0107] (Formula 1-3)

[0108] Among them, f and g are both activation functions. Among them, f can be tanh, relu, sigmoid and other activation functions, and g is usually softmax or other.

[0109] Time moves forward, and the state s1 at this time, as the memory state of time 1, will participate in the prediction activity of the next moment, as shown in (Equation 1-4):

[0110]

[0111]

[0112] (Formula 1-4)

[0113] By analogy, the final output value can be obtained as shown in (Formula 1-5):

[0114]

[0115]

[0116] (Formula 1-5)

[0117] Here W, U, and V are equal at every moment, that is, weight sharing.

[0118] This cyclic structure enables RNN to process sequences of arbitrary length and perform well in fields such as natural language processing, speech recognition, and time series prediction. In this way, RNN can learn long-term and short-term dependencies between elements in the sequence and perform effective prediction or classification tasks.

[0119] Long short-term memory (LSTM) is a special type of recurrent neural network (RNN) that solves the gradient vanishing or gradient exploding problems that occur in traditional RNNs when processing long sequences by introducing a gating mechanism (including input gate, forget gate, and output gate).

[0120] Each unit of LSTM decides at each time step what information to retain from the input data, what outdated information to discard, and how to update its internal state, so that it can learn long-term dependencies in long sequence data. Specifically, the model is implemented through forget gate, input gate, output gate and cell state.

[0121] Forget gate: Through the operation of x and ht and the sigmoid function, a vector of 0, 1 is obtained. 0 corresponds to a part of the previous memory that needs to be forgotten, and 1 corresponds to the part of the previous memory that needs to be retained, as shown in (Formula 1-6).

[0122] (Formula 1-6)

[0123] The input gate obtains a new memory state by adding the previous information that needs to be retained and the current information that needs to be remembered, as shown in (Equation 1-7).

[0124]

[0125] (Formula 1-7)

[0126] The cell state is the memory state, as shown in (Equation 1-8).

[0127] (Formula 1-8)

[0128] The output gate is integrated , and obtain an output as shown in (Formula 1-9).

[0129]

[0130] (Formula 1-9)

[0131] This structure makes LSTM perform well in processing tasks such as natural language, speech recognition, and time series prediction that require memorizing long-distance dependencies. In short, LSTM, through its unique gating unit design, can effectively maintain and update information flow, making it more powerful and flexible in sequence modeling.

[0132] 3) Model deployment and integration: Deploy the trained model to the actual industrial control system to identify the protocol type in the network message in real time. Integrate the identification results into the knowledge graph and update the communication protocol information between devices.

[0133] In some embodiments, see Figure 5 , Figure 5 It is a flow chart of steps S501-S503 provided in an embodiment of the present application. The acquisition of network traffic data in a real industrial control network environment can be achieved through steps S501-S503, which will be explained in combination with each step.

[0134] In step S501, ports through which traffic is generated by devices communicating and interacting through a switching node are screened based on data in the network as mirror ports.

[0135] In step S502, a port is selected on the switching node as an observation port, and an analysis device is connected to the port.

[0136] In step S503, the switching node of the network is configured to forward the traffic of the mirror port on the node to the observation port.

[0137] As an example, steps S501-S503 specifically include:

[0138] 1) Mirror port selection: By analyzing the data in the network, filter the ports through which the devices that exchange information through switching nodes (routers, switches, etc.) generate traffic.

[0139] 2) Observation port selection: Select a port on the switching node as the observation port and connect the analysis device to the port.

[0140] 3) Traffic mirroring: By configuring on the switching nodes (routers, switches, etc.) of the network, the traffic of the mirror port on the node is forwarded to the observation port.

[0141] In some embodiments, see Figure 6 , Figure 6 It is a flow chart of steps S601-S604 provided in an embodiment of the present application, wherein the protocol type in the network traffic data is identified based on the protocol type identification model, and the first device type and the second device type are determined by the association between the industrial control protocol and the specific device. This can be achieved through steps S601-S604, which will be explained in conjunction with each step.

[0142] In step S601, the protocol identification model is applied to an actual network traffic analysis environment.

[0143] In step S602, the captured network traffic is analyzed based on the protocol identification model to identify the type of communication protocol used.

[0144] In step S603, for a typical industrial control protocol, based on a first association between the typical industrial control protocol and a first specific device type, a first device type of the device is directly identified.

[0145] In step S604, based on a second association with a second specific device type, a possible second device type is identified, and a candidate set is constructed based on the second device type, wherein the association degree of the first association is greater than that of the second association.

[0146] As an example, steps S601-S604 specifically include:

[0147] 1) Model deployment: Deploy the trained protocol type identification model to the actual network traffic analysis environment to ensure that the model can process the captured network data in real time.

[0148] 2) Protocol type identification: Use the model to analyze the captured network traffic and identify the type of communication protocol used. This step is the basis for device type identification.

[0149] 3) Device type association: For typical industrial control protocols, the strong association between them and specific device types can be used to directly identify the device type. For example, the Modbus protocol is usually associated with PLC and RTU devices.

[0150] 4) Candidate set generation: For those protocols that are weakly associated with a specific device type, the model will identify multiple possible device types to form a candidate set. This requires further analysis and verification to determine the final device type.

[0151] In some embodiments, see Figure 7 , Figure 7 It is a flow chart of steps S701-S703 provided in an embodiment of the present application. The deep processing of the second device type based on the knowledge graph can be achieved through steps S701-S703, which will be explained in combination with each step.

[0152] In step S701, deep data mining is performed based on the device types, attributes, connection relationships, and communication protocols stored in the knowledge graph to obtain association links and feature interaction patterns between device types;

[0153] In step S702, based on the graph neural network of the knowledge graph, the relationship and interaction mode between devices are determined;

[0154] In step S703, the consistency between the actual interaction data in the network traffic and the pattern stored in the knowledge graph is identified through pattern matching.

[0155] As an example, steps S701-S703 specifically include:

[0156] 1) Knowledge graph deep mining: Use the device type, attributes, connection relationship, communication protocol and other information stored in the knowledge graph to conduct deep data mining. Analyze the association links and feature interaction patterns between device types.

[0157] 2) Application of graph reasoning algorithm: A graph neural network (GNN) based on knowledge graph is used to analyze the complex relationships and interaction patterns between devices. This algorithm can help identify implicit connections between device types. Graph convolution is mainly used to infer device types based on the number of network packets.

[0158] In the basic framework of graph convolution, the input layer inputs the adjacency matrix, and the hidden layer multiplies the feature matrix by each node for training. Each time the network is updated, the feature matrix is ​​updated, and the graph structure remains unchanged. After multiple trainings, the feature matrix reaches convergence (that is, the median value of the feature matrix remains basically unchanged). Finally, a classification is performed to give a probability to determine which category the node in the graph belongs to. During the entire network training process, a point product is performed on the node and the feature, realizing the transfer of nodes and features between communities.

[0159] 3) Pattern matching technology: Use pattern matching technology to identify the consistency between the actual interaction data in the network traffic and the patterns stored in the knowledge graph. This helps to verify and infer the device type.

[0160] In summary, the embodiments of the present application have the following beneficial effects:

[0161] (1) Improving recognition accuracy and efficiency: By building a cross-industry knowledge graph, a wealth of device entity, attribute, and relationship information is integrated, making it possible to use more comprehensive data for in-depth analysis and reasoning during the device recognition process, thereby significantly improving recognition accuracy and efficiency.

[0162] (2) Enhanced cross-industry adaptability: The method is not limited to a single industry, but can collect and process industrial control equipment data across industries, construct a knowledge graph with strong versatility, and enhance the applicability and portability of the method in different industries.

[0163] (3) High degree of automation and intelligence: It can automatically extract key device attributes, build relationship types, train protocol type recognition models, and identify protocol types in real time under a real industrial control network environment, significantly improving the level of automation and intelligence in industrial control equipment recognition.

[0164] (4) Support device identification in complex environments: In a real industrial control network environment, it can cope with complex network traffic data, effectively identify unknown or possible device types through deep data mining and pattern matching, and improve the recognition ability in complex environments.

[0165] (5) Real-time and dynamic: By capturing and analyzing network traffic data in real time and combining it with the dynamic update of the knowledge graph, it can reflect the changes in equipment in the industrial control network in real time, providing timely and accurate information support for security monitoring and troubleshooting.

[0166] (6) Improve system security: Accurately identifying the type of industrial control equipment helps to timely discover and prevent potential security threats, and improve the security and stability of industrial control systems. By deeply mining the relationships and interaction patterns between devices, potential security vulnerabilities and attack paths can also be discovered, providing strong support for the formulation of security strategies.

[0167] (7) Reduce operation and maintenance costs: Through automated and intelligent identification methods, the reliance on manual intervention is reduced, and the operation and maintenance costs are reduced. At the same time, accurate identification of equipment types also helps to optimize resource allocation, improve equipment utilization and overall operation and maintenance efficiency.

[0168] In summary, this knowledge graph-based industrial control equipment identification method has shown significant beneficial effects in improving identification accuracy, enhancing cross-industry adaptability, improving automation and intelligence levels, supporting complex environment identification, improving system security, and reducing operation and maintenance costs.

[0169] Based on the same inventive concept, the embodiment of the present application also provides an industrial control equipment identification device based on a knowledge graph corresponding to the industrial control equipment identification method based on a knowledge graph in the first embodiment. Since the principle of solving the problem by the device in the embodiment of the present application is similar to the above-mentioned industrial control equipment identification method based on a knowledge graph, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0170] like Figure 8 As shown, Figure 8 800 is a schematic diagram of the structure of an industrial control equipment identification device 800 based on a knowledge graph provided in an embodiment of the present application. The industrial control equipment identification device 800 based on a knowledge graph includes:

[0171] The data processing module 801 is used to obtain cross-industry industrial control equipment data and pre-process the cross-industry industrial control equipment data;

[0172] A graph construction module 802 is used to construct a knowledge graph based on the preprocessed cross-industry industrial control equipment data, wherein the knowledge graph represents all equipment entities, attributes and relationship information;

[0173] A model training module 803 is used to train and construct a protocol type recognition model based on the network communication data in the cross-industry industrial control equipment data;

[0174] The traffic acquisition module 804 is used to acquire network traffic data in a real industrial control network environment;

[0175] A type determination module 805, configured to identify the protocol type in the network traffic data based on the protocol type identification model, and determine a first device type and a second device type through the association between the industrial control protocol and the specific device, wherein the first device type represents a determined device type, and the second device type represents a possible device type or an unknown device type;

[0176] The deep processing module 806 is used to perform deep processing on the second device type based on the knowledge graph, determine the unique type of possible device types, and infer unknown device types.

[0177] Those skilled in the art should understand that Figure 8 The implementation functions of each unit in the industrial control equipment identification device 800 based on knowledge graph shown can be understood by referring to the relevant description of the aforementioned industrial control equipment identification method based on knowledge graph. Figure 8 The functions of each unit in the knowledge graph-based industrial control equipment identification device 800 shown can be implemented by a program running on a processor, or by a specific logic circuit.

[0178] In a possible implementation, the data processing module 801 obtains cross-industry industrial control equipment data, including:

[0179] Determine the overall architecture of different industrial control networks, and perform statistical processing on the devices included in each layer of the overall architecture to build a candidate set for the results of industrial control device identification;

[0180] Collecting data for the devices included in each layer to determine basic attributes of different devices, wherein the data collection includes at least one of technical parameter collection, performance index collection, and functional feature collection;

[0181] Determine the connection relationship and communication protocol between devices in different industrial control networks, and collect business traffic, wherein the business traffic includes data transmission volume and transmission rate, data type, transmission mode and time characteristics, the data type includes at least one of sensor data, control commands, status information, and log records, the transmission mode includes at least one of periodic transmission, event-triggered transmission and real-time streaming data, and the time characteristics include at least one of transmission frequency, peak time and data timeliness;

[0182] The data processing module 801 pre-processes the cross-industry industrial control equipment data, including:

[0183] Perform at least one of the following processing on the cross-industry industrial control equipment data: cleaning, sorting, conversion, and standardization;

[0184] The data processing module 801 also includes:

[0185] The preprocessed data is organized logically and functionally and stored in a matching database.

[0186] In a possible implementation, the graph construction module 802 constructs a knowledge graph based on the preprocessed cross-industry industrial control equipment data, including:

[0187] Automatically obtain and identify the names of various industrial control devices based on named entity recognition;

[0188] Extracting key attributes of the industrial control device from the device description and related documents of the industrial control device, and associating the key attributes with the corresponding industrial control device, wherein the key attributes include at least one of technical parameters, performance indicators, and functional characteristics;

[0189] Identify and define the relationship type between industrial control devices, wherein the relationship type includes at least one of a physical connection, a logical connection, and a business connection, and classify the relationship type, wherein the classification category includes at least one of a direct connection, an indirect connection, and a dependency relationship, so as to represent it in the knowledge graph;

[0190] Extracting connectivity relationships between industrial control devices based on rule matching, and converting the connectivity relationships into edges in the knowledge graph, wherein the rules are determined based on device type, communication protocol, and connection mode;

[0191] The device entities, attributes and relationship information of all industrial control devices are integrated into the knowledge graph. The connectivity of the knowledge graph is determined by checking whether there is a path between any two nodes, represented by the connection matrix A. The density D of the graph of the knowledge graph is the possible maximum value of the number of edges E of the graph and the number of edges in the complete graph. The ratio is:

[0192]

[0193] Where N is the total number of nodes in the graph.

[0194] In a possible implementation, the model training module 803 trains and constructs a protocol type recognition model based on the network communication data in the cross-industry industrial control equipment data, including:

[0195] Based on the collected network message data of known protocols, key features in the network message are extracted, wherein the key features include at least one of the port number, transport layer protocol, application layer protocol, IP address, message length, and byte sequence of payload content;

[0196] Constructing a deep learning model based on the key features, wherein the deep learning model is used to identify the protocol type of the message data, and the deep learning model is a hybrid model of three models: a convolutional neural network (CNN), a recurrent neural network (RNN), and a long short-term memory (LSTM) network;

[0197] The trained deep learning model is deployed as the protocol type recognition model in the target industrial control system to identify the protocol type in the network message in real time, and the recognition result is integrated into the knowledge graph to update the communication protocol information between devices.

[0198] In a possible implementation, the traffic acquisition module 804 acquires network traffic data in a real industrial control network environment, including:

[0199] Based on the data screening in the network, the port through which the traffic is generated by the device communicating and interacting through the switching node is used as the mirror port;

[0200] Select a port on the switching node as the observation port and connect the analysis device to the port;

[0201] The switching node of the network is configured to forward the traffic of the mirror port on the node to the observation port.

[0202] In a possible implementation, the type determination module 805 identifies the protocol type in the network traffic data based on the protocol type identification model, and determines the first device type and the second device type through the association between the industrial control protocol and the specific device, including:

[0203] Applying the protocol identification model to an actual network traffic analysis environment;

[0204] Analyze the captured network traffic based on the protocol identification model to identify the type of communication protocol used;

[0205] For a typical industrial control protocol, directly identifying a first device type of the device based on a first association between the typical industrial control protocol and a first specific device type;

[0206] Based on a second association with a second specific device type, a possible second device type is identified, and a candidate set is constructed based on the second device type, wherein the association degree of the first association is greater than the second association.

[0207] In a possible implementation, the deep processing module 806 performs deep processing on the second device type based on the knowledge graph, including:

[0208] Perform deep data mining based on the device types, attributes, connection relationships, and communication protocols stored in the knowledge graph to obtain association links and feature interaction patterns between device types;

[0209] Determine the relationship and interaction mode between devices based on the graph neural network of the knowledge graph;

[0210] The consistency between actual interaction data in network traffic and the patterns stored in the knowledge graph is identified through pattern matching.

[0211] The above-mentioned industrial control equipment identification device based on knowledge graph has the following beneficial effects:

[0212] (1) Improving recognition accuracy and efficiency: By building a cross-industry knowledge graph, a wealth of device entity, attribute, and relationship information is integrated, making it possible to use more comprehensive data for in-depth analysis and reasoning during the device recognition process, thereby significantly improving recognition accuracy and efficiency.

[0213] (2) Enhanced cross-industry adaptability: The method is not limited to a single industry, but can collect and process industrial control equipment data across industries, construct a knowledge graph with strong versatility, and enhance the applicability and portability of the method in different industries.

[0214] (3) High degree of automation and intelligence: It can automatically extract key device attributes, build relationship types, train protocol type recognition models, and identify protocol types in real time under a real industrial control network environment, significantly improving the level of automation and intelligence in industrial control equipment recognition.

[0215] (4) Support device identification in complex environments: In a real industrial control network environment, it can cope with complex network traffic data, effectively identify unknown or possible device types through deep data mining and pattern matching, and improve the recognition ability in complex environments.

[0216] (5) Real-time and dynamic: By capturing and analyzing network traffic data in real time and combining it with the dynamic update of the knowledge graph, it can reflect the changes in equipment in the industrial control network in real time, providing timely and accurate information support for security monitoring and troubleshooting.

[0217] (6) Improve system security: Accurately identifying the type of industrial control equipment helps to timely discover and prevent potential security threats, and improve the security and stability of industrial control systems. By deeply mining the relationships and interaction patterns between devices, potential security vulnerabilities and attack paths can also be discovered, providing strong support for the formulation of security strategies.

[0218] (7) Reduce operation and maintenance costs: Through automated and intelligent identification methods, the reliance on manual intervention is reduced, and the operation and maintenance costs are reduced. At the same time, accurate identification of equipment types also helps to optimize resource allocation, improve equipment utilization and overall operation and maintenance efficiency.

[0219] In summary, this knowledge graph-based industrial control equipment identification method has shown significant beneficial effects in improving identification accuracy, enhancing cross-industry adaptability, improving automation and intelligence levels, supporting complex environment identification, improving system security, and reducing operation and maintenance costs.

[0220] like Fig. 9 As shown, Fig. 9 The present invention provides a schematic diagram of the structure of an electronic device 900, wherein the electronic device 900 includes:

[0221] A processor 901, a storage medium 902 and a bus 903, wherein the storage medium 902 stores machine-readable instructions executable by the processor 901. When the electronic device 900 is running, the processor 901 communicates with the storage medium 902 through the bus 903, and the processor 901 executes the machine-readable instructions to perform the steps of the industrial control equipment identification method based on the knowledge graph described in the embodiment of the present application.

[0222] In actual application, the components in the electronic device 900 are coupled together via the bus 903. It is understood that the bus 903 is used to realize the connection and communication between these components. In addition to the data bus, the bus 903 also includes a power bus, a control bus, and a status signal bus. However, for the sake of clarity, Fig. 9 Various buses are labeled as bus 903.

[0223] The above electronic device has the following beneficial effects:

[0224] (1) Improving recognition accuracy and efficiency: By building a cross-industry knowledge graph, a wealth of device entity, attribute, and relationship information is integrated, making it possible to use more comprehensive data for in-depth analysis and reasoning during the device recognition process, thereby significantly improving recognition accuracy and efficiency.

[0225] (2) Enhanced cross-industry adaptability: The method is not limited to a single industry, but can collect and process industrial control equipment data across industries, construct a knowledge graph with strong versatility, and enhance the applicability and portability of the method in different industries.

[0226] (3) High degree of automation and intelligence: It can automatically extract key device attributes, build relationship types, train protocol type recognition models, and identify protocol types in real time under a real industrial control network environment, significantly improving the level of automation and intelligence in industrial control equipment recognition.

[0227] (4) Support device identification in complex environments: In a real industrial control network environment, it can cope with complex network traffic data, effectively identify unknown or possible device types through deep data mining and pattern matching, and improve the recognition ability in complex environments.

[0228] (5) Real-time and dynamic: By capturing and analyzing network traffic data in real time and combining it with the dynamic update of the knowledge graph, it can reflect the changes in equipment in the industrial control network in real time, providing timely and accurate information support for security monitoring and troubleshooting.

[0229] (6) Improve system security: Accurately identifying the type of industrial control equipment helps to timely discover and prevent potential security threats, and improve the security and stability of industrial control systems. By deeply mining the relationships and interaction patterns between devices, potential security vulnerabilities and attack paths can also be discovered, providing strong support for the formulation of security strategies.

[0230] (7) Reduce operation and maintenance costs: Through automated and intelligent identification methods, the reliance on manual intervention is reduced, and the operation and maintenance costs are reduced. At the same time, accurate identification of equipment types also helps to optimize resource allocation, improve equipment utilization and overall operation and maintenance efficiency.

[0231] In summary, this knowledge graph-based industrial control equipment identification method has shown significant beneficial effects in improving identification accuracy, enhancing cross-industry adaptability, improving automation and intelligence levels, supporting complex environment identification, improving system security, and reducing operation and maintenance costs.

[0232] The embodiment of the present application also provides a computer-readable storage medium, which stores executable instructions. When the executable instructions are executed by at least one processor 901, the industrial control equipment identification method based on the knowledge graph described in the embodiment of the present application is implemented.

[0233] In some embodiments, the storage medium can be a magnetic random access memory (FRAM), a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a flash memory, a magnetic surface storage, an optical disk, or a compact disc read-only memory (CD-ROM); it can also be various devices including one or any combination of the above memories.

[0234] In some embodiments, executable instructions may be in the form of a program, software, software module, script or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as a stand-alone program or as a module, component, subroutine or other unit suitable for use in a computing environment.

[0235] As an example, executable instructions may, but do not necessarily, correspond to a file in a file system, may be stored as part of a file that stores other programs or data, for example, in one or more scripts in a HyperText Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple coordinated files (for example, files storing one or more modules, subroutines, or code portions).

[0236] By way of example, executable instructions may be deployed to be executed on one computing device, or on multiple computing devices located at one site, or on multiple computing devices distributed across multiple sites and interconnected by a communication network.

[0237] The computer-readable storage medium has the following advantages:

[0238] (1) Improving recognition accuracy and efficiency: By building a cross-industry knowledge graph, a wealth of device entity, attribute, and relationship information is integrated, making it possible to use more comprehensive data for in-depth analysis and reasoning during the device recognition process, thereby significantly improving recognition accuracy and efficiency.

[0239] (2) Enhanced cross-industry adaptability: The method is not limited to a single industry, but can collect and process industrial control equipment data across industries, construct a knowledge graph with strong versatility, and enhance the applicability and portability of the method in different industries.

[0240] (3) High degree of automation and intelligence: It can automatically extract key device attributes, build relationship types, train protocol type recognition models, and identify protocol types in real time under a real industrial control network environment, significantly improving the level of automation and intelligence in industrial control equipment recognition.

[0241] (4) Support device identification in complex environments: In a real industrial control network environment, it can cope with complex network traffic data, effectively identify unknown or possible device types through deep data mining and pattern matching, and improve the recognition ability in complex environments.

[0242] (5) Real-time and dynamic: By capturing and analyzing network traffic data in real time and combining it with the dynamic update of the knowledge graph, it can reflect the changes in equipment in the industrial control network in real time, providing timely and accurate information support for security monitoring and troubleshooting.

[0243] (6) Improve system security: Accurately identifying the type of industrial control equipment helps to timely discover and prevent potential security threats, and improve the security and stability of industrial control systems. By deeply mining the relationships and interaction patterns between devices, potential security vulnerabilities and attack paths can also be discovered, providing strong support for the formulation of security strategies.

[0244] (7) Reduce operation and maintenance costs: Through automated and intelligent identification methods, the reliance on manual intervention is reduced, and the operation and maintenance costs are reduced. At the same time, accurate identification of equipment types also helps to optimize resource allocation, improve equipment utilization and overall operation and maintenance efficiency.

[0245] In summary, this knowledge graph-based industrial control equipment identification method has shown significant beneficial effects in improving identification accuracy, enhancing cross-industry adaptability, improving automation and intelligence levels, supporting complex environment identification, improving system security, and reducing operation and maintenance costs.

[0246] In the several embodiments provided in the present application, it should be understood that the disclosed methods and electronic devices can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0247] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0248] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0249] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a platform server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard drives, ROM, RAM, magnetic disks, or optical disks.

[0250] The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for identifying industrial control equipment based on knowledge graph, characterized in that: The method comprises: Acquire cross-industry industrial control equipment data, and pre-process the cross-industry industrial control equipment data; Constructing a knowledge graph based on the preprocessed cross-industry industrial control equipment data, wherein the knowledge graph represents all equipment entities, attributes, and relationship information; Training and building a protocol type recognition model based on network communication data in the cross-industry industrial control equipment data; Obtain network traffic data in a real industrial control network environment; Identify the protocol type in the network traffic data based on the protocol type identification model, and determine a first device type and a second device type through the association between the industrial control protocol and the specific device, wherein the first device type represents a determined device type, and the second device type represents a possible device type or an unknown device type; Deeply processing the second device type based on the knowledge graph to determine a unique type of possible device types and infer unknown device types; The deep processing of the second device type based on the knowledge graph includes: Perform deep data mining based on the device types, attributes, connection relationships, and communication protocols stored in the knowledge graph to obtain association links and feature interaction patterns between device types; Determine the relationship and interaction mode between devices based on the graph neural network of the knowledge graph; The consistency between actual interaction data in network traffic and the patterns stored in the knowledge graph is identified through pattern matching.

2. The method according to claim 1, characterized in that The acquisition of cross-industry industrial control equipment data includes: Determine the overall architecture of different industrial control networks, and perform statistical processing on the devices included in each layer of the overall architecture to build a candidate set for the results of industrial control device identification; Collecting data for the devices included in each layer to determine basic attributes of different devices, wherein the data collection includes at least one of technical parameter collection, performance index collection, and functional feature collection; Determine the connection relationship and communication protocol between devices in different industrial control networks, and collect business traffic, wherein the business traffic includes data transmission volume and transmission rate, data type, transmission mode and time characteristics, the data type includes at least one of sensor data, control commands, status information, and log records, the transmission mode includes at least one of periodic transmission, event-triggered transmission and real-time streaming data, and the time characteristics include at least one of transmission frequency, peak time and data timeliness; The preprocessing of the cross-industry industrial control equipment data includes: Perform at least one of the following processing on the cross-industry industrial control equipment data: cleaning, sorting, conversion, and standardization; The method further comprises: The preprocessed data is organized logically and functionally and stored in a matching database.

3. The method according to claim 1, characterized in that: The constructing of a knowledge graph based on the preprocessed cross-industry industrial control equipment data includes: Automatically obtain and identify the names of various industrial control devices based on named entity recognition; Extracting key attributes of the industrial control device from the device description and related documents of the industrial control device, and associating the key attributes with the corresponding industrial control device, wherein the key attributes include at least one of technical parameters, performance indicators, and functional characteristics; Identify and define the relationship type between industrial control devices, wherein the relationship type includes at least one of a physical connection, a logical connection, and a business connection, and classify the relationship type, wherein the classification category includes at least one of a direct connection, an indirect connection, and a dependency relationship, so as to represent it in the knowledge graph; Extracting connectivity relationships between industrial control devices based on rule matching, and converting the connectivity relationships into edges in the knowledge graph, wherein the rules are determined based on device type, communication protocol, and connection mode; The device entities, attributes and relationship information of all industrial control devices are integrated into the knowledge graph. The connectivity of the knowledge graph is determined by checking whether there is a path between any two nodes, represented by the connection matrix A. The density D of the graph of the knowledge graph is the possible maximum value of the number of edges E of the graph and the number of edges in the complete graph. The ratio is: Where N is the total number of nodes in the graph.

4. The method according to claim 1, characterized in that: The training and building of a protocol type recognition model based on the network communication data in the cross-industry industrial control equipment data includes: Based on the collected network message data of known protocols, key features in the network message are extracted, wherein the key features include at least one of the port number, transport layer protocol, application layer protocol, IP address, message length, and byte sequence of payload content; Constructing a deep learning model based on the key features, wherein the deep learning model is used to identify the protocol type of the message data, and the deep learning model is a hybrid model of three models: a convolutional neural network (CNN), a recurrent neural network (RNN), and a long short-term memory (LSTM) network; The trained deep learning model is deployed as the protocol type recognition model in the target industrial control system to identify the protocol type in the network message in real time, and the recognition result is integrated into the knowledge graph to update the communication protocol information between devices.

5. The method according to claim 1, characterized in that The obtaining of network traffic data in a real industrial control network environment includes: Based on the data screening in the network, the port through which the traffic is generated by the device communicating and interacting through the switching node is used as the mirror port; Select a port on the switching node as the observation port and connect the analysis device to the port; The switching node of the network is configured to forward the traffic of the mirror port on the node to the observation port.

6. The method according to claim 1, characterized in that The identifying the protocol type in the network traffic data based on the protocol type identification model, and determining the first device type and the second device type through the association between the industrial control protocol and the specific device, includes: Deploy the protocol identification model into an actual network traffic analysis environment; Analyze the captured network traffic based on the protocol identification model to identify the type of communication protocol used; For a typical industrial control protocol, directly identifying a first device type of the device based on a first association between the typical industrial control protocol and a first specific device type; Based on a second association with a second specific device type, a possible second device type is identified, and a candidate set is constructed based on the second device type, wherein the association degree of the first association is greater than the second association.

7. An industrial control equipment identification device based on knowledge graph, characterized in that: The device comprises: A data processing module, used to obtain cross-industry industrial control equipment data and pre-process the cross-industry industrial control equipment data; A graph construction module, used to construct a knowledge graph based on the preprocessed cross-industry industrial control equipment data, wherein the knowledge graph represents all equipment entities, attributes and relationship information; A model training module, used for training and building a protocol type recognition model based on the network communication data in the cross-industry industrial control equipment data; Traffic acquisition module, used to obtain network traffic data in a real industrial control network environment; a type determination module, configured to identify the protocol type in the network traffic data based on the protocol type identification model, and determine a first device type and a second device type through an association between an industrial control protocol and a specific device, wherein the first device type represents a determined device type, and the second device type represents a possible device type or an unknown device type; A deep processing module is used to perform deep processing on the second device type based on the knowledge graph, determine the unique type of possible device types, and infer unknown device types; the deep processing of the second device type based on the knowledge graph includes: performing deep data mining processing based on the device types, attributes, connection relationships and communication protocols stored in the knowledge graph to obtain the associated links and characteristic interaction patterns between device types; determining the relationship and interaction pattern between devices based on the graph neural network of the knowledge graph; identifying the consistency between the actual interaction data in the network traffic and the pattern stored in the knowledge graph through pattern matching.

8. An electronic device, characterized in that: include: A processor, a storage medium and a bus, wherein the storage medium stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the storage medium through the bus, and the processor executes the machine-readable instructions to execute the industrial control equipment identification method based on the knowledge graph as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the industrial control equipment identification method based on the knowledge graph as described in any one of claims 1 to 6 is executed.