Methods, systems, equipment, and media for generating attack chains in industrial control systems

By constructing a trust graph and identifying abnormal logs using behavioral baselines, and combining this with reverse recursive traversal to generate an attack path graph, the problem of existing technologies being unable to identify covert attacks and automatically reconstruct attack paths has been solved, thus achieving efficient and secure monitoring of industrial control systems.

CN123122864APending Publication Date: 2026-11-17HUANENG (FUJIAN) ENERGY DEVELOPMENT LIMITED COMPANY FUZHOU BRANCH +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611036254.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-13
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies cannot effectively identify and automatically reconstruct covert attack paths in industrial control systems, especially attacks using unknown vulnerabilities or slow penetration methods. This results in a large number of covert attack behaviors being ignored and attack paths not being fully reconstructed.

Method used

By receiving trusted logs, a trust graph is constructed. Abnormal logs are filtered based on the behavioral baseline of entities. An attack path graph is generated by recursively traversing the abnormal logs in reverse. Combined with digital signatures and graph neural networks to learn behavioral patterns, the attack path is automatically traced.

Benefits of technology

It can effectively identify low-frequency, slow-penetrating, and other covert attacks that do not match any known features, and automatically reconstruct the complete attack path without the need for manual retrieval, thereby improving the security and emergency response efficiency of industrial control systems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123122864A_ABST
    Figure CN123122864A_ABST
Patent Text Reader

Abstract

This invention relates to the field of information security, specifically to a method, system, device, and medium for generating attack chains for industrial control systems. The method includes: receiving trusted logs sent by all entities; constructing a trust graph using the entities and operation objects corresponding to all trusted logs as nodes and the corresponding operation types as edges; concatenating all received trusted logs into a trusted log sequence based on the trust graph in chronological order; filtering out abnormal logs in the trusted log sequence based on a preset behavioral baseline for each entity; and recursively traversing all nodes in the trust graph that can be associated with the endpoint through edges, using the node corresponding to the operation object involved in the abnormal log as the endpoint, to generate an attack path graph. This invention can automatically trace all preceding operations without manual retrieval, thus solving the problem that existing methods cannot identify covert attacks and cannot automatically reconstruct the complete attack path.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of information security, specifically to a method, system, device, and medium for generating attack chains for industrial control systems. Background Technology

[0002] Industrial control systems are critical cyber-physical systems used to monitor and control industrial production processes. Typical examples include distributed control systems (DCS), programmable logic controllers (PLCs), and supervisory control and data acquisition (SCADA) systems. These systems are widely deployed in critical national infrastructure sectors such as power, petrochemicals, metallurgy, and water treatment, and their safe and stable operation is directly related to national economic development and public safety. In recent years, with the deep integration of the Industrial Internet and information technology, traditionally relatively closed industrial control systems have gradually been exposed to open network environments, facing increasingly severe cybersecurity threats. Attackers often follow a complete attack chain model to intrude into industrial control systems. This model typically includes multiple stages such as reconnaissance and detection, initial intrusion, weaponized delivery, vulnerability exploitation, backdoor implantation, remote control, lateral movement, and the final target operation (such as disrupting production, tampering with process parameters, or triggering a safety accident). Accurately tracing the attack path of industrial control systems can not only reconstruct the attacker's complete intrusion trajectory and identify weaknesses in the security protection system, but also provide direct evidence for emergency response and system hardening, thereby effectively blocking the subsequent evolution of the attack chain and preventing catastrophic production accidents.

[0003] Currently, attack attribution methods for industrial control systems typically involve collecting and centrally storing system logs scattered across various devices. These logs are then retrieved and analyzed based on pre-defined static rules (such as feature matching and threshold alarm rules) to attempt to identify abnormal behavior and trace the attack source. However, this method has a significant drawback: its analytical capabilities heavily rely on pre-defined static rules. For unknown attacks, zero-day exploits, or attacks implemented through slow penetration or low-frequency operations, the system cannot effectively identify and warn of attacks because these attack patterns do not match any existing rule features. Furthermore, it struggles to automatically extract the inherent correlations between multi-stage attacks from massive amounts of logs, resulting in many covert attack behaviors being overlooked and attack paths not being fully reconstructed. Summary of the Invention

[0004] To address the aforementioned problems, this invention provides a method, system, device, and medium for generating attack chains for industrial control systems.

[0005] The first aspect of this invention discloses a method for generating attack chains for industrial control systems, comprising: Receive trusted logs sent by all entities; wherein the trusted logs include operation type and operation object; A trust graph is constructed using the entities and operation objects corresponding to all trusted logs as nodes and the corresponding operation types as edges. Based on the trust graph, all received trusted logs are concatenated into a trusted log sequence in chronological order. Based on a preset behavioral baseline for each entity, abnormal logs in the trusted log sequence are filtered out. Using the node corresponding to the operation object involved in the abnormal log as the endpoint, all nodes that can be associated with the endpoint through edges are recursively traversed in reverse in the trust graph to generate an attack path graph.

[0006] Furthermore, the trusted log also includes a timestamp; And, the step of concatenating all received trusted logs into a trusted log sequence based on the trust graph in chronological order includes: Based on the timestamp, and according to the position of the entity and operation object corresponding to each trusted log in the node of the trust graph, trusted logs with the same node are connected in chronological order to generate a trusted log sequence.

[0007] Furthermore, the step of obtaining the behavioral baseline includes: Obtain the historical logs of the industrial control system; The security history log is obtained by filtering the historical logs to obtain the logs of the industrial control system when it was not attacked. The security history logs are parsed to extract the security operation model corresponding to each security history log; wherein, the security operation model refers to an entity's operation on an operation object; The safety operation model is learned using a graph neural network to obtain the behavioral baseline of each entity.

[0008] Furthermore, the trusted log also includes the entity's digital signature on the trusted log and the entity's identity identifier; Furthermore, the step of filtering out anomalous logs from the trusted log sequence based on a preset behavioral baseline for each entity includes: For each trusted log in the trusted log sequence: Obtain the public key of the entity corresponding to the trusted log based on the identity identifier; The digital signature is verified using the public key, and the verification result is determined. If the verification is successful, the trusted log is determined to be an abnormal log based on the preset behavior baseline of the entity corresponding to the trusted log. If the verification fails, the trusted log is determined to be an abnormal log.

[0009] Furthermore, the digital signature is generated by the entity using a national cryptographic signature algorithm within its trusted execution environment or built-in secure chip.

[0010] Furthermore, based on the preset behavioral baseline of the entity corresponding to the trusted log, the steps for determining whether the trusted log is an abnormal log include: Determine whether the operation type corresponding to the trusted log belongs to the preset behavior baseline of the entity corresponding to the trusted log; If it does not belong to the category, then the trusted log is determined to be an abnormal log.

[0011] Furthermore, the step of generating an attack path graph by recursively traversing all nodes that can be associated with the endpoint through edges in the trust graph, using the node corresponding to the operation object involved in the abnormal log as the endpoint, includes: Taking the node corresponding to the operation object involved in the abnormal log as the endpoint, all nodes that can be associated with the endpoint through edges are recursively traversed in the trust graph to generate multiple attack paths. The confidence level of each attack path is calculated based on the risk level of each operation type on each attack path and the degree of deviation of each operation type from the behavioral baseline of the corresponding entity. Select the attack path with the highest confidence level and generate an attack path graph.

[0012] A second aspect of this invention discloses an industrial control system attack chain generation system, comprising: A receiving module is used to receive trusted logs sent by all entities; wherein, the trusted logs include operation types and operation objects; The building module is used to construct a trust graph with the entities and operation objects corresponding to all trusted logs as nodes and the corresponding operation types as edges. The concatenation module is used to concatenate all received trusted logs into a trusted log sequence according to the trust graph in chronological order. The filtering module is used to filter out abnormal logs in the trusted log sequence based on a preset behavioral baseline for each entity. The generation module is used to generate an attack path graph by recursively traversing all nodes that can be associated with the endpoint through edges in the trust graph, with the node corresponding to the operation object involved in the abnormal log as the endpoint.

[0013] A third aspect of the present invention discloses an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the steps of any of the industrial control system attack chain generation methods disclosed in the first aspect of the present invention.

[0014] The fourth aspect of the present invention discloses a storage medium storing a computer program that, when executed by a processor, implements the steps of any of the industrial control system attack chain generation methods disclosed in the first aspect of the present invention.

[0015] Since this invention no longer relies on preset static rules for anomaly detection, but instead identifies any behavior that deviates from the normal operating mode based on the behavioral baseline of each entity, even low-frequency, slow-penetration, and other covert attacks that do not match any known features can be effectively detected. At the same time, by establishing operational semantic associations between nodes and edges in the trust graph, combined with reverse recursive traversal ending at the anomaly log, all preceding operations can be automatically traced without manual retrieval, thus solving the problem that existing methods cannot identify covert attacks and cannot automatically reconstruct the complete attack path. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating a method for generating an attack chain for an industrial control system, as disclosed in an embodiment of the present invention. Figure 2 This is a schematic diagram of the structure of an industrial control system attack chain generation system disclosed in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of the electronic device disclosed in the embodiments of the present invention. Detailed Implementation

[0018] To enable those skilled in the art to better understand the present invention, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, or product comprising a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, apparatus, or products.

[0020] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0021] Please see Figure 1 As shown, Figure 1 This is a flowchart illustrating a method for generating attack chains for industrial control systems, as disclosed in an embodiment of the present invention. Figure 1 As shown, the industrial control system attack chain generation method may include the following operations: S101. Receive trusted logs sent by all entities; wherein, the trusted logs include operation type and operation object; In this trusted embodiment, an entity refers to a pre-determined key device or software module in the industrial control system that is configured with a digital certificate. This entity can be a user, device, or process. Examples include controllers, engineering workstations, operator workstations, server cryptographic machines, signature verification servers, VPN gateways, critical application software, and maintenance USB keys. The digital certificate preferably uses a national cryptographic standard digital certificate. Operation objects include, for example, data files, controllers, network connections or configuration files, and devices. Operation types include, for example, login, configuration download, parameter modification, communication connection, or certificate verification.

[0022] Each entity communicates with the security management platform through a secure transmission channel based on national cryptographic algorithms. All trusted logs are transmitted within this encrypted channel to prevent eavesdropping or tampering. This embodiment of the invention does not limit the specific format of the trusted logs. As an example, an extended Syslog format can be used, which embeds the entity's digital signature value and identity identifier in the header or custom fields of a standard Syslog message, and includes core content such as operation type, operation object, and key parameters as the log body.

[0023] S102. Construct a trust graph using the entities and operation objects corresponding to all trusted logs as nodes and the corresponding operation types as edges. In this trusted embodiment, the trust graph is used to record the operational relationships between elements in the industrial control system. The construction process of the trust graph is as follows: For each received trusted log, the entity performing the operation and the operation object are extracted from the trusted log, and these two are used as two nodes in the trust graph. If an entity or operation object has not yet appeared in the graph, a new corresponding node is added; if it already exists, the existing node is reused. Subsequently, the operation type in the trusted log is extracted, and this operation type is used as the label of the directed edge, pointing from the node corresponding to the entity to the node corresponding to the operation object. At the same time, the timestamp, key parameters, and other information of the trusted log are stored as attributes of the edges. As new trusted logs continuously arrive, the above process continues, and the nodes and edges in the trust graph will be continuously expanded or updated, thereby forming a dynamic trust graph that reflects the system's operating history.

[0024] For example, a trusted log from the engineer station (entity identifier ENG_01) is received, containing the message "Modify the pressure setting value of controller PLC_101 to 120". The parsed entity is identified as ENG_01, the operation object as PLC_101, and the operation type as "Set pressure value". Two nodes, ENG_01 and PLC_101, are located or created in the trust graph, and a directed edge from ENG_01 to PLC_101 is added, labeled "Set pressure value", along with the timestamp and other parameters of the trusted log. If a subsequent trusted log from the operator station (OP_02) is received, containing the message "Start pump P_001", the nodes OP_02 and P_001 are similarly created or located, and a "Start" edge is added. When different trusted logs share nodes, these edges naturally connect into a chain, providing a structural foundation for subsequent operation sequence chaining and attack path tracing.

[0025] S103. Based on the trust graph, concatenate all received trusted logs into a trusted log sequence according to the time order. In a trusted embodiment, the trusted log further includes a timestamp; And, the step of concatenating all received trusted logs into a trusted log sequence based on the trust graph in chronological order includes: Based on the timestamp, and according to the position of the entity and operation object corresponding to each trusted log in the node of the trust graph, trusted logs with the same node are connected in chronological order to generate a trusted log sequence.

[0026] For example, the system received and parsed the following three trusted log entries (arranged in the actual order of receipt): Trusted Log A: Timestamp 10:00:05, Entity is Operator Station OP_01, Operation Object is Valve V001, Operation Type is "Open"; Trusted Log B: Timestamp 10:00:00, Entity is Operation and Maintenance UKey KEY_01, Operation Object is Operator Station OP_01, Operation Type is "Authentication Login"; Trusted Log C: Timestamp 10:00:10, Entity is Valve V001, Operation Object is Pipeline PIPE_01, Operation Type is "Transporting Medium".

[0027] First, the three trusted logs are sorted according to their timestamps: the earliest is trusted log B, followed by trusted log A, and the latest is trusted log C. Then, the nodes in the constructed trust graph are connected. In the trust graph, the operation target node of trusted log B is OP_01, and the entity node of trusted log A is also OP_01. Since they share the same node OP_01, trusted log B and trusted log A are connected chronologically, forming a partial sequence: KEY_01 → OP_01 → V001. Next, the operation target node of trusted log A is V001, and the entity node of trusted log C is also V001. Again, they share the same node V001, so trusted log C is connected to the end of the sequence, forming the complete sequence: KEY_01 → OP_01 → V001 → PIPE_01. If a trusted log does not share a node with other trusted logs (e.g., a completely independent device operation record), it will not be added to that sequence but will instead be the first record of another independent sequence. In this way, scattered trusted log events can be organized into trusted log sequences with causal relationships, providing structured input for subsequent behavioral baseline comparisons.

[0028] By introducing timestamps into trusted logs and linking trusted logs with the same nodes into a sequence according to time order, originally scattered log events from different entities can be organized according to causal logic. Since an operation in an industrial control system often triggers subsequent operations, and these operations share the same nodes, connecting them using timestamps and node positions can automatically reconstruct the dependencies between operations. Unlike simple time-based sorting, this optional embodiment only links optional logs with the same nodes, avoiding the incorrect chaining of irrelevant events, thereby generating a semantically correct sequence of operations and improving the accuracy of attack chain reconstruction.

[0029] S104. Based on the preset behavior baseline of each entity, filter out the abnormal logs in the trusted log sequence; In an optional embodiment, the step of obtaining the behavioral baseline includes: Obtain the historical logs of the industrial control system; The security history log is obtained by filtering the historical logs to obtain the logs of the industrial control system when it was not attacked. The security history logs are parsed to extract the security operation model corresponding to each security history log; wherein, the security operation model refers to an entity's operation on an operation object; The safety operation model is learned using a graph neural network to obtain the behavioral baseline of each entity.

[0030] This optional embodiment extracts the security operation model of each subject to the object of operation by acquiring historical logs when the system is not under attack, and uses a graph neural network to learn the behavioral baseline of each subject. The graph neural network can capture complex interaction patterns between subjects and objects of operation, including frequently occurring operation types, operation sequences, and implicit constraints, rather than simply recording isolated individual operations. Compared to methods based on fixed rules or simple statistics, this learning approach based on historical normal behavior can automatically adapt to the business characteristics of different industrial sites, forming personalized baselines. When the system suffers an unknown attack, even if the attacker bypasses traditional rule detection, their operation pattern will be identified because it deviates from this baseline, thereby improving the detection capability of zero-day attacks and covert threats.

[0031] In an optional embodiment, the trusted log further includes the entity's digital signature on the trusted log and the entity's identity identifier; Furthermore, the step of filtering out anomalous logs from the trusted log sequence based on a preset behavioral baseline for each entity includes: For each trusted log in the trusted log sequence: Obtain the public key of the entity corresponding to the trusted log based on the identity identifier; The digital signature is verified using the public key, and the verification result is determined. If the verification is successful, the trusted log is determined to be an abnormal log based on the preset behavior baseline of the entity corresponding to the trusted log. If the verification fails, the trusted log is determined to be an abnormal log.

[0032] In this optional embodiment, the identity identifier may be a certificate serial number or a fixed-length hash value obtained by hashing the certificate serial number, which is uniquely bound to the entity and cannot be changed during system operation.

[0033] This optional implementation combines digital signature verification with behavioral baseline checks to filter anomalous trusted logs. This dual verification mechanism addresses different types of threats: signature verification failure typically indicates that the log has been tampered with, forged, or originated from an unauthorized entity, constituting an integrity or authenticity attack; while successful signature verification but deviation from the behavioral baseline suggests that the legitimate entity's credentials may have been stolen, or that an insider has performed unauthorized operations. The two complement each other, avoiding the blind spots of single verification and thus more comprehensively identifying anomalies.

[0034] In an optional embodiment, the digital signature is generated by the entity using a national cryptographic signature algorithm within its trusted execution environment or built-in secure chip.

[0035] This optional implementation ensures the security of the private key from the source. A trusted execution environment or secure chip prevents attackers from reading the private key stored within it. Even if the entity is remotely controlled, attackers cannot forge the signatures of other entities or tamper with the signature generation process. Simultaneously, the national cryptographic algorithm complies with national cryptographic management requirements and is suitable for critical infrastructure scenarios. This feature makes subsequent signature verification results highly reliable: once signature verification passes, it can be assured that the log content was indeed generated by the entity and has not been modified. This trust foundation supports the reliability of the entire attack chain generation method, avoiding misjudgments due to key leakage or signature forgery.

[0036] In an optional embodiment, the step of determining whether the trusted log is an anomalous log based on a preset behavioral baseline of the entity corresponding to the trusted log includes: Determine whether the operation type corresponding to the trusted log belongs to the preset behavior baseline of the entity corresponding to the trusted log; If it does not belong to the category, then the trusted log is determined to be an abnormal log.

[0037] This optional embodiment simplifies anomaly detection based on behavioral baselines to determining whether the operation type in the trusted log belongs to the entity's preset behavioral baseline. Operation type is the most crucial semantic element in the trusted log, and different types of operations have vastly different impacts on industrial control systems. If an entity has never been recorded as performing a certain type of operation by the behavioral baseline, but such an operation appears in the trusted log, it should be considered highly suspicious, regardless of the reasonableness of the operation parameters. This blacklist / whitelist approach based on operation type has low computational overhead and is suitable for scenarios involving real-time processing of large volumes of logs.

[0038] S105. Taking the node corresponding to the operation object involved in the abnormal log as the endpoint, recursively traverse all nodes that can be associated with the endpoint through edges in the trust graph to generate an attack path graph.

[0039] In an optional embodiment, the step of generating an attack path graph by recursively traversing all nodes that can be associated with the endpoint through edges in the trust graph, using the node corresponding to the operation object involved in the anomaly log as the endpoint, includes: Taking the node corresponding to the operation object involved in the abnormal log as the endpoint, all nodes that can be associated with the endpoint through edges are recursively traversed in the trust graph to generate multiple attack paths. The confidence level of each attack path is calculated based on the risk level of each operation type on each attack path and the degree of deviation of each operation type from the behavioral baseline of the corresponding entity. Select the attack path with the highest confidence level and generate an attack path graph.

[0040] In this optional embodiment, the risk level is preset based on the criticality of the operation object, the severity of the operation consequences, and industry safety standards. For example, the risk level of modifying the controller firmware is higher than that of reading status information, and the risk level of adjusting safety interlock parameters is higher than that of viewing non-critical parameters.

[0041] The deviation of each operation type from the corresponding entity behavior baseline is determined by comparing the current operation type with the set of allowed operation types for that entity recorded in the baseline. The deviation is highest if the operation type is not in the baseline; if it is in the baseline, the deviation can be further quantified based on the matching degree of operation sequence, time window, or frequency. The deviation can also be calculated using the statistical distribution of historical normal operation data. For example, the distance between the feature vector corresponding to the current operation type and the normal pattern vector in the baseline can be measured; a larger distance indicates a higher deviation.

[0042] Ultimately, the confidence level of each attack path can be obtained by weighted summation or product of the risk and deviation levels of all operations on the path.

[0043] For example: Assume the following directed edges (in chronological order) exist in the trust graph: Edge 1 is "Operation and Maintenance UKeyKEY_01 Authentication Login Operator Station OP_01", Edge 2 is "Operator Station OP_01 Modify Parameters Controller PLC_101", Edge 3 is "Controller PLC_101 Execute Command Valve V001", and Edge 4 is "Hacker Host HACK_01 Remote Login Operator Station OP_01". If there is an exception log entry whose operation object is "Controller PLC_101", then the PLC_101 node will be the endpoint. Reverse recursive traversal: Find all edges with PLC_101 as the object, i.e., edge 2 (object is OP_01), and obtain the upstream node OP_01; then, with OP_01 as the current node, find all edges with OP_01 as the object, including edge 1 (object is KEY_01) and edge 4 (object is HACK_01), and obtain the upstream nodes KEY_01 and HACK_01 respectively; continuing with KEY_01 and HACK_01 as the current nodes, there are no more upstream edges. This generates two attack paths: path A is "KEY_01 → OP_01 → PLC_101", and path B is "HACK_01 → OP_01 → PLC_101". Based on the final calculated confidence level, path B is selected as the attack path graph.

[0044] In this optional embodiment, the risk level can reflect the inherent harmfulness of the operation, while the deviation reflects the distance between the operation and normal behavioral patterns. Combining these two aspects can distinguish between accidental minor anomalies and malicious attack chains. Through sorting and filtering, security operations personnel can prioritize addressing the most likely attack paths, improving incident response efficiency.

[0045] In an optional embodiment, after generating the attack path graph, the method further includes using the pre-injected network topology and service dependencies in the trust graph to determine the range of other nodes affected by the anomaly logs, and outputting this range as additional information for the attack path graph.

[0046] In an optional embodiment, after generating the attack path diagram, the method further includes graphically displaying the complete attack path on a security management platform interface, and annotating evidence of each step of the path and other potential devices or data affected by the attack.

[0047] Please see Figure 2 As shown, Figure 2 This is a schematic diagram of the structure of an industrial control system attack chain generation system disclosed in an embodiment of the present invention, including: The receiving module 201 is used to receive trusted logs sent by all entities; wherein the trusted logs include operation types and operation objects; Module 202 is used to construct a trust graph with the entities and operation objects corresponding to all trusted logs as nodes and the corresponding operation types as edges. The concatenation module 203 is used to concatenate all received trusted logs into a trusted log sequence according to the trust graph in chronological order. Filtering module 204 is used to filter out abnormal logs in the trusted log sequence based on a preset behavioral baseline for each entity. The generation module 205 is used to generate an attack path graph by recursively traversing all nodes in the trust graph that can be associated with the endpoint through edges, using the node corresponding to the operation object involved in the abnormal log as the endpoint. Specific limitations regarding the industrial control system attack chain generation system can be found in the limitations of the industrial control system attack chain generation method described above, and will not be repeated here. Each module in the aforementioned industrial control system attack chain generation system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of the electronic device in hardware format or independent of it, or stored in the memory of the electronic device in software format, so that the processor can call the corresponding operations of each module.

[0048] It should be noted that, in order to highlight the innovative aspects of this invention, this embodiment does not include modules that are not closely related to solving the technical problems proposed by this invention, but this does not mean that there are no other modules in this embodiment.

[0049] like Figure 3 As shown, the electronic device 1 provided by the present invention may include a memory 12, a processor 13 and a bus, and may also include a computer program stored in the memory 12 and executable on the processor 13, such as an industrial control system attack chain generation program.

[0050] The memory 12 includes at least one type of readable storage medium, such as flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 12 can be an internal storage unit of the electronic device 1, such as a portable hard drive. In other embodiments, the memory 12 can be an external storage device of the electronic device 1, such as a plug-in portable hard drive, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the electronic device 1. Furthermore, the memory 12 can include both internal and external storage units of the electronic device 1. The memory 12 can be used not only to store application software and various types of data installed on the electronic device 1, such as code generated by an industrial control system attack chain, but also to temporarily store data that has been output or will be output.

[0051] In some embodiments, the processor 13 may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits packaged with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor 13 is the control unit of the electronic device 1, connecting various components of the electronic device 1 through various interfaces and lines. It executes programs or modules stored in the memory 12 (e.g., industrial control system attack chain generation programs) and calls data stored in the memory 12 to perform various functions of the electronic device 1 and process data.

[0052] The processor 13 executes the operating system of the electronic device 1 and various installed applications. The processor 13 executes the applications to implement the steps in the above-described industrial control system attack chain generation method.

[0053] For example, the computer program may be divided into one or more modules, which are stored in the memory 12 and executed by the processor 13 to complete this application. The one or more modules may be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the electronic device 1. For example, the computer program may be divided into a receiving module 201, a building module 202, a concatenation module 203, a filtering module 204, and a generating module 205.

[0054] The integrated unit implemented as a software functional module described above can be stored in a computer-readable storage medium, which can be non-volatile or volatile. The software functional module stored in the storage medium includes several instructions to cause a computer device (which may be a personal computer, computer equipment, or network device, etc.) or processor to execute some functions of the industrial control system attack chain generation method described in the various embodiments of this application.

[0055] In summary, the industrial control system attack chain generation method, system, device, and medium disclosed in this invention, because it no longer relies on preset static rules for anomaly detection but instead identifies any behavior deviating from the normal operating mode based on the behavioral baseline of each entity, can effectively detect even low-frequency, slow-penetrating, and other covert attacks that do not match any known features. Furthermore, by establishing operational semantic associations between nodes and edges in the trust graph, combined with a reverse recursive traversal ending at the anomaly log, all preceding operations can be automatically traced without manual retrieval, thus solving the problems of existing methods being unable to identify covert attacks and automatically reconstruct the complete attack path. Therefore, this invention effectively overcomes the various shortcomings of the prior art and has high industrial application value.

[0056] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A method for generating attack chains for industrial control systems, characterized in that, The method includes: Receive trusted logs sent by all entities; wherein the trusted logs include operation type and operation object; A trust graph is constructed using the entities and operation objects corresponding to all trusted logs as nodes and the corresponding operation types as edges. Based on the trust graph, all received trusted logs are concatenated into a trusted log sequence in chronological order. Based on a preset behavioral baseline for each entity, abnormal logs in the trusted log sequence are filtered out. Using the node corresponding to the operation object involved in the abnormal log as the endpoint, all nodes that can be associated with the endpoint through edges are recursively traversed in reverse in the trust graph to generate an attack path graph.

2. The method for generating an attack chain for an industrial control system according to claim 1, characterized in that, The trusted log also includes a timestamp; And, the step of concatenating all received trusted logs into a trusted log sequence based on the trust graph in chronological order includes: Based on the timestamp, and according to the position of the entity and operation object corresponding to each trusted log in the node of the trust graph, trusted logs with the same node are connected in chronological order to generate a trusted log sequence.

3. The method for generating an attack chain for an industrial control system according to claim 1, characterized in that, The steps for obtaining the behavioral baseline include: Obtain the historical logs of the industrial control system; The security history log is obtained by filtering the historical logs to obtain the logs of the industrial control system when it was not attacked. The security history logs are parsed to extract the security operation model corresponding to each security history log; wherein, the security operation model refers to an entity's operation on an operation object; The safety operation model is learned using a graph neural network to obtain the behavioral baseline of each entity.

4. The method for generating an attack chain for an industrial control system according to claim 1, characterized in that, The trusted log also includes the entity's digital signature on the trusted log and the entity's identity identifier; Furthermore, the step of filtering out anomalous logs from the trusted log sequence based on a preset behavioral baseline for each entity includes: For each trusted log in the trusted log sequence: Obtain the public key of the entity corresponding to the trusted log based on the identity identifier; The digital signature is verified using the public key, and the verification result is determined. If the verification is successful, the trusted log is determined to be an abnormal log based on the preset behavior baseline of the entity corresponding to the trusted log. If the verification fails, the trusted log is determined to be an abnormal log.

5. The method for generating an attack chain for an industrial control system according to claim 4, characterized in that, The digital signature is generated by the entity using a national cryptographic signature algorithm within its trusted execution environment or built-in security chip.

6. The method for generating an attack chain for an industrial control system according to claim 4, characterized in that, Based on the preset behavioral baseline of the entity corresponding to the trusted log, the steps to determine whether the trusted log is an abnormal log include: Determine whether the operation type corresponding to the trusted log belongs to the preset behavior baseline of the entity corresponding to the trusted log; If it does not belong to the category, then the trusted log is determined to be an abnormal log.

7. The method for generating an attack chain for an industrial control system according to claim 1, characterized in that, Taking the node corresponding to the operation object involved in the abnormal log as the endpoint, the steps of recursively traversing all nodes that can be associated with the endpoint through edges in the trust graph to generate the attack path graph include: Taking the node corresponding to the operation object involved in the abnormal log as the endpoint, all nodes that can be associated with the endpoint through edges are recursively traversed in the trust graph to generate multiple attack paths. The confidence level of each attack path is calculated based on the risk level of each operation type on each attack path and the degree of deviation of each operation type from the behavioral baseline of the corresponding entity. Select the attack path with the highest confidence level and generate an attack path graph.

8. An attack chain generation system for industrial control systems, characterized in that, include: A receiving module is used to receive trusted logs sent by all entities; wherein, the trusted logs include operation types and operation objects; The building module is used to construct a trust graph with the entities and operation objects corresponding to all trusted logs as nodes and the corresponding operation types as edges. The concatenation module is used to concatenate all received trusted logs into a trusted log sequence according to the trust graph in chronological order. The filtering module is used to filter out abnormal logs in the trusted log sequence based on a preset behavioral baseline for each entity. The generation module is used to generate an attack path graph by recursively traversing all nodes that can be associated with the endpoint through edges in the trust graph, with the node corresponding to the operation object involved in the abnormal log as the endpoint.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the industrial control system attack chain generation method as described in any one of claims 1 to 7.

10. A storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the industrial control system attack chain generation method as described in any one of claims 1 to 7.