Attack response method and device based on large model of power system, equipment and medium

Through the attack response method based on the power system large model, network attack characteristics are extracted and responses are generated using the large language model, the problem that the existing technology is difficult to meet the needs of diversified security scenarios of the power system is solved, and effective power system network security is achieved promote.

CN118944910APending Publication Date: 2024-11-12许浩
View PDF 0 Cites 3 Cited by

Patent Information

Application Number
CN202410866527.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-01
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

The existing network security shooting range is difficult to meet the needs of diversified security scenarios of power systems and cannot effectively simulate and respond to network attacks in power systems.

Method used

The attack response method based on the power system large model is adopted, by obtaining network attack data, extracting features and determining the attack intention, the preset network shooting range big model is used to generate response responses, and the emergence capabilities of the Bert model and the large language model are analyzed.

Benefits of technology

It has realized effective analysis and response solutions for diversified security scenarios of power systems, improved the training effect of power system network security, and solved the problem that existing technology is difficult to meet the needs of power system network security improvement.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118944910A_ABST
    Figure CN118944910A_ABST
Patent Text Reader

Abstract

The invention relates to an attack response method, device and equipment based on a large model of a power system and a medium, and the method can be used for carrying out feature extraction on network attack data for a security scene of the power system, then determining an attack intention according to features, and determining a response corresponding to the attack intention based on a preset large model of a network target range. According to the method, in combination with the emergence capability of the large model, the diversified security scenes of the power system can be effectively analyzed, and a response coping scheme is given, so that the diversified power system network security scene training is realized, and the technical problem that the existing network security target range is difficult to meet the network security improvement requirement of the power system is solved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The embodiments of the present application relate to the technical field of power system network protection, and in particular to an attack response method, device, equipment and medium based on a large power system model. Background Art

[0002] The power system is the national infrastructure, and the network security of the power system is an important aspect of national security. At present, the concepts of automation, informatization and networking have penetrated into all aspects of the power system. The master stations at all levels, substations of various voltage levels and power plants are highly interconnected to form a huge network space. Network attacks on the power system will affect the safe and stable operation of the power system. At present, in order to practice the contingency plan of the power system in the face of network attacks, there is an urgent need for a network security range for the power system. This range platform should be able to simulate the actual situation of the current power system's network space, reproduce security issues in a virtual environment, and provide attack and protection drills. Summary of the invention

[0003] The following is a summary of the subject matter described in detail herein. This summary is not intended to limit the scope of the claims.

[0004] The main purpose of the embodiments of the present application is to propose an attack response method, device, equipment and medium based on a large model of the power system, which can combine the emergence capability of the large model and effectively analyze the diverse security scenarios of the power system.

[0005] To achieve the above-mentioned purpose, a first aspect of an embodiment of the present application provides an attack response method based on a large power system model, and the attack response method based on a large power system model includes:

[0006] Obtain data on cyber attacks against power systems;

[0007] Extracting network attack features from the network attack data, and determining attack intent based on the network attack features;

[0008] The response corresponding to the attack intention is determined based on a preset network target range model.

[0009] The method provided by the present application includes the following beneficial effects:

[0010] This method can extract features from network attack data for power system security scenarios, and then determine the attack intent based on the features, and determine the corresponding response to the attack intent based on the preset network target range large model. This method can combine the emergence capability of the large model, effectively analyze the diverse security scenarios of the power system, and provide corresponding response plans, thereby realizing diversified power system network security scenario training and solving the technical problem that the existing network security target range is difficult to meet the needs of improving the network security of the power system.

[0011] In some implementations of the present application, determining the attack intention according to the network attack feature includes:

[0012] Obtain the attack intention understanding model trained based on the Bert model;

[0013] The network attack feature is input into the attack intention understanding model to extract a sentence vector from the network attack feature through the attack intention understanding model, and the sentence vector is linearized and normalized to obtain the attack intention to which the sentence vector belongs.

[0014] In some implementations of the present application, obtaining the attack intention to which the sentence vector belongs includes:

[0015] Determining the probability that the sentence vector belongs to each type of the attack intention according to the similarity function;

[0016] The type with the highest probability is selected as the attack intention to which the sentence vector belongs.

[0017] In some embodiments of the present application, the type of attack intention includes at least one of identity impersonation, malware, service attack, security vulnerability and service interruption;

[0018] The similarity function includes:

[0019]

[0020] Wherein, T is the similarity function of the identity impersonation, the malware, the service attack, the security vulnerability, and the service interruption, α is a constant, and x 1 to x 5 Map the attack features of the identity impersonation, the malware, the service attack, the security vulnerability, and the service interruption to features in the same dimension, S is the similarity function, [x] + =max(x,0),[x] + It means taking the maximum value between x and 0.

[0021] In some embodiments of the present application, the network attack behavior is at least one of an attack behavior extracted from a safety report of the power system, an attack behavior collected by security software of the power system, and an attack behavior generated by simulation.

[0022] In some embodiments of the present application, the network range big model is a big model obtained by fine-tuning an open source big language model, or a big model trained based on multiple attack intent training samples that can predict the response corresponding to the attack intent.

[0023] In some embodiments of the present application, the network range big model is a big model that integrates the network security knowledge graph.

[0024] To achieve the above-mentioned purpose, a second aspect of an embodiment of the present application provides an attack response device based on a large model of a power system, and the attack response device based on a large model of a power system includes:

[0025] An attack data acquisition unit, used to acquire network attack data against the power system;

[0026] An attack intention determination unit, configured to extract network attack features from the network attack data and determine attack intentions based on the network attack features;

[0027] The response generation unit is used to determine the response corresponding to the attack intention based on a preset network target range model.

[0028] To achieve the above-mentioned purpose, the electronic device provided in the third aspect of an embodiment of the present application includes: at least one control processor and a memory for communicating with the at least one control processor; the memory stores instructions that can be executed by the at least one control processor, and the instructions are executed by the at least one control processor so that the at least one control processor can execute an attack response method based on a large model of a power system according to the above-mentioned first aspect.

[0029] To achieve the above-mentioned purpose, the fourth aspect of an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute an attack response method based on a large power system model of the above-mentioned first aspect.

[0030] It can be understood that the beneficial effects of the second to fourth aspects compared with the related art are the same as the beneficial effects of the first aspect compared with the related art. Please refer to the relevant description in the first aspect, and no further details will be given here. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0032] Figure 1 It is a flowchart of an attack response method based on a large power system model provided in an embodiment of the present application;

[0033] Figure 2 It is a flowchart of another attack response method based on a large power system model provided in an embodiment of the present application;

[0034] Figure 3 It is a structural schematic diagram of an attack response device based on a large power system model provided in an embodiment of the present application;

[0035] Figure 4 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0037] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0038] 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 only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0039] The power system is the national infrastructure, and the network security of the power system is an important aspect of national security. At present, the concepts of automation, informatization and networking have penetrated into all aspects of the power system. The master stations at all levels, substations of various voltage levels and power plants are highly interconnected to form a huge network space. Network attacks on the power system will affect the safe and stable operation of the power system. Therefore, in order to practice the contingency plan of the power system in the face of network attacks, there is an urgent need for a network security range for the power system. This range platform should be able to simulate the actual situation of the current power system's network space, reproduce security issues in a virtual environment, and provide attack and protection drills. At the same time, it should be able to update and iterate scenarios based on new security vulnerabilities.

[0040] In terms of artificial intelligence, deep learning has made great progress in various AI research fields such as computer vision (CV) and natural language processing (NLP). For example, deep residual networks (ResNets) have surpassed human performance in image classification. In the past two years, the emergence of large models such as chatGPT and GPT-4 has promoted the rapid development of natural language processing technology. The language model GPT-4 also outperforms humans in several natural language understanding tasks on multiple benchmark test sets.

[0041] Reference Figure 1 and Figure 2 Some embodiments of the present application provide an attack response method based on a large power system model. The attack response method based on a large power system model includes the following steps S110 to S130:

[0042] Step S110: Acquire network attack data on the power system.

[0043] In this step, the power system is an electric energy production and consumption system consisting of power plants, power transmission and transformation lines, power distribution stations and power consumption. Its function is to convert primary energy in nature into electric energy through power generation devices, and then supply electric energy to various users through transmission, transformation and distribution. To achieve this function, the power system also has corresponding information and control systems at various links and levels to measure, regulate, control, protect, communicate and dispatch the production process of electric energy to ensure that users obtain safe and high-quality electric energy.

[0044] The network attack data is data extracted from network attack behaviors on the power system. In some embodiments of the present application, the network attack behaviors are at least one of attack data extracted from the security report of the power system, attack data collected by the security software of the power system, and attack data generated by simulation.

[0045] The types of network attack behaviors can cover all current attack types, including but not limited to: password intrusion, Trojan horse, WWW spoofing, node attack, network monitoring, IP spoofing, hacker software and flood attack.

[0046] Step S120: extract network attack features from the network attack data, and determine attack intent based on the network attack features.

[0047] In this step, the attack intention includes but is not limited to: identity impersonation, malware, service attack, security vulnerability, and service interruption.

[0048] In this step, the network attack features can be extracted from the attack data through the feature encoder of multiple attack means. The attack intention to which the network attack features belong will be determined later.

[0049] In some embodiments of the present application, determining the attack intention according to the network attack characteristics includes the following steps S210 and S220:

[0050] Step S210: Obtain an attack intention understanding model trained based on the Bert model.

[0051] Step S220: input the network attack features into the attack intention understanding model to extract sentence vectors from the network attack features through the attack intention understanding model, and linearize and normalize the sentence vectors to obtain the attack intention to which the sentence vectors belong.

[0052] In some embodiments of the present application, step S220 obtains the attack intent to which the sentence vector belongs, including the following steps S310 and S320:

[0053] Step S310: Determine the probability that the sentence vector belongs to each type of attack intent according to the similarity function.

[0054] Step S320: Select the type with the highest probability as the attack intention to which the sentence vector belongs.

[0055] In some implementations, the similarity function includes:

[0056]

[0057] Where T is the similarity function of identity impersonation, malware, service attack, security vulnerability, and service interruption, α is a constant, and x 1 to x 5 Maps the attack features of identity impersonation, malware, service attacks, security vulnerabilities, and service interruptions to features in the same dimension. S is the similarity function, [x] + =max(x,0),[x] + It means taking the maximum value between x and 0.

[0058] Because identity impersonation, malware, service attacks, security vulnerabilities, service interruptions and other types of attacks do not exist independently, their features need to be aligned through a similarity function. Intent attack recognition unifies diverse intentions into a feature vector scale to facilitate the accuracy of subsequent operations.

[0059] Step S130: Determine the response corresponding to the attack intention based on the preset network target range model.

[0060] In some embodiments, the network target range big model is a big model obtained by fine-tuning an open source big language model, and the trained power network target range big model can be a big model fine-tuned based on an open source big language model (such as LLama, ChatGLM, etc.) combined with certain power network security-related data.

[0061] Alternatively, it is a large model that can predict the response to the attack intention obtained by training based on multiple attack intention training samples. For example, ChatGLM-6b can be used as a large model for the power network target range. Since the basic model has 6 billion parameters, training is very difficult. Therefore, the basic model is frozen during training and the parameters of the basic model are not updated to reduce training time and cost.

[0062] The large model of the power network target range can include a multi-layer perceptron layer and a Transformer encoder layer. The multi-layer perceptron layer enables the model to capture more feature information, and the Transformer encoder layer is used to convert the input sequence information into a set of representation vectors through the self-attention mechanism.

[0063] In some implementations of the present application, the large model of the power network range can be added with a control network, which is used to introduce factual knowledge of the knowledge graph into the basic model output to control model generation. This fully utilizes the emergence and flexibility of the large language model, and can also use the knowledge accuracy of the network security knowledge graph to solve the technical problem that hallucinations (commonly known as "serious nonsense") will occur when using a single large model.

[0064] In some embodiments of the present application, the large network range model outputs attack behavior warnings / reports based on actual conditions, or cooperates with other systems to complete the handling of attack behaviors.

[0065] This method can extract features from data based on feature encoders for multiple attack methods for security scenarios of power systems, and then determine attack intent based on the features, and determine the corresponding response to the attack intent based on a preset network target range large model. This method can combine the emergence capabilities of large models, effectively analyze diverse security scenarios, and provide corresponding response plans, thereby realizing diverse power system network security scenario training and solving the technical problem that the existing network security target range is difficult to meet the needs of improving network security in power systems.

[0066] Reference Figure 3 Some embodiments of the present application provide an attack response device based on a large power system model. The attack response device based on the large power system model includes an attack data acquisition unit 1100, an attack intention determination unit 1200, and a response generation unit 1300. Specifically:

[0067] The attack data acquisition unit 1100 is used to acquire network attack data on the power system.

[0068] The attack intention determination unit 1200 is used to extract network attack features from network attack data and determine attack intentions according to the network attack features.

[0069] The response generation unit 1300 is used to determine the response corresponding to the attack intention based on the preset network range model.

[0070] This embodiment and the above method embodiment are based on the same inventive concept and will not be described in detail here.

[0071] like Figure 4 The present application also provides an electronic device, which includes:

[0072] at least one memory;

[0073] at least one processor;

[0074] at least one program;

[0075] The program is stored in the memory, and the processor executes at least one program to implement the attack response method based on the large power system model described above in the present disclosure.

[0076] The electronic device may be any intelligent terminal including a mobile phone, a tablet computer, a personal digital assistant (PDA), a vehicle-mounted computer, etc.

[0077] The electronic device according to the embodiment of the present application is described in detail below.

[0078] The processor 1600 may be implemented by a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;

[0079] The memory 1700 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 1700 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the program code is stored in the memory 1700, and the processor 1600 calls and executes the attack response method based on the large power system model in the embodiment of the present application.

[0080] Input / output interface 1800, used to implement information input and output;

[0081] The communication interface 1900 is used to realize the communication interaction between the device and other devices. The communication can be realized through a wired manner (such as USB, network cable, etc.) or a wireless manner (such as mobile network, WIFI, Bluetooth, etc.);

[0082] Bus 2000 , which transmits information between various components of the device (e.g., processor 1600 , memory 1700 , input / output interface 1800 , and communication interface 1900 );

[0083] The processor 1600 , the memory 1700 , the input / output interface 1800 , and the communication interface 1900 are connected to each other in communication within the device via the bus 2000 .

[0084] An embodiment of the present application also provides a storage medium, which is a computer-readable storage medium. The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the above-mentioned attack response method based on the large model of the power system.

[0085] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0086] The embodiments described in this application are intended to more clearly illustrate the technical solutions of the embodiments of this application and do not constitute a limitation on the technical solutions provided by the embodiments of this application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0087] Those skilled in the art will appreciate that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0088] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0089] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0090] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0091] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0092] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of 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 integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0093] The units described as separate components may or may not be physically separated, and the components shown as units 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.

[0094] 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. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0095] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including multiple instructions to enable an electronic device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of each embodiment method of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), disk or optical disk and other media that can store programs.

[0096] The above is a specific description of the preferred implementation of the embodiments of the present application, but the embodiments of the present application are not limited to the above-mentioned implementation methods. Technical personnel familiar with the field can also make various equivalent modifications or substitutions without violating the spirit of the embodiments of the present application. These equivalent modifications or substitutions are all included in the scope defined by the claims of the embodiments of the present application.

Claims

1. An attack response method based on a large power system model, It is characterized in that The attack response method based on the power system large model includes: Obtain data on cyber attacks against power systems; Extracting network attack features from the network attack data, and determining attack intent based on the network attack features; The response corresponding to the attack intention is determined based on a preset network target range model.

2. The attack response method based on the power system large model according to claim 1, It is characterized in that Determining the attack intention according to the network attack characteristics includes: Obtain the attack intention understanding model trained based on the Bert model; The network attack feature is input into the attack intention understanding model to extract a sentence vector from the network attack feature through the attack intention understanding model, and the sentence vector is linearized and normalized to obtain the attack intention to which the sentence vector belongs.

3. The attack response method based on the power system large model according to claim 2, It is characterized in that The obtaining the attack intention to which the sentence vector belongs includes: Determining the probability that the sentence vector belongs to each type of the attack intention according to the similarity function; The type with the highest probability is selected as the attack intention to which the sentence vector belongs.

4. The attack response method based on the power system large model according to claim 3, It is characterized in that The type of attack intention includes at least one of identity impersonation, malware, service attack, security vulnerability and service interruption; The similarity function includes: Wherein, T is the similarity function of the identity impersonation, the malware, the service attack, the security vulnerability, and the service interruption, α is a constant, and x 1 to x 5 Map the attack features of the identity impersonation, the malware, the service attack, the security vulnerability, and the service interruption to features in the same dimension, S is the similarity function, [x] + =max(x,0),[x] + It means taking the maximum value between x and 0.

5. The attack response method based on the power system large model according to claim 1, It is characterized in that The network attack data is at least one of attack data extracted from a security report of the power system, attack data collected by security software of the power system, and attack data generated by simulation.

6. The attack response method based on the power system large model according to claim 1, It is characterized in that The network range large model is a large model obtained by fine-tuning an open source large language model, or a large model trained based on multiple attack intent training samples that can predict the response corresponding to the attack intent.

7. The attack response method based on the power system large model according to claim 6, It is characterized in that The network range big model is a big model that integrates the network security knowledge graph.

8. An attack response device based on a large power system model, It is characterized in that The attack response device based on the power system large model includes: An attack data acquisition unit, used to acquire network attack data against the power system; An attack intention determination unit, configured to extract network attack features from the network attack data and determine attack intentions based on the network attack features; The response generation unit is used to determine the response corresponding to the attack intention based on a preset network target range model.

9. An electronic device, It is characterized in that include: at least one control processor and a memory for communicatively coupling with the at least one control processor; The memory stores instructions that can be executed by the at least one control processor, and the instructions are executed by the at least one control processor so that the at least one control processor can execute the attack response method based on the large power system model as described in any one of claims 1 to 7.

10. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the attack response method based on the power system large model as described in any one of claims 1 to 7.

Citation Information

Cited By

  • Power monitoring system attack scene processing method and device based on attack chain matching

    CN119766569A

  • Attack scenario processing method and device for power monitoring system based on attack chain matching

    CN119766569B

  • Network security intelligent association analysis method based on generative large model

    CN119903142A