Security protection method and device, storage medium and computer program product

By capturing real-time traffic data in industrial control systems and using Pareto distribution and machine learning algorithms to build a normal behavior model, and combining extreme value theory to calculate event probability, the detection and protection problems faced by industrial control systems are solved. Real-time security protection of the system is realized.

CN119026120AInactive Publication Date: 2024-11-26BEIJING LIUFANG CLOUD INFORMATION TECH CO LTD +1
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202411507218.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-28
Publication Date
2024-11-26
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Industrial control systems face unknown threats and abnormal behaviors that cannot be effectively detected and protected. The singularity and complexity of existing industrial control firewalls lead to the inability to effectively respond to the ever-changing logical control and unknown threats in industrial scenarios.

Method used

By capturing real-time traffic data, using Pareto distribution and machine learning algorithms to analyze and model historical sample data, build a normal behavior model, calculate event probability based on extreme value theory, match and judge whether the traffic behavior is abnormal in real time, and thus perform security protection .

Benefits of technology

Real-time security protection for industrial control systems is achieved, timely discover and respond to unknown threats and abnormal behaviors, and ensure the safe and stable operation of the system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119026120A_ABST
    Figure CN119026120A_ABST
Patent Text Reader

Abstract

The invention discloses a security protection method and device, a storage medium and a computer program product, and relates to the technical field of industrial control, and the method comprises the steps: capturing real-time traffic data based on the traffic behavior of an industrial control system; analyzing and modeling pre-acquired historical sample data by applying Pareto distribution and a machine learning algorithm to obtain a normal behavior model; matching the real-time flow data with the normal behavior model, and calculating an event probability in combination with an extreme value theory; and performing safety protection on the industrial control system according to the event probability. By establishing a safety behavior model, real-time flow data deviating from the safety behavior model is regarded as an abnormal behavior or an unknown threat, and detection and protection of the unknown threat are realized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of industrial control technology, and in particular to a security protection method, device, storage medium and computer program product. Background Art

[0002] Industrial control systems have been widely used in major infrastructure and play a huge role in factory operations. With the deep integration of industrialization and informatization, security threats from information networks are gradually affecting industrial control systems. Mainstream industrial control firewalls are mostly integrated hardware and software devices, which mainly match and alarm known vulnerabilities and features, and can only basically meet the detection and blocking of solidified processes, known threats and abnormal behaviors in industrial scenarios. However, it is impossible to effectively detect the ever-changing logical control, unknown threats, and abnormal behaviors in industrial control systems in industrial scenarios, and there is a lack of effective protection measures.

[0003] Therefore, how to provide security protection against unknown threats and abnormal behaviors faced by industrial control systems has become an urgent problem to be solved in this application.

[0004] The above contents are only used to assist in understanding the technical solution of the present application and do not constitute an admission that the above contents are prior art. Summary of the invention

[0005] The main purpose of this application is to provide a security protection method, device, storage medium and computer program product, aiming to solve the technical problem of how to provide security protection against unknown threats and abnormal behaviors faced by industrial control systems.

[0006] To achieve the above objectives, the present application proposes a security protection method, which includes: capturing real-time flow data based on flow behavior of the industrial control system; Apply Pareto distribution and machine learning algorithms to analyze and model the historical sample data obtained in advance to obtain a normal behavior model; Matching the real-time traffic data with the normal behavior model, and calculating event probability in combination with extreme value theory; The industrial control system is protected according to the event probability.

[0007] In one embodiment, before the step of applying Pareto distribution and machine learning algorithm to analyze and model the pre-acquired historical sample data to obtain the normal behavior model, the step further includes: A data communication model is built according to the production process of the industrial control system and the instruction flow in the data traffic, and a normal production process data model is established.

[0008] In one embodiment, the step of applying Pareto distribution and machine learning algorithm to analyze and model the pre-acquired historical sample data to obtain a normal behavior model includes: Collecting historical sample data from a data source of the industrial control system based on the data communication model and the normal production process data model, wherein the historical sample data includes historical traffic data and known threat label data; Cleaning the historical traffic data and the known threat label data; Construct a multi-dimensional data set based on the data characteristics of the cleaned historical traffic data and known threat label data; Apply Pareto distribution to analyze the cleaned historical traffic data and obtain Pareto analysis results; Based on the Pareto analysis results, the multidimensional data set is modeled using a machine learning algorithm to obtain a normal behavior model.

[0009] In one embodiment, the step of modeling the multidimensional data set using a machine learning algorithm based on the Pareto analysis result to obtain a normal behavior model comprises: Selecting a corresponding machine learning algorithm according to the multidimensional data set; In combination with the Pareto analysis results, a corresponding machine learning algorithm is used to model the multi-dimensional data set to obtain an initial normal behavior model; The multi-dimensional data set is used to perform data training and model iteration on the initial normal behavior model to obtain a normal behavior model.

[0010] In one embodiment, the step of matching the real-time traffic data with the normal behavior model and calculating the event probability in combination with extreme value theory includes: Matching the real-time traffic data with the normal behavior model; According to the Pareto analysis result, calculating the matching degree between the real-time traffic data and the normal behavior model; Based on the matching degree and in combination with extreme value theory, the event probability of the real-time traffic data is evaluated.

[0011] In one embodiment, the step of performing security protection on the industrial control system according to the event probability includes: When the event probability is lower than a preset threshold, the real-time traffic data is marked as an abnormal event and enters the alarm processing stage; Performing security protection on the industrial control system by entering the alarm processing stage; When the event probability is not lower than a preset threshold, the flow data newly flowing into the industrial control system is monitored in real time to safely protect the industrial control system.

[0012] In one embodiment, when the event probability is lower than a preset threshold, the real-time traffic data is marked as an abnormal event, and the steps of entering the alarm processing stage include: After marking the real-time traffic data as an abnormal event, an administrator is notified in a visual manner; The security problem existing in the industrial control system is repaired according to the abnormal event, and the relevant parameters of the normal behavior model are adjusted.

[0013] In addition, to achieve the above purpose, the present application also proposes a safety protection device, which includes: Data acquisition module for capturing real-time flow data based on the flow behavior of industrial control systems; The analysis and modeling module is used to apply Pareto distribution and machine learning algorithms to analyze and model the pre-acquired historical sample data to obtain a normal behavior model; A model matching module, used to match the real-time traffic data with the normal behavior model and calculate event probability in combination with extreme value theory; An early warning module is used to provide security protection for the industrial control system according to the probability of the event.

[0014] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by the processor, the steps of the security protection method described above are implemented.

[0015] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the security protection method described above.

[0016] One or more technical solutions proposed in this application have at least the following technical effects: Based on the traffic behavior of the industrial control system, real-time traffic data is captured. The real-time traffic data reflects the current operating status and network traffic of the industrial control system. A series of analyses are then performed based on the real-time traffic data to achieve real-time security protection of the industrial control system. The Pareto distribution and machine learning algorithms are used to analyze and model the pre-acquired historical sample data to obtain a normal behavior model. The normal behavior model reflects the traffic behavior characteristics of the industrial control system under normal operating conditions. Through the normal behavior model, the traffic behavior of the industrial control system is predicted and judged. The real-time traffic data is matched with the normal behavior model, and the event probability is calculated in combination with the extreme value theory. By comparing the real-time traffic data with the normal behavior model, it can be determined whether the current traffic behavior meets the normal behavior characteristics. In the matching process, the event probability is calculated in combination with the extreme value theory, so as to further determine whether the traffic behavior has abnormal behavior or unknown threats. Finally, according to the calculated event probability, the industrial control system is protected, and the unknown threats and abnormal behaviors faced by the industrial control system are discovered and responded to in a timely manner, ensuring the safe and stable operation of the industrial control system. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0018] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0019] Figure 1 A schematic diagram of a process flow provided for the first embodiment of the security protection method of the present application; Figure 2 A flowchart of the second embodiment of the security protection method of the present application is provided; Figure 3 A brief flowchart of the security protection method provided for this application; Figure 4 A schematic diagram of the structure of the industrial control system firewall provided for this application; Figure 5 A schematic diagram of the process of building a normal behavior model; Figure 6 This is a schematic diagram of the module structure of the safety protection device according to an embodiment of the present application; Figure 7 A schematic diagram of the device structure of the hardware operating environment involved in the security protection method in the embodiment of the present application.

[0020] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0021] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0022] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0023] The main solution of the embodiment of the present application is: based on the traffic behavior of the industrial control system, real-time traffic data is captured; the pre-acquired historical sample data is analyzed and modeled using the Pareto distribution and machine learning algorithms. Pareto distribution is a mathematical model that describes the distribution of extreme values, which is suitable for describing data with thick tail characteristics. In network security scenarios, Pareto distribution can be used to describe the distribution of extreme values ​​of asset traffic behavior. By calculating the Pareto distribution parameters of the traffic behavior of each asset, a normal behavior model for each asset is established. Machine learning algorithms (such as logistic regression, support vector machine, random forest, etc.) are also used to train historical data to improve the accuracy of the data model; after collecting real-time traffic data, it is matched with the established normal behavior model. By comparing the difference between real-time traffic data and the normal behavior model, it can be preliminarily determined whether there is abnormal behavior. In order to further improve the accuracy of detection, the extreme value theory is combined to calculate the probability of events. The extreme value theory believes that low-probability events often represent abnormal or unknown threats. Therefore, when calculating the probability of an event, if the probability of a data point is very low (for example, less than 5%), it is regarded as abnormal behavior. In this way, potential threats can be identified and warned more accurately. Industrial control systems can be protected based on the calculated event probability. Any behavior that deviates from the normal security baseline is considered abnormal behavior or unknown threat and triggers an alarm mechanism.

[0024] In this embodiment, for ease of description, the following description is made by taking identification of an industrial control system firewall as the execution subject.

[0025] The embodiment of the present application takes into account that: since industrial control systems have been widely used in major domestic infrastructure, they play a huge role in factory operation. However, the computing resources of traditional industrial control system technology are limited. When it comes to industrial control systems, only efficiency and real-time characteristics are considered, and security is not considered as the main indicator. With the promotion of informatization and the acceleration of the industrialization process, more and more computer and network technologies are applied to industrial control systems. While greatly promoting industrial production, they are also facing the problem of security protection of industrial control systems. The singleness of protection of general firewalls and the complexity of industrial control systems make general firewalls incapable of facing the security protection of industrial control systems. Most of the mainstream industrial control firewalls on the market are integrated hardware and software devices, which match alarms based on known vulnerabilities and features. The industrial control firewalls with integrated detection functions can basically meet the solidified processes, known threats and normal behavior detection and blocking in industrial scenarios, but the industrial firewall detection module cannot effectively detect the ever-changing logical control, unknown threats and abnormal behaviors in the industrial scene control system. Therefore, the industrial control firewall lacks effective protection means for unknown threats.

[0026] Therefore, the present application provides a solution to capture real-time traffic data based on the traffic behavior of the industrial control system. The real-time traffic data reflects the current operating status and network traffic of the industrial control system. A series of analyses are subsequently performed based on the real-time traffic data to achieve real-time security protection of the industrial control system. The Pareto distribution and machine learning algorithms are used to analyze and model the pre-acquired historical sample data to obtain a normal behavior model. The normal behavior model reflects the traffic behavior characteristics of the industrial control system under normal operating conditions. Through the normal behavior model, the traffic behavior of the industrial control system is predicted and judged; the real-time traffic data is matched with the normal behavior model, and the event probability is calculated in combination with the extreme value theory. By comparing the real-time traffic data with the normal behavior model, it can be determined whether the current traffic behavior meets the normal behavior characteristics. In the matching process, the event probability is calculated in combination with the extreme value theory, so as to further determine whether the traffic behavior has abnormal behavior or unknown threats; finally, according to the calculated event probability, the industrial control system is protected, and the unknown threats and abnormal behaviors faced by the industrial control system are discovered and responded to in a timely manner, so as to ensure the safe and stable operation of the industrial control system.

[0027] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions, an industrial control system firewall, etc. The following takes an industrial control system firewall as an example to illustrate this embodiment and the following embodiments.

[0028] Based on this, the present application embodiment provides a security protection method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the security protection method of the present application.

[0029] In this embodiment, the safety protection method includes steps S10 to S40: Step S10, collecting real-time flow data based on the flow behavior of the industrial control system; In the environment of massive real-time data from multiple sources and heterogeneity in industrial control systems, industrial control system firewalls not only monitor and filter network traffic in and out of the industrial control system, but are also responsible for collecting and analyzing this traffic data to identify potential security threats or abnormal behaviors.

[0030] As a key point for data collection, the industrial control system firewall captures all new real-time traffic data flowing into the industrial control system. For example, the industrial control system firewall can parse and extract traffic data through technologies such as deep packet inspection (DPI) to obtain detailed information on traffic behavior, such as source address, destination address, protocol type, packet size, transmission time, etc., so as to promptly discover and respond to potential security incidents.

[0031] Step S20, applying Pareto distribution and machine learning algorithm to analyze and model the pre-acquired historical sample data to obtain a normal behavior model; Applying Pareto distribution means modeling the behavior of each asset in an industrial control system. Pareto distribution is a mathematical model that describes the distribution of extreme values ​​and is suitable for describing data with thick-tail characteristics. In network security scenarios, especially in industrial control systems, the distribution of traffic data often exhibits thick-tail characteristics, that is, most of the traffic comes from a few sources, while the remaining traffic is scattered among many sources. Therefore, using Pareto distribution to model historical sample data can accurately capture the distribution law of such extreme values ​​and provide a basis for subsequent anomaly detection.

[0032] Applying machine learning algorithms means selecting appropriate machine learning algorithms based on the classification problem of normal behavior and abnormal behavior. The machine learning algorithms available for selection in this application include but are not limited to: logistic regression, support vector machine (SVM), random forest, etc. These machine learning algorithms can process high-dimensional data and have good classification performance.

[0033] Therefore, Pareto distribution is used to analyze historical sample data, especially to model the distribution of extreme values ​​of traffic behavior. It helps identify the most important traffic sources or patterns, thereby building the minimum behavior boundary of each device or user. Machine learning algorithms can identify common features of normal behavior and issue warnings for behaviors that deviate from these features. Combined with the analysis results of applying Pareto distribution, machine learning algorithms are used to model the pre-acquired historical sample data, and appropriate algorithms (such as logistic regression, support vector machine, random forest, etc.) are selected to learn normal behavior and obtain a normal behavior model, thereby establishing a normal security baseline.

[0034] Step S30, matching the real-time traffic data with the normal behavior model, and calculating the event probability in combination with extreme value theory; Extreme value theory refers to the distribution law of maximum or minimum values ​​among a large number of independent and identically distributed random variables. In the network security scenario, the behavior of each asset in the industrial control system is regarded as a random variable, and the maximum or minimum value of its traffic behavior is observed to determine whether there is an anomaly.

[0035] When new real-time traffic data flows into the industrial control system firewall, the industrial control system firewall matches the real-time traffic data with the normal behavior model. Normal behavior always falls within the boundary, and abnormal behavior always deviates from the normal range in some imperceptible dimension. Therefore, traffic and behavior that deviate from the normal model are considered to be abnormal behavior or unknown threats, and alarms are issued to achieve detection and protection of unknown threats.

[0036] Specifically, the real-time traffic data is matched with the normal behavior model, the matching degree between the newly-incoming real-time traffic data and the normal behavior model of each asset is calculated, and the probability of the event is calculated based on the matching degree. Combined with the extreme value theory, low-probability events are judged as abnormal, and high-probability events are judged as normal.

[0037] Step S40: performing security protection on the industrial control system according to the event probability.

[0038] Determine whether there is abnormal behavior in the industrial control system based on the event probability. Once abnormal behavior or potential threats are detected, the alarm mechanism is triggered immediately. Through the early warning display module of the industrial control system firewall, the abnormal information is displayed to the administrator or relevant personnel so that timely response measures can be taken.

[0039] Furthermore, according to the security policy, the industrial control system firewall automatically or manually blocks abnormal traffic to prevent the threat from spreading further; or, isolates abnormal devices or users from the network to ensure the safe and stable operation of the entire industrial control system.

[0040] In the process of handling abnormal traffic, the industrial control system firewall records all abnormal behaviors and their processing processes for subsequent security audits and analysis; continuously monitors the traffic data of the industrial control system to ensure that new threats can be discovered and responded to in a timely manner; analyzes the characteristics of abnormal behaviors, continuously improves and optimizes normal behavior models, and improves the security protection capabilities of industrial control systems.

[0041] This embodiment provides a security protection method, which captures real-time traffic data based on the traffic behavior of the industrial control system. The real-time traffic data reflects the current operating status and network traffic of the industrial control system. A series of analyses are subsequently performed based on the real-time traffic data to achieve real-time security protection of the industrial control system; the pre-acquired historical sample data is analyzed and modeled using the Pareto distribution and machine learning algorithm to obtain a normal behavior model. The normal behavior model reflects the traffic behavior characteristics of the industrial control system under normal operating conditions. The traffic behavior of the industrial control system is predicted and judged through the normal behavior model; the real-time traffic data is matched with the normal behavior model, and the event probability is calculated in combination with the extreme value theory. By comparing the real-time traffic data with the normal behavior model, it can be determined whether the current traffic behavior meets the normal behavior characteristics. In the matching process, the event probability is calculated in combination with the extreme value theory, so as to further determine whether the traffic behavior has abnormal behavior or unknown threats; finally, according to the calculated event probability, the industrial control system is protected, and the unknown threats and abnormal behaviors faced by the industrial control system are discovered and responded to in a timely manner, so as to ensure the safe and stable operation of the industrial control system.

[0042] Based on the first embodiment of the present application, the second embodiment of the present application is proposed. In the second embodiment of the present application, the same or similar contents as those of the above-mentioned embodiment 1 can be referred to the above introduction, and will not be repeated in the following.

[0043] On this basis, please refer to Figure 2 Before step S20, the safety protection method further includes step S12: Step S12, building a data communication model according to the production process of the industrial control system and the instruction flow in the data traffic, and establishing a normal production process data model.

[0044] The main principle of this application is to use an unsupervised algorithm to comprehensively analyze the network traffic between each asset flowing through the firewall, automatically discover each business asset cluster, learn the communication behavior between and within asset clusters, and construct the minimum behavior boundary of each device or user.

[0045] Therefore, according to the unsupervised algorithm, the industrial control system firewall captures the traffic flowing through the industrial system firewall and obtains real-time traffic data, so as to comprehensively analyze the network traffic between each asset passing through the firewall; the normal behavior model automatically discovers each business asset cluster, learns the communication behavior between and within the asset clusters, and constructs the minimum behavior boundary of each device or user.

[0046] However, since the construction of the normal behavior model relies on a large amount of historical normal behavior data, which often comes from the actual operation of the industrial control system, and contains key information such as data flow and instruction flow in each link of the production process, it is necessary to build a data communication model and establish a normal production process data model to provide a solid data foundation for the construction of the normal behavior model.

[0047] Exemplarily, data that can fully reflect the production process and data flow instruction flow of the industrial control system is collected from various data sources of the industrial control system (such as network traffic logs, system logs, application logs, etc.). Furthermore, the collected data is cleaned, noise is removed, missing values ​​are processed, and standardization and normalization are performed to improve the accuracy of subsequent modeling. Furthermore, based on the collected data, the instruction flow in the data flow is analyzed to identify the communication relationship between various production equipment or users. A data communication model is built using the communication relationship, and the data communication model describes the data flow and instruction transmission process in the industrial control system.

[0048] On the basis of the data communication model, combined with the production process of the industrial control system, the behavioral characteristics of each production equipment or user at different production stages are analyzed, and the data in the normal production process is modeled based on the behavioral characteristics to establish a data model of the normal production process. The normal production process data model fully considers each link in the production process and the instruction flow in the data flow, and is a model that expresses all normal production process values.

[0049] In this embodiment, a data communication model is built according to the production process of the industrial control system and the instruction flow in the data flow, and a normal production process data model is established. Building a data communication model and establishing a normal production process data model actually provides a solid data foundation for the construction of a normal behavior model, better understands and tracks changes in the production process, and thus ensures that the resulting normal behavior model can always be consistent with the actual operating status of the industrial control system.

[0050] Based on the first embodiment and / or the second embodiment of the present application, the third embodiment of the present application is proposed. In the third embodiment of the present application, the same or similar contents as those of the first embodiment and / or the second embodiment can be referred to the above introduction, and will not be repeated in the following.

[0051] In this embodiment, the step S20 of applying the Pareto distribution and the machine learning algorithm to analyze and model the pre-acquired historical sample data to obtain the normal behavior model may include steps S21 to S25: Step S21, collecting historical sample data from the data source of the industrial control system based on the data communication model and the normal production process data model, the historical sample data including historical flow data and known threat label data; Using the data communication model and the normal production process data model as a guide, historical sample data is collected from various data sources of industrial control systems (such as network traffic logs, system logs, application logs, etc.).

[0052] It should be noted that historical sample data includes historical traffic data and known threat label data. The known threat label data comes from previous security event records or external threat intelligence sources.

[0053] Step S22, cleaning the historical traffic data and the known threat label data; Clean historical traffic data and known threat label data. The cleaning process includes removing noise (such as invalid or redundant data), processing missing values ​​(such as by interpolating or deleting missing records), standardizing and normalizing data, etc. Since dirty data may cause the model to learn incorrect features or patterns, the data cleaning step is crucial to the accuracy of the subsequent normal behavior model.

[0054] Step S23, constructing a multi-dimensional data set according to the data characteristics of the cleaned historical traffic data and known threat label data; After data cleaning, a data set containing multiple dimensions is constructed based on the data's attributes, identification, and source. These dimensions include time, source address, destination address, protocol type, traffic size, etc., in order to more comprehensively reflect the network traffic behavior of the industrial control system.

[0055] Step S24, applying Pareto distribution to analyze the cleaned historical traffic data to obtain Pareto analysis results; The cleaned historical traffic data is analyzed using Pareto distribution to identify imbalances and extreme value distributions in the data. Specifically, by analyzing the traffic behavior of each asset in the industrial control system, its Pareto distribution parameters, such as shape parameters and scale parameters, can be calculated to obtain Pareto analysis results.

[0056] By analyzing the Pareto analysis results, we can understand the characteristics of each asset or user's traffic behavior, including which traffic is normal and which may be abnormal. The parameters in the Pareto analysis results will be used for the subsequent normal behavior model construction.

[0057] Step S25, modeling the multidimensional data set using a machine learning algorithm based on the Pareto analysis result to obtain a normal behavior model.

[0058] Since the Pareto distribution has been used to analyze the cleaned historical traffic data, the imbalance and extreme value distribution in the data have been identified. Further, the Pareto analysis results are combined with machine learning algorithms (such as logistic regression, support vector machine, random forest, etc.) to model the multi-dimensional data set, learn the characteristics of normal traffic behavior, and build a normal behavior model that can predict and distinguish normal from abnormal behavior.

[0059] Specifically, in a feasible implementation manner, step S25 may include steps S251 to S253: Step S251, selecting a corresponding machine learning algorithm according to the multi-dimensional data set; Multidimensional data sets include network traffic logs, system logs, application logs, etc., which together constitute a comprehensive description of network traffic, user behavior, host logs and other data in industrial scenarios. Select appropriate machine learning algorithms based on the data characteristics of the collected multidimensional data sets and the nature of the problem. For example, for a binary classification problem (normal behavior vs. abnormal behavior), you can choose algorithms such as logistic regression, support vector machine (SVM), and random forest.

[0060] Step S252, combining the Pareto analysis result, using a corresponding machine learning algorithm to model the multi-dimensional data set to obtain an initial normal behavior model; Use Pareto analysis results to assist machine learning algorithms in modeling multidimensional data sets. Pareto analysis is a statistical method used to identify imbalances in data, that is, a few key factors often have a significant impact on the results. In network security scenarios, Pareto analysis can help the system identify the most important traffic sources or patterns.

[0061] Combined with the results of the Pareto analysis, the selected machine learning algorithm is used to model the multidimensional data set. In this process, the machine learning algorithm will learn the characteristics of normal traffic behavior, including the size, frequency, source and destination of the traffic, and build a model that can distinguish between normal and abnormal behavior. Through this step, an initial normal behavior model is obtained.

[0062] Step S253: Use the multi-dimensional data set to perform data training and model iteration on the initial normal behavior model to obtain a normal behavior model.

[0063] Furthermore, in order to ensure the accuracy of the normal behavior model prediction, the collected multi-dimensional data set is used to perform data training and model iteration on the initial normal behavior model, allowing the model to learn from a large amount of data and optimize its parameters.

[0064] During the training process, new traffic data points are matched with the model, and the model parameters are adjusted based on the matching results. Through multiple iterations of training, the model gradually adapts to and learns the complex characteristics of normal traffic behavior. Ultimately, an optimized and verified normal behavior model is obtained, which can accurately identify and predict the difference between normal and abnormal behavior.

[0065] In this embodiment, the Pareto distribution and machine learning algorithm are applied to perform deep learning and feature extraction on normal behaviors in historical sample data, thereby accurately identifying the behavior patterns of normal traffic. Based on the learned normal behavior features, a model that can describe and match normal behaviors is constructed, and the initial normal behavior model is trained to ensure that the normal behavior model can more accurately detect abnormal behaviors or unknown threats.

[0066] Based on the above embodiments of the present application, a fourth embodiment of the present application is proposed. In the fourth embodiment of the present application, the same or similar contents as those of the above embodiments can be referred to the above introduction, and will not be described in detail later.

[0067] In this embodiment, the step S30 of matching the real-time traffic data with the normal behavior model and calculating the event probability in combination with the extreme value theory may include steps S31 to S33: Step S31, matching the real-time traffic data with the normal behavior model; Match the real-time traffic data with the normal behavior model to match the normal behavior model of the corresponding asset. For example, before matching, the real-time traffic data may also be preprocessed to ensure the accuracy and consistency of the data. The preprocessing step includes data cleaning (removing noise, handling missing values), standardization and normalization.

[0068] Step S32, calculating the matching degree between the real-time traffic data and the normal behavior model according to the Pareto analysis result; The process of matching real-time traffic data with the normal behavior model is not a simple comparison of data values, but refers to the characteristics of the analysis results of the Pareto distribution. Specifically, the industrial control system firewall calculates the degree of match between the new traffic data and the normal behavior model of each asset, and the calculation of this degree of match involves the results of the Pareto analysis. By comparing the characteristics of the real-time traffic data with the distribution law of extreme values ​​described in the Pareto distribution, it is evaluated whether the real-time traffic data conforms to the normal behavior model.

[0069] Step S33: Based on the matching degree and in combination with extreme value theory, the event probability of the real-time traffic data is evaluated.

[0070] Extreme value theory is a statistical method used to study the distribution of extreme values ​​in random variables. In network security scenarios, the behavior of each asset can be regarded as a random variable, and whether there is an anomaly can be determined by observing the maximum or minimum value of its traffic behavior.

[0071] Combining the matching degree between real-time traffic data and normal behavior model and the idea of ​​extreme value theory, the industrial control system firewall can evaluate the event probability of real-time traffic data. The event probability reflects the possibility of whether the real-time traffic data is normal behavior or abnormal behavior.

[0072] For example, if the real-time traffic data matches the normal behavior model very well, then the event probability will be higher and it is more likely to be considered as normal behavior; conversely, if the match is very low, then the event probability will be lower and it is more likely to be considered as abnormal behavior or an unknown threat.

[0073] In this embodiment, real-time traffic data is matched with a normal behavior model, and abnormal behavior in network traffic can be monitored in real time. By comparing real-time traffic with a preset normal behavior model, the system quickly identifies traffic data that deviates from the normal behavior pattern. The accuracy of abnormal behavior detection can be further improved by combining extreme value theory to calculate event probabilities. The application of extreme value theory in network security scenarios enables industrial control system firewalls to determine whether traffic behavior is abnormal based on the distribution law of the maximum or minimum values ​​in a large number of independently distributed random variables (i.e., asset behaviors). This not only improves the protection capabilities of industrial control firewalls against known threats, but more importantly, enhances the detection and protection capabilities against unknown threats and abnormal behaviors, making the security protection of industrial control systems more comprehensive and effective.

[0074] Based on the above embodiments of the present application, a fifth embodiment of the present application is proposed. In the fifth embodiment of the present application, the same or similar contents as those of the above embodiments can be referred to the above introduction, and will not be described in detail later.

[0075] In this embodiment, step S40 of performing security protection on the industrial control system according to the event probability may include steps S41 to S43: Step S41, when the event probability is lower than a preset threshold, the real-time traffic data is marked as an abnormal event and enters the alarm processing stage; The event probability reflects the degree to which the real-time traffic data deviates from the normal behavior model. When this probability is lower than the preset threshold (for example, 5%), the industrial control system firewall considers this real-time traffic data to be abnormal, which may be a manifestation of unknown threats or abnormal behavior. Therefore, the industrial control system firewall will mark this real-time traffic data as an abnormal event and trigger the alarm processing stage.

[0076] It should be noted that the preset thresholds are set based on the security requirements and historical data of the industrial control system, aiming to ensure that the system can accurately identify abnormal behavior while avoiding false positives and negative negatives.

[0077] Step S42, performing security protection on the industrial control system by entering the alarm processing stage; During the alarm processing stage, the industrial control system firewall takes a series of measures to provide security protection for the industrial control system.

[0078] Specifically, in a feasible implementation manner, the step S42 of performing security protection on the industrial control system by entering the alarm processing stage may include steps S421-S422: Step S421, marking the real-time traffic data as an abnormal event and notifying the administrator in a visual manner; When the event probability of real-time traffic data is lower than the preset threshold and is determined to be an abnormal event by the industrial control system firewall, the industrial control system firewall will generate an alarm or notification on the user interface, clearly indicating which traffic data is marked as abnormal and the possible causes or characteristics of the abnormality. In this way, administrators can quickly understand the potential threats in the industrial control system and take appropriate measures to deal with them.

[0079] The advantage of visual notifications is that they can intuitively display abnormal events, helping administrators to understand the nature of the problem more quickly and make accurate decisions. In addition, by providing detailed abnormal event information, administrators can also more easily track and locate the root cause of the problem, thereby solving security problems more effectively.

[0080] Step S422: repair the security problem existing in the industrial control system according to the abnormal event, and adjust the relevant parameters of the normal behavior model.

[0081] In order to further improve the security protection capabilities of industrial control systems, the relevant parameters of the normal behavior model are adjusted according to abnormal events. This is because abnormal events may reveal some unknown or unexpected behavior patterns in industrial control systems, and these patterns need to be taken into account in the normal behavior model. By adjusting the model parameters, the normal behavior model can more accurately reflect the actual behavior of the industrial control system, thereby improving the detection and protection capabilities of the industrial control system against unknown threats.

[0082] Step S43, when the event probability is not lower than a preset threshold, real-time monitoring of the flow data newly flowing into the industrial control system to safely protect the industrial control system; When the event probability of real-time traffic data is not lower than the preset threshold, the industrial control system firewall considers this real-time traffic data to be normal and does not deviate from the normal behavior model. However, this does not mean that the industrial control system firewall can relax its vigilance. Because network attacks and abnormal behaviors are constantly changing, the industrial control system firewall needs to continuously monitor the new traffic data flowing into the industrial control system to ensure the security protection of the industrial control system.

[0083] Specifically, the industrial control system firewall will continue to collect and process new inflow traffic data, update the normal behavior model, and monitor event probabilities in real time. If a new abnormal event is found, the industrial control system firewall will immediately enter the alarm processing stage and take corresponding security protection measures. In this way, the industrial control system firewall can continuously adapt to changes in the network environment and ensure the safe and stable operation of the industrial control system.

[0084] In this embodiment, when the event probability of real-time traffic data is lower than the preset threshold, the industrial control system firewall can automatically mark it as an abnormal event, timely and accurately identify traffic data that does not conform to normal behavior patterns, and provide key information for subsequent security protection; by marking abnormal events, normal traffic and potential threats are distinguished, thereby enhancing the security monitoring capability of the industrial control system. The start of the alarm processing phase means that the industrial control system firewall has identified potential risks and is ready to take further actions to ensure the safe operation of the industrial control system. According to the abnormal events, the security issues existing in the industrial control system are repaired, and the relevant models are adjusted to eliminate security risks in a timely manner, thereby improving the security protection capability of the industrial control system. By repairing security issues and adjusting models, new security threats are continuously learned and adapted to maintain effective protection of the industrial control system and respond to unknown threats and abnormal behaviors faced by the industrial control system.

[0085] For example, in order to help understand the implementation process of the security protection method obtained by combining the present embodiment with the above-mentioned embodiments 1 to 4, please refer to Figure 3 , Figure 3A brief flowchart of a safety protection method is provided, specifically: "Start" is the starting point of the process, marking the start of the modeling and analysis process based on traffic behavior. In this stage, the real-time traffic data flowing through the industrial control system firewall is captured for subsequent processing.

[0086] Modeling based on traffic behavior: Industrial control system firewalls collect historical traffic data and use this data to build models. The purpose of the model is to simulate or predict future traffic behavior and help understand traffic patterns and trends.

[0087] Specifically, Figure 4 As shown, Figure 4 The schematic diagram of the structure of the industrial control system firewall provided in this application. The industrial control system firewall based on unknown threat protection consists of a data acquisition module, a data processing module, an analysis and modeling module, a data training module, and an early warning display module. Figure 5 , Figure 5 The following is a flow chart of building a normal behavior model. The data acquisition module collects control system PLC, firewall traffic and data; the data processing module processes the collected data according to the attributes, identification and source of the data; the analysis and modeling module relies on the built-in behavior algorithm to model the processed data; the data training module relies on the normal behavior data model for training to establish a normal security baseline. The normal behavior model is built through the data acquisition module, data processing module, analysis and modeling module, and data training module of the industrial control system firewall to establish a normal security baseline.

[0088] Pareto Distribution: Apply Pareto distribution to analyze real-time traffic data. Pareto distribution is a statistical distribution that is often used to identify imbalances in data, such as 80% of the impact comes from 20% of the causes. This step is used to identify the most important traffic sources or patterns.

[0089] It should be noted that during the analysis and modeling process, the industrial control system firewall may receive new traffic data points. This new data is critical to updating the model and maintaining the accuracy of the analysis.

[0090] Determine the probability size: Based on the analysis results of the Pareto distribution, evaluate the probability size of each data point. This helps to determine which data points are critical and which may be abnormal or unimportant. For example, if the probability of a data point is small, that is, it is unlikely to be the main source of traffic or mode, it is marked as "small probability". This helps to focus on the key factors that are more likely to affect the system. Group the data points with higher probabilities into a large cluster. This cluster represents the most important source of traffic or mode in the industrial control system and requires special attention.

[0091] Judged as normal: If all data points are analyzed and considered normal, that is, no abnormal or unexpected behavior is found, then the analysis of this real-time traffic data ends.

[0092] Judged as abnormal: If data points that do not meet expectations are found during the data analysis process, they are marked as "abnormal" and the alarm processing stage needs to be entered.

[0093] Alarm processing: For data points marked as abnormal, further processing is required. Processing includes fixing problems, adjusting model parameters, notifying administrators, etc., to ensure the normal operation and security of industrial control systems.

[0094] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the security protection method of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0095] This application also provides a safety protection device, please refer to Figure 6 , the safety protection device comprises: A data acquisition module 10, for capturing real-time flow data based on flow behavior of the industrial control system; The analysis and modeling module 20 is used to apply Pareto distribution and machine learning algorithm to analyze and model the pre-acquired historical sample data to obtain a normal behavior model; A model matching module 30, for matching the real-time traffic data with the normal behavior model, and calculating event probability in combination with extreme value theory; The early warning module 40 is used to perform security protection on the industrial control system according to the event probability.

[0096] The safety protection device provided by the present application adopts the safety protection method in the above embodiment to solve the technical problem of safety protection. Compared with the prior art, the beneficial effects of the safety protection device provided by the present application are the same as the beneficial effects of the safety protection method provided by the above embodiment, and the other technical features in the safety protection device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0097] The present application provides a security protection device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the security protection method in the above-mentioned embodiment one.

[0098] Reference below Figure 7, which shows a schematic diagram of the structure of the safety protection equipment suitable for implementing the embodiment of the present application. The safety protection equipment in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 7 The safety protection device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0099] like Figure 7 As shown, the safety protection device may include a processing device 1001 (such as a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the safety protection device are also stored. The processing device 1001, ROM1002 and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the safety protection device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a safety protection device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or have alternatively.

[0100] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0101] The safety protection device provided by the present application adopts the safety protection method in the above embodiment to solve the technical problem of safety protection. Compared with the prior art, the beneficial effects of the safety protection device provided by the present application are the same as the beneficial effects of the safety protection method provided by the above embodiment, and the other technical features in the safety protection device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0102] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0103] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present 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.

[0104] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, wherein the computer-readable program instructions are used to execute the security protection method in the above-mentioned embodiment.

[0105] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM: Random Access Memory), a read-only memory (ROM: Read Only Memory), an erasable programmable read-only memory (EPROM: Erasable Programmable Read Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM: CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency: Radio Frequency), etc., or any suitable combination of the above.

[0106] The computer-readable storage medium may be included in the safety protection device; or it may exist independently without being assembled into the safety protection device.

[0107] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the security protection device, the security protection device: captures real-time traffic data based on the traffic behavior of the industrial control system; applies Pareto distribution and machine learning algorithms to analyze and model the pre-acquired historical sample data to obtain a normal behavior model; matches the real-time traffic data with the normal behavior model, and calculates the event probability in combination with extreme value theory; and performs security protection on the industrial control system based on the event probability.

[0108] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0109] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0110] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0111] The readable storage medium provided in this application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned security protection method, and can solve the technical problems of security protection. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the security protection method provided in the above-mentioned embodiment, and will not be repeated here.

[0112] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned security protection method when executed by a processor.

[0113] The computer program product provided by this application can solve the technical problem of security protection. Compared with the prior art, the beneficial effects of the computer program product provided by this application are the same as the beneficial effects of the security protection method provided by the above embodiment, which will not be repeated here.

[0114] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A safety protection method, characterized in that: Applied to industrial control systems, the method comprises: capturing real-time flow data based on flow behavior of the industrial control system; Apply Pareto distribution and machine learning algorithms to analyze and model the historical sample data obtained in advance to obtain a normal behavior model; Matching the real-time traffic data with the normal behavior model, and calculating event probability in combination with extreme value theory; The industrial control system is protected according to the event probability.

2. The method according to claim 1, characterized in that Before the step of applying the Pareto distribution and the machine learning algorithm to analyze and model the pre-acquired historical sample data to obtain the normal behavior model, the following steps are also included: A data communication model is built according to the production process of the industrial control system and the instruction flow in the data traffic, and a normal production process data model is established.

3. The method according to claim 2, characterized in that The step of applying Pareto distribution and machine learning algorithm to analyze and model the pre-acquired historical sample data to obtain a normal behavior model includes: Collecting historical sample data from a data source of the industrial control system based on the data communication model and the normal production process data model, wherein the historical sample data includes historical traffic data and known threat label data; Cleaning the historical traffic data and the known threat label data; Construct a multi-dimensional data set based on the data characteristics of the cleaned historical traffic data and known threat label data; Apply Pareto distribution to analyze the cleaned historical traffic data and obtain Pareto analysis results; Based on the Pareto analysis results, the multidimensional data set is modeled using a machine learning algorithm to obtain a normal behavior model.

4. The method according to claim 3, characterized in that The step of using a machine learning algorithm to model the multi-dimensional data set based on the Pareto analysis result to obtain a normal behavior model comprises: Selecting a corresponding machine learning algorithm according to the multidimensional data set; In combination with the Pareto analysis results, a corresponding machine learning algorithm is used to model the multi-dimensional data set to obtain an initial normal behavior model; The multi-dimensional data set is used to perform data training and model iteration on the initial normal behavior model to obtain a normal behavior model.

5. The method according to claim 2, characterized in that The step of matching the real-time traffic data with the normal behavior model and calculating the event probability in combination with extreme value theory includes: Matching the real-time traffic data with the normal behavior model; According to the Pareto analysis result, calculating the matching degree between the real-time traffic data and the normal behavior model; Based on the matching degree and in combination with extreme value theory, the event probability of the real-time traffic data is evaluated.

6. The method according to claim 1, characterized in that The step of performing security protection on the industrial control system according to the event probability comprises: When the event probability is lower than a preset threshold, the real-time traffic data is marked as an abnormal event and enters the alarm processing stage; Performing security protection on the industrial control system by entering the alarm processing stage; When the event probability is not lower than a preset threshold, the flow data newly flowing into the industrial control system is monitored in real time to safely protect the industrial control system.

7. The method according to claim 6, characterized in that When the event probability is lower than a preset threshold, the real-time traffic data is marked as an abnormal event and the step of entering the alarm processing stage includes: After marking the real-time traffic data as an abnormal event, an administrator is notified in a visual manner; The security problem existing in the industrial control system is repaired according to the abnormal event, and the relevant parameters of the normal behavior model are adjusted.

8. A safety protection device, characterized in that: The safety protection device comprises: Data acquisition module for capturing real-time flow data based on the flow behavior of industrial control systems; The analysis and modeling module is used to apply Pareto distribution and machine learning algorithms to analyze and model the pre-acquired historical sample data to obtain a normal behavior model; A model matching module, used to match the real-time traffic data with the normal behavior model and calculate event probability in combination with extreme value theory; An early warning module is used to provide security protection for the industrial control system according to the probability of the event.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the security protection method according to any one of claims 1 to 7 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the security protection method according to any one of claims 1 to 7 are implemented.

Citation Information

Patent Citations

  • Smart city network equipment performance abnormity threshold evaluation method based on extreme value theory

    CN115001997A

  • Mixed-flow type runner fatigue life calculation method based on actually measured dynamic stress

    CN116663258A

  • Multi-dimensional time sequence anomaly detection method based on time variable double-attention mechanism

    CN118378139A

  • Computer information security management system

    CN118468291A

  • Systems and methods for probabilistic forecasting of extremes

    EP4293541A1