Network traffic anomaly sensing system and method based on big data

Through big data technology, the industrial network traffic is split into sub-sequences, and the perception sensitivity level is classified, mined and adjusted, which solves the real-time monitoring and prediction problems of industrial network traffic anomaly detection in the existing technology, and achieves high accuracy of industrial network traffic Improved monitoring and robustness.

CN118972140APending Publication Date: 2024-11-15CHANGZHOU CHUXIN CULTURE TECH CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202411158632.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-22
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

Existing industrial network traffic anomaly detection methods are difficult to monitor and predict network security risks in real time, and are sensitive to external noise, poorly robust, and difficult to adapt to dynamically changing network traffic data.

Method used

Using a network traffic abnormality perception system based on big data, the industrial production process is split into process subsequences through the data monitoring module, the classification matching module divides data types, the data mining module mines attack characteristics, the sensitivity level matching module predicts perception sensitivity level, and the sensitivity The adjustment module adjusts the perception sensitivity level in real time to achieve accurate monitoring and prediction of industrial network traffic.

Benefits of technology

It significantly improves the accuracy of abnormal traffic perception in industrial networks, can explore the abnormal traffic attack rules in real time, predict security events, improves network situation awareness capabilities, and significantly improves robustness when facing massive traffic data.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118972140A_ABST
    Figure CN118972140A_ABST
Patent Text Reader

Abstract

A network traffic anomaly sensing system based on big data relates to the technical field of network traffic monitoring and comprises a monitoring center which is in communication connection with a data monitoring module, a classification matching module, a data mining module, a sensitivity grade matching module and a sensitivity adjusting module. The data monitoring module is used for acquiring monitoring data of each flow sub-sequence; the classification matching module is used for dividing the data types of the monitoring data into normal data, main attack data and attack data; the data mining module is used for performing real intention data mining on the attack data; the sensitivity grade matching module is used for obtaining the prediction perception sensitivity grade of each process sub-sequence in the current monitoring period in the remaining different moments after each process sub-sequence generates the attack data in the current monitoring period; and the sensitivity adjusting module is used for adjusting the sensing sensitivity level of the flow sub-sequence, so that the flow anomaly sensing accuracy of the industrial network is remarkably improved in the face of mass data of the industrial network.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of network traffic monitoring, and in particular to a network traffic anomaly perception system and method based on big data. Background Art

[0002] When the industrial control system is operating normally, the various devices are not completely synchronized in their collaborative work, and the amount of network traffic consumed by each device is not exactly the same. When the industrial control system is attacked by the network, there are also jumps in the network traffic. Attackers often seize this opportunity to attack the industrial control system and confuse it. At the same time, they will also use this opportunity to insert various attack risk factors, causing long-term and difficult-to-solve problems for the industrial control system. In terms of network traffic security monitoring, the current technology is mainly aimed at changes in the operating performance of the equipment or equipment vulnerabilities. Continuous product updates require a lot of R&D and production costs; Comparative document CN115102790A "Network traffic anomaly perception system and method based on big data" establishes a normal perception time model and an abnormal perception time model for network traffic; through the normal perception time model and the abnormal perception time model for network traffic, it further predicts the unknown risk state perception network traffic jump time; and then combines the predicted normal, abnormal and unknown risk perception times to predict the risk value of network traffic, and judges whether the network traffic is normal based on the risk value. At the same time, combined with the predicted time model, it locates the specific time when the network traffic anomaly may occur, realizes real-time monitoring of the network traffic status, and provides accurate prevention and maintenance time preparation for relevant responsible personnel.

[0003] Comparative document CN111695823A "An abnormality assessment method and system based on industrial control network traffic" This method includes the following steps: abnormality verification step, filtering the misreported abnormal information through known multi-source security information; abnormality aggregation step, reducing the number of abnormal traffic through aggregation algorithm, and realizing the standardization of industrial network abnormal information; abnormality association step, analyzing, perceiving and predicting security events through abnormal association. This abnormality assessment system is conducive to improving the interpretability and predictability of industrial control network traffic, enhancing network situation awareness capabilities, and timely avoiding security risks in industrial networks.

[0004] Current anomaly detection methods urgently need to solve the following problems: when industrial networks are attacked, a large amount of abnormal traffic is generated, and it is difficult to explore the patterns of a large amount of data through manual means. In addition, existing methods are based on post-evaluation of security incidents, and lack methods to display current network security risks in real time and predict possible security incidents. At the same time, anomaly detection methods usually analyze traffic distribution and set traffic classification thresholds to detect abnormal traffic in the network. Although this method is simple to implement, it has a fixed method of use and is difficult to adapt to a variety of dynamically changing network traffic data. In addition, the performance of this method is sensitive to external noise, so its robustness is poor. Summary of the invention

[0005] In order to solve the above technical problems, the object of the present invention is to provide a network traffic anomaly perception system based on big data, including a monitoring center, wherein the monitoring center is communicatively connected with a data monitoring module, a classification matching module, a data mining module, a sensitivity level matching module and a sensitivity adjustment module; The data monitoring module is used to split the industrial production process into several process sub-sequences and obtain monitoring data of each process sub-sequence; The classification matching module is used to classify the data types of the monitoring data collected by the industrial network monitoring points of each process subsequence into normal data, main attack data and feint attack data; The data mining module is used to perform real intention data mining on feint attack data of various types of attack features generated in several historical monitoring cycles of each process subsequence; The sensitivity level matching module is used to obtain the predicted perception sensitivity level of each process subsequence at the remaining different moments in the current monitoring cycle according to the probability of generating main attack data of each type of attack feature in the remaining different moments of each process subsequence after each process subsequence generates feint attack data in the current monitoring cycle and the attack feature type of the main attack data; The sensitivity adjustment module is used to adjust the perception sensitivity level of the process subsequence in real time according to the predicted perception sensitivity level and the data type of the real-time monitoring data.

[0006] Furthermore, the data monitoring module splits the industrial production process into several process sub-sequences, and the process of obtaining monitoring data of each process sub-sequence includes: Obtain the process flow characteristics of the current industrial network production equipment, extract process information based on the process flow characteristics, split the industrial production process according to the process information, and divide it into several process sub-sequences; Industrial network monitoring points are set in each process subsequence, and performance monitoring indicators and flow monitoring indicators of each process subsequence are obtained according to the functional characteristics of the process flow characteristics of each process subsequence; The industrial network monitoring point obtains corresponding monitoring data according to the performance monitoring index and the flow monitoring index, marks the monitoring time, and sets the monitoring cycle.

[0007] Furthermore, the classification matching module divides the data types of the monitoring data collected by the industrial network monitoring points of each process subsequence into normal data, main attack data and feint attack data, including: Acquire the monitoring data collected by the industrial network monitoring points of each process subsequence, obtain the performance indicator threshold and flow indicator threshold corresponding to each process subsequence, and compare the performance indicator monitoring data and flow indicator monitoring data in the monitoring data of each process subsequence with the corresponding performance indicator threshold and flow indicator threshold respectively; The monitoring data of the flow index monitoring data not meeting the corresponding flow index threshold and the monitoring data of the performance index monitoring data meeting the corresponding performance index threshold are marked as feint data; The monitoring data of the flow index monitoring data that does not meet the corresponding flow index threshold and the monitoring data of the performance index monitoring data that does not meet the corresponding performance index threshold are marked as the main attack data; The monitoring data whose flow index monitoring data meets the corresponding flow index threshold and whose performance index monitoring data meets the corresponding performance index threshold are marked as normal data.

[0008] Furthermore, the process of the data mining module performing real intention data mining on the feint attack data of various types of attack features generated in several historical monitoring cycles of each process subsequence includes: Obtain historical monitoring data collected by industrial network monitoring points of each process subsequence for several historical monitoring cycles, obtain feint attack data of various types of attack features generated by the target process subsequence in several historical monitoring cycles, and main attack data of various types of attack features generated by the target process subsequence and other process subsequences at different remaining moments in the corresponding several historical monitoring cycles; An abnormal association model is constructed based on deep learning, and the feint data of various types of attack features generated by each process subsequence in several historical monitoring cycles and the main attack data of various types of attack features generated by each process subsequence at different remaining moments in the corresponding several historical monitoring cycles are used as historical training sets. The fault detection model is trained in real time by dividing the historical data set into a training set and a test set, and the trained abnormal association model is output; After obtaining the feint attack data of different types of attack features generated by the target process subsequence in the current monitoring period through the abnormal association model, the time series distribution probability of the main attack data of each type of attack features generated by the target process subsequence and other process subsequences in the current monitoring period at the remaining different moments is obtained.

[0009] Furthermore, the sensitivity level matching module obtains the predicted perception sensitivity level of each process subsequence at the remaining different moments in the current monitoring cycle according to the probability of generating the main attack data of each type of attack feature of each process subsequence at the remaining different moments and the attack feature type of the main attack data after each process subsequence generates the feint attack data in the current monitoring cycle, and the process includes: After each process subsequence generates feint attack data in the current monitoring cycle, the probability of each process subsequence generating main attack data of each type of attack feature in the remaining different moments and the attack feature type of the main attack data are used as evaluation indicators, the indicator weights of the evaluation indicators are set, the perception sensitivity level is preset, and the membership matrix of each process subsequence to the perception sensitivity level is judged through fuzzy comprehensive evaluation; After each process subsequence generates feint data in the current monitoring period, the predicted perception sensitivity level of each process subsequence at different remaining moments in the current monitoring period is obtained according to the membership matrix and the indicator weight.

[0010] Furthermore, the process of the sensitivity adjustment module adjusting the perception sensitivity level of the process subsequence in real time according to the predicted perception sensitivity level and the data type of the real-time monitoring data includes: Acquire the real-time monitoring data collected by the industrial network monitoring points of the target process subsequence in the current monitoring cycle, classify the data types of the real-time monitoring data, and if the real-time monitoring data is the main attack data, generate a network isolation and repair signal and send it to the monitoring center, and increase the perception sensitivity level of other process subsequences in the current monitoring cycle to the highest sensitivity level; If the real-time monitoring data is feint data, then the predicted perception sensitivity levels of the target process subsequence and other process subsequences at different remaining moments in the current monitoring cycle are obtained, and the perception sensitivity levels of the target process subsequence and other process subsequences at different remaining moments in the current monitoring cycle are adjusted according to the predicted perception sensitivity levels.

[0011] Furthermore, when the monitoring center receives the network isolation and repair signal, it performs network isolation on the process subsequence according to the network isolation and repair signal, and generates network attack emergency measures to feed back to the management personnel. The management personnel performs network repair or defense upgrade on the process subsequence according to the network attack emergency measures.

[0012] The network traffic anomaly perception method based on big data includes the following steps: Step S1: split the industrial production process into several process sub-sequences, and obtain monitoring data of each process sub-sequence; divide the data type of the monitoring data collected by the industrial network monitoring points of each process sub-sequence into normal data, main attack data and feint attack data; Step S2: Perform real intention data mining on the feint attack data of various types of attack features generated in several historical monitoring cycles of each process subsequence; obtain the predicted perception sensitivity level of each process subsequence at the remaining different moments in the current monitoring cycle according to the probability of generating the main attack data of various types of attack features of each process subsequence at the remaining different moments after each process subsequence generates the feint attack data in the current monitoring cycle and the attack feature type of the main attack data; Step S3: adjusting the perception sensitivity level of the process subsequence in real time according to the predicted perception sensitivity level and the data type of the real-time monitoring data.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: the data types of monitoring data collected by the industrial network monitoring points of each process subsequence are divided into normal data, main attack data and feint data, and real intention data mining is performed on the feint data of various types of attack features generated in several historical monitoring cycles of each process subsequence; according to the probability of generating main attack data of various types of attack features of each process subsequence in the remaining different moments after each process subsequence generates feint data in the current monitoring cycle and the attack feature type of the main attack data, the predicted perception sensitivity level of each process subsequence in the remaining different moments in the current monitoring cycle is obtained, and the perception sensitivity level of the process subsequence is adjusted in real time according to the predicted perception sensitivity level and the data type of the real-time monitoring data. When the industrial network encounters an attack and generates a large amount of abnormal traffic, the abnormal traffic attack law in the industrial network is explored by means of the analysis of a large amount of data, the probability of occurrence of subsequent main attack data is predicted based on the existing feint data, and the possible security incidents are inferred, and the perception sensitivity level of the process subsequence is adjusted in real time. When facing the massive traffic data of the industrial network, the accuracy of traffic anomaly perception of the industrial network is significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a schematic diagram of a network traffic anomaly perception system based on big data according to an embodiment of the present application.

[0015] Figure 2 This is a schematic diagram of a network traffic anomaly perception method based on big data according to an embodiment of the present application. DETAILED DESCRIPTION

[0016] The following is a clear and complete description of the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0017] like Figure 1As shown, a network traffic anomaly perception system based on big data includes a monitoring center, and the monitoring center is communicatively connected with a data monitoring module, a classification matching module, a data mining module, a sensitivity level matching module and a sensitivity adjustment module; The data monitoring module is used to split the industrial production process into several process sub-sequences and obtain monitoring data of each process sub-sequence; The classification matching module is used to classify the data types of the monitoring data collected by the industrial network monitoring points of each process subsequence into normal data, main attack data and feint attack data; The data mining module is used to perform real intention data mining on feint attack data of various types of attack features generated in several historical monitoring cycles of each process subsequence; The sensitivity level matching module is used to obtain the predicted perception sensitivity level of each process subsequence at the remaining different moments in the current monitoring cycle according to the probability of generating main attack data of each type of attack feature in the remaining different moments of each process subsequence after each process subsequence generates feint attack data in the current monitoring cycle and the attack feature type of the main attack data; The sensitivity adjustment module is used to adjust the perception sensitivity level of the process subsequence in real time according to the predicted perception sensitivity level and the data type of the real-time monitoring data.

[0018] It should be further explained that, in the specific implementation process, the data monitoring module splits the industrial production process into several process sub-sequences, and the process of obtaining the monitoring data of each process sub-sequence includes: Obtain the process flow characteristics of the current industrial network production equipment, extract process information based on the process flow characteristics, split the industrial production process according to the process information, and divide it into several process sub-sequences; Industrial network monitoring points are set in each process subsequence, and performance monitoring indicators and flow monitoring indicators of each process subsequence are obtained according to the functional characteristics of the process flow characteristics of each process subsequence; The industrial network monitoring point obtains corresponding monitoring data according to the performance monitoring index and the flow monitoring index, marks the monitoring time, and sets the monitoring cycle.

[0019] It should be further explained that, in the specific implementation process, the classification matching module divides the data types of the monitoring data collected by the industrial network monitoring points of each process subsequence into normal data, main attack data and feint attack data, including: Acquire the monitoring data collected by the industrial network monitoring points of each process subsequence, obtain the performance indicator threshold and flow indicator threshold corresponding to each process subsequence, and compare the performance indicator monitoring data and flow indicator monitoring data in the monitoring data of each process subsequence with the corresponding performance indicator threshold and flow indicator threshold respectively; The monitoring data of the flow index monitoring data not meeting the corresponding flow index threshold and the monitoring data of the performance index monitoring data meeting the corresponding performance index threshold are marked as feint data; The monitoring data of the flow index monitoring data that does not meet the corresponding flow index threshold and the monitoring data of the performance index monitoring data that does not meet the corresponding performance index threshold are marked as the main attack data; The monitoring data whose flow index monitoring data meets the corresponding flow index threshold and whose performance index monitoring data meets the corresponding performance index threshold are marked as normal data.

[0020] It should be further explained that, in the specific implementation process, the process of the data mining module performing real intention data mining on the feint attack data of various types of attack features generated in several historical monitoring cycles of each process subsequence includes: Obtain historical monitoring data collected by industrial network monitoring points of each process subsequence for several historical monitoring cycles, and obtain feint attack data of various types of attack features generated by the target process subsequence in several historical monitoring cycles. Attack feature types include but are not limited to: traffic size, traffic peak, traffic duration, connection frequency, traffic attack location, and main attack data of various types of attack features generated by the target process subsequence and other process subsequences at different remaining moments in the corresponding several historical monitoring cycles; An abnormal association model is constructed based on deep learning, and the feint data of various types of attack features generated by each process subsequence in several historical monitoring cycles and the main attack data of various types of attack features generated by each process subsequence at different remaining moments in the corresponding several historical monitoring cycles are used as historical training sets. The fault detection model is trained in real time by dividing the historical data set into a training set and a test set, and the trained abnormal association model is output; After obtaining the feint attack data of different types of attack features generated by the target process subsequence in the current monitoring period through the abnormal association model, the time series distribution probability of the main attack data of each type of attack features generated by the target process subsequence and other process subsequences in the current monitoring period at the remaining different moments is obtained.

[0021] It should be further explained that, in the specific implementation process, the sensitivity level matching module generates feint attack data for each process subsequence in the current monitoring period, and the generation probability of main attack data of each type of attack feature in the remaining different moments of each process subsequence and the attack feature type of the main attack data. The process of obtaining the predicted perception sensitivity level of each process subsequence in the remaining different moments of the current monitoring period includes: After each process subsequence generates feint attack data in the current monitoring cycle, the probability of generating main attack data of each type of attack feature in the remaining different moments of each process subsequence and the attack feature type of the main attack data are used as evaluation indicators, and the indicator weights of the evaluation indicators are set. The weight vector is determined based on the experience of experts, in order to reduce the uncertainty in the fuzzy comprehensive evaluation model; Preset perception sensitivity levels, including monitoring granularity and alarm thresholds. The monitoring granularity indicates the frequency of data collection at the monitoring points of the industrial network. More detailed monitoring of each part of the industrial network can be achieved through more frequent monitoring. At the same time, by adjusting the alarm threshold of the system, it can become more sensitive to potential threats. The membership matrix of each process subsequence to the perception sensitivity level is judged through fuzzy comprehensive evaluation; After each process subsequence generates feint data in the current monitoring period, the predicted perception sensitivity level of each process subsequence at different remaining moments in the current monitoring period is obtained according to the membership matrix and the indicator weight.

[0022] It should be further explained that, in a specific implementation process, the process in which the sensitivity adjustment module adjusts the perception sensitivity level of the process subsequence in real time according to the predicted perception sensitivity level and the data type of the real-time monitoring data includes: Acquire the real-time monitoring data collected by the industrial network monitoring points of the target process subsequence in the current monitoring cycle, classify the data types of the real-time monitoring data, and if the real-time monitoring data is the main attack data, generate a network isolation and repair signal and send it to the monitoring center, and increase the perception sensitivity level of other process subsequences in the current monitoring cycle to the highest sensitivity level; If the real-time monitoring data is feint data, then the predicted perception sensitivity levels of the target process subsequence and other process subsequences at different remaining moments in the current monitoring cycle are obtained, and the perception sensitivity levels of the target process subsequence and other process subsequences at different remaining moments in the current monitoring cycle are adjusted according to the predicted perception sensitivity levels.

[0023] It should be further explained that, during the specific implementation process, when the monitoring center receives a network isolation and repair signal, it performs network isolation on the process subsequence according to the network isolation and repair signal, and generates network attack emergency measures and feeds them back to the management personnel. The management personnel performs network repair or defense upgrade on the process subsequence according to the network attack emergency measures.

[0024] like Figure 2 As shown, the network traffic anomaly perception method based on big data includes the following steps: Step S1: split the industrial production process into several process sub-sequences, and obtain monitoring data of each process sub-sequence; divide the data type of the monitoring data collected by the industrial network monitoring points of each process sub-sequence into normal data, main attack data and feint attack data; Step S2: Perform real intention data mining on the feint attack data of various types of attack features generated in several historical monitoring cycles of each process subsequence; obtain the predicted perception sensitivity level of each process subsequence at the remaining different moments in the current monitoring cycle according to the probability of generating the main attack data of various types of attack features of each process subsequence at the remaining different moments after each process subsequence generates the feint attack data in the current monitoring cycle and the attack feature type of the main attack data; Step S3: adjusting the perception sensitivity level of the process subsequence in real time according to the predicted perception sensitivity level and the data type of the real-time monitoring data.

[0025] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A network traffic anomaly perception system based on big data, including a monitoring center, characterized in that: The monitoring center is communicatively connected with a data monitoring module, a classification matching module, a data mining module, a sensitivity level matching module and a sensitivity adjustment module; The data monitoring module is used to split the industrial production process into several process sub-sequences and obtain monitoring data of each process sub-sequence; The classification matching module is used to classify the data types of the monitoring data collected by the industrial network monitoring points of each process subsequence into normal data, main attack data and feint attack data; The data mining module is used to perform real intention data mining on feint attack data of various types of attack features generated in several historical monitoring cycles of each process subsequence; The sensitivity level matching module is used to obtain the predicted perception sensitivity level of each process subsequence at the remaining different moments in the current monitoring cycle according to the probability of generating main attack data of each type of attack feature in the remaining different moments of each process subsequence after each process subsequence generates feint attack data in the current monitoring cycle and the attack feature type of the main attack data; The sensitivity adjustment module is used to adjust the perception sensitivity level of the process subsequence in real time according to the predicted perception sensitivity level and the data type of the real-time monitoring data.

2. According to the big data-based network traffic anomaly perception system according to claim 1, it is characterized in that: The data monitoring module splits the industrial production process into several process sub-sequences, and the process of obtaining monitoring data of each process sub-sequence includes: Obtain the process flow characteristics of the current industrial network production equipment, extract process information based on the process flow characteristics, split the industrial production process according to the process information, and divide it into several process sub-sequences; Industrial network monitoring points are set in each process subsequence, and performance monitoring indicators and flow monitoring indicators of each process subsequence are obtained according to the functional characteristics of the process flow characteristics of each process subsequence; The industrial network monitoring point obtains corresponding monitoring data according to the performance monitoring index and the flow monitoring index, marks the monitoring time, and sets the monitoring cycle.

3. The network traffic anomaly perception system based on big data according to claim 2 is characterized in that: The process in which the classification matching module divides the data types of the monitoring data collected by the industrial network monitoring points of each process subsequence into normal data, main attack data and feint attack data includes: Acquire the monitoring data collected by the industrial network monitoring points of each process subsequence, obtain the performance indicator threshold and flow indicator threshold corresponding to each process subsequence, and compare the performance indicator monitoring data and flow indicator monitoring data in the monitoring data of each process subsequence with the corresponding performance indicator threshold and flow indicator threshold respectively; The monitoring data of the flow index monitoring data not meeting the corresponding flow index threshold and the monitoring data of the performance index monitoring data meeting the corresponding performance index threshold are marked as feint data; The monitoring data of the flow index monitoring data that does not meet the corresponding flow index threshold and the monitoring data of the performance index monitoring data that does not meet the corresponding performance index threshold are marked as the main attack data; The monitoring data whose flow index monitoring data meets the corresponding flow index threshold and whose performance index monitoring data meets the corresponding performance index threshold are marked as normal data.

4. The network traffic anomaly perception system based on big data according to claim 3 is characterized in that: The process of the data mining module performing real intention data mining on the feint attack data of various types of attack features generated in several historical monitoring cycles of each process subsequence includes: Obtain historical monitoring data collected by industrial network monitoring points of each process subsequence for several historical monitoring cycles, obtain feint attack data of various types of attack features generated by the target process subsequence in several historical monitoring cycles, and main attack data of various types of attack features generated by the target process subsequence and other process subsequences at different remaining moments in the corresponding several historical monitoring cycles; An abnormal association model is constructed based on deep learning, and the feint data of various types of attack features generated by each process subsequence in several historical monitoring cycles and the main attack data of various types of attack features generated by each process subsequence at different remaining moments in the corresponding several historical monitoring cycles are used as historical training sets. The fault detection model is trained in real time by dividing the historical data set into a training set and a test set, and the trained abnormal association model is output; After obtaining the feint attack data of different types of attack features generated by the target process subsequence in the current monitoring period through the abnormal association model, the time series distribution probability of the main attack data of each type of attack features generated by the target process subsequence and other process subsequences in the current monitoring period at the remaining different moments is obtained.

5. The network traffic anomaly perception system based on big data according to claim 4 is characterized in that: The process of obtaining the predicted perception sensitivity level of each process subsequence at different remaining moments in the current monitoring cycle according to the probability of generating main attack data of each type of attack feature of each process subsequence at different remaining moments and the attack feature type of the main attack data by the sensitivity level matching module after each process subsequence generates feint attack data in the current monitoring cycle includes: After each process subsequence generates feint attack data in the current monitoring cycle, the probability of each process subsequence generating main attack data of each type of attack feature in the remaining different moments and the attack feature type of the main attack data are used as evaluation indicators, the indicator weights of the evaluation indicators are set, the perception sensitivity level is preset, and the membership matrix of each process subsequence to the perception sensitivity level is judged through fuzzy comprehensive evaluation; After each process subsequence generates feint data in the current monitoring period, the predicted perception sensitivity level of each process subsequence at different remaining moments in the current monitoring period is obtained according to the membership matrix and the indicator weight.

6. The network traffic anomaly perception system based on big data according to claim 5 is characterized in that: The process of the sensitivity adjustment module adjusting the perception sensitivity level of the process subsequence in real time according to the predicted perception sensitivity level and the data type of the real-time monitoring data includes: Acquire the real-time monitoring data collected by the industrial network monitoring points of the target process subsequence in the current monitoring cycle, classify the data types of the real-time monitoring data, and if the real-time monitoring data is the main attack data, generate a network isolation and repair signal and send it to the monitoring center, and increase the perception sensitivity level of other process subsequences in the current monitoring cycle to the highest sensitivity level; If the real-time monitoring data is feint data, then the predicted perception sensitivity levels of the target process subsequence and other process subsequences at different remaining moments in the current monitoring cycle are obtained, and the perception sensitivity levels of the target process subsequence and other process subsequences at different remaining moments in the current monitoring cycle are adjusted according to the predicted perception sensitivity levels.

7. The network traffic anomaly perception system based on big data according to claim 6 is characterized in that: When the monitoring center receives the network isolation and repair signal, it performs network isolation on the process subsequence according to the network isolation and repair signal, and generates network attack emergency measures to feed back to the management personnel. The management personnel perform network repair or defense upgrade on the process subsequence according to the network attack emergency measures.

8. A method for perceiving network traffic anomalies based on big data, specifically applied to a system for perceiving network traffic anomalies based on big data as claimed in any one of claims 1 to 7, characterized in that: The following steps are involved: Step S1: split the industrial production process into several process sub-sequences, and obtain monitoring data of each process sub-sequence; divide the data type of the monitoring data collected by the industrial network monitoring points of each process sub-sequence into normal data, main attack data and feint attack data; Step S2: Perform real intention data mining on the feint attack data of various types of attack features generated in several historical monitoring cycles of each process subsequence; obtain the predicted perception sensitivity level of each process subsequence at the remaining different moments in the current monitoring cycle according to the probability of generating the main attack data of various types of attack features of each process subsequence at the remaining different moments after each process subsequence generates the feint attack data in the current monitoring cycle and the attack feature type of the main attack data; Step S3: adjusting the perception sensitivity level of the process subsequence in real time according to the predicted perception sensitivity level and the data type of the real-time monitoring data.

Citation Information

Patent Citations

  • Real-time flow monitoring alarm system based on Internet of Things

    CN117768235A

  • Defense resource grouping method for blue party intention orientation

    CN118296421A

  • Intelligent monitoring and early warning method and system for pultrusion process

    CN118410438A

  • Network information security supervision method and system

    CN118473777A

  • Fused neural network model-based prediction system and method for engine surge fault

    WO2022068587A1